NATO
COSM
A
•OTAN CCMS
NATO-CCMS
United States
Environmental Protection
Agency
Office of
Drinking Water
Washington DC 20460
EPA 570/9-84-005
CCMS-112
Committee on the
Challenges of Modern Society
(NATO/CCMS)
Adsorption Techniques in
Drinking Water Treatment
NATO/CCMS Drinking Water
Pilot Project Series
CCMS 112
-------
ADSORPTION TECHNIQUES IN DRINKING WATER TREATMENT
Papers and Discussion from the NATO/CCMS
Symposium held in Reston, Virginia, USA
April 30 through May 2, 1979
Edited by:
Dr. Paul V. Roberts and R. Scott Summers
Stanford University
Department of Civil Engineering
Stig Regli, Ross Pickford, and Frank Bell
Office of Drinking Water
U.S. Environmental Protection Agency
NATO/CCMS Symposium Co-chairmen:
Dr. Heinrich Sontheimer
Dr. Joseph A. Cotruvo
Sponsored by
Office of Drinking Water
U.S. Environmental Protection Agency
Washington, D.C. 20460
1984
-------
DISCLAIMER
This report has been reviewed by the Criteria and Standards
Division, Office of Drinking Water, U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection
Agency, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
11
-------
FOREWORD
Many of the industrial Nations today face problems related
to population, energy, and protection of the environment. In
order to optimize use of the scientific and technical expertise
from different countries, the Committee on the Challenges of
Modern Society (CCMS) was created between the Allied Nations
of the North Atlantic Treaty Organization (NATO). This inter-
national society of scientists strengthens ties among members
of the North Atlantic Alliance and permits NATO to fill a
broader social role with non-member countries. CCMS has been
responding to the increasingly complex, technological problems
facing modern society.
The Drinking Water Pilot Study was initiated by the U.S.
Environmental Protection Agency (EPA) in order to address a
broad spectrum of drinking water quality and health related
issues. Six subject areas have been studied by a number of
groups representing individuals from eleven NATO countries
and three non-alliance countries with technical participation
from many others. The conclusions and recommendations reached
by the participants are hoped to allow national programs to
focus on specific areas of water supply research and to bring
out the most up-to-date technology and practices. However,
the recommendations and conclusions do not necessarily reflect
the policy of the U.S. or any other participating countries.
The work of the Pilot Project has been covered in a summary
report.
In addition two international symposiums, "Oxidation
Techniques in Drinking Water Treatment," and "Adsorption
Techniques in Drinking Water Treatment," were held in Karlsruhe,
Federal Republic of Germany., between September 9-13, 1978, •
and in Reston, Virginia, United States on April 30 through
May 2, 1979, respectively. The first of these was supported
by the Federal Republic of Germany and the second by the
United States EPA. This volume contains the manuscripts of
all the lectures and related discussions from the symposium
held in Reston. In addition a summary paper was added by the
editor. Herein is a comprehensive survey on the practical
application of adsorption techniques for removing organics
from drinking water.
Current uses of granular activated carbon (GAC) for
controlling synthetic organic chemicals are discussed by
representatives from the U.S., Canada and six European
countries. GAC beds vary from shallow sand-replacement carbon
filters (1 meter deep) with short contact times, to deep
carbon contactors (as much as 4 meters deep) with long contact
times, often following sand filters and ozonation. Performance
of the GAC depends on the type of organics desired to be
removed, the quality of the water, the preceding treatment,
the depth and type of carbon used, and the throughput prior
to carbon replacement or regeneration.
iii
-------
Laboratory techniques, adsorption models, and pilot
testing can be used for predicting GAC performance. Organic
removal effectiveness can be measured by total organic carbon
(TOO, organic halogen (TOX), U.V. absorption, bioassays, and
specific compound analysis using gas chromatography and mass
spectrometry. Each of these methods, or the concurrent use
of several methods, is appropriate for different concerns of
treatment effectiveness.
GAC is often used most cost-effectively with on-site
regeneration. Factors which determine the appropriateness of
this include labor, energy, carbon costs, carbon loss, and
reactivation frequency requirements. Different furnace types
and operational modes need to be considered.
Although emphasis is given to GAC and biological activated
carbon {i.e., GAC preceded by ozone or other oxidant), powdered
activated carbon and ion exchange resins are also discussed.
In particular situations, either of these may be a more
suitable treatment alternative.
This workshop is a tribute to the efforts of all the
participants involved. Paul Roberts, Scott Summers and Stig
Regli were editors of the proceedings with assistance from
Ross Pickford and Frank Bell. It is hoped that the ties
established, and the good spirit of international cooperation
that has prevailed during the symposium and through the
finalization of this report, will continue in the development
of future related projects.
Joseph A. Cotruvo, Ph.D. Heinrich Sontheimer
U.S. Environmental Protection Engl'er-Bunte Institute
Agency University of Karlsruhe
Washington, D.C. 20460 Karlsruhe
Federal Republic of Germany
IV
-------
CONTENTS
Page
Foreword ill
Table of Contents v
List of Authors x
SYMPOSIUM OVERVIEW 1
Roberts, Paul V.; Summers, R. Scott
OVERVIEW OF PROBLEM - USA PERSPECTIVE 33
Jorling, Thomas C.
The Protection of Drinking Water Quality 35
Discussion 38
Kimm, Victor
U.S.A. Position - Historical and Regulatory .... 39
Suffet, I.H.
National Academy of Science Report - An
Evaluation of Activated Carbon for
Drinking Water Treatment 43
Moser, Richard H.
U.S.A. Experience with Granular Activated
Carbon Adsorption 77
Discussion 98
Gage, Stephen J.
Luncheon Presentation 103
EUROPEAN EXPERIENCE IN THE USE OF ACTIVATED CARBON
TREATMENT
Schulof, Pierre
An Evolutionary Approach to Activated Carbon
Treatment in France Ill
Meijers, Adriaan P.; Rook, Johan J.; Schultink, Bart;
Smeenk, Johan G.N.M.; van der Laan, Johan; Poels,
Cees L.M.
Objectives and Procedures for GAC Treatment in
the Netherlands 137
Masschelein, W.J.
Practical Applications of Adsorption Techniques
in Drinking Water - Belgium Experiences 168
-------
Page
Schalekamp, Maarten
The Use of GAC Filtration to Ensure Quality in
Drinking Water from Surface Sources - Swiss
Experiences 180
Goodall, J.B.; Hyde, R.A.
Current United Kingdom Practice in the Use of
Granular Activated Carbon in Drinking Water
Treatment 201
Sontheimer, Heinrich
Design Criteria and Process Schemes for GAC
Filters - German Experiences 215
Discussion 227
DESIGN CRITERIA FOR ACTIVATED CARBON ADSORBERS 233
Merk, Werner
Prediction of Multicomponent Adsorption
Behavior in Activated Carbon Adsorbers:
Kinetic Aspects 235
Frick, Bernd
Prediction of Multicomponent Adsorption
Behavior: Equilibrium Aspects 256
Holzel, Gerd
Laboratory Activated Carbon Test Methods
for Water Utilities 270
Baldauf, Gunther
Prediction of Breakthrough Patterns of
Organics from Laboratory Tests 285
DeMarco, Jack; Brodtmann, Noel
Prediction of Full Scale Plant Performance
from Pilot Columns 295
Heilker, Ewald
The Mulheim Process for Treating Ruhr River
Water 313
Discussion 325
OPERATIONAL EXPERIENCES WITH ACTIVATED CARBON
ADSORBERS 329
Osborne, D.J.; Kennett, C.A.
Operational Experiences with the Carbon
Adsorption Plant at Church Wilne 331
vi
-------
Page
Wood, Paul R.j DeMarco, Jack
Treatment of Goundwater with Granular
Activated Carbon 348
Miller, Richard
Treatment of Ohio River Water 374
Miltner, Richard J.
Results from Ohio River Valley Water Sanitation
Commission Studies 396
Cairo, Patrick R.; Suffet, Irwin H.
Design and Operational Experience with
Activated Carbon Adsorbers: Treatment of
Delaware River Water, U.S.A 416
McCarty, Perry L.; Argo, David; Reinhard, Martin
Operational Experiences with Activated Carbon
Adsorbers at Water Factory 21 456
Sander, Richard
Effect of Prechlorination on Activated Carbon
Adsorption 478
Fokken, B.; Kurz, R.
Removal of Purgeable Organic Chlorine Compounds
by Activated Carbon Adsorption 493
Stevens, A.A.; Seeger, D.R.; DeMarco, J.; Moore, L.
Removal of Higher Molecular Weight Organic
Compounds by the Granular Activated Carbon
Adsorption Unit Process 506
MONITORING TECHNIQUES FOR THE CONTROL OF THE
ADSORPTION PROCESS 519
Takahashi, Y.
A Review of Analysis Techniques for Organic
Carbon and Organic Halide in Drinking Water .... 521
Fuchs, Friedrich; Sontheimer, Heinrich
Use of Ultraviolet (UV) Adsorption for Control
of Adsorption Processes 550
Bull, R.J.; Pereira, M.A.; Blackburn, K.L.
Bioassay Techniques for Evaluating the Possible
Carcinogenicity of Adsorber Effluents 562
vii
-------
Page
Reinhard, Martin; Schreiner, Joan E.; Everhart, Tom;
Graydon, James
Specific Analysis of Trace Organics in Water
Using Gas Chromatography and Mass
Spectroscopy 582
Discussion 606
ACTIVATED CARBON REACTIVATION AND COSTS 609
Klein, Jurgen; Juntgen, Harald
Theory and Practice of Regeneration 611
Osborne, D.J.; Kennett, C.A.
Operational Experiences with the Carbon
Regeneration Furnace at Church Wilne 631
Poggenburg, Wilhem
Experience on the Reactivation of Activated Carbon
with a Double-Stage Fluidized-Bed Furnace at the
Holthausen Treatment Plant, Dusseldorf, FRG .... 642
Strack, B.; Martin, H.
Experience with a Fluidized Bed Furnace at
Benrath Treatment Plant 658
Love, 0. Thomas; Inhoffer, Wendell R.
Experience with an Infrared Furnace for
Reactivating Granular Activated Carbon 668
Gumerman, Robert C.; Gulp, Russell L.; Clark, Robert M.
Cost of Granular Activated Carbon Adsorption
Treatment in the U.S 680
Clark, Robert M.; Dorsey, Paul
Influence of Operating Variables on the Cost of
Granular Activated Carbon Treatment 705
Fiessinger, F.
Cost of Granular Activated Carbon Treatment
in France 743
Discussion 762
NEW DEVELOPMENTS IN ADSORPTION TECHNIQUES 767
DiGiano, Francis, A.
General Considerations in Assessing the
Beneficial Aspects of Microbial Activity on
Granular Activated Carbon 769
Vlll
-------
Paqe
Benedek, Andrew
Considerations in the Sequencing of Chemical
Oxidation and Activated Carbon Adsorption 789
Discussion 803
Kolle, Walter
Humic Acid Removal with Macroreticular Ion
Exchange Resins at Hannover 805
Werner, P.; Klotz, M.; Schweisfurth, R.
Investigations Concerning the Microbiology of
GAC-Filtration for Drinking Water Treatment .... 812
Discussion 826
Singley, J.E.; Beaudetr B.A.; Ervin, A.L.; Zegel, W.C.
Use of Powdered Activated Carbon to Reduce
Organic Contaminant Levels 827
Symons, James M.; Carswell, J. Keith; DeMarco, Jack;
Love, O. Thomas, Jr.
Removal of Organic Contaminants from Drinking
Water Using Techniques Other than Granular
Activated Carbon Alone - A Progess Report 844
van der Kooij, D.
A Contribution to the Discussion on Biological
Processes in Granular Activated Carbon
Filters 880
CLOSING REMARKS 887
ix
-------
LIST OF AUTHORS
D.G. Argo
Dr. G. Baldauf
B.A. Beaudet
Dr. A. Benedek
K.L. Blackburn
N. Brodtmann
Dr. R.J. Bull
P.R. Cairo
Chief Engineer, Orange Co. Water
District
10500 Ellis Ave., P.O. Box 8300
Fountain Valley, California 92708
U.S.A.
Engler-Bunte-Institute d. Universitat
Karlsruhe, Postfach 6380, D-7500
Karlsruhe 1, F.R.G.
Environmental Science and
Engineering, Inc.
P.O. Box 13454
Gainesville, Florida 32604, U.S.A.
Associate Professor
Dept. of Chemical Engineering
McMaster University
Hamilton, Ontario L8S 4L7, Canada
Toxicological Assessment Branch
Health Effects Research Laboratory
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Director, Jefferson Parish Dept.
of Water
P.O. Box 10007
Jefferson, Louisiana 70181, U.S.A.
Toxicological Assessment Branch
Health Effects Research Laboratory
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Chief, Research and Development
Division
Philadelphia Water Dept.
1180 Municipal Service Bldg.
Philadelphia, Pennsylvania 19107
U.S.A.
-------
J.K. Carswell
Dr. R.M. Clark
R.L. Gulp
J. DeMarco
Dr. F.A. DiGiano
P. Dorsey
A.L. Ervin
T. Everhart
F. Fiessinger
Research Engineer, MERL-DWRD
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Chief, Economics Analysis Section
MERL-WSRD
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Vice-President, Culp-Wesner-Culp
P.O. Box 40
El Dorado Hills, California 95630
U.S.A.
Research Sanitary Engineer, MERL-DWRD
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Associate Professor of Civil
Engineering, Marston Hall
University of Massachusetts
Amherst, Massachusetts 01002, U.S.A.
Research Assistant, MERL-DWRD
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Environmental Science and
Engineering, Inc.
P.O. Box 13454
Gainesville, Florida 32604, U.S.A.
Water Quality Control Research
Laboratory
Dept. of Civil Engineering
Stanford University
Stanford, California 94305, U.S.A.
Societe Lyonnaise des Eaux,
et de L'Eclairage
4544up Cortambert,
70516-Paris, France
B. Fokken
Bauassessor Diplomingenieur
Gas-, Elektrizitats-, and Wasserwerke
Postbox 100890, 5000 Cologne, F.R.G.
XI
-------
B. Prick
Engler-Bunte-Institute d. Universitat
Karlsruhe
Postfach 6380
D-7500 Karlsruhe, 1, F.R.G.
Dr. F. Fuchs
Engler-Bunte-Institute d. Universitat
Karlsruhe
Postfach 6380
D-7500 Karlsruhe, 1, F.R.G.
S.J. Gage
Assistant Administrator for Research
and Development
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460 U.S.A.
Dr. J.B. Goodall
Water Research Center, P.O. Box 16
Medmenham, Marlow, Bucks, SL7 2HD
Buckinghamshire, England
J. Graydon
Water Quality Control Research Lab.
Dept. of Civil Engineering
Stanford University
Stanford, California 94305 U.S.A
Dr. R.C. Gumerman
Vice-President, Culp-Wesner-Culp
2232 S.E. Bristol, #210
Santa Ana, California 92707 U.S.A.
E. Heilker
Technical Director, R.W.W., Am Schloss
Broich 123, 433 Mulheim, F.R.G.
G. Holzel
Engler-Bunte-Institute d. Universitat
Karlsruhe
Postfach 6380
D-7500 Karlsruhe, 1, F.R.G.
R.A. Hyde
Water Research Center, P.O. Box 16
Medmenham, Marlow, Bucks, SL7 2HD
Buckinghamshire, England
xn
-------
W.R. Inhoffer
General Superintendent and Chief
Engineer
Passaic Valley Water Commission
1525 Main Ave.
Clifton, New Jersey, 07011 U.S.A.
Dr. T.C. Jorling
Professor of Environmental Science
Williams College
Williamstown, Massachusetts 01267 U.S.A.
Prof. Dr. H. Juntgen
Berbau-Forschung GMBH
Franz-Fisher-Weg 61
4300 Essen-Kray, F.R.G,
C.A. Kennett
Lower Trent Division
Severn Trent Water Authority
Mapperley Hall, Lucknow Ave.
Mapperley, Nottingham, NG3 SBN
United Kingdom
V. Kimm
Director, Office of Drinking Water
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460 U.S.A.
Dr. J. Klein
Bebau-Forschung GMBH, Franz-Fisher-
WEG 61
4300 Essen-Kray, F.R.G.
M. Klotz
Institut fur Hygiene and Mikrobiologie
Universitat des Saarlandes, Bau 43
D-6650 Homburg/Saar, F.R.G.
Dr. W. Kolle
Chief Chemist, Hanover Waterworks
Stadtwerke Hanover Ag.
Postfach 5747
3000 Hanover 1, F.R.G.
Dr. D. van der Kooij
The Netherlands Waterworks
Testing and Research Institute
KIWA Ltd.
2280 ab Rijswijk, Postbus 70
Netherlands
Xlll
-------
Dr. R. Kurz
Dipl. Chem., GEW-Werke Koln AG
Augustasta 4a
5000 Koln, F.R.G.
Dr. J. van der Laan
Chief Chemist, Waterworks
Midden-Haber and Utrecht, Holland
Reactorweg 47 Postbus 2124
3500 GC Utrecht
Netherlands
Dr. O.T. Love, Jr.
Research Engineer, MERL-DWRD
Office of Research and Development
U.S. Environmental Protection Agency
26 W. St. Clair Street
Cincinnati, Ohio 45268 U.S.A.
H. Martin
Wuppertaler Stadtwerke AG
Postfach 20 16 16
5600 Wuppertal 2, F.R.G.
Dr. W.J. Masschelein
Director, Brussels Intercommunal
Waterboard
Manager of the Laboratories
764 Chaussee de Waterloo
B-1180 Brussels, Belgium
Dr. P.L. McCarty
Dept. of Civil Engineering
Stanford University
Stanford, California 94305 U.S.A.
Dr. A.P. Meijers
Research Chemical Engineer
KIWA N.V., Postbus 70, Rijswikj 2109,
Wiers 17, Nieuwegein, Netherlands
xiv
-------
Dr. W. Merk
R. Miller
R. J. Miltner
L. Moore
R.H. Moser
D. J. Osborne
M.A. Pereira
Dr. C.L.M. Poels
W. Poggenburg
Dr. M. Reinhard
Engler-Bunte-Institut d. Universitat
Karlsruhe
Postfach 6380
D-7500 Karlsruhe, 1, F.R.G.
Superintendent of Water Works
4747 Spring Grove Avenue
Cincinnati, Ohio 45232, U.S.A.
Ohio River Valley Water Sanitation
Commission
414 Walnut Street
Cincinnati, Ohio 45202, U.S.A.
Chemist, MERL-DWRD
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Assistant Director, System Water
Quality
American Water Works Service Co., Inc.
Haddon Heights, New Jersey 08035
U.S.A.
Humphreys and Glasgow, Ltd.
22 Carlisle Place
London SW 1, P1JA
United Kingdom
Toxicological Assessment Branch
Health Effects Research Laboratory
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Toxicologist
KIWA N.V., Postbus 70, Rijswijk 2109
Weirs 17, Nieuwegein
Netherlands
Stadtwerke Diisseldorf AG
Luisenstrasse 105
D-4000 Dusseldorf 1, F.R.G.
Senior Research Associate
Department of Civil Engineering
Stanford University
Stanford, California 94305, U.S.A.
xv
-------
Dr. p.v. Roberts
Dr. J.J. Rook
R. Sander
M. Schalekamp
J.E. Schreiner
P. Schulhof
Adjunct Professor
Department of Civil Engineering
Stanford University
Stanford, California 94305, U.S.A.
Chief Chemist
Drinkwaterleiding Rotterdam
Postbus 1166
Rotterdam, Netherlands
Engler-Bunte-Institut d. Universitat
Karlsruhe
Postfach 6380
D-7500 Karlsruhe, 1, F.R.G.
Director, Zurich Waterworks
Wasserversorgunq Zurich
Bahnhofcruai 5,.
CH-8023 Zurich 1, Switzerland
Water Quality Research Laboratory
Department of Civil Engineering
Stanford University
Stanford, California 94305, U.S.A.
Manager of the Equipment Department
Compagnie Generale des Eaux, Siege
Social
52 rue d'Anjou
F-75384 Paris Cedex 08, France
Dr. B. Schultink
R. Schweisfurth
D.R. Seeger
Dr. J.E. Singley
Chemist, Provincial Waterworks of
North Holland
van tssenlaan
10 Postbus 5, 2060 BA
Bloemendaal, Netherlands
Institut fiir Hygiene and Mikrobiologie
Universitat des Saarlandes, Bau 43
D-6650 Homburg/Saar, F.R.G.
Research Chemist, MERL-DWRD
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Environmental Science and Engineering,
Inc.
P.O. Box 13454
Gainesville, Florida 32604, U.S.A.
xvi
-------
J.G.M. Smeenk
Prof. Dr. H. Sontheimer
Dr. A.A. Stevens
Dr. B. Strack
Dr. I.H. Suffet
R.S. Summers
Senior Chemist
Amsterdam Municipal Water Works
Condensatorweg 54
Postbus 8169..
1005 AD Amsterdam, Holland
Engler-Bunte-Institut d. Universitat
Karlsruhe
Postfach 6380
D-7500 Karlsruhe, 1, F.R.G.
MERL-DWRD
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Wuppertaler Stadtwerke AG
Postfach 20 16 16
5600 Wupppertal 2, F.R.G.
Environmental Studies Institute
Dept. of Chemistry, Abbott Building
Drexel University
32nd and Chestnut Streets
Philadelphia, PA 19104, U.S.A.
Department of Civil Engineering
Stanford University
Stanford, California 94305, U.S.A.
Dr. J.M, Symons
Y. Takahashi
Chief, Physical and Chemical Removal
Branch, Water Supply Research Div.
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268, U.S.A.
Research and Applications Manager
Dohrmann Division, Envirotech Corp.
3240 Scott Blvd.
Santa Clara, California 95050, U.S.A.
P. Werner
P.R. Wood
Institut fur Hygiene und Mikrobiologie
Universitat des Saarlandes, Bau 43
D-6650 Homburg/Saar, F.R.G.
Associate Professor, Division of
Environmental and Urban Systems
Florida International University
Miami, Florida 33199, U.S.A.
xvii
-------
W.C. Zegel Environmental Science and Engineering,
Inc.
P.O. Box 13454
Gainesville, Florida 32604, U.S.A.
XVlll
-------
SYMPOSIUM OVERVIEW
-------
vvEPA
NATO CMM
•OTAN CCMS
NATO-CCMS
-------
SYMPOSIUM OVERVIEW
Paul V. Roberts and R. Scott Summers
INTRODUCTION
This Symposium on Adsorption Techniques in Drinking Water
Treatment was intended to bring together researchers and practi-
tioners from a number of NATO member countries for an exchange
of experiences regarding the application of adsorption tech-
nology in solving modern challenges in water supply. Foremost
among these challenges is that of providing safe potable water.
Our growing awareness of the health threat represented by
organic trace pollutants, both those present in contaminated
water supplies and those formed during treatment, is a reminder
that meeting this challenge will not be a simple task.
Not surprisingly, a diversity of views was represented at
the Symposium. Indeed, the value of the Symposium stems largely
from the wide variety of nationalities, backgrounds, and views
represented. To the extent that these diverse views can be
synthesized, the proper role of activated carbon treatment in
solving modern challenges to water supply can be defined more
confidently.
It was not a goal of the Symposium to reach formal agree-
ment in the form of resolutions or the like. Nonetheless, a
substantial consensus was developed with respect to a number
of important issues. The purpose of this overview is to point
out areas of consensus, identify questions on which there is
strong disagreement, and to compare the results of the investi-
gations reported at the Symposium. Only information presented
at the meeting is referenced; the reference numbers are keyed to
the list appended to this summary.
We have chosen to emphasize a comparative evaluation of
data on granular activated carbon (GAC) performance from a
unified point of view, rather than to digest the arguments so
well stated in the original papers. For the comparative evalu-
ation of GAC column performance data presented at this Sym-
posium, it is necessary to focus on the removal of total organic
carbon (TOC) and total trihalomethanes (TTHM), because only
these parameters were reported in a sufficient number of papers.
In keeping with the spirit of the Symposium, the intent is to
Delineate the implications for practice.
-------
GAG COLUMN PERFORMANCE
In a number of papers in this Symposium, data were pre-
sented that illustrate the breakthrough of organics, as measured
by TOC, TTHM, UV-absorbance, and the concentrations of specific
organic compounds in GAC columns. These data were derived from
studies that were conducted under widely differing conditions—
using raw waters of diverse composition, vastly different scales
of operation size, various pretreatment configurations, diverse
types of GAC, and differing values of empty bed contact time and
bed depth. Hence, it is of interest to compare the results of
the studies of organics breakthrough under such a wide range of
conditions, with a goal of gaining insight into the range of
variability that may be expected if GAC treatment is broadly
implemented in practice.
TOC Removal
TOC is thought to be a useful, quantitative measure of the
total amount, of organic constituents present in a water. Hence,
complete removal of TOC—or the equivalent measure, dissolved
organic carbon (DOC), for filtered samples—is tantamount to
complete removal of organic constituents. Virtually complete
removal of TOC is one conceivable objective of GAC treatment.
In practice, however, it is found that complete removal of
TOC by GAC treatment is infeasible under water treatment condi-
tions. An immediate, partial breakthrough of TOC occurs when a
column filled with fresh GAC is commissioned. This indicates
that a portion of the influent TOC is not amenable to removal by
GAC treatment. With increasing service time, the GAC becomes
saturated with organics; the effluent concentration (CE)
rises and eventually assumes a steady state value. According to
the theory of fixed-bed adsorption, the concentration in the
effluent at steady state ought to equal the influent concentra-
tion (Cj). However, it has been observed that under water
treatment conditions the effluent concentration of TOC seldom
reaches the influent concentration; rather the effluent concen-
tration rises to a steady state value lower than the influent
concentration (41). In effect, the GAC column continues to
remove a fraction of the influent TOC virtually indefinitely.
This more-or-less constant steady state removal is attributed to
biodegradation (8, 20, 36, 41).
Evidence of these phenomena—immediate partial breakthrough
and indefinite steady state removal—can be inferred from the
TOC breakthrough data presented at this Symposium. Table 1
includes all results for GAC columns which attained steady state
removal or complete breakthrough. To facilitate reference, we
have cited both the number of the paper and the figure number
within that paper from which the referenced data were drawn.
We have used letters as postscripts to the paper number to
-------
Table 1
GAC Performance for TOC Removal
STEADY STATE
Data Initial Average
Desig- Fraction Fraction Throughput Empty Bed Hydraulic Influent
nation Figure No. Remaining Remaining Bed Volumes Contact Time Loading Concentra- Pretreat-
J'aper/ In Paper (C/CT) (CL/CT)- 0 EBCT vo tion mentii
Expt. Cited E l ° ^ l ss ss (min) (ffl/hr) C^mg/1)
4A 1 0.35 1.0 12,000 15 64 B.P.C./CLAR
3B 6 0.25 0.65 20,000 10 tt 3 OZ/CLAR/FIL
4A 4 0.40 0.80 13,000 12 20 2-4 CLAR/OZ/FIL
4B 8 tt 0.80 8,600 15 12 4-5.5 j BPC/CLAR/
4C 8 tt 0.65 4,300 30 12 4-5. 5 j FIL
4D * tt 0.75 11,500 12 15 5
4E * tt 0.70 7,000 24 7.5 5
CLAR/FIL/
CHL
7A 4 0.5 0.90 9,000 5.4 10 4-5 ) BPC/CLAR/
7B 5 0.1 0.85 6,000 6 10 3-4 J FIL
8 4 0.55 0.90 11,000 tt tt 1.5 tt
13A 3 0.1 0.65## 6,500 11 5.4 3-4
13B 3 0.1 0.60
-------
Table 1
GAG Performance for TOG Removal
(Continued)
ilesig-
nation
Paper/
?.xpt.
17H
171
17J
17K
20A
20B
21
25
32A
32B
41A
41B
41C
41D
STEADY STATE
Fraction Fraction Throughput Empty Bed Hydr.
Figure No. Remaining Remaining Bed Volumes Contact Time Loac
In Paper (CL/r) (VCT>«i« e EBCT v<
Cited ^ l ° ^ Z SS ss (min) (ra/1
11 0.15 0.95 13,500 3.2 17
11 0.15 0.85 11,000 7.5 17
11 0.1 0.80 10,000 12 17
11 0.1 0.75 7,200 16. 17
9 0.1 0.70 5,300 34. 14
10 0.15 0.60 3,000 34. 14
9 0.1 0.70 9,000 20 7.
7 0.5 0.85« 5,100 tt tt
2 0 0.75 9,000 8 tt
2 0 0.65 9,000 8 tt
11 0.1 0.65 24,000 9 tt
11 0.1 0.45** 29,000 9 tt
15 0.1 0.70 30,000 10 19.
15 0.1 0.75 30,000 10 19.
RANGES . 0.45-1.0 1,600-30,000 3-34 1.9-20
RANGES MINUS EXTREMES 0.6-0.9 3,000-14,000 5.4-24 2.6-17
*
IN RANGE 87% 87% 84% 80%
Average
uilic Influent
ling Concentra- Pretreat-
j tion ment§§
ir) C (mg/1)
2.5-1.5*
2.5-1.5* SET/
, , , .* CLAR/
VTT
2.5-1.5* J
16t }
lot §
5 3.2 CHL/CLAR/FIL
2.0-3.5 tt
4.4 } CHL/
} CLAR/
2-7 ' FIL
1-2 CLAR
1-2 CLAR/OZ
1-2 CLAR/OZ/FIL
1-2 CLAR/FIL
1-16
2-6
86%
FOOTNOTES
* Data given in text of Paper No. 4.
t Constituent regularly measured and reported was chemical oxygen demand (COD) .
TOC calculated on given COD/TOC ratio of ~2.5.
4 Wastewater treatment. TRICKLING FILTER/CLAR/AIR STRIPPING/RECARBONATION/FIL
§ Wastewater treatment. ACTIVATED SLUDGE/CLAR/AIR STRIPPING/RECARBONATION/FIL
S Steady decrease in influent concentration with time.
** Influent concentration measured before 0 addition.
tt Data insufficient, to estimate this quantity.
' Hit General increase over a long period In CE/C! after rapid initial breakthrough
§ § Key to abbreviations regarding pretreatment :
B.P.C. — Breakpoint chlorination CLAR — Clarification
Chi — Non-breakpoint chlorination FIL — Sand Filtration
SET — Pre-settling or storage OZ — Ozonation
-------
distinguish among the individual experiments within a given
paper. The influent TOC concentration was generally in the
range of 3 to 6 mg/1. Exceptions were: two papers—(17) and
(41)—concerning the treatment of Ohio River water, for which
the influent TOC concentration was approximately 1 to 2.5 mg/1;
and one paper (20), which dealt with the advanced treatment of
wastewater having a TOC concentration in excess of 10 mg/1.
Immediate Breakthrough
The ratio of effluent concentration to influent concentra-
tion, defined as fraction remaining (cE/Cj), generally was
observed to be in the range of 0.1 to 6.5 immediately after
start-up (Table 1). In only one case (32) was the initial
effluent concentration reported as zero.
The initial breakthrough (CE/CT^O can ^e exPected to
depend on the composition of the organic constituents, as well
as empty bed contact time (EBCT) or bed depth. Values of
(CE/C,.) are plotted against bed depth in Figure 1. For the v
most part, (C /C,) £ 0.2; only one-fourth (9 of 36) of the
values exceed O.z, and 7 of these 9 positively deviating values
of (CR/CI)O were observed in experiments in which the GAC bed
depth was less than 1 m.
The plot of Cp/Cj) versus EBCT (Figure 2) shows sub-
stantial scatter. It can only be said that the maximum value of
(C /C,.) , observed at a given value of EBCT, decreases with
increasing EBCT up to EBCT = 20 minutes.
Exhaustion of GAC and Steady-State Removal
With increasing service time, the (Cp/C-,) value begins to
rise. The value of (C /Cj) continues to rise until an approxi-
mate steady state is reacned. We have expressed the length of
service in terms of the volume of water treated, normalized by
dividing by the volume of the empty bed. This normalized
quality is referred to as the throughput ratio:
Qt _ t ,.
GAC " VGAC " (
where 0 = throughput ratio, or cumulative number of bed
volumes treated as of time t (dimensionless)
Q = volumetric flow rate (m /hours)
t = elapsed time (hours)
-------
1.0
-P0.8
0.6
< 2 0.4
t z
2 <
IU
0.2
o
o
• Water Treatment - Typical Conditions
A Low Influent Concentration, Cj< 2.5 mg/l TOC
• Wostewater Treatment , Cj>IO mg/l TOC '
o,A Preozonation
• ••
2.0 4.0 6.0
GAC DEPTH, (m)
8.0
Figure 1. Dependence of Initial Fraction Remaining (cE/Cj)o
on GAC Bed Depth. Typical conditions are taken
to be 2.5 < C,. < 10 mg/l TOC, without preozonation
1.0
o _,0.8 -
t|0.4
ui
a:
o 0.2
INITIAL BREAKTHROUGH
X
• Water Treatment-Typical Conditions
i A Low Influent Concentration,Ci<2.5mg/l TOC
• Wastewoter Treatment, Cj>IOmg/l TOC -
O,A Preozonation
PPER BOUND
A »AA
A »A \
* A •• • V
0 IO 20 30 40
EMPTY BED CONTACT TIME , EBCT (MINUTES)
Figure 2. Dependence of Initial Fraction Remaining
(C /CT)_ on Empty Bed Contact Time (EBCT)
H* J, O
8
-------
GAC
EBCT
= bulk volume of GAC in the adsorbent bed (m )
= empty bed contact time (hours)
- cumulative volume of water as of time t (m )
Hence, the throughput ratio serves as a dimensionless time
parameter.
The value of (Cg/Cj) is shown as a function of the through-
put ratio (9) in Figure 3 for several representative data sets.
The data shown are representative of the full spectrum of results
presented at this symposium; the data from 16B constitute a
relatively rapid approach to steady-state; the data from 32A
represent approximately the median rate of approach to steady-
state; and the 41A data indicate a much slower rate of exhaustion
than do the data from the other studies (Table 1). The tendency
in Figure 3 is clear: for most experiments (as in the examples
of 16B and 32A), the fraction remaining rises at first rapidly,
then ever more slowly toward an approximate steady state. There
are some exceptions (such as 41A), in which the value of (C /c )
shows a more symmetrical S-shaped breakthrough response: (C /C )
at first remains nearly constant for an appreciable time period,
before beginning to rise.
REPRESENTATIVE TOC BREAKTHROUGH CURVES
0 ' i .0, , , , rA —
10,000 20,000 ^ 50,000
THROUGHPUT -BED VOLUMES (m3/m3)
Figure 3. Representative TOC Breakthrough Curves
The point at which steady state is reached is an important
characteristic of the GAC column's performance; it indicates the
number of bed volumes that can be treated before the adsorptive
capacity is exhausted. Thereafter, the removal of TOC is
-------
achieved primarily by biodegradation. For purposes of compari-
son, we have estimated this exhaustion p"bint for all cases shown
in Table 1. Fluctuations in the rate of approach to the steady
state condition will have a major impact on the throughput ratio
at steady state (8 ). To account for these variations and
their effects on 8 ^, an estimation of steady state is taken
to be within 5 percent of (CE/^I)SSI designated by the downward
arrows in Figure 3. This arbitrary choice reflects an attempt
to make a reasonable estimate of the point at which the remain-
ing adsorption capacity becomes negligible.
Values of (C-j/Cj) at steady state and 6 (the number of
throughput bed volumes at which (cE/Cj)ss is approached within
5 percent) are given for all GAG breaJctnrough runs in Table 1.
These values are plotted as (cE/Cj) versus 8 in Figure 4
with the source of the data designated to facilitate references,
and without designation to permit trends to be seen. The values
of (CE/CT^SS ^e ma*-nly in tne range of 0»6 to 0.9; only
13 percent 16 of 45) of the data points are outside that range.
The mean and standard deviation for (CE/CT)SS are 0.76 + 0.11.
The values of 8 are clustered in the range of 3,000 to 14,000
bed volumes; only 13 percent (6 of 45) of the data points lie
outside that range. Four of the six outlying values of 6
stem from a single investigation (41). The low outlying §§int
(16D) corresponds to a GAC type that is seemingly ill-suited for
TOC removal. The mean and standard deviation of 8 for all
data are 10,200 + 6,600 bed volumes. These valuessire averages
of a large number of experiments representing a wide range of
conditions, and hence must be interpreted with caution. None-
theless, it can be said with some confidence that the capacity
of GAC to adsorb TOC under typical water treatment conditions is
exhausted before 14,000 bed volumes have been treated. The
removal effectiveness thereafter is approximately 25 percent of
the influent TOC, a level which presumably can be maintained
indefinitely.
There is no apparent strong effect of preozonation on" the
value of (CE/CI^SS* Tne ^ata from 3B suggest that preozonation
may increase 8 , effectively increasing the amount of TOC
removed prior to exhaustion of the GAC's adsorptive capacity,
but the direct comparison of 4 1C and 41D does not support that
interpretation.
The value of 8 depends weakly on the empty bed contact
time (Figure 5). Tnis should be interpreted in the light of the
definitions of 8 and EBCT
a _ _ tss (2)
ss
10
-------
1.2
10,000 20,000 30,000
THROUGHPUT-BED VOLUMES (m3/m3)
Figure 4. TOC Fraction Remaining at Steady State (Cg/C^gg
and Throughput Ratio at Which It is Reached
A. (Above) - with designation of experimental data
B. (Below) - without designation of data
11
-------
30
o
O
rr
10
B.
Water Treatment -Typical Conditions
A Low Influent Concentration, Cj<2.5mg/l TOC
• Wastewater Treatment, C > IOmg/1 TOC
o,£ Preozonation
0 4 8 12 16 20 24 28 32 36
EMPTY BED CONTACT TIME, EBCT (MINUTES)
Figure 5. Throughput Ratio at Steady State QSS versus EBCT
A. (Above) - with designation of experimental data
B. (Below) - without data designation
12
-------
wnere t is the time at which steady state is reached within
5 percent. It can be seen from Figure 5" that 6 tends to
vary inversely with EBCT. The dashed lines replisent approxi-
mate upper and lower limits, ignoring data in which the influent
concentration or GAG type (16D) were abnormal, or preozonation
was practiced. In other words, t increases with increasing
EBCT, but in less than direct proportion. This is shown in
Figure 6, in which t is plotted versus EBCT, omitting the
points that are anomalous because of abnormal GAC type or
preozonation. From a regression of log(t ) on log(EBCT), it
is evident that the data are correlated bfsan expression of the
form
t^ " constant x (EBCT)n (3)
s s
The estimate of n is 0.62 + 0.20 (95 percent confidence interval)
Hence there is a positive correlation between t and EBCT
under typical conditions of water treatment, but the dependence
is weaker than linearity (n
-------
I203
40 OTlOO
liJ 20
10
I I
Cj.mg/l TOC
I.5to2.5
3 to 6
10 to 16
0.62-
. x(EBCT)
j—I—i i i i i 1 1 ... i
12 5 10 20 50
EMPTY BED CONTACT TIME, EBCT(MINUTES)
Figure 6. Observed Run Length at Which Steady
State is Reached
14
-------
0.5
Ul
O
INDIVIDUAL
CONTACTOR
5,000
10,000
15,000
33
(
THROUGHPUT-BED VOLUMES (m/m)
O
cr
u.
O 0.5
O
- ASSUMED
B.
INDIVIDUAL
CONTACTOR
_ REGENERATI
- CRITERION
III 111
0 5,000 10,000 15,000
THROUGHPUT-BED VOLUMES(m3/m3)
Figure 7. Integral Breakthrough Curve for
Characterizing Performance of Multiple,
Parallel Contactor Operation
15
-------
Suppose that the individual contactors are regenerated
regularly after 5,000 bed volumes have -been treated. Assuming
that EBCT = 30 minutes, the regeneration interval (run length)
will be 2500 hours. Under these circumstances, the overall GAC
plant performance will be superior to that predicted by the
value of the fraction remaining f = (C /C.,), at the chosen
value of throughput ratio 6. The effluent average fraction
remaining is given by
? fi * f*
7* *• 1 I
CT? i-1 \ ^T
V
R
fd6 (for sufficiently large values of m) (5)
0
where f^= (cE/Cj) = fraction remaining = (1 - fraction removed)
f = plant effluent average fraction remaining
Cp = average effluent concentration
8 = regeneration interval in throughput bed volumes
m = number of contactors in parallel
subscript i refers to the i contactor
The integral in Equation 5 can be evaluated by numerical inte-
gration, as shown in Figure 7A where the breakthrough curve in
the interval 0<8<5000 is divided into ten increments. In the
example shown, the average fraction remaining f of the combined
GAC effluents would be only 0.31 (Figure 7B), if the GAC were
regularly regenerated after a throughput of 5,000 bed volumes.
It is useful to estimate the limiting throughput ratio that
would result in a desired average concentration of the combined
effluent. This can be found by calculating the value of the
integral
e
f = JR fd8 (5)
0
as a function of 6, plotting the result as in Figure 7B. For
the example chosen, the limiting throughput ratio is 11,000 bed
volumes estimated from the integral curve, compared to a value
of 5,000 bed volumes estimated from the individual breakthrough
16
-------
curve. This implies that, for the example shown, the GAG in
a large-scale plant with multiple contactors will need to be
regenerated slightly less than half as frequently as would be
predicted by simply reading the value of the limiting throughput
ratio for the individual contactor.
Integral TOC breakthrough curves for two experiments—16B
and 32A (Table 1 and Figure 2)—are plotted in Figures 8 and 9,
respectively. These experiments are chosen by virtue of their
TOC breakthrough curves being typical of one of two types—rapid
(16B), or average (32A) breakthrough. In both cases, the
integral breakthrough is significantly slowe_r than the instanta-
neous breakthrough, as expected. Choosing f = 0.5 as a hypo-
thetical criterion, the run length would be increased by a fac-
tor of two to three (Table 2) when judged according to the
integral curve rather than the instantaneous curve for eight
cases from Table 1.
Table 2
Comparison of Instantaneous and Integral Breakthrough Curves
for Estimating GAC Run Length for 50 Percent TOC Removal
Paper
16B
15B
32A
17B
21
17A
41A
41D
Figure No.
in Paper
Cited
10
5
3
10
9
10
11
15
Throughput 0 At 50% Removal
Instantaneous
(Bed Volumes)
1,000
1,800
2,700
6,000
3,000
4,000
17,000
18,000
integral
(Bed Volumes)
2,700
3,800
7,500
12,400
8,600
10,000
-53,000
-38,000
/fl \
/°Integral\
9 1
\ Instan. /
2.7
2.1
2.8
2.1
2.9
2.5
3.1
2.1
This indicates that the design regeneration frequency may
be overestimated when based on breakthrough data from individual
columns, as is commonly done. Because regeneration frequency is
one of the most important determinants of the cost of GAC treat-
ment, it is crucial to carefully consider how pilot plant data
17
-------
cf
gO.8
<0.6
2
UJ
oc
0.4
z
o
I0-2
u_
1 1 r
INSTANTANEOUS
1 1 1 1—-T o i—i r
I6B
50% REMOVAL
i i i I i
I
5,000
10,000
15,000
THROUGHPUT -BED VOLUMES (m3/m3)
Figure 8. Importance of Multiple, Parallel Contactor
Operation - Experiment 16B
0 5,000 10,000 15,000
THROUGHPUT-BED VOLUMES (m3/m3 )
Figure 9. Importance of Multiple, Parallel Contactor
Operation - Experiment 32A
18
-------
should be used in estimating the regeneration frequency required
to satisfy a given effluent standard.
Trihalomethane Removal
Trihalomethanes have significance as organic contaminants of
public health concern, and also as indicators of the chlorinated
organic hydrocarbons formed during water treatment by chlorina-
tion. The papers in this Symposium offer abundant evidence that
trihalomethanes occur in appreciable concentrations in treated
waters, and that they can be removed by GAC treatment, albeit
for a limited time before the GAC is exhausted. The data are
summarized in Table 3. The experiments are identified citing
both the paper and the specific figure from which the data were
taken; the code is consistent with Table 1. The results are
difficult to compare. Some authors have reported measured total
trihalomethane concentrations (TTHM); others have reported total
trihalomethane formation potential (TTHMFP) data obtained at
differing conditions of chlorine dose, reaction time, pH and
temperature. It must be recognized that the collective para-
meters TTHM and TTHMFP are inherently different; therefore,
direct comparison of their behavior may be misleading.
Immediate Breakthrough
The values of (CE/C,.) for TTHM and TTHMFP in the early
stages of the experiments tend to cluster near zero. The ini-
tial concentration was significantly above zero for only six of
the sixteen data points. Hence, the initial removal of TTHMs
and TTHMFPs was in general substantially more complete than that
of TOC.
Exhaustion of TTHM Removal Capacity
All of the TTHM and TTHMFP data in Table 3 exhibit an
ultimate steady state fraction remaining in excess of 75 percent
(Figure 10), except for 41A and 41B, experiments in which TTHMFP
was measured and the influent TOC concentration was unusually
low. It is useful to divide the data into two groups: one
comprising data sets for which the influent TOC concentration
was 3 to 6 mg/1 (4, 13, 16) and a second set for which the
influent TOC concentration was 2.5 mg/1 or less (17, 41). The
latter group, corresponding to low influent TOC concentration,
exhibits substantially more prolonged TTHM and TTHMFP removal
compared to the main group of data, for which the influent
concentration was greater than 3 mg/1 TOC. The values of 6
were two to three times higher for the group having the lowlf
influent TOC concentration. This difference may be explained by
lesser competition for adsorption sites between TTHMs and other
organics when the TOC concentration is low.
19
-------
Table 3
GAC Performance for TTHM arid TTHMFP Removal
Data Des-
ignation
Paper/
Experiment
AB
AC
13A
13B
13C
13D
16G
16H
161
16A
16B
16C
16D
17A
17B
17C
ISA
18B
23
A1A
A IB
Figure No Initial
in Paper Fraction
cited Analysis Remaining
ss
1.0
0.75
1.0
0.65
0.85
0.75
•0.95
1.0
1.0
1.0
1.0
1.0
1.0
0.9
0.9
0.8
1.0
1.0
0.75
0.55
0.35**
STATE
Throughput
Bed Volume
ess
11,500
5,800
7,800
7,000
~5,700
-A, 700
9.AOO
12,300
6,500
A, 900
A,AOO
A, 900
A, 000
27,800
18.AOO
19,800
9,900
8,900
9,360
-19,000
-29,000
Hydr.
Empty Bed Loa<
Contact Time vo
EBCT (min) (m/hi
15 12
30 12
11 5. A
22 5. A
17 2.8
21 2.2
19 7
6.2 7
6.2 7
6.2 7
6.2 7
6.2 7
6.2 7
A. 5 6
7.5 6
7.5 6
7.1 6.1
11 3.1
20 tt
9 tt
9 tt
, . Average
£lic Influent
^ Concentration Pretreatment
} CjCug/1)
130-30 A BPC/CLAR/FIL
130-30 1
300-100
300-100
300-100
300-100
100-200
126
AOO-600
200-600
200-600
200-600
200-600 .
20-80
20-80
20-80
CLAR/BPC/FIL
SET/CLAR/CHL
2°-15° } CHL/CLAR
10-90 J
100-300 CLAR/CHL/FIL
AO-80 CLAR
AO-80 CLAR/OZ
Footnotes consistent with Table 1
-------
o
1.2
UJ
O
i.o
< 0.8
UJ
(T
0.6
O 0.4
(T
0.2
IT
I JL IIIIIIIIIIIIII\I
Cj -TOC (mg/l)
<2.5
O
O
>3
•
•
unknown
*
*
TTHM
TTHMFP
I I I I I I I I I I I I I I I I I I I I I I I I
5,000 10,000 15,000 20,000 25,000
THROUGHPUT -BED VOLUMES (m3/m3)
30,000
Figure 10. TTHM and TTHMFP Fraction Remaining at Steady
State (Cp/CT) and Throughput Ratio at Which
It is Reached3
21
-------
REPORTING OF GAC COLUMN PERFORMANCE DATA
In reviewing the papers presented at this Symposium, the
need for a standard format for reporting GAC data has become
apparent. Authors with different research objectives report GAC
data in a variety of forms, suitable and sufficient for their
individual purposes. If, for example, an author is interested
in reactivation frequency based on a maximum contaminant level
(MCL) criterion, then the graph of most interest is the effluent
concentration C versus time t. Others may be interested in
the mass loading of an activated carbon or in the time to
complete exhaustion of a carbon, and accordingly report their
data in a form that best satisfies their objective. While
reporting data in this objective-specific fashion satisfies the
immediate needs of the author, it limits the interpretation of
the data by others with different objectives. By including
values for a few additional parameters and using a standard
format for graphical presentation, some papers could be rendered
more generally useful. Values of the following parameters
should be specified if the potential of the data is to be fully
realized.
Water Quality Characteristics
Knowledge of water quality parameters, in addition to the
constituent of concern, is often helpful in the interpretation
of the data, especially if abnormal conditions exist. Data on
organic content (TOC or COD), chlorine residual, pH, dissolved
oxygen, nitrogen species, and suspended solids should be reported.
The measurements of collective organic parameters such as TOC or
COD are helpful even if the objective of the study is to quantify
the removal of specific organic pollutants. In reporting data
from analytical methods that are not yet standardized (such as
specific organic analysis), the detection limit and precision of
the determination should be given.
GAC Physical Characteristics
GAC grain size, specific surface area, and bulk density p
(g GAC per m bed volume) are useful information for comparison
of GAC performance. The value of PB is needed for comparing
mass loading per unit mass GAC. Carbon age, number of regenera-
tions, has been shown to influence adsorptive capacity (28,32)
and should be reported. Also, the manufacturer and type of
activated carbon should be specified.
Pretreatment
This is normally reported, but often the level of pretreat-
ment is not given. Sander (21) and Symons (41) report that
chlorination and ozone-oxygen pretreatment have a substantial
Affect on GAC adsorption systems. Dosages and contact times for
22
-------
pretreatment steps should be given if they are suspected of
having an effect on the compounds of interest.
GAC Process Parameters
The following data are needed:
Empty bed contact time, EBCT;
Surface loading rate (superficial linear velocity), v ;
GAC bed volume, VGAC;
Volumetric flow rate, Q; and
GAC bed depth, D.
Values of three of the last four parameters can provide enough
information so that the other process parameters can be estimated
and complete data analysis is possible. These parameters are
related as follows:
EBCT = VGAC
-------
the latter two forms are used. When monitoring the constituents
of water or wastewater, influent concentration C often varies
considerably, which affects the breakthrough pattern of the GAG
system. The extent of influent quality variation cannot be
appraised if the concentration data are normalized. When C
is reported as an average value and an MCL criterion is of
interest, the latter two methods give only an estimate of C .
If the data are to be transformed into a mass loading format,
cumulative amount removed versus cumulative amount applied,
then both C and CE are required.
Laboratory studies differ from investigations at full-scale
operating plants in that the influent concentration can be
reasonably constant by dosing the influent with controlled
amounts of solute. If the influent concentration can be main-
tained steady in this fashion, then the fraction remaining or
percent removed can be used to represent the concentration
variable. However, as a rule the individual data for influent
and effluent concentration should be given, to permit inter-
pretation of concentration fluctuations and trends and to
facilitate estimation of the approach to steady state effluent
quality.
The independent variable is normally reported as: time t
in service; volume of water treated V • or throughput ratio 6,
defined in Equation 1. The advantage of using throughput ratio
6 over the other two alternatives is that GAC contractors with
different EBCT values can be compared on an equalized basis.
The interrelationships between 9 , t, and V are defined in
Equation 1.
RELATING GAC USE RATIO TO COLUMN PERFORMANCE
The rate at which GAC must be replaced or regenerated is
an important factor influencing the cost of GAC treatment (33,
34). The GAC use ratio (34) is defined as the amount of GAC
that must be replaced or regenerated per unit volume of water
treated. The use ratio influences strongly the level of
effluent quality resulting from GAC treatment. Conversely, the
required degree of treatment dictates the necessary use ratio.
For economic reasons, the use ratio will be chosen as small as
possible, just large enough to satisfy the effluent quality
criterion. The units of use ratio are chosen in (34) as pounds
GAC per million gallons water treated; the corresponding metric
units are g/m .
The use ratio can be defined in terms of the throughput
ratio, as follows
PR
UR = H- (8)
24
-------
r,~ ,. • 9 GAG regenerated
where UR = GAC use ratio, ^ - 2 -
m water treated
How can the use ratio be estimated from GAC column per-
formance data? Consider the example of the data for TOC removal
from Passaic River water (32). We chose the data for previously
reactivated GAC (32A in Tables 1 and 2 and in Figures 3 and 9)
as being representative of long-term operation. The values of
and UR corresponding to different levels of treatment for that
example are summarized in Table 4. The values of the use ratio
range from 49 to 160 g/m (equivalent to mg/liter), or approxi-
mately 400 to 1300 pounds per million U.S. gallons, depending on
the effluent quality criterion and the question of single or
multiple columns. Any large water treatment plant would closely
approximate multiple, parallel-column operation. Hence, for the
example chosen, the use ratio to achieve an average TOC removal
of 50 percent in a large treatment plant would be approximately
59 g/m , or 490 pounds per million gallons. The corresponding
run length tR (time between regenerations) would be:
tR = 9R x EBCT (days) = 7,500 x jf^ = 42 days. (9)
This example is intended for purposes of illustration only, and
not as a general guideline. The methodology should prove useful
in establishing a connection between papers that present GAC
column performance data (3, 4, 1, 8, 13, 15-18, 20, 21, 23, 25,
32, 41) and those concerned with the costs of treatment and
reactivation (33-35).
SUMMARY AND CONCLUSIONS
We have attempted to synthesize rather than summarize
the Proceedings of this Symposium, by identifying areas of
consensus and by applying a unified approach to interpreting
GAC performance data.
There appeared among the experts, from the European and
North American nations present at this Symposium, a remarkably
strong consensus in favor of GAC treatment for control of
dissolved organic contaminants, including those causing taste
and odor as well as those of health concern.
There was general agreement in the following areas. GAC
treatment has been shown to be effective for removing organic
contaminants from water (1-8, 13, 15-23). The available tech-
nologies for GAC treatment and reactivation are felt to be
relatively reliable and trouble-free (2-8, 14-20, 28-32). The
understanding of GAC performance is reasonably well founded in
theory (1, 9-12); pilot study results have improved our under-
standing of the breakthrough behavior of GAC columns (9, 13, 19,
25
-------
Table 4
Estimating The GAC Use Ratio From
Column Performance for TOC Removal
Example—Data from (32), EBCT = 8 min,(C ) =4.4 mg/1 TOC
-L 3VS
Criterion for
Reactivation
Steady State
reached in a
single column
50% TOC
removal in a
single column
50% TOC
removal in multiple
columns operated in
parallel
Throughput
Ratio at
Reactivation
6R
(bed volumes)
9,000*
2,700+
7,500f
GAC
Use Ratio1'
UR
(g/»3)
49
160
59
Resulting
Fraction
Remaining.'
CE/CI
Single
0.75
0.5
0.7ft
Multiple
**
0.53
A*
0.3
0.5
* Estimated from Figure 3 of this Overview
t Estimated from Figure 9 of this Overview _ -
# Calculated from Equation 8 using pfi = 4.4 x 10 g/m
Units conversion: (g/m3) x 8.34 = Ibs/million U.S. gallons,
§ See the text of this Overview for a discussion of steady-state GAC
column performance. For this case (32A, Table 1), (CT,/CT')__ a 0.75.
For this case (32A, Table 1), (C.,/CT)
£j J. SS
75W
Equivalent to the value .of (Cg/Cj) for a single column over
the entire period between reactivations.
tt Equivalent to the value of (C /C ) at which an individual column
would have to be reactivated!.
26
-------
41). The concurrent removal of organic solutes by biodegrada-
tion and adsorption in GAG columns is widely agreed upon; the
GAC appears to serve a dual function as an adsorbent and an
attachment surface for microorganisms (8f 20, 36, 37, 39, 42).
GAC is almost universally accepted as being superior to powdered
activated carbon for removal of collective organic substance
(e.g., TOC) as well as specific organic contaminants; powdered
carbon may be chosen as a short-term solution or as a low-cost
means of controlling periodic taste and odor problems (3, 40).
It is widely believed that GAC is justified for public health
reasons in many cases where water supplies are prone to con-
tamination by synthetic organic chemicals (1, 3-6, 8) despite
the incremental cost (33-35) of GAC treatment; for such cases
GAC presently is the only practical alternative to developing a
new water supply (2-8). Macroreticular resins (38, 41) show
promise for the long term, but have not as yet been demonstrated
to be effective in practice.
Conflicting opinions were expressed in the following areas.
Some speakers from European nations and a number of discussants
voiced reservations regarding the feasibility and desirability
of uniform, formal regulations for GAC treatment of water
polluted by organic chemicals. Recommendations on a case-by-
case basis are preferred by many. Also, concern was expressed
about the rapid breakthrough of some specific compounds, notably
chloroform. Furthermore, it was noted that chromatographic
displacement can occur, resulting in effluent concentrations for
some compounds being temporarily greater than the corresponding
influent concentrations. The phenomena of elution and satura-
tion observed in GAC columns, as well as the complex nature of
the mixture of sorbing solutes, require that special attention
be paid to characterizing effluent quality (1, 19, 20). Rapid
progress is being made in developing and implementing chemical
analytical techniques (24, 25, 27), but it is as yet unclear how
bioassay techniques might be used effectively (26). It has been
demonstrated how modern techniques of organic chemical analysis
can be applied to quantitative characterization of the reli-
ability of removal of specific pollutants in GAC treatment (20).
It was observed that some organic priority pollutants continue
to be removed effectively even after the steady state condition
of TOC removal has been attained (20). This may be caused by
either strong preferential adsorption or biodegradation, depend-
ing on the substance in question and other conditions. There
was general enthusiasm for enhancing the microbiological
activity in GAC columns to improve and prolong the removal of
organics (3, 8, 14, 36, 37), but there was disagreement on the
extent to which this can be accomplished (37, 42). Indeed, it
is questioned whether GAC beds are more effective than beds of
granular, non-sorbing media such as sand or coal (42). The
process combination of preozonation and GAC has strong pro-
ponents (3, 8, 36); the technical arguments in favor are
27
-------
convincing, but it is not yet clear that preozonation effec-
tively improves GAC performance (41). These areas of divergent
opinion are the subjects of intense research intended to clarify
the issues.
A consistent approach to analyzing GAC column performance
results was applied herein to the data on removal of TOC and
trihalomethanes presented at this Symposium. The analysis
showed order-of-magnitude agreement among studies conducted
under an extremely broad range of conditions. Considering the
disparities in conditions/ the agreement should be considered
heartening. The results of European investigations fell for
the most part in the mid-range of the spectrum of results of
studies conducted in the U.S.A. This too is encouraging; given
the substantially greater body of experience with GAC in
Europe/ it would be an advantage to North American specialists
to be able to draw on the European experience, in confidence
that conditions are sufficiently similar. Indeed, the opening
of communicatioa channels for transmitting such experience was
one of the main motivations of the Symposium.
28
-------
REFERENCES TO SYMPOSIUM PAPERS
1. Suffet ., I.H., National Academy of Science Report
An Evaluation of Activated Carbon for Drinking Water
Treatment.
2. Moser, R. H., U.S.A. Experience with Granular Activated
Carbon Adsorption.
3. Schulhof, P., An Evolutionary Approach to Activated Carbon
Treatment.in France
4. Meijers, A.P., et al., Objectives and Procedures for GAC
Treatment, in the Netherlands.
5. Masschelein, W. J., Practical Applications of Adsorption
Techniques in Drinking Water - Belgian Experiences.
6. Schalekamp, M., The Use of GAC Filtration to Ensure Quality
in Drinking Water from Surface Sources.- Swiss Experiences.
7. Goodall, J. B.r and R. A. Hyde, Current United Kingdom
Practice in the Use of Granular Activated Carbon in
Drinking Water Treatment.
8. Sontheimer, H., Design Criteria and Process Schemes for GAC
Filters, - German Experiences.
9. Merk, W., Prediction of Multicomponent Adsorption Behavior
in Activated Carbon Adsorbers: Kinetic Aspects.
10. Frick, B., Prediction of Multi-Component Adsorption
Behavior: Equilibrium Aspects.
11. Holzel, G., Laboratory Activated Carbon Test Methods for
Water Utilities.
12. Baldauf, G., Prediction of Breakthrough Patterns of
Organics from Laboratory Tests.
13. DeMarco, J., and N. Brodtmann, Prediction of Full Scale
Plant Performance from Pilot Columns.
29
-------
14. Heilker, E., The Mulheim Process for Treating Ruhr River
Water.
/ and Kennett, C.A.
15. Osborne, D. J., Operational Experience with the Carbon
Adsorption Plant at Church Wilne.
16. Wood, P., and J. DeMarco, Treatment of Groundwater with
Granular Activated Carbon.
17. Miller, R., Treatment of Ohio River Water.
18. Miltner, R. J., Results from Ohio River Valley Water
Sanitation Commission Studies.
19. Cairo, P. R., and I. H. Suffet, Design and Operational
Experiences with Activated Carbon Adsorbers: Treatment
of Delaware River Water, U.S.A.
20. McCarty, P. L., D. Argo, and M. Reinhard, Operational
Experiences with Activated Carbon Adsorbers at Water
Factory 21.
21. Sander, R., Effect of Prechlorination on Activated Carbon
Adsorption.
22. Fokken, B., and R. Kurz, Removal of Purgeable Organic
Chlorine Compounds by Activated Carbon Adsorption.
23. Stevens, A. A., et al., Removal of Higher Molecular Weight
Organic Compounds by the Granular Activated Carbon
Adsorption Unit Process.
24. Takahashi, Y., A Review of Analysis Techniques for Organic
Carbon and Organic Halide in Drinking Water.
/ and Sontheimer, H.
25. Fuchs, F., Use of Ultraviolet (UV) Absorption for Control
of Adsorption Processes.
26. Bull, R. J., M. A. Pereira, and K. L. Blackburn, Bioassay
Techniques for Evaluating the Possible Carcinogenicity of
Adsorber Effluents.
27. Reinhard, M., et al., Specific Analysis of Trace Organics
in Water Using Gas Chromatography and Mass Spectroscopy.
28. Klein, J., and H. Juntgen, Theory and Practice of
Regeneration.
29. Osborne, D. J., and C. A. Kennett, Operational Experiences
with the Carbon Regeneration Furnace at Church Wilne.
30
-------
30. Poggenburg, W., Experience on the Reactivation of Activated
Carbon with a Double-Stage Fluidized-Bed Furnace at the
Holthausen Treatment Plant, Dusseldorf, FRG.
31. Strack, B., H. Martin, Experience with a Fluidized Bed
Furnace at Benrath Treatment Plant.
32. Love, O. T.f and W. R. Inhoffer, Experience with an
Infrared Furnace for Reactivating Granular Activated
Carbon.
33. Gumerman, R. C., R. L. Gulp, and R. M. Clark, Cost of
Granular Activated Carbon Adsorption Treatment in the U.S.
34. Clark, R. M., and P. Dorsey, Influence of Operating
Variables on the Cost of Granular Activated Carbon
Treatment.
35. Fiessinger, F., Cost of Granular Activated Carbon Treatment
in France.
36. DiGiano, F., General Considerations in Assessing the
Beneficial Aspects of Microbial Activity on Granular
Activated Carbon.
37. Benedek, A., Considerations in the Sequencing of Chemical
Oxidation and Activated Carbon Adsorption.
38. KoJle, W., Humic Acid Removal with Macroreticular Ion
Exchange Resins at Hannover-
/ Schweisf-Urth' R. .
39. Werner, P., and M. Klotz, Investigations Concerning
the Microbiology of GAC-Filtration for Drinking Water
Treatment.
40. Singley, J. E., et al., Use of Powdered Activated Carbon
to Reduce Organic Contaminant Levels.
41. Symons, J. M., et al., Removal of Organic Contaminants
from Drinking Water Using Techniques other than Granular
Activated Carbon Alone h - A Progress Report
42. van der Kooij, D., A Contribution to the Discussion on
Biological Processes in Granular Activated Carbon Filters.
31
-------
xvEPA
MATO CMM
•OTAN CCMS
NATO-CCMS
32
-------
OVERVIEW OF PROBLEM - USA PERSPECTIVE
33
-------
f/EPA
NATO
•OT AM CCMS
NATO-CCMS
34
-------
THOMAS C. JORLING
ASSISTANT ADMINISTRATOR FOR
WATER AND WASTE MANAGEMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
THE PROTECTION OF DRINKING WATER QUALITY
Tom Jorling is the Assistant Administrator for Water and
Hazardous Materials from the Environmental Protection
Agency. He has the overall responsibility for implemen-
tation of all environmental legislation relating to
water—both drinking water and surface water protection—
the construction of public waste treatment facilities,
and solid waste disposal activities. That is a broad
responsibility. Mr. Jorling has a very appropriate
background to handle- that position. He is an attorney
and has an advanced degree in Biology, in effect a com-
bination of law and technology. He was instrumental in
the development of clean water legislation during his
term working in the Congress.
Good morning. My purpose is primarily to welcome the guests
from abroad—Dr. Sontheimer and his colleagues, as well as those
from many other countries. No one disputes the fact that one of
the first reponsibilities of government is to assure people that
the water they are consuming is safe. In the United States we
have a particularly good track record in assuring safety from
conventional pollutants and pathogens. Like many others through-
out the world, we now are coming to grips with the problems
represented by the chemical revolution on one hand, and the grow-
ing awareness on the other of the pathways and risks associated
with long-term exposure to those chemicals. That theme, I think,
in large measure provides the motivation for a conference such as
this.
Approaches to Drinking Water Quality Regulations
Drinking water is unique. It can be contaminated at its source,
during the treatment process itself, and in the distribution
system. Two of those—the treatment and distribution system—
35
-------
are obviously within direct control. The source, on the other
hand, may not be. But we must look at the overall implications
for drinking water in the context of protecting the life support
systems of our people in general.
I would like to take a moment to put the safe drinking water
program, which comes under the Safe Drinking Water Act, into the
context of the overall protection programs of the United States
as administered by EPA and the States. I think it's clear that
everyone is moving more and more toward source control of con-
taminants, especially synthetic contaminants. In the United
States, we now have some people who are beneficiaries and
others, especially those in my position, who are burdened by the
very marginal legislative enactments directed at one or another
of the threats and risks associated with protection of our
environment. Over the last 10 years we have had 16 statutes
enacted or amended in the United States, dealing with this
question. Thus, we have been evolving into an overall program
of environmental protection.
The task of the Environmental Protection Agency, at present,
is to integrate these multiple authorities into a coherent,
efficient theme and program of environmental protection. I
believe, and I think most people agree, that we now have about
as much authority as we need, with the possible exception of
legislation that the administration will be recommending on
abandoned hazardous waste sites. Our task now is to use that
authority to implement the intent of the legislation.
Toxic Substances Control
We start with TOSCA. For those who are not familiar with our
acronyms, TOSCA is the Toxic Substances Control Act. In princi-
ple, it is directed at preventing from entry into commerce those
chemicals that pose an unreasonable risk. It also has residual
authority, but it is designed to operate if other regulatory
programs are inadequate to deal with the problems. A similar
statute is designed to control the release into the environment
of those chemicals used for pesticide purposes. Under the Solid
Waste Act, amended and now called the Resource Conservation and
Recovery Act (RCRA), we have full authority to regulate solid
waste, with particular authority for that portion of the solid
waste stream that is characterized as hazardous. The Clean
Water Act is directed at release of chemicals into the environ-
ment that reach surface waters. The Clean Air Act is designed
to control chemical release ijito the ambient air. So we have a
full range of authority. And into that range of authority fits
the Safe Drinking Water Act programs that will be covered by
some of the subjects to be discussed at this conference.
36
-------
The United States is implementing the Safe Drinking Water Act in
several ways. We have established maximum contaminant levels
for pollutants. We are in the process of dealing with the
problems of additives associated with drinking water treatment
and distribution. We are presently working on an agreement with
the Food and Drug Administration that will enable us to address
these problems. We are addressing the questions concerning
corrosion, in both treatment and distribution processes. But
today's concern, obviously, is the problem of synthetic organic
materials, and those organic materials which are formed in the
treatment process.
Regulations
All of you know that in February, 1978, the Environmental Protec-
tion Agency proposed regulations for control of trihalomethanes.
These compounds, as well as being harmful themselves, are in-
dicators of a multitude of other chemical contaminants formed
during the treatment of water by reaction of chlorine with the
naturally present organic substances. Moreover, EPA proposed
requiring the use of treatment technology, such as granular acti-
vated carbon (GAC), to control the presence of other synthetic
organic chemicals that reach the public drinking water through
contaminated surface and ground-water supplies. There has been
intense controversy concerning those two sets of regulations.
The public comment period closed last fall. We have been evalu-
ating those comments along with the written and oral comments we
have received, and we will be translating these comments into a
formal promulgation.
A great deal of controversy surrounds the subject of this meet-
ing, so I personally look forward to the dialogue here. I must
say one of the most frustrating aspects of my present job is my
inability to stay through a session like this. I must, in fact,
return to my office to convince the Governor of Mississippi that
EPA was not responsible for the floods in Mississippi. I'm not
sure I'll be able to succeed, but I do have to leave.
I do want to welcome you. I look forward to the proceedings of
this conference because they are extremely important. Dealing
with drinking water is a very controversial and yet a very
significant responsibility of public officials. It's one that
no one takes lightly, certainly no one in EPA. We will be as
thorough as we can in the regulatory process, and much of what
is to be said here bears on that kind of regulatory program.
I welcome you and hope that you have a pleasant and very inform-
ative session. Thank you very much.
37
-------
DISCUSSION
DR. COTRUVO: What would you say is your impression of the
feelings of Congress and the present Administration on dealing
with environmental quality matters? As we all know, there are
pressures developing which include energy problems and inflation.
What would you say is the mood of the Congress and the Adminis-
tration?
JORLING: That's an open-ended question and one that I
would have to answer personally rather than from my official
position, although my perception is a function of that position
and bears on the answer. In my opinion, the Administration's
posture toward environmental protection is unchanged, unchanged
from the President's earliest commitment before taking office.
I also think that, in general, that position will prevail in
Congress. That's not to say that there aren't many who are
looking for easy answers to some of the more difficult problems
that are facing our industrial society and all industrial
societies, namely inflation and shortages of energy. There is a
lot of rhetoric and dialogue, especially from very outspoken
individuals both in Congress and the Executive Branch, concern-
ing benefits and costs. But as a general matter, I do not
expect any change in Congress or in the Administration to the
commitment that both have given, historically, to environmental
protection and public health problems. I think we will go
through a phase when people will raise concerns about the
benefits and costs, and empirical benefits versus the empirical
coses. I do not, however, think it will change the basic thrust
of our regulatory" programs, including those under the Safe
Drinking Water Act.
38
-------
VICTOR KIMM
DEPUTY ASSISTANT ADMINISTRATOR
FOR DRINKING WATER
U.S. ENVIRONMENTAL PROTECTION AGENCY
U.S.A. POSITION—HISTORICAL AND REGULATORY
Mr. Victor Kimm is the Deputy Assistant Administrator
for Drinking Water for EPA. He reports to Mr. Jorling
and is specifically responsible for applying the Safe
Drinking Water Act in all of its facets. Mr. Kimm has
been in Government service for approximately 15 years.
He is an engineer, has had experience and training in
public policy in many respects, has been in the Safe
Drinking Water Program since its beginning in 1974,
and has carried it through to the present controversies
in which we find ourselves.
I would like to welcome all those in attendance. To our good
friend Professor Sontheimer, I would like to say that there's no
way on earth that we're going to come close to matching the hos-
pitality we received at the Oxidation Conference at Karlsruhe in
1978. Nevertheless, we hope to enlighten as well as entertain
the participants. When we reviewed the program on Friday, we
concluded that we had crammed at least a week's conference into
a few days, and I suspect that those of you who stay with us for
our evening and other activities, will discover this may become
somewhat of an endurance test. At any rate, we hope that we can
make these days as meaningful as possible for everyone.
Historical Perspective
There is a historical link to the challenges now facing water
supply professionals throughout the industrial world and, in
particular, the United States; fate has us at a very important
crossroad. The primary public health concern at the turn of
the century was about contaminants that had acute, immediate
effects on public health. In the present context, we find our-
selves becoming increasingly concerned about those contaminants
39
-------
which do not have immediate effects, for which there are no
clear linkages, but which nevertheless, have potentially serious
risks that are a challenge to the public health world today. In
1900, in the United States, 36 deaths out of every 100,000 were
attributed to typhoid. On a per capita basis, that corresponds
to more than we currently lose to automobile accidents. It
clearly was a major cause of death in the United States. With
the advent of filtration and disinfectant practices that started
in the early part of this century, and as those practices became
more widely used, typhoid was virtually eliminated. This is
often cited as one of the great public health benefits of the
20th century.
In regard to the current set of concerns in public health, the
application of more sophisticated analytical techniques, begin-
ning in the early 1970's, enabled us to inspect the organic
fraction in drinking water and to identify and quantify a
variety of contaminants. We then began investigating both the
long-term health significance and the sources of such contami-
nants, as well as seeking ways of reducing the public's exposure
to these materials. Simply stated, the contaminants have often
been categorized into two basic groups: (1) trihalomethanes and
other halogenated organic materials which arise as byproducts
of conventional disinfection with chlorine, and (2) other man-
made chemicals that are finding their way into drinking water
through contamination of water sources. The second category is
a byproduct of our chemical revolution; as mentioned earlier,
approximately 20,000 chemicals are produced, commercially, in
significant quantities. Enormous amounts of chemicals are
transported via our major waterways, which are often used as
drinking water sources. Spills and unplanned discharges pose
serious problems that we face periodically. The effects are
clearly significant and of major concern. The work done in the
last few years has given us a better picture of the levels of the
contaminants that confront us and of the almost overwhelming
dimensions of the trace organic contaminant problems, when
considered nationwide. Several papers in this Conference will
give current data related to these problems.
When reviewing what we really know about the problem, we must
realize that only a small portion is uncovered—at best,
30 percent of the total organic fraction can be identified as
specific compounds. In many instances, we are dealing with
chemical and physical phenomena that are just now beginning to
be studied carefully and understood.
Regulatory Approaches with Respect to Technology
A regulatory problem can be approached by considering maximum
contaminant levels or required treatment practices. I do not
40
-------
wish or plan to rekindle the heated debate which we have had,
over the last year, on the regulations, but merely to provide a
context for our discussion of technology during the next few
days. The need for reducing exposure to trihalomethanes has been
generally accepted. The technology for reducing trihalomethanes
is very similar to what we have used in the past and does not
seem to present a problem as serious as the proposed use of
granular carbon or equivalent alternatives. Adsorption tech-
niques have not been widely used for drinking water treatment in
the United States thus far; they have been used more elsewhere
in the world. We will hear a great deal more about these tech-
nologies during this Conference. They are effective and they
offer, in our judgment, a significant step in the direction of
eliminating objectionable materials from our drinking water.
The National Academy of Sciences has listed some 22 carcinogens
that have been detected, at least once, in drinking waters in the
United States; two additional carcinogens have been added to the
list since it was published. We anticipate that the list of
chemicals will grow over the next few years as more and more
specific materials are tested. We are relying on what has been
the health community's guiding principle for at least a decade,
namely to reduce human exposure to carcinogens to the degree
that is feasible. Over the next few years, we must expect that
the number of compounds of specific and immediate health concern
will continue to grow, barring a major breakthrough on the health
side. Faced with that, and in collaboration with other water
supply professionals from all over the world, we are trying to
determine the most effective means of reducing exposure to these
materials. The effects are not immediate, and it is difficult
to establish a link to adverse health effects. Nonetheless,
these effects are deeply serious and of major public health
concern.
I hope that in the next few days, as we review the experience
of other countries, we may come to realize that the European
experience with regard to use of granular activated carbon is
somewhat different from what we are familiar with here in the
States. I think that there will be many examples brought forth
to show that many of the technical problems with deep-bed GAG
filters and on-site regeneration facilities have, in fact, been
overcome in a number of countries. There have been many, many
years of experience in implementing this type of process, even
though this has not been practiced extensively in the United
States.
Much is known, yet much needs to be developed. I think that
solving the technical problems associated with controlling trace
organic contaminants is a major concern to all of us. EPA's role
is: to support long-term research; to disseminate information;
and, as we move into utility level experimentation and devel-
opment, to provide technical assistance. The water utility
41
-------
industry also will need to conduct pilot studies and research, to
apply innovatively new technologies, and to maintain a continuing
dialogue with the rest of the water supply profession. Together,
Federal and State Government, industry, utilities and suppliers,
and the professional engineers who support these activities can
learn to effectively control trace organic contaminants in our
drinking water. Thank you.
42
-------
NATIONAL ACADEMY OF SCIENCE REPORT
AN EVALUATION OF ACTIVATED CARBON
FOR DRINKING WATER TREATMENT *
I.E. Suffet
INTRODUCTION
The National Academy of Science report on the use of acti-
vated carbon for drinking water treatment was divided into six
major areas:
I. Adsorption Efficiency of Granular Activated Carbon
(GAC) - discusses removal of organic compounds of health
concern, effectiveness of pretreatment for removal of
organic compounds, and competitive adsorption between
organics in laboratory, pilot plant, and full scale
plant operations.
II. Microbial Activity on GAC - discusses the extent of
interaction of microbial activity with the adsorption
process, the effect of pretreatment disinfectant
processes, and the impact of microbial colonization
on the water treated.
III. Production of Nonbiological Substances By or Vvithin
the GAC Bed - discusses the carbon surface in relation
to.catalytic reactions and reaction with disinfectants.
*Safe Drinking Water Committee of the National Research Council •
Subcommittee on Adsorption - Chairman - I.H. Suffet - Members -
M. Alexander, Cornell University, Ithaca, N.Y.; J.T. Cookson,
JTC Environmental Consultants, Washington, D.C.; F. DiGiano, U.
Mass., Amherst, Mass; R. Kunin, Consultant, Yardley,Pa.; J.
Shands, U. Florida, Gainesville, Fl.; V.I. Snoeyink, U.
Illinois, Urbana, 111; M.J. McGuire, Brown and Caldwell,
Pasadena, Ca. (Consultant!
Safe Drinking Water Committee: Chairman - J. Doull, U. Kansas
Medical Center, Kansas City, Mo. ; Safe Drinking Committee Staff
- R. Housewright, R. Golden and R. Widdus
See complete report for all data (NAS, 1979)
43
-------
IV. Regeneration of GAC - discusses the change of GAG
including chemical leaching of GAC.
V. Adsorption Efficiency of Other Adsorbents - discusses
the removal of humic substances by ion exchange resins
and the removal of organic compounds of health concern
by polymeric adsorbents.
VI. Analytical Methods to Monitor Adsorbent Unit Processes
in Water Treatment - discusses monitoring of adsorbent
processes by specific and nonspecific organic analysis
for chemicals of health concern.
The results of the report are presented with the objective
of defining the efficacy of adsorbents. The report did not
evaluate toxicological, epidemiological, or economic and plant
design aspects of the problem.
It is known that raw water sources and disinfected water
supplies may contain organic compounds that have been demon-
strated to be carcinogenic or otherwise toxic to experimental
animals or in epidemiological studies. Many other compounds
that are also present either have not been identified or their
effects on health have not been characterized. At present, the
compounds of known concern are synthetic organic chemicals or
compounds produced by disinfection of potable water supplies.
Lists of chemicals of health concern have been defined in a
series of sources: (1) Categories of Known or Suspected Organic
Chemical Carcinogens Found in Drinking Water (NASf 1977, p. 794),
(2) Carcinogens and Suspect Carcinogens in Drinking Water (NCI,
1978), (3) List of Chemicals Submitted by EPA for Evaluation by
the Safe Drinking Water Committee Toxicology Subcommittee (NAS,
1979), (4) Interagency Regulatory Liaison Group Target List
(IRLG, 1978), (5) Pesticides with established maximum contami-
nant levels (MCL's) (US EPA Interim Drinking Water Standards,
1975) and (6) Trihalomethanes (THM's) with proposed composite
maximum contaminant level (MCL) values (US EPA 1978a). Table 1
shows an edited compilation from these lists.
The list of chemicals submitted by the EPA for evaluation
by the Safe-Drinking Water Subcommittee on Toxicology added 5
compounds (Table 2), to the 22 Known or Suspected Organic
Chemical Carcinogens Found in Drinking Water (NAS, 1977). Also,
this subcommittee developed Suggested No Adverse Response Levels
(SNARL's) on acute exposure for 27 synthetic organic chemicals.
(Table 2 lists the SNARL's.)
44
-------
Table 1. Available Adsorption Data for Organics of Health Concern*
Organic Molecular
compounds weight
Acrylonltrlle
Aldrln
Benzene
Benzo(a)pyrene
o-BHC
{'-Bficaindane)
Bta (2-chloroethyl)ether
Bronodlchloronethane
Broinofora
Carbon tetrachlorlde •
Chlordone
Chlorobenzene
Chlorodlbronomethane
Chlorofom
53. 06
365
78.11
252.3
290.85
290.85
143.01
163.83
252.75
153.82
764
112.56
208.29
119.38
Number
of data
points
2
6
5
2
Unknown
1
11
7
15
11
3
7
7
7
Carbon
type
Unknown
F300
F400
Nor it NK
Unknown
OU-A
F300
F300
F400
F400
~F300
F300
F400
F400
Adsorption
test
method
Unknown
Isotherm
Isotherm
Z reduction
Z reduction
Z reduction
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Maximum
Equilibrium surface
concentration concentration,
range mol/1 mol/g
1.4xlO~3-9.2xlO~3
2.5xlO~'-7.1xlO~8
2.6xlO~5-1.3xlO~*
Unknown
Unknown
7xlO"8
4.2xlO~5-1.2xlO~*
7.9xlO~'-3.5xlO~7
2xlO~8-1.3xlO~6
3xlO~*-2.6xlO~7
1.0xlO~'-2.3xlO~*
1.2xlO"5-1.5xlO"*
7.2xlO"'-1.9xlO"7
2.8xlO"8-lxlO~6
9.6x10"*
3.0xlO~5
3.4xlO"3
99. 8X
removal
80Z removal
lxlO~*
1.2x10"*
1.3xlO"5
1.0x10"*
2.6xlO"S
S.lxlO"5
l.SxlO"2
l.SxlO"5
lxlO~5
Comments Reference
Procedure Diihti et^ at. ,
unknown 1974
Hahon, 1979
Fochtnan and
Dobbs, 1979
Review Borneff,
article 1979
Abatract Schmidt,
1974
Translated Shevehenko
abstract et_ al^. , 1974
pH 7 and 9 Dobbs, et
data pooled al., 1978
Dobbs et
aj_., 1978
Weber et
-------
Table 1. Available Adsorption Data for Organics of Health Concern9 (continued)
Nunbor
Organic Molecular of data
compounds weight points
DDE
DDT
Dlchlorobeniene
1,2-Dlchloroe thane
2,4-Dlchlorophenoxy-
ocetlc acid (2.4-D)
Dleldrln
1,4-Dloxane
1,1-Dlphenylhydrazlne
Endrln
Halogcnated phenols
(eg. prntachlorophenol)
Ileptachlor
lleptachlor epoxldc
Hcxachloroe thane
Methoxychlor
Methylene Chloride
(Dlchloronvothane)
318
354.5
147.01
98.96
221
381
88.12
184
381
266.4
373.3
389
236.74
346
84.9
5
2
3
7
Unknown
6
8
5
6
5
2
Unknown
1
Unknov-n
17
Carbon
type
F300
F300
F300
F300
Aqua
Nuchar A
F300
F400
F400
F300
F300
F300
"Unknown
F300
Unknown
F400
Adsorption
test
method
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Isotherm
Z reduction
Column
recovery
Z reduction
Isotherm
Maximum
Equilibrium surface
concentration concentration,
range mol/1 mol/g
3.5xlO~9-l.lxlO~7
1.0xlO"8-5.9xlO"8
2.7xlO"5-6:8xlO~S
1.7xlO"7-6.6xlO"7
—7 -&
2x10 -2x10
2.1xlO~10-1.7xlO"8
2.5xlO"5-1.5xlO"3
2.2xlO"6-4.9xlO"5
1.8xlO"10-3.9xlO"8
1.7xlO"6-2.9xlO"5
1.6xlO"8-1.5xlO"7
Unknown
Unknown
Unknown
1.7xlO"5-5.0xlO"5
1.6xlO"5
2.3xlO"5
1.4xlO"3
6.9x10"'
4x10"*
1.7*10~5
2.6x10"*
8.1x10"*
6.2xlO~5
2.1xlO"3
3.3xlO"5
>80Z removal
Amount of
carbon not
specified
>80Z removal
S.BxlO"5
Comments Reference
Mahon, 1979
Mahon, 1979
1,4-dlchloro- M«hon, 1979
benzene
Dobbs et
• 1.., 1978
Sodium Aly and
salt Faust, 1965
Mahon, 1979
McCulre et
• 1.., 1978
Fochtman and
Dol.bs, 1979
Mahon, 1979
Pentachloro- Dobbs et
phenol, pH3 a_l. , 1978
Mahon, 1979
Abstract Schmidt, 1974
100Z removal Chrlswell
100 ug/l et^ a_K. 1977
influent
Abstract Schmidt, 1974
Weber et
al_., 1977
-------
Table 1. Available Adsorption Data for Organics of Health Concern3 (continued)
Organic
conrpounda
Maximum
Kuaber Adsorption Equilibrium surface
Molecular of data Carbon test concentration concentration,
weight points type not hod ranae mol/1 mol/g . Coraraente Reference
Nltrosamlnei 198.07 17 F300 Isotherm 5.0xlO"8-3.5xlO"S 2.2xlO~3
(eg N-Nltrosodlphenylamlne)
Polychlorlnated
blphcnyla (PCB)
(eg Arochlor 12S4)
Polynuclear
aroeatlc hydro-
carbons
Slnazlne
Tetrachloroethylene
Toluene
Toxaphene
1,1. 1-Trlchloroethane
Trlchloroethylene
2,4,5-Trlchloro-
phenoxyproplonlc
acid (2.4.5-TP)
Xylene
Mixture 6 F300 Isotherm 0.5ug/l-37ug/l 6.4 ng/g
Eight - 10 types Z reduction ppb level Variable
compounds
201.69 1 OU-A Z reduction 7xlO~7 4xlO~*
165.83 10 F400 Isotherm 6xlo"10-2.5xlO"8 6xlo"S
92.15 4 F300 Isotherm 1.7xlO"5-4.8xlO"5 9.5xlO"5
412 Unknown F300 Isotherm 5xlO~8-4xlO~7 lxlO~*
133.41 - F300 Isotherm 2.7xHf'-1.2xlO~3 2.7xlO"3
131.39 7 F300 Isotherm I.lxl0~'-4.8xl0~7 4.2xlO"5
Columbia
269 Unknown ••}£ Isotherm Unknown Unknown
106.2 4 F300 Isotherm 1.4xlO"5-l.lxlO~* 1.2xlO~3
N-Nltroso- Dobb* et
dlphenylamlne al . . 1978
Arochlor Nation, 1979
1254
99Z removal; Borneff,
Review Article 1979
Translated Shevenhenk
abstract it £l- • 1974
Weber et
al.. 1977
Mahon, 1979
Hager and
Rlzzo, 1974
b
Hansen. 1979
Dobbs et
al. . 1978
Follows Langmulr Weber and
Isotherm Could, 1966
Para laooer Dobbs et
al., 1978
Note: All values are for neutral forms of the molecules unless stated otherwise.
"There are no CAC data available for the following organic chemical* of concern
to health: B-BHC, bls(2-chloropropyl)ethcr, butyl bromide, dlbromochloropropane, 1,2-dlbromoethane, dtchlorofluoro-
DCthane, l,l,-dlchloroethene(vlnylldene chloride), cplchlorohydrln, ethylenethlourea (ETU), ethylene dibromlde, ethylene
oxide, kepone, methyl iodide, pentachloronltrobenzene (PCNB), polybromtnated blphenyla(PBB), 1,1,2-trlchloroethane,
. trlchloroftiioromethane.and vlnvl chloride.
ftr^»nal Co.i«nunicatlon.
-------
Table 2. Summation of Acute and(Chronic Exposure
Levels and Carcinogenic Risk Estimates
of Chemicals Reviewed
Chemical
Acrylonitrile
Benzene
Benzenehexachlor ide
Cadmium
Carbon tetrachloride
Dichlorodifluorome thane
1-2-Dichloroe thane
Epichlorhydrin
Ethylene dibrcmide
Methylene chloride
Polychlorinated biphenyl
Tetrachloroethylene
1,1, 1-Trichloroethylene
Trichlorof luorone thane
Toluene
Uranium
Xylenes
Bromide
Catechol
Chlorine dioxide
Chlorite
Chloroform
Dibronochlorame thane
2-4-Dichlorophenol
Hexachlorobenzene
Iodide
Resorcinol
Suggested No Adverse Response
Level (SNARL) mg/1
Exposure Period
24-hour 7-day Chronic
3.5
14
350
0.84
35
0.35
172
105
88
420
3.5
21
1400
2.2
22
18
115.5
11.7
12.6
0.5
0.08 0.005
2.0
5.6
0.53
5.
0.05
24.5
15
8
35 0.34
0.21
11.2
224 2.3
0.38
0.21
3.2
0.7
0.03
16.5 1.19
0.5
Upper 95%
Confidence
Estimate3 of
Lifetime Cancer
Risk Per ug/1
1.3 x 10~6
7.0 x 10"7
—f.
9.1 x 10 b
_7
1.4 x 10 '
_c
2.9 x 10 D
Note: See Drinking Water and Health (NAS, 1977) for details;
Safe Drinking Water Committee - Subcommittee on Toxicology,
headed by S. Murphy, U. Texas Medical School, Houston, Texas
(NAS, 1979).
aAn addition to the List of Known or Suspected Organic Chemical
Carcinogens found in Drinking Water (NAS, 1979).
48
-------
Human exposure data and sublethal animal data were used
to develop single agent SNARL exposures.
ADSORPTION EFFICIENCY OF GAC
The adsorption efficiency of GAC includes the following
major topics of study: Efficiency of GAC to Adsorb ug/1 Quan-
tities of Organic Compounds of Concern to Health, Effects of
Pretreatment on GAC Treatment, and Competitive Effects on GAC.
This discussion includes competitive effects between
trace organics, competition between humics and organics, and
displacement vs. re-equilibrium effects.
Efficiency of GAC to Adsorb Organics of Concern to Health
The efficiency of adsorption of many organic compounds is
shown in Figure 1 for over seven orders of magnitude of equilib-
rium concentration (McGuire and Suffet, 1979). The shaded area
represents an attempt to describe a general isotherm. At rela-
tively high equilibrium concentrations, C (depending on how well
a compound is adsorbed), the slope of theeisotherm is relatively
flat. As the equilibrium concentration decreases, the slope in-
creases until it becomes equal to 1.0, indicating compliance with
Henry's law of adsorption (Radke and Prausnitz, 1972). The more
poorly adsorbed compounds (urea for example) have a slope of 1.0
at high C values. The compounds which are more strongly ad-
sorbed have a slope of 1.0 at much lower C values. For example,
the isotherm for 2,4,6-trichlorophenol wasedetermined over five
orders of magnitude and the maximum slope is only 0.592 at 10
moles/1. Figure 1 also shows that for the isotherms determined
on a comparative basis, substituted phenols adsorb much better
than low molecular weight chlorinated organic compounds.
Our understanding of how specific organic compounds are
adsorbed is generally derived from the work on well adsorbed,
substituted phenols at mg/1 concentrations. Figure 1 indicates
that the poorly adsorbed and even the moderately well adsorbed
compounds have steep slopes at the yg/1 level (1x10 to
10 x 10~ moles/1). Reducing compound concentrations two orders
of magnitude into the ng/1 range requires a much larger carbon
dose or longer column contact time than a two-order-of-magnitude
decrease at the mg/1 level. Thus, reduction to subtrace levels
of these specific organics of health concern is more difficult
than would normally be expected based on the current understand-
ing of the adsorption of organic compounds.
Although the isotherms in Figure 1 were determined by
different investigators using different techniques and different
carbons, there is surprising agreement between isotherms for
the same compound. Clearly, other aspects of the experimental
-------
U1
o
Equilibrium Concentration, moles/1
Figure 1. General and Specific Adsorption Isotherms
a)(Di)-n-butyl phthalate b)Bis(2-chloroethyl) ether
c)Dimethyl phthalate d)Dibromochloromethane
-------
conditions including carbon type and water quality, affect? the
positions of the isotherms (McGuire, et al, 1978).
Available adsorption data for organics of health concern
are shown in Table 1. Table 1 indicates that there are adsorp-
tion data pertaining to 40 of the 58 specific organics and
classes of organic compounds on the lists. The data illustrate
that a great number of the chemicals of concern are adsorbable
on activated carbon; essentially all of the compounds tested
have been shown to be adsorbable to some extent. However, com-
petitive effects will be important in determining the capacity
observed in specific water treatment applications. The isotherm
data base should be expanded to provide the basic data for the
adsorption characteristics of chemicals that are potentially
harmful to health. Isotherms and mass transfer models have been
reviewed by McGuire and Suffet (1978). These models, hopefully,
will one day enable prediction of breakthrough curves of single
solutes in multi-solute systems.
Effects of Pretreatment Steps on GAC Treatment
Because pretreatment processes remove adsorbable organic
matter and consequently reduce the subsequent rate of saturation
of the GAC column, these processes should be used to their
maximum extent. Lowering the organic concentration will also
result in a lower GAC regeneration frequency, less competitive
adsorption, and lower demand for a disinfectant which could form
undesirable end products. Table 3 shows the general positive
effects of pretreatment steps on Total Organic Carbon (TOC), THM
precursors, and specific organics. The application of powdered
activated carbon (PAC) before GAC may be advisable in situations
involving spills of specific chemicals. It is recommended that
an adsorption process should be evaluated together with a clari-
fication process that precedes it in a water treatment plant.
In some cases, the removal of THM precursors by flocculation and
clarification may be sufficient, eliminating the need for an
adsorption process.
Competitive Effects on GAC
Competitive adsorption effects on GAC are created by the
variability of a wide spectrum of trace organics which are found
together in water supplies and generated within the treatment
process (Suffet, et al, 1978a). Dynamic situations in GAC beds
can be complicated due to a variety of competing species which
have different adsorbabilities and concentrations. The varia-
bility of specific organics which can enter a water treatment
process, their relative concentrations, their degree of competi-
tion, i.e., stronger vs. weaker adsorbing components, and their
relative diffusion properties will influence displacement
effects. Even if organics do not compete, shifts in influent
51
-------
Table 3. The Effect of Pretreatment on the
Removal of Adsorbable Organic Matter
METHOD
COAGULATION - SEDIMENTATION AND FILTRATION
A. Lower loading rates (25%-65% TOC), (Kavanaugh, 1978;
Semmens, et al, 1978)
B. Lower THM precursors (J> 70%) (Babcock & Singer, 1977)
C. Lower specific organics -
Phenols (Sridharan & Leer 1972)
Pesticides (Robeck, et al, 1965)
Polyaromatic (Andelman, 1973)
Hydrocarbons
AERATION-VOLATILES - Needs study
BANK FILTRATION - (< 65-76% DOC) McCreary and Snoeyink, 1977)
PAC - Needs study
concentration will cause the effluent concentration to re-
equilibrate with it. The extent of the re-equilibration will be
determined by adsorption equilibria in each case. Depending upon
the relative concentrations of these organics/ displacement also
can lead to temporarily higher concentrations in the effluent of
the GAC bed than in the influent for any given species.
Competition can be anticipated between trace organics, which
account for only a small fraction of the TOC, and the majority of
TOC, which is comprised of humic substances. It must be noted,
however, that humic substances will vary in adsorption proper-
ties, thereby making competitive interactions more difficult to
predict. Table 4 shows examples of competitive interactions
between humic substances and trace organics (Snoeyink, et al,
1977). The data show why musty odors of methylisoborneol (MIB)
and geosmin can be removed for many years, as these compounds
compete favorably with humic substances. Because the source of
humic substances may be different for each water supply and,
in fact, seasonally variable, pilot plant studies at specific
locations over the seasons of the year will yield the most
useful information on the effects of competition. More basic
data are needed in this area.
52
-------
Table 4. Competitive Effects Between Humic
Substances and Organics
COMPONENTS
Trichlorophenol (TCP)-
Humic Substances3
Humic Substances-
Methylisoborneol (MIB)
Humic Substances-
Geosmin
EQUILIBRIUM
CONCENTRATION
RANGE
10-50 mg/1.
10~b - 10"5 M
10-100 mg/1
0.1-100 yg/1
10-40 mg/1
0.1-100 yg/1
RELATIVE EXTENT OF
COMPETITIVE
DISPLACEMENT
TCP » Humic
Substances
MIB » Humic
Substances
Geosmin » Humic
Substances
Humic Substances-
Benzanthracene
10-100 mg/1
1-10 yg/1
None
Aldrich Co.
After snoeyink, et aL, 1977.
The wide spectrum of organics that have been identified
in water supplies represent an equally wide spectrum of adsorp-
tion behavior. This should also lead to competition among
organics of differing adsorbabilities for sites on the carbon
surface. However, there is little information on the mutual
reduction in adsorption capacity for competing organics in che
truly multicomponent mixtures that are encountered in drinking
water treatment. The direct information provided by the pilot
plant study seems to be an alternative to the more scientific,
but as yet underdeveloped, approach that is provided by mathe-
matical modeling of competitive adsorption.
The most immediate and presssing need is to determine
whether significant displacement Of any organics of health con-
cern occurs in the operation of GAC beds for treatment of drink-
ing water. The subcommittee recommends that even with the
identification of those trace organics of concern in the raw
water supply and measurement of their concentrations, including
the concentration variability, pilot plant studies will still be
necessary to confirm the relative order of breakthrough of each
contaminant and to assess the importance of any displacement
effect. This displacement effect can be caused both by a
variable influent composition and by competitive adsorption.
Although it is difficult to distinguish between these two causa-
tive factors in pilot plant studies, every effort should be
made to determine which is more important.
53
-------
MICROBIAL ACTIVITY ON GAC
The microbial activity on GAC includes the following major
topics of study:
• Interaction of Microbial Activity with Adsorption on
GAC
• Effects of 0., vs. Cl, Pretreatment on Microbial Activity
on GAC J z
• Impact of Microbial Colonization on Drinking Water,
including:
- Microbial Contamination of Drinking Water
- Microbial Products
- Microbial Toxins
Interaction of Microbial Activity with Adsorption on GAC
Electron micrographs provide clear evidence of the presence
of microorganisms on the surface of GAC (Weber, et al., 1978;
McElhany and McKeon, 1978). The types and number of microorgan-
isms will depend upon the amount and nature of the available
substrates. Table 5 divides organics in four categories of
adsorbability and degradability. While the classification is
simple, it also is ambiguous because the term nonbiodegradable
is not easily defined. The term recalcitrant, rather than
non-biodegradable, has been used by Alexander (1973) to describe
substances that persist for extended periods under all environ-
mental conditions thus far tested. Alexander (1973) lists 15
possible mechanisms to explain recalcitrance.
The organics that are most likely to be attacked microbially
fall into the non-adsorbable, biodegradable category (Table 5).
In treatment with GAC, these organics could be removed by the
biofilm surrounding the carbon granules without affecting the
adsorption process. The adsorbable, biodegradable category
(Category 4, Table 5) of organics is of more interest because of
the strong interaction expected between biological and adsorption
processes (Ying and Weber, 1978; Tien, 1978; Benedek, 1977).
This sets GAC apart from sand or other nonadsorbing media.
In water treatment, the removal of adsorbable, biodegradable
organics (Category 4, Table 5) by microbial activity, rather than
by adsorption, enhances the opportunity for the GAC bed to remove
the adsorbable, nonbiodegradable organics (Category 3, Table 5).
Category 3 is generally of the most concern because it contains
many of the synthetic organics that are suspected carcinogens.
Thus, the question of whether or not GAC can be regenerated
biologically needs to be addressed. The evidence reported thus
far is incomplete and contradictory.
54
-------
Table 5. Interaction of Microbial Activity
with Adsorption on GAC
Category
b
1
2b
3a
4b
Ad sor bab i 1 ity
Non
Non
High
High
Degr ad ab i li ty
Non
High
Non
High
a Synthetic Organic Category-Chlorinated
Organics
b
More Sites for cateoory £, the Organics of health
concern shown in Table 1.
While the GAC bed is a suitable environment for microbial
activity, questions regarding which specific organics are biode-
gradable and at what rate also remain unanswered by research.
In addition to accessibility, the concentration of the substrate
is an important factor. Clearly, many organics of concern are
present in drinking water in such low concentrations that the
opportunities for microbial action may be quite limited unless
adsorption occurs first; however, adsorption into the pores of
GAC may prevent biodegradation if the substrate becomes inacces-
sible to the microorganism. One factor which could influence
substrate availability is re-equilibration when influent concen-
tration decreases in a dynamic situation. In this case, sorbed
substrate could be released and once again become available to
microorganisms.
Category 3 from Table 5 (adsorbability and nondegradability)
is the area of particular interest for organics of health con-
cern. No direct evidence has been given for removal of specific
organics of health concern by microbial activity. Only one study
(at the Jefferson Parish Water Works, which was reported by
Brodtmann, et al., 1979), implied that precursors to THM forma-
tion were removed by microbial action. An indirect benefit of
microbial activity may be the lengthening of GAC service time by
removing organics that would otherwise occupy adsorption sites,
but this in itself does not imply more effective removal of
organics of health concern.
55
-------
Much of the recent interest on the effect of microbial
activity and GAC performance stems from European research and
experience. While operation of GAC beds for 6 months to 2 years
without regeneration has shown that microbial activity removes
organics as measured by group parameters/ such as TOC, potassium
permanganate demand, chemical oxygen demand (COD), and UV-
absorbance, the efficiency of biodegradation is less than that
of adsorption. Removal of about 1.5 mg/1 TOC by microbial degra-
dation requires about 30 minutes of GAC contact (Eberhardt, et
al., 1974; Jekel, 1977; Benedek, 1977) while adsorption only
requires approximately 10 to 15 minutes.
More careful investigation is needed to determine if
adsorbed organics, which are more resistant to biodegradation,
can be acted upon by microbes.
Effects of Ozone vs. Chlorine Pretreatment on Microbial Activity
on GAC
Miller, et al. (1978) and Benedek (1977) have reviewed pilot
scale tests of preozonation before GAC at Bremen, AmstPrdam, and
Morsang-Sur-Seine and full scale tests at Dusseldorf, Mulheim,
and Rouen-la-Chapelle. These tests have shown that organics,
measured by collective organic parameters (e.g., TOC:UV) are
efficiently removed through the total system, including the
flocculation and adsorption steps; however, the ability of micro-
bial action to aid in removal of specific organics of health
concern was not reported. Thus far, one study, at the U.S. EPA
Cincinnati pilot plant (U.S. EPA 1978b) implied that precursors
to THM formation were somehow altered by ozonation in such a
manner that their removal was made more effective.
The role ot ozone in promoting microbial activity remains
unclear. Conversion of organics to more biodegradable forms and
addition of oxygen to the biofilm are two advantages implied by
Rice, et al. (1978) and Sontheimer, et al. (1978). On the nega-
tive side, less adsorbable, polar organics, such as carboxylic
acid and aldehydes have been identified in drinking water (Safe
Drawing Water Subcommittee on the Chemistry of Disinfectants
NAS, 1979). Benedek (1977) also shows that ozone may decrease
the adsorbability of some organics.
While prechlorination does not stop microbial growth on
GAC, there is some evidence to suggest that it results in the
formation of chlorinated organics (Safe Drinking Water Subcom-
mittee on the Chemistry of Disinfectants, NAS, 1979). These
organics are considered much more resistant to biodegradation on
the GAC surface (Sontheimer, et al., 1978). Studies are needed
to determine the changes in biodegradability and adsorbability
which are brought about by ozonation or chlorination of specific
organics of concern to public health.
56
-------
Impact of Microbial Colonization on Drinking Water
The attachment of microbes to GAC surfaces has been
described by Marshall (1976) and Daniels (1972). Data from
studies on microbial numbers in effluents of GAC filters indi-
cate detachment is a negligible problem, controllable by back-
washing (Eberhardt, et al., 1974; Klotz, et al., 1976; Mueller
and Bernhardt, 1976; Sontheimer, et al., 1978; Schalekamp 1976).
Microbial Contamination of Drinking Water—
The bacteria that have been reported on GAC beds are con-
sidered nonpathogenic. Genera identified include Flavobacterium
and Xanthomonas (U.S. EPA, 1978b); Acinetobacter, Pseudomonas,
Caulobacter, coryneforms, Flavobacterium, AlcalFgenes, actino-
mycetes. Bacillus, Planctomyces, and Moraxella (Ben Blanken,
1978); Pseudomonas, Enterobacter agglomerans, Acinetobacter,
Alcaligenes faecalis, Moraxella, and Flavobacterium (Love and
Symons, 1978, reporting the work of Parsons), and ten species
of Pseudomonas (McElhaney and McKeon, 1978). Although it is
unlikely, the possibility that some pathogens may be able to
colonize GAC beds should be tested. Of primary importance are
enteric pathogens, such as Salmonella, Shigella, Vibrio, Yersinia
enterocolitica, and the enterotoxigenic E. coTT.
Microbial Products—
The published research does not indicate which compounds
are or are not acted upon by microbial activity on GAC and what
compounds may be generated as a consequence of bacterial growth.
Attention has been focused on bulk organic materials in water
and not on discrete classes of molecules.
Microorganisms can generate a variety of highly potent,
low molecular weight toxicants in culture. Studies of several
model environments and some natural systems indicate that
toxicants may be or are indeed formed during GAC treatment
(Boethling and Alexander, 1979).
Microbial Toxins—
Endotoxins are lipopolysaccharides from gram-negative
bacteria. Endotoxin levels in water that have been filtered
through GAC are either not increased or not significantly
increased (Love and Symons, 1978 and Jorgenson, et al., 1976).
The measured levels are very low and should pose no risk.
Discussion of Microbial Activity on GAC
If the use of GAC is to become widespread, research is
necessary to identify the factors that are responsible for the
initiation of microbial growth on the GAC surface and the
microbiological generation of organic compounds during carbon
treatment. Attention should be given to the compounds that are
present and retained in the bed as well as to those that are
57
-------
released and could appear in the effluent. Research should be
directed toward compounds generated from innocuous, natural
precursors as well as toward those that may be formed as a
result of microbial action on synthetic organic chemicals. The
investigation should identify the specific organic molecules
thus formed and not simply the quantity of organic carbon that
is emitted from GAG. The toxicological significance of such
compounds and the possibility of their occurrence in concentra-
tions sufficiently high to be of concern to public health should
be assessed.
PRODUCTION OF NONBIOLOGICAL SUBSTANCES BY OR WITHIN THE GAG BED
One potential source of chemical breakthrough from a GAG
column is chemical change on GAG surfaces. This section evalu-
ates the significance of this potential source of toxic organic
chemicals in drinking water. Few studies have addressed the
ability of carbon to catalyze reactions producing new chemical
components and their release to water that is being purified for
human consumption. Emphasis has been placed on adsorption and
removal. Table 6 shows examples of physiochemical interactions
on GAG. It appears that reactions on GAG involve surface oxides
and other impurities in the carbon structure (Cookson, 1978).
Surface oxides are formed for example by the reduction of dis-
infectants.
Table 6. Physiochemical Interactions on GAG
1. CATALYSIS BY GAG - Bush -S-S- (Ishizaki and Cookson, 1974)
2. OXIDATION OR REDUCTION REACTION ON
GAG - MALONIC ACID (Garten and Weiss, 1959)
3. CATALYSIS BY METALS ON GAG - Fe (Rideal and Wright, 1926)
4. RELEASE OF CHEMICALS FROM GAG - Oxidation of GAG itself
5. REDUCTION OF DISINFECTANTS - Cl, BY GAC - C* CO + Cl~
(Magee, 1956)
0- BY GAC - C* CO + 02
(Dietz and Bitner, 1972)
6. RELEASE OF CARBON FINES - 2.5-25 m diameter (McCarty,
et al, 1979)
58
-------
Because there is a limited amount of data on catalysis on
activated carbon, one is handicapped when attempting to draw
specific conclusions. In addition, generalizations are extremely
risky, making it difficult to project the possible significance
of catalysis during the process of adsorption of organics on
GAG. Reaction products that were not originally present can be
released to the water solution and behave as typical solutes in
their adsorption equilibrium and competition for carbon sites.
This subject needs further research.
There is a possibility that carbon may undergo significant
changes via its patterns of use and regeneration procedures.
The studies that were reviewed point to the effects of adsorbed
ions on carbon's catalytic potential for a number of reactions.
Thus, it appears possible that the recycling and regeneration
of activated carbon can result in a material that is radically
different than the virgin material. This may indicate the need
for a control program that prevents the carbon utilized for
drinking water from being used for any other purpose.
Chlorine, chlorine dioxide, and ozone react readily with
carbon and may react with compounds adsorbed on carbon. There is
no evidence to indicate that such reactions, under the conditions
that exist in water treatment plants, will or will not produce
potentially hazardous compounds. Research is needed to identify
the end products of reactions between activated carbon and disin-
fectants, especially chlorine and chlorine dioxide. Also the
interaction of oxygen with the carbon surface must be evaluated.
Carbon fines can be released during use. McCarty, et al.
(1979) found four times the number of fines released during up-
flow filtration versus downflow filtration. Studies should be
conducted to determine levels of carbon fines that pass into
drinking water and their biological significance.
REGENERATION OF GAC
Regeneration procedures influence the chemical properties
of GAC and this in turn will also influence adsorption, catalytic
properties, and leachable chemicals. The ash content, surface
oxides, and pore size distribution of GAC are examples of the
properties which can change by regeneration processes and which
can effect adsorption (Cookson, 1978). Sufficient data are not
available to determine differences in adsorption between regener-
ated and virgin GAC. At present, no significant health problems
have been documented.
The leaching of toxic organics from carbon to water has not
been studied sufficiently to draw general conclusions. Although
small amounts of polycyclic aromatic hydrocarbons (PAH's) have
been detected (Borneff, 1979), they are well adsorbed and do not
appear to readily leach off. The analytical studies are few,
59
-------
emphasizing the need for evaluation of various regenerated carbon
types and virgin carbons.
The leaching of metals from carbon to water appears to be
slight, producing concentrations too low to affect health (Love
and Symons, 1978).
Pollutants are discharged during the GAC regeneration
process. Few studies have sufficiently characterized the air
pollutant discharges to evaluate their health effects. These
studies are also needed.
ADSORPTION EFFICIENCY OF OTHER ADSORBENTS
Two distinctly different processes were considered: The
use of anion exchange resins for the removal of humic material
and the use of polymeric adsorbents, e.g., Ambersorb XE-340 (Rohm
and Haas Co.), for removal of low molecular weight organics of
health concern. The physical and chemical nature of the resins
must be considered. Attrition of resin by abrasion or osmotic
forces should be minimized to maintain long-lived beds and to
minimize escape into the water distribution system. Chemical
degradation by oxidants such as Cl and ozone should be minimized
by reduction to Cl~ and O~ since oxidants can alter the activity
of a resin by changing the nature of the surface (Kunin, 1976).
Removal of Humic Substances by Ion Exchange Resins
The use of ion exchange resins other than cation exchange
resins for softening potable water has been quite limited. Most
data on the use of anion exchange resins for the removal of humic
material comes from treatment of cooling water for power utili-
ties (Kunin, 1973, 1976). There is sufficient evidence indicat-
ing that the technology can be applied to drinking water prior
to chlorination to remove precursors of CHC1., production, the
so-called trihalomethane formation potential (THMFP) and humic
material in general. Davis (1977) and Boening, et al. (1979)
showed that strong base anion exchange resins were the best in
adsorbing humics (Table 7). Thirty to 64 % of the exchange capa-
city was regenerated with NaOH by both authors. Boening et al.
(1979) cautioned that although strong base resins remove more
organics, they are that much harder to regenerate than weak base
resins and regeneration is critical for the use of anion
exchangers.
Pilot plant studies by Ko'lle (1976), Gauntlett (1975), Jayes
and Abrams (1968), Tilsworth (1974), and Wood and DeMarco (1979)
have shown the ability of anion exchangers to remove color, total
organics, and THM precursors from ground and surface waters.
Wood and DeMarco (1979) found that the anion exchange resin IR-
904 removed more THM precursor (=50%) from raw and lime softened
60
-------
Table 7. Adsorption of Humic Acids by Anion Exchange Resins
Humic Acidc Removed, %
Resin
IRA-410a
IRA-410a
A-57b
A-30Bb
ES-340b
pH 2.2
97
92
79
71
82
pH 7
70
52
22
44
59
Source: Davis (1977)
f Rohm and Haas, Corp. - Strong base resin
Diamond Shamrock, Corp. - Intermediate base resin
c Commercially available humic acids
water than GAC or Ambersorb XE-340, but GAC removed slightly more
from finished water. The anion exchanger did not remove 22 pur-
geable organics and in fact, in the presence of chlorine the
anion exchange process catalyzed formation of high concentrations
of these organics [e.g., chloroform (1.75X), and trichloroethy-
lene (10X)].
Present studies indicate that nitrosamines may be generated
by passage of water over a strongly basic resin (Fiddler, et al.,
1977; Cohen and Backman 1978; Gough, et al., 1977). The exact
mechanism has yet to be demonstrated. This is not surprising
since the resins concentrate the potential reactants for forming
nitrosamines and also accumulate potential catalysts (iron and
copper). The published data on the formation of nitrosamines on
anion exchange resins suggest that the use of such resins for
potable water should be reconsidered. It may be possible to use
anion exchange resins in conjunction with GAC; the resin would
remove the bulk of the humic material and the GAC would remove
the remainder of the humic material as well as any nitrosamines
that had been generated. However, there are no data for such
possible systems.
More research on anion exchange resins and their use in
potable water treatment is needed. Present data on the use of
anion exchange resins are' based upon those that were regenerated
only a few times or have been used in their virgin states. Such
data are highly unrealistic for the water treatment industry.
61
-------
An acceptable loss of capacity upon regeneration can be addressed
only after many regenerations are completed.
Adsorption Efficiency of Polymeric Adsorbents
Experience with polymeric adsorbents to remove organic com-
pounds from drinking water and regeneration of these materials
are limited in comparison with the use of GAG. Laboratory
studies and a few recent pilot plant studies indicate that the
synthetic carbonaceous adsorbents of the Ambersorb XE-340 type
do not have the broad-spectrum adsorption properties of GAC, and
that they selectively adsorb lower molecular weight organics but
apparently do not readily adsorb humic material (Chudyk, et al.,
1979; Neely, 1979). However, the data indicate some promise for
special situations, such as the removal of toxic organics from
sources contaminated with low molecular weight organics (Wood and
DeMarco, 1979).
Compounds of health concern for which isotherm data are
available include only chloroform (XAD-4, XE-340), bromodichloro-
methane (XE-340) and dieldrin (XAD-4, XE-340) (Neely, 1979; Rohm
and Haas, 1977; and Chudyk, et al., 1979). Chudyk, et al., (1979)
found 10 mg/1 of a commercially available humic acid did not
affect adsorption of >ig/l amounts of CHC1- on XE-340. CHC1.,
adsorption on XE-340, relative to GAC, has been shown to be
3:1 to 5:1 at high CHC1- concentrations {Neely, 1979) and
approximately 1:1 at low CHC13 concentrations (Chudyk,et al.,
1979). The degree of competition between organic compounds needs
to be studied.
In a series of small column studies, extremely small adsorp-
tive capacities were exhibited by the non-carbonaceous resins
XAD-2 and XAD-7 in comparison to GAC and XE-340 (McGuire, 1977).
This is in agreement with results from pilot plant studies
(Suffet, et al., 1978 a and b) in which they compared Filtrasorb
400, XAD-2 and XE-340 as post treatment adsorbers for chlorinated
Philadelphia drinking water from the Delaware River. XAD-2 was
shown to be of much lower capacity (Suffet, et al., 1978a).
XE-340 removed low molecular weight organics, whereas GAC removed
high and low molecular weight organics. Caution was expressed
that the data were largely qualitative and interpretation is
complicated by the highly variable nature of the organic content
at the influent to the adsorption columns.
Wood and DeMarco (1979), during a pilot scale study of a
groundwater supply at Miami, reported that XE-340 did not remove
THMFP where GAC removed 20 percent from finished water. How-
ever, XE-340 did remove 24 percent of the THMFP from raw water.
XE-340 always removed a greater amount of individual volatile
compounds than GAC per unit weight, e.g., chloroform (4X), cis-1,
2-dichloroethane (3X), bromodichloromethane (2.8X), and
dibromochloromethane (1.5X) from finished water.
62
-------
To further define resin uses, studies on resin regeneration
systems are needed to determine the efficiency of different types
of regenerants, attrition rates/ and oxidant effects on the
adsorbent. The resin to be cycled should be used for treating
contaminated well water to provide direct interpretations of
removal efficiencies. Slejko and Meigs (1979) state that XE-340
may be regenerated by steam with periodic alcohol washes but only
preliminary data are available. Crucial questions remain con-
cerning how to regenerate efficiently, what type regenerant to
use, and the actual rate of loss of capacity.
ANALYTICAL METHODS TO MONITOR ADSORBENT UNIT PROCESSES IN WATER
TREATMENT
The focus of this report has been on the individual com-
pounds of potential harm to health. Thus, the primary purpose
of this section is to evaluate if analytical methods are avail-
able to monitor the operation of adsorbent processes for these
compounds at a water treatment plant.
The variability of the complement of organic compounds
(Table 8) in the influent to the GAC column affects the effi-
ciency of the adsorptive process for compounds and must be moni-
tored. Chemical compounds in the effluent may be the same as
those that entered the carbon bed or they may have been changed
Table 8. Organic Compounds in Drinking Water Functionally
Defined by the U.S. EPA, (1978b)
Class
I CAUSES OF TASTE AND ODOR
b
II SYNTHETIC ORGANIC COMPOUNDS
III PRECURSORS-CLASS IV REACTIVE0
IV DISINFECTANT BYPRODUCTS5
V NATURAL ORGANIC COMPOUNDS-NOT CLASS III3
a90% by weight of the TOC are high molecular weight compounds
of Class V.
Compounds of health concern are present in these classes.
These compounds are represented in Table 2.
°Precursors primarily of Class IV are represented by THMFP.
63
-------
by chemical or microbial action within the bed. The variability
of species type and concentration in the column influent are also
reflected in variable effluent concentrations from competitive
and reequilibration effects discussed in the Competitive Effects
section of this report.
Thus, infrequent grab sampling might not be sufficient for
an adequate evaluation of material entering or leaving a GAG
column. Since minute-to-minute sampling is presently imprac-
tical, the variability of these organics in the influent and
effluent can best be followed by on-line composite sample col-
lection augmented by grab samples.
Two different analytical approaches can be used to determine
specific organic compounds of health concern. In one approach,
individual compounds are analyzed that have been selected because
of health implications. In the second, a general "screening"
procedure for the isolation of organic compounds is used and the
compounds of interest are selectively determined from the same
sample matrix.
Table 2 lists specific pollutants of potential health con-
cern that should be monitored. Specific analytical methods for
each compound or groups of compounds are being developed at the
present time and are reviewed by Keith and Telliard (1979).
The second approach to analysis uses screening procedures.
The screening procedure consists of a qualitative analysis for
individual compounds of health concern and subsequent quantita-
tive evaluation of the constituents that were found. A general
sequence that is followed for screening of trace organics is:
Sample > Isolation Method > Concentration >
Chromatographic Separation > Qualitative
Identification > Quantification
Suffet and Radziul (1976) have described the methods that
can isolate chemicals with health implications. For example,
the less polar volatile organics of health concern are isolated
by methods such as the purge and trap volatile organic analysis
(b.p. £ 150°C) (Grob, 1976), liquid-liquid extraction (Yohe, et
al., 1979), and macroreticular resin accumulators (Chriswell, et
al.r 1977). These methods must be supplemented by specific
analysis for compounds that are not isolated by the specific
extraction methods.
The complex group of trace organics in the extracted sample
is chromatographed and a sample profile or fingerprint is obtain-
ed. This consitutes an information pattern in each chromato-
graphic analysis. When the profile is compared to a profile of
a set of standards of the compounds of health concern, a tenta-
tive identification of the sample components can be made. When
64
-------
many profiles of influents and effluents are plotted in the same
manner, chromatographic profiles can be compared and differences
noted. GC/MS analyses must be used to confirm the identification
of the compounds tentatively identified by the isolation method.
Examples of the screening have been reported for purge and trap
analyses (Steiglitz, et al., 1976, and Wood and DeMarco, 1979),
resin accumulators (Suffet, et al., 1978a), and liquid-liquid
extraction (Yohe, et al., 1979). When monitoring an adsorption
process for volatile pollutants of health concern that can be
detected by GC techniques, only selected GC/MS analyses may be
necessary. GC analysis can be used for monitoring purposes if
defined concentration limits for pollutants are established.
Nonspecific Organic Analyses
Nonspecific organic analyses lump together a large number
of organic compounds into one collective measurement heap. The
analysis of specific organic compounds with suspected health
implications (Table 2) should be the criteria by which all of
the nonspecific analyses are evaluated.
Nonspecific organic analytical techniques for monitoring the
performance of adsorbent beds need further development. Correla-
tions between nonspecific measurements and compounds of health
concern should be completed on a site-specific basis, because the
specific organic compounds which must be controlled will probably
vary among water supplies. The correlation of such measurements
as UV absorbance, fluorescence, or analyses of organic carbon
(e.g., TOC) with GAC-breakthrough of specific organic compounds
of health concern has not been observed (Brodtmann et al., 1979;
Wood and DeMarco, 1979; Symons, et al., 1975). However, TOC
remains an important measure of the mass loading of an adsorption
column.
The correlation of Total Organic Halogen (TOX) in the
aqueous or absorbed phases with organic compounds of health con-
cern has not been sufficiently studied. This should be developed
for routine use as many compounds of health concern are chlori-
nated, Table 2.
THM precursors which participate in the haloform reaction
have not been shown to correlate with nonspecific organic analy-
sis. Although fluorescence has been used to monitor humic mate-
rial (Snoeyink, et al., 1977; Sylvia and Donlon, 1979), it has
not been shown to correlate with THMFP. TOC and THMFP did not
correlate well in pilot column studies (Wood and DeMarco, 1979;
and Cairo, et al., 1979). Organic carbon is a collective measure
of all organic constituents and only a few of these are haloform
precursors.
65
-------
GENERAL CONCLUSIONS AND RECOMMENDATIONS
Raw water sources and disinfected water supplies may contain
organic compounds that have been demonstrated to be carcinogenic
or otherwise toxic in experimental animals or in epidemiological
studies. Also present are a large number of compounds that
either have not been identified or their effects on health have
not been characterized. Properly operated GAC systems can
remove or effectively reduce the concentration of many of the
compounds described above. Less is known about synthetic resins
than about GAC, but it is known that they can be applied to
remove certain types of organic contaminants.
The information available as of this date on the treatment
of water with GAC provides no evidence that harmful health
effects are produced by the process under proper operating con-
ditions. However, there are incomplete studies on the possible
production of such effects with virgin or regenerated carbon
through reactions that may be catalyzed by the GAC surface,
reactions of disinfectants with GAC or compounds adsorbed on it,
reactions mediated by microorganisms that are part of the pro-
cess, or by the growth of undesirable microorganisms on GAC.
Studies are also needed on the properties of regenerated acti-
vated carbons and on the adsorption of additional contaminants
with potential health effects.
The frequency of GAC regeneration is determined by the
organic compounds in the water and their competitive interac-
tions. The types and concentrations of organic compounds may
vary widely in different locations and seasons of the year.
Competitive interactions are complex and presently cannot be pre-
dicted without data from laboratory and/or pilot scale tests on
the water to be treated.
While there is ample evidence for the effectiveness of GAC
in removing many organics of health concern, more data are needed
on the quantification of any harmful health effects related to
the use of GAC. This need, however, should not prevent the pre-
sent use of GAC at locations where analysis of the water supply
clearly indicates the existence of a potential health hazard
greater than that which would result from the use of GAC.
Clarification processes (coagulation, sedimentation, filtra-
tion) remove significant amounts of some organics, especially
some types of THM precursors and relatively insoluble compounds
that may be associated with particulates. In some cases, the
removal of THM precursors by clarification may be sufficient to
eliminate the need for an adsorption process.
66
-------
REFERENCES
1. Alexander, M. 1973. Biotechnology report: Nonbiodegrad-
able and other recalcitrant molecules. Biotechnol. Bioeng.
15:611-647.
2. Aly, O.M. and S.D. Faust. 1965. Removal of 2,4-dichloro-
phenoxyacetic acid derivatives from natural waters. J. Am.
Water Works Assoc. 57:221-230.
3. Andelman, J.B. 1973. World Health Organization, European
standards for organic matter in drinking water, p. 55-61
in V.L. Snoeyink, ed. Organic Matter in Water Supplies:
Occurrence, Significance, and Control, Proceedings of the
15th Water Quality Conference, Univ. of Illinois at Urbana-
Champaign, Urbana, 111.
4. Babcock, D.B. and P.C. Singer. 1977. Chlorination and
coagulation of humic and fulvic acids. Proceedings of the
97th Annual American Water Works Association Conference,
Anaheim, Ca. Paper No. 16-6, 16 pp.
5. Benedek, A. 1977. The effect of ozone on activated carbon
adsorption - A mechanistic analysis of water treatment
data. Presented at 101 Symposium on Advanced Ozone Tech-
nology. International Ozone Institute, Toronto, Ontario.
26 pp.
6. Boening, P.H., D.D. Beckmann, and V.L. Snoeyink. 1979.
Activated carbon vs. resin adsorption of humic substances.
J. Am. Water Works Assoc. (In press.)
7. Boethling, R.S. and M. Alexander. 1979. Effect of concen-
tration of. organic chemicals on their biodegradation by
natural microbial communities. Appl. Environ. Microbiol.
(In press.)
8. Borneff, J. 1979. Elimination of carcinogens (excluding
haloforms) by active carbon. In I.H. Suffet, and M.J.
McGuire, eds. Activated Carbon Adsorption of Organics from
the Aqueous Phase, Proceedings of the 1978 ACS Symposium in
Miami Beach, FL. Ann Arbor Science Publ., Ann Arbor, MI.
(In press.)
67
-------
9. Brodtmann, N. , J. DeMarco, and D. Greenberg. 1979. Crit-
ical study of large-scale granular activated carbon filter
units for the removal of organic substances from drinking
water. In I.H. Suffet, and M.J. McGuire, eds. Activated
Carbon Adsorption of Organics from the Aqueous Phase,
Proceedings of the 1978 ACS Symposium in Miami Beach, FL.
Ann Arbor Science Publ., Ann Arbor, MI. (In press.)
10. Cairo, P.R., J.V. Radziul, I.H. Suffet, and M.J. McGuire.
1979. The application of bench-scale and pilot-scale
studies for control of organic chemical contaminants in
drinking water. In I.H. Suffet, and M.J. McGuire, eds.
Activated Carbon Adsorption of Organics from the Aqueous
Phase, Proceedings of the 1978 ACS Symposium in Miami
Beach, FL. Ann Arbor Science Publ., Ann Arbor, MI. (In
press.)
11. Chriswell, C.D., R.L. Ericson, G.A. Junk, K.W. Lee, J.S.
Fritz, and H.J. Svec. 1977. Comparison of macroreticular
resin and activated carbon as sorbents. J. Am. Water Works
Assoc. 69:669-674.
12. Chudyk, W.A., V.I. Snoeyink, D.D. Beckman, and T.J.
Temperly. 1979. Activated carbon vs. resin adsorption of
2-methylisoborneol and chloroform. J. Am. Water Works Assoc,
(In press.)
13. Cohen, J.B. and J.D. Backman. 1978. Measurement of En-
vironmental Nitrosamines. pp. 357-372, in E.A. Walker, L.
Grichute, M. Casteguaro, and R.E. Lyle, Eds. Environmental
Aspects of N-Nitroso Compounds. World Health Organization,
International Agency for Research on Cancer, Lyon IARC Sci,
Pub #19.
14. Cookson, J.T., Jr. 1978. Adsorption mechanisms: The
chemistry of organic adsorption on activated carbon.
Pp. 241-279 in P.N. Cheremisinoff and F. Ellerbusch eds.
Carbon Adsorption Handbook. Ann Arbor Science Publ.
Ann Arbor, Mich.
15. Dahm, D.B., R.J. Pilie, and J.P. Lafornara. 1974.
Technology for managing spills on land and water. Envir.
Sci. Tech. 8:1076-1079.
16. Daniel, S.L. 1972. The Adsorption of Microorganisms onto
Solid Surfaces: A Review, pp. 211-253 in chapt. 19.
Developments in Industrial Microbiology Process of the 28th
General Meeting of the Society for Industrial Microbiology.
Published by Am.Inst.Biol.Sci.Washington,D.C.
68
-------
17. Davis, T.A. 1977. Electro-regenerated ion-exchange
deionization of drinking water. Final report. U.S.
Environmental Protection Agency, Health Effects Research
Lab., Cincinnati, Ohio. (Report No.) EPA/600/1-77-035.
83 pp.
18. den Blanken, J.G. 1978. Research on the bacteria popula-
tion and its activity in activated carbon filters and its
influence on drinking water quality. Annual Report,
Laboratory of Sanitary Engineering, Dept. of Civil
Engineering Delft University, The Netherlands.
19. Dietz, V.R. and J.I. Bitner. 1972. The reaction of ozone
with adsorbent charcoal. Carbon 10:145-154.
20. Dobbs, R.A., R.J. Middendorf, and J.M. Cohen. 1978.
Carbon adsorption isotherms for toxic organics. U.S.
Environmental Protection Agency. Office of Research and
Development. Municipal Environmental Research Laboratory,
Cincinnati, Ohio. 131 pp.
21. Eberhardt, M. 1976. Experience with the use of biologi-
cally effective activated carbon. Pp. 331-347 in H.
Sontheimer, ed. Translation of Reports on Special Problems
of Water Technology. Volume 9 - Adsorption. Conference
held in Karlsruhe, Federal Republic of Germany, 1975.
U.S. Environmental Protection Agency, Municipal Environ-
mental Research Lab., Cincinnati, Ohio. [Report No.]
EPA-600/9-76-030.
22. Fiddler, W., J.W. Pensabene, P.C. Doerr and C.J. Dooley.
1977. The presence of dimethyl- and diethylnitrosamines
in deionized water. Food Cosmet. Toxicol. 15:441-443.
23. Fochtman, E.G., and R.A. Dobbs. 1979. Adsorption of car-
cinogenic compounds by activated carbon. In I.H. Suffet
and M.J. McGuire, eds. Activated Carbon Adsorption of
Organics from the Aqueous Phase, Proceedings of the 1978
ACS Symposium in Miami Beach, FL. Ann Arbor Sci. Publ.,
Ann Arbor, MI. (In press.)
24. Garten, V.A., and D.E. Weiss. 1959. Functional oroups in
activated carbon and carbon black with ion- and elec££Qn-
exchanqe properties. Proc. Third Conf. on Carbon, Univ. of
Buffalo, pp 295-313.
25. Gauntlett, R. 1975. A comparison between ion-exchanqe
resins and activated carbon for the ranoval of'
from water. Water Research Technology Report TRIO,
Medmenham Laboratory, England.
69
-------
26. Gough, T.A., K.S. Webb and M.F. McPhail. 1977. Volatile
Nitrosamines from Ion Exchange Resins. Food. Cosmet.
Toxicol. 15: 432-440.
27. Grob, K. and F. Zurcher. 1976. Stripping of trace organic
substances from water. Equipment and Procedures. J.
Chromatog. 117:285-294.
28. Hager, D.G. and J.I. Rizzo. 1974. Removal of toxic
organics from wastewater by adsorption with granular
activated carbon. Presented at USEPA Technology Transfer
Session on Treatment of Toxic Chemicals/ Atlanta, Georgia.
29. Hansen, 1979 - Private Communication Calgon Corp.
30. IRLG - Interagency Regulatory Liaison Group. 1978.
Regulations Release Chemical Hit List. Chem. and Eng. News
56:50.
31. Ishizaki, C. and J.T. Cookson, Jr. 1974. Influence of
surface oxides on adsorption and catalysis with activated
carbon. Pp. 201-231 in A.J. Rubin, ed. Chemistry of Water
Supplyf Treatment and Distribution. Ann Arbor Science
Publ., Ann Arbor, MI.
32. Jayes, D.A., and I.A. Abrams. 1968. A new method of color
removal: Development report. N. Eng. Water Works Assoc.
82:15-25.
33. Jekel, M. 1977. Biological treatment of surface waters in
activated carbon filters. Presented at the meeting of the
Water Research Center (England), KIWA (Netherlands) and EBI
(Fed. Rep. Germany), Engler-Bunte Institute, Karlsruhe
University, Fed. Rep. Germany.
34. Jorgensen, J.H., J.C. Lee, and H.R. Pahren. 1976. Rapid
detection of bacterial endotoxins in drinking water and
renovated wastewater. Appl. Environ. Microbiol. 32:347-
351.
35. Kavanaugh, M.C. 1978. Coagulation for removal of trihalo-
methane precursors at the Ralph D. Bollman Treatment Plant.
Presented at the 98th Annual Conference of the American
Water Works Association, Atlantic City, New Jersey. Paper
No. 15-3, 21 pp.
36. Keith, L.H. and W.A. Telliard. 1979. Priority pollutants
I. A perspective view. Environ. Sci. Tech. 13(4):416-
423.
70
-------
37. Klotz, M., P. Werner, and R. Schweisfurth. 1976. Investi-
gations concerning the microbiology of activated carbon
filters. Pp. 312-330 in H. Sontheimer, ed. Translation of
Reports on Special Problems of Water Technology. Volume 9
- Adsorption. Conference held in Karlsruhe, Federal
Republic of Germany, 1975. U.S. Environmental Protection
Agency, Municipal Environmental Research Lab., Cincinnati,
Ohio. (Report No.) EPA-600/9-76-030.
38. Kolle, W. 1976. Use of macroporous ion exchangers for
drinking water purification. Pp. 405-413 in H. Sontheimer,
ed. Translation of Reports on Special Problems of Water
Technology. Volume 9 - Adsorption. Conference held in
Karlsruhe, Federal Republic of Germany, 1975. U.S.
Environmental Protection Agency, Municipal Environmental
Research Lab., Cincinnati, Ohio. [Report No.] EPA-600-9-
76-030.
39. Kunin, R. 1973. Helpful hints in ion exchange technology
(conclusion). Amber-Hi-Lites No. 132. Rohm and Haas Co.,
Philadelphia. 5 pp.
40. Kunin, R. 1972. Ion Exchange Resins. Krieger Publ. Co.,
Huntington, N.Y. 504 pp.
41. Love, T.O. and J.M. Symons. 1978. Operational aspects of
granular activated carbon adsorption treatment. Unpub-
lished paper of U.S. Environmental Protection Agency.
Drinking Water Research Division.
42. Marshall, K.C. 1976. Interfaces in Microbial Ecology.
Harvard Univ. Press, Cambridge, Mass. 156 pp.
43. Magee, V. 1956. The application of granular active carbon
for dechlorination of water supplies. Proc. Soc. Water
Treat. Exam. 5:17-40. Chem. Abs. 54:11342f, 1960
44. Mahon, J.H. 1979. Selected carbon isotherms conducted on
single organic compounds. Calgon Corp. Pittsburgh, PA.
45. McCarty, P.L., D. Argo and M. Reinhard. 1979. Operational
experiences with activated carbon adsorbers at Water
Factory 21. pp. 456-477 of these Proceedings.
46. McCreary, J.J., and V.L. Snoeyink. 1977. Granular
activated carbon in water treatment. J. Am. Water Works
Assoc. 69:437-444.
71
-------
47. McElhaney, J. and W.R. McKeon. 1978. Enumeration and
identification of bacteria in granular activated carbon
columns. In Proc. 6th Ann. Am. Water Works Assoc. Water
Qual. Tech. Conf.f Louisville, Kentucky, (Dec. 1978).
48. McGuire, M.J., I.H. Suffet, and J.V. Radziul. 1978.
Assessment of unit processes for the removal of trace
organic compounds from drinking water. J. Am. Water Works
Assoc. 70:565-572.
49. McGuire, M.J., and I.H. Suffet. 1978. Adsorption of
organics from domestic water supplies. J. Am. Water Works
Assoc. 70:621-636.
50. McGuire, M.J. and I.H. Suffet. 1979 "The Calculated Net
Adsorption Energy Concept" in I.H. Suffet, and M.J.
McGuire, eds. Activated Carbon Adsorption of Organics from
the Aqueous Phase, Proceedings of the 1978 ACS Symposium in
Miami Beach, FL. Ann Arbor Science Publ., Ann Arbor, MI.
(In press.)
51. McGuire, M.J. 1977. The optimization of water treatment
unit processes for the removal of trace organic compounds
with an emphasis on the adsorption mechanism, Ph.D. Thesis,
Drexel University, Philadelphia, Pennsylvania.
52. Miller, G.W., P.G. Rice, C.M. Robson, R.I. Scullin, and W.
Kuhn* 1978. An assessment of ozone and chlorine dioxide
technologies for treatment of municipal water supplies.
Municipal Environmental Res. Lab., Water Supply Res. Div.,
U.S. Environmental Protection Agency, Cincinnati, Ohio.
(Report No.) EPA-600/2-78-147. 588 p.
53. Mueller, G. and H. Bernhardt. 1976. Comparative bacteri-
ological examinations on activated charcoal filter using
chlorinated and ozonized ground and reservoir waer. Forum
Umwelt Hyg. 2(27):393-396. Translated TR-77-311 by EPA.
54. National Academy of Sciences. 1979. Drinking Water and
Health. Safe Drinking Water Committee, National Research
Council. NAS, Washington, D.C. (In press.)
55. National Academy of Sciences. 1977. Drinking Water and
Health. Safe Drinking Water Committee, National Research
Council. Washington, D.C. 939 pp.
56. National Cancer Institute. 1978. Human health considera-
tions of carcinogenic organic chemical contaminants in
drinking water. Fed. Reg. 43:29148-29150.
72
-------
57. Neely, J.W. 1979. A model for the removal of trihalo-
methanes from water by Ambersorb XE-340. In I.H. Suffet
and M.J. McGuire, eds. Activated Carbon Adsorption of
Organlcs from the Aqueous Phase, Proceedings of the 1978
ACS Symposium in Miami Beach/ FL. Ann Arbor Sci. Publ.,
Ann ARbor, MI. (In press.)
58. Radke, C.J., and J.M. Prausnitz. 1972 . Thermodynamics of
multi-solute adsorption from dilute liquid solutions.
AIChE J. 18:761-768.
59. Rice, R.G., G.W. Miller, C.M. Robson, and W. Kuhn. 1978.
A review of the status of preozonation of granular
activated carbon for removal of dissolved organics and
ammonia from water and wastewater. Pp. 485-537 in P.N.
Cheremisinoff and F. Ellerbusch, eds. Carbon Adsorption
Handbook. Ann Arbor Science Publishers/ Inc., Ann Arbor,
Mich.
60. Rideal, E.K., and W.M. Wright. 1926. Low temperature
oxidation at charcoal surfaces. Part II. The behavior of
charcoal in the presence of promoters. J. Chem. Soc.
128:1813-1821.
61. Robeck, G.G., K.A. Dostal, J.M. Cohen, and J.F. Kreissl.
1965. Effectiveness of water treatment processes in
pesticide removal. J. Am. Water Works Assoc. 57:181-200.
[Chem. Abs. 62:11532e, 1965]
62. Rohm and Haas Co. 1977. Ambersorb carbonaceous adsorb-
ents. Bulletin 1E-231. Philadelphia. 19 pp.
63. Schalekamp, M. 1976. Use of activated carbon in the
treatment of lake water. Pp. 128-159 in H. Sontheimer,
ed. Translation of Reports on Special Problems of Water
Technology. Volume 9 - Adsorption. Conference held in
Karlsruhe, Federal Republic of Germany, 1975. U.S.
Environmental Protection Agency, Municipal Environmental
Research Lab., Cincinnati, Ohio. (Report No.) EPA-600/
9-76-030.
64. Schmidt, K. 1974. Effectiveness of standard drinking
water preparation for eliminating pesticides and other
pollutants. Gas. Wasserfach Wasser-Abwasser. 115:72-76.
65. Semmens, M., J.K. Edzwald, M. Taylor, and R. Sanks. 1978.
Organics Removal by Coagulation - A Review and Research
Needs. Presented at the 98th Annual AWWA Conf., Atlantic
City, NJ, 35 pp.
73
-------
66. Shevehenko, M.A., P.V. Marchenko, P.N. Tarah, and E.V.
Kravets. 1974. Removal of some pesticides from drinking
water by adsorption methods. Vodosnabzh. Sanit. Tekn.
(USSR) 10:31-32.
67. Slejko, F.L. and G. Meigs. 1979. Economic analysis of
employing Ambersorb XE-340 carbonaceous adsorbent in trace
organics removal from drinking water. In I.H. Suffet and
M.J. McGuire, eds. Activated Carbon Adsorption of Organics
from the Aqueous Phase/ Proceedings of the 1978 ACS
Symposium in Miami Beach, PL. Ann Arbor Sci. Publ., Ann
Arbor, MI.(In press.)
68. Snoeyink, V.L. and J.J. McCreary. 1978. Reaction of
activated carbon with aqueous chlorine and other disinfec-
tion agents. Technical Progress Report, U.S. Environmental
Protection Agency, Grant No. R 805-293-01, Univ. of 111. at
Urbana-Champaign. Dept. of Civil Engineering, Urbana, 111.
33 pp.
69. Snoeyink, V.L., J.J. McCreary, and C.J. Murin. 1977.
Activated carbon adsorption of trace organic compounds.
Final report. U.S. Environmental Protection Agency,
Municipal Environmental Research Lab., Cincinnati, Ohio.
Water Supply Research Div. (Report No.) EPA/600/2-77/223.
129 p.
70. Sontheimer, H., E. Heilker, M.P. Jekel, H. Nolte and F.H.
Vollmer. 1978. The Muelheim process. J. Am. Water Works
Assoc. 70:393-396.
71. Sridharan, N. and G.F. Lee. 1972. Coprecipitation of
organic compounds from lake water by iron salts. Environ.
Sci. Technol. 6:1031-1033.
72. Steiglitz, L. , W. Roth, W. Kuhn, and W. Leger. 1976.
Behavior of organic halogen compounds in the purification
of drinking water. Vom Wasser 47:347-377.
73. Suffet, I.H., L. Brenner, J.T. Coyle and P.P. Cairo. 1978a
Evaluation of the capability of granular activated carbon
and XAD-2 resin to remove trace organics from treated
drinking water. Environ. Sci. Technol. 12:1315-1322.
74. Suffet, I.H. and J.V. Radziul. 1976. Guidelines for
the quantitative and qualitative screening of organic
pollutants in water supplies. J. Am. Water Works Assoc.
68:5230-524, Addendum: 69: 174, 1977.
74
-------
75. Suffet, I.H., Radziul, J.V., Cairo, P.R. and Coyle, J.T.
1978b. Evaluation of the Capability of Granular Activated
Carbon and Resins to Remove Chlorinated and Other Trace
Organics from Treated Drinking Water. In R. Jolley et al.,
Eds. Water Chlorination; Env. Impact & Health Effect.
Vol. 2 Ann Arbor Sci. Publ., Ann Arbor, MI.
76. Sylvia, A.E. and R.J. Donlan. 1979. The use of fluorescence
in monitoring removal of organics in drinking water. In
I.H. Suffet, and M.J. McGuire, eds. Activated Carbon
Adsorption of Organics from the Aqueous Phase, Proceedings
of the 1978 ACS Symposium in Miami Beach/ FL. Ann Arbor
Science Publ., Ann Arbor, MI.(In press.)
77. Symons, J.M., T.A. Bellar, J.K. Carswell, J. DeMarco, K.L.
Kropp, G.G. Robeck, D.R. Seeger, C.J. Slocum, E.L. Smith,
and A.A. Stevens. 1975. National organics reconnaissance
survey for halogenated organics. J. Am. Water Works Assoc.
67:634-648.
78. Tien, C. 1979. Bacterial growth on adsorption interac-
tions in granular activated carbon column. In I.H. Suffet,
and M.J. McGuire, eds. Activated Carbon Adsorption of
Organics from the Aqueous Phase, Proceedings of the 1978
ACS Symposium in Miami Beach, FL.Ann Arbor Science Publ.,
Ann Arbor, MI. (In press.)
79. Tilsworth, T. 1974. Organic and Color Removal from Water
Supplies by Synthetic Resinous Adsorbents. Inst. of Water
Resources, U.S. Dept. Interior/ Office of Water Resource
Research. Report #IRW-50, pp. 56 (NTIS # 233068).
80. U.S. Environmental Protection Agency. 1978a. Interim
primary drinking water regulations: control of organic
chemical contaminants in drinking water. Fed. Reg. 43
(28):5756-5780.
81. U.S. Environmental Protection Agency. 1978b. Interim
treatment guide for controlling organic contaminants
in drinking water using granular activated carbon.
Symons, J.M., ed. Support document for U.S. Environmental
Protection Agency Interim Primary Drinking Water Regula-
tions, Control of Organic Chemical Contaminants in Drinking
Water; Fed. Reg. 43 (28):5756-5780, Feb. 9, 1978.
Water Supply Research Div., Municipal Environmental Res.
Lab., U.S. Environmental Protection Agency, Cincinnati,
Ohio. 207 pp.
82. U.S. Environmental Protection Agency. 1975. National
Interim Primary Drinking Water Regulations. Fed. Reg.
Dec. 24, 1975. pp. 59566-59588.
75
-------
83. Weber, W.J., Jr. and J.P. Gould. 1966. Sorption of
organic pesticides from aqueous solution. Iri Organic
Pesticides in the Environment, Advances in Chemistry Series
60, American Chemical Society, Washington, B.C.
84. Weber, W.J., Jr., M. Pirbazari, M. Herbert and R. Thompson.
1977. Effectiveness of activated carbon for removal of
volatile halogenated hydrocarbons from drinking water, pp.
125-141 in J.A. Borchardt, J.K. Cleland, W.J. Pedman and G.
Olivier, eds. Viruses and Trace Contaminants in Water and
Wastewater. Ann Arbor Science Publishers,Inc., Ann Arbor,
Mich.
85. Weber, W.J., Jr., M. Pirbazari and G.L. Melson. 1978.
Biological growth on activated carbon: An investigation
by scanning electron microscopy. Envir. Sci. Tech.
12:817-819.
86. Wood, P. and J. DeMarco. 1979. Effectiveness of various
adsorbents in removing organic compounds from water. Part
II: Removing Total Organic Carbon and Trihalomethane
Precursors Substance. In I.H. Suffet and M.J. McGuire,
eds. Activated Carbon Absorption of Organics from the
Aqueous Phase, Proceedings of the 1978 ACS Symposium in
Miami Beach, FL.Ann Arbor Science Publ., Ann Arbor, MI.
(In press.)
87. Ying, W.C. and W.J. Weber, Jr. 1978. Bio-physiochemical
adsorption systems for wastewater treatment: predictive
modeling for design and operation. Presented at the 33rd
Annual Purdue Industrial Waste Conference Purdue Univ.
88. Yohe, T.L., I.H. Suffet, and R.J. Grochowski. 1979.
Development of a teflon helix continuous liquid-liquid
extraction apparatus and its application for the analysis
of organic pollutants in drinking water. ASTM Special
Technical Publ. #686. Symposium on Organic Pollutants,
Denver, Colo., June 19-20, 1978. (In press.)
76
-------
U.S.A. EXPERIENCE WITH
GRANULAR ACTIVATED CARBON ADSORPTION
Richard H. Moser
INTRODUCTION
In 1975, the United States Environmental Protection Agency
conducted a survey of water treatment plants using granular
activated carbon for taste and odor control. The results have
been summarized by Symons . They found 33 plants using GAC;
30 on a routine basis and three on an experimental basis.
Thirteen other plants have converted some, but not all, of their
filters to GAC. The average quantity of water treated at these
plants was about 8 mgd, ranging from 0.3 to 33.5 mgd. The
average media depth was 28 inches with the range of 11 to 84
inches. A sand layer was being used under the GAC at 77 percent
of these plants. In 95 percent of these cases, water was
gravity fed to the GAC but only 14 percent had the water pre-
filtered prior .to passing through the adsorbent. At 35 percent
of these plants, 2 gpm/ft was the filtration rate, with the
remainder about equally divided above and below this value.
Finally, 82 percent of the plants were installed in the years
1970 to 1975.
More recently, larger GAC plants have been placed into
service, with Chattanooga, Tennessee (part of the American Water
Works System) being the largest at 72 mgd nominal capacity.
Of the 46 plants in the EPA survey, 13 were part of the AWW
System; therefore, our experience constitutes at least one third
of the USA experience with granular carbon in drinking water
plants.
We did not know exactly why GAC did a better job of con-
trolling taste and odor than did powdered carbon and the chemi-
cal oxidants, nor did we feel it was essential to know. The
material worked—and that's what counted. Still, today, the
precise mechanics of GAC adsorption are a subject of contro-
versy. Researchers did ascertain that this "magical" granular
black material phenomenally removed dissolved organics from
water, especially those large molecule organic chemicals
77
-------
associated with taste and odor problems. Yet it was not enough
to know that GAG did work, we had to know how it worked, and why
in some cases it did not work. We wanted to be able to predict
how well it would work and for how long. Lastly, we had to know
if any problems were being created by its use. All of these
concerns are rightfully addressed by the GAG developers and
researchers. However, those of us who had experience with GAG
also became aware of these concerns and are now able to assist
in the development of the answers.
TASTE AND ODOR CONTROL
The American Water Works System began its GAG experience at
Hopewell, Virginia, in 1960 and currently has 19 systems with
GAG. Table 1 shows the current list. Hopewell, interestingly^
is the only one which uses GAG as a post contactor; even as long
ago as 1960, it was recognized that post contactors are better,
especially in cases of heavy organic -loading where bed life is
a matter of weeks. Such was the case at Hopewell, where only 30
to 60 days of carbon life could be expected to remove the earthy
odor now known to be caused by geosmin.* Figure 1 shows the
Hopewell plant flow diagram. The Appomattox River, source of
supply for Hopewell, receives back water from the nutrient-rich
James River during high tide, thereby providing ideal conditions
for growth of Actinomycetes. Thankfully, the James River water
quality has improved over the years so that we now get a one
year service life.
Regeneration was not installed at Hopewell because this was
our first GAG experience. After the first few exchanges of GAG,
the operating personnel decided to replace some sand with
partially spent GAG to get a longer contact time. They found
that the carbon could still dechlorinate after it was exhausted
for taste and odor control. Since they had been dechlorinating
with sodium bisulfite, enough sand filters were converted to
spent carbon filters to accomplish the same degree of dechlori-
nation. The carbon was tested in parallel with sand for sus-
pended solids removal and it was found that carbon did slightly
better than sand without affecting its dechlorinating proper-
ties. (See Table 2). Today, some carbon has been in service
for almost 10 years and is still effectively dechlorinating.
Since then, many other locations have verified GAC's ability to
remove suspended matter and now it is routinely accepted as a
filter medium.
Over the years, we found that the iodine number (an indica-
tion of available surface area) of GAG decreases with use, which
means that the carbon spaces are becoming filled with organic
compounds. Experience taught us that when the iodine number
Identified by. Dr. J.K.G. Silvey of North Texas State University
78
-------
Table 1
American Water Works System Locations Presently Using GAC
Bed Depth (In.) LBS. GAC
Illinois
E. St. Louis 18 412,200
Granite Oity 18 174,000
Peoria 30 165,000
Indiana
Kokomo 20 162,000
Muncie 21 90,500
Richmond 24 100,900
Terre Haute 30 127,000
Iowa
Davenport 24 331,700
Kentucky
Lexington 14 177,600
Ohio
Ashtabula 24 75,000
Marion 18 80,000
Pennsylvania
Mew Castle 30 190,800
Pittsburgh 30 1,947,400
Tennessee
Chattanooga 30 927,400
Virginia
Hopewell 60 81,000
West Virginia
Huntington 30 337,500
Madison 18 7,300
Princeton 18 24,300
Weston 18 1*1,600
Total 19 — 5,126,200
79
-------
I
Apponattox River
Low Service Pump
2-Stage High Energy Mix
X^
r T"
pT»pne feT*
Pump
z Virgin
'ilters,
50" Deep
sd Water
Storage
ributive
Pumps
J.
r
' ^
)Five Settling
Basins
HO Filters at 0.67 MGD each
with soent carbon medium
I
I
1
Filtered Mate
Storage
Distributive
Pumps
w
, TQ
Industrial
System
To
Domestic
System
Figure 1. Virginia-American Water Company
Hopewell, Virginia
80
-------
Table 2
Impurity Removal by Adsorption - Filtration by
Carbon Beds* and by Sand Medium (Hopewell, Virginia)
Impurity
Threshold Odor
Number
Color
Manganese (ppm)
Iron (ppm)
Turbidity (JTU)
Chlorine (ppm)
Influent
35-140
4-14
0.066-0.15
o;2-0.37
0.45-1.4
1.4-2.8
Carbon Bed
Effluent
0-4 +
0-2
0.008-0.017
0.006-0.025
0.07-0.15
0-0.25
Sand Medium
Effluent
35-70+
1-2
0.008-0.017
0.012-0.087
.0.10-0.25
1.4-2.8
*24" Deep at 2
+0dor samples over 60 day period; other data 150 day period,
Extracted from: "Removal of Organic Contaminants by Granular
Carbon Filtration" by Hager and Flentje,
JAWWA, November, 1965.
81
-------
is reduced to about 300 (virgin material ranges from 650 to
1100) that the carbon is no longer effective enough to assure
taste and odor control. While the iodine number drops, the
apparent density (the weight of carbon per unit volume) in-
creases, also indicating pick-up of organic chemicals. An
increase of about 20 percent (0.42 to 0.50 gm/cm ) seems to
accompany the drop in the iodine number. Figures 2 and 3 show
these effects. A threshold odor number of three seems to
correlate with the iodine number and the apparent density as
described above, so they make good operating parameters.
When the American Water Works Co. purchased several pro-
perties in West Virginia, the plant serving the City of Nitro
was a prime candidate for GAC. Figure 4 shows the plant layout.
The Kanawha River contained innumerable organic wastes because
of the high density of chemical industry in the Charleston area.
Not only was GAC considered to be necessary, but tests showed
that taste and odor removal could only be assured for a few
months. During these tests, some specific compounds were
identified and shown to be adsorbed by virgin GAC (Tables 3 and
4). Data on CCE and COD was also generated, but not included
here because these parameters are no longer considered signifi-
cant. To minimize capital expenditure, GAC was used as a sand
replacement. With the cooperation of Calgon Corp., a multiple
hearth furnace (Figure 5) was leased and installed at the Nitro
Water Plant for exclusive use by our people for regeneration
every few months. The furnace was larger than needed and was
therefore used periodically. Probably because of the start/stop
operation, the regeneration facility performed very unsatisfac-
torily. The furnace was difficult to start up, carbon losses in
handling were about 10 percent, and temperature was hard to
control.
Because the furnace was so difficult to operate, regenera-
tion was not conducted as often as needed. Threshold odor
became a secondary concern as regeneration frequency was changed
to once per year. Regeneration efficiency was determined only
by periodic tests for apparent density, so data is inconclusive
on that question. Table 5 shows a regeneration cost breakdown
done after a few years1 operation. Finally in 1973, the Nitro
Plant was replaced by a new facility using an unpolluted stream.
The Nitro installation constitutes our only (and until recently
the nation's only) plant scale regeneration experience in a
drinking water plant.
Since then, we have learned considerably more about GAC,
how it is used, how it works, and its pitfalls. The remaining
17 systems that we have in service were installed as filter/
adsorbers, because it was known or guaranteed that the GAC would
have a multiyear life for taste and odor removal. The difficult
task of removing GAC from a conventional filter box need not be
82
-------
M
0)
0)
C
•H
•O
o
1000
900
800
700
600
500
uoo
300
200
Years
Figure 2. Typical Iodine Number vs. Time Relationship
12315
Years
Figure 3. Typical Apparent Density vs. Time Relationship
83
-------
STEAM BOILER
FURNACE
CONTROL PANEL
10
ill PUMP HOUSE]
y
£
V
RB<
!
w
>
JE
)N
J
9
7-
T
FILTER
BUILDING
A
FI.TERS
*
i
0
•
^—^
SECONDARY
DO
CARBON SLURRY
TANK
QUENCH TANK
CARBON MAKE-UP
HOPPER
REGENERATION
FURNACE
SETTLING
BASIN
SETTLING BASIN
AERATION
BASIN
AERATION |
PUMP
. STATION //
PUMP STATION
50 1OO
SCALE - FEET
Figure 4. West Virginia Water Company Plant Layout Showing
Location of Carbon Filters and Carbon Reactivation
Equipment
84
-------
00
1 BEIMC FILL EP WITH REACTIVATED CAKBOM.
FlLTEB B«'Wb BACKWASMED To EITHER: —
CI.VAM OUT CARBON @Eo To RCMO
TURBIDITY-.
B. SETTLE < EveM OUT CAQBOM Beo
THAT HA.S Been FILLED WITH CARBOM
AND WHICH is Now> BCIKI& u5ro AS A
OPCRATIOIsl TO REMOVE OOJCCT-
IOMABUE T^S,TE AMD ODOR FRoM H'.
WATER. RRloO. To IT& BEIMG. Pon«PSO To
Fll-Te« IN WMICf) Tile &PCKIT CARBOM IS
F«>R QEACTIVATIOM AFTER
WHICH IT £»>»J BE KCU&eo l»l THE SYSTEM
AS MOTTiO I
Figure 5. Granular Carbon Purification System, West Virginia
Water Company Nitro, West Virginia
-------
Table 3
Concentrations of Suspected Contaminants in Parts per Billion
00
en
Con taminant
Ethylbenzene
Styreme
Bis (2-chloroethyl) ether
2-Ethyl hexanol
Bis (2-rchlorolsopropyl) ether
Methyl benzyl alcohol
Acetophenone
Isophorone
Tetralin
Total
Threshold odor number (25°C)
Nov. 13, 1963
Raw
Kanawha
11
18
55
110
2
19
90
-------
Table 4
Influence of Contact Time on Passage of Specific Contaminants,
Experiment 1
Contaminant
Ethylbenzene
Bis (2-chloroethyl)
ether
2-Ethyl hexanol
Bis (2-chloro-
Isopropyl -ether
Methyl benzyl
alcohol
Acetophenone
Isophorone
Total
Threshold odor
number (25°C)
Contact Tiaie-min
1.9
5
3-8
5.6
7-5
Depth-ft
10
15
20
Concentration*-ppb
20
9*
57
26
62
11
_12.
285
32
18
kn
20
10
13
-------
Table 5
Direct Costs of Granular Carbon Reactivation*
Estimated Daily Costst
Fuel ($21.20/1000 m3)
Power ($0.01/kWh)
Steam ($4.40/1000 kg)
Make-up Carbon (5% loss/
cycle)
Labor (existing personnel
normally adequate)
U.S.
Dollars
5
3
5
13
25
Pounds
Sterling
2.08
1.25
2.08
5.41
10.42
38
15.83
Cost per kg** 3.3 cents 1.4 pence
*Excludes amortization, insurance, and taxes (if any).
tBased on a 75 cm I.D. x 6 hearth furnace
**Based on 1135 kg/day of carbon being reactivated.
Extracted from:
"Adsorption and Filtration with Granular
Activated Carbon" by Hager and Fulker,
Journal of the Society for Water Treatment
and Examination, Vol. 17, 1968.
88
-------
done so often as to be a burden. Service life of GAC beds has
ranged from two to six years, except for Hopewell.
GAC can provide benefits beyond taste and odor removal as
we found out when the James River was contaminated with kepone
in 1974. Naturally, Hopewell's supply was affected. Hundreds
of tests were conducted by the Commonwealth of Virginia and
ourselves which showed without a doubt that the carbon removed
virtually all of that contaminant (Table 6). Thanks to GAC, we
had no urgent need to be concerned about kepone in the drinking
water.
Table 6
Kepone Removal by Granular Carbon at Hopewell, Virginia
River Water Pre GAC Post GAC
Kepone (ppb)
Avg.* 0.017 0.011 0.001
*60 samples taken over a two year period.
HANDLING OF CARBON
During these past 20 years, we have handled millions of
pounds of carbon. We have used hand shovels, pails, drums,
eductors, eductors on hoists with trolleys, cranes with buckets,
and vacuum trucks—all of which work. Undoubtedly, the Hopewell
plant personnel had more experience than anywhere else in our
company. The method employed for carbon removal from the filters
uses the water eductor with a periodic backwash at low rate. By
maneuvering the suction hose around the filter, virtually all the
carbon can be removed, even where gravel is supporting it. When
sand is present, it is more difficult to remove all the carbon
without getting the sand mixed with it. But the system works
very well, and with a little equipment and a little experience
removing 800 ft from a 1 mgd filter, it can be done in less
than four man-hours.
PROBLEMS WITH GAC USE
Most of our experience has been with bituminous material,
but lately we have had good success with lignite. Not all
the experiences with GAC have been good. Since GAC is lighter
than other media, backwash rates had to be lowered to prevent
loss of material. Not only lower backwash rates, but the rate
89
-------
as a function of water temperature, suddenly became a crucial
operating parameter. A more precise wash flow control had to be
installed to prevent loss of the relatively expensive material.
Figure 6 shows that relationship. Recently, we have found that
some carbons require higher wash rates as shown in Figure 7,
which also shows the effect of particle size. This may demon-
strate a change in carbon raw materials or in processing, but
with variation like this, filter washing requires closer oper-
ational control. Even with good control, some carbon gets washed
away - perhaps one inch per year.
Another problem is the possible accumulation of mud in the
media. This does not happen in all cases, but where floe
particles are particularly cohesive, as in a high polymer-dosed
water, the floe does not readily wash away at these low wash
rates, even when wash times have been extended. Contrary to
what might be expected, wash water consumption does not decrease
with use of GAG (because of lower rates). In actuality, wash
water consumption often increases due to the extended times
required to clean the bed.
Another operating problem is due to carbon's affinity for
chlorine. In many cases, the absence of chlorine has caused
bacteria to multiply in the bed causing the filtered water to
contain bacteria many-fold higher than the inlet water. Total
plate count bacteria have been controlled by backwashing with
water containing up to 50 ppm of combined chlorine residual in
all but one case. Pittsburgh is the exception, where bacteria
counts in filtered water are tens of thousands per ml during the
summer. Fortunately, post chlorination lowers the total plate
count to less than 10 colonies/ml. Since this is not a national
problem, I presume there is not much concern. Contrarily,
there are some who contend that the bacteria should be encour-
aged to multiply. Those of us who have for years strived to
reduce general bacteria to as close to zero as possible,
throughout the treatment process, have difficulty in accepting
that concept, although it seems to have merit in reducing
organics.
Despite having some problems, granular activated carbon is
an excellent operating tool - at a cost which is not exorbitant
for taste and odor control. Table 7 gives some recent cost data
for GAC treatment after several years operation. These costs do
not reflect the decrease in other taste and odor chemicals,
changes in wash water, changes in chlorine consumption, or any
savings in sludge disposal. The overall cost reduction for just
taste and odor control at these locations has been 34 percent.
SPECIFIC ORGANIC REMOVAL
While we installed granular carbon only where severe,
chronic taste or odor problems occurred, we recognized that
90
-------
80
70
60
Z
2 50
CO
Z
< 40
Q.
X
m 30
20
10
0
35° F
6 8 10 12
LINEAR FLOW RATE (gpm/ft2)
16
Figure 6. Bed Expansion of Hydrodarco 1030 (10x30 Mesh)
At Various Temperatures (Water)
-------
Witter nt 77° F (26°C)
•' = HYOnODANCO 1030
HYDMODARCO 810
I) 10 12 M 10
Supni ficiiil Vnlocily <|pni/si| It
Effective size of 1030 - 0.8 to 0.9 mm
Effective size of 8l6 = 1.2 to 1.3 mm
Figure 7. Bed Expansion Curves for
Hydrodarco 1030 and Hydrodarco 816
-------
Table 7
Cost Data
vo
Plant
Ashtabula, OH
Chattanooga, TN
Daveppqrt, IA
E. St. Louis, IL
Granite City, IL
Kokomo, IN
Lexington, KY
Marion, Oil
Muncie, IN
Peoria, IL
Richmond, IN
Terre Haute, IN
Total
Capacity
(MOD)
I)
81
30
35
13
11
20
7
0
30
B
6
253
Raw Water Source
Lake Erie
Tenneasee River
Mississippi River
Mississippi River
Mississippi River
Wildcat Creek
Private Reservoir
Wells and Soioto River
White River
Illinois River
Private Reservoir
Wells and Wabash River
$/MG
3-70
5.73
16.30
3.38
8.35
1.63
3.95
3-71
6.57
11.68
6.19
3- "7
6.117
AV.
-------
other organics were being removed. However, it was not until
recently that such data have been generated. We participated in
the ORSANCO study, replenishing some filters with virgin GAG
at Huntington and also evaluated the partially spent GAG at
Pittsburgh. Mr. Miltner will present that data later in this
conference.
We have conducted trihalomethane analyses throughout our
entire system and have found that most GAG installations were
already exhausted. However, at Hopewell, after 9 months,
considerable THMs were still being removed (see Table 8). The
same phenomenon occurred at Davenport, Iowa after 7 months,
as Table 9 reveals. Therefore, we can say that one cannot
predict bed life for THM removal without specific, case-by-case
evaluation.
I have also reviewed some of our data looking for possible
desorption effects. Table 10 is an example of how a particular
compound can be at a higher concentration in the carbon-filtered
water than in the applied water. Researchers now say one can
expect this to occur when carbon is used long after its removal
life, which can explain the results in this table. However,
with all the compounds which granular carbon can remove, how
does one know when (and even if) desorption occurs? This
question and others cannot be answered in this paper. Future
studies will hopefully answer this and other questions being
raised about applications of granular activated carbon in water
treatment.
REFERENCES
1. Symons, J.M. "Summary of Granular Activated Carbon Prac-
tice Data"; Water Supply Research Division, U.S. Environ-
mental Protection Agency, Cincinnati, Ohio; February 17,
1976, mimeo, 10 pp.
2. Dostal, Kenneth A., Rex C. Pierson, Donald G. Hager, and
Gordon G. Robeck. 1965. "Carbon Bed Design Study at Nitro,
West Virginia"; Journal of American Water Works
Association; 57.
3. Hager, D.G. and R.D. Fulker. 1968. "Adsorption and Fil-
tration with Granular Activated Carbon"; Journal of the
Society for Water Treatment and Examination; 17.
4. Blanck, Clarence A. "Taste and Odor Control Utilizing
Granular Activated Carbon in the Plains Region of the
American Water Works System", In: Central Illinois
Section Meeting-American Society of Civil Engineers,
Illinois, March 22, 1979.
94
-------
Table 8
Trihalomethanes at Hopewell, Virginia Plant Effluent
Chloroform
Dibromochloromethane
Dichlorobroraomethane
Bromofonn
TTHM
% Removal TTHM
Concentration, ug/1
7 month old (GAC
Pre GAC
77
22
17
20
136
Post GAC
55
5
-------
Table 9
Davenport Water Company
Date
a\
Location
Trlhalomethanes
(UK/1)
Reduction
November
1978*
December
1978
January
1979
Filter Influent
Filter Effluent
Filter Influent
Filter Effluent
Filter Influent
Filter Effluent
Quenched
152
120
93
97
71
62
Formation
Potential
87
15
56
8
26
7
Terminal
239
135
l'»9
105
97
69
4'»
30
29
-------
Table 10
Example of Possible Competitive Adsorption and Desorption
Western Pennsylvania Water Company - Pittsburgh, Pa.
Concentration, ug/1
Chloroform
Dibromochloromethane
Dichlorobromomethane
Bromofonn
TTHM *
Filte
Influent
30
3
1
28
62
r #18
Effluent
16
23
9
15
63
Filt
Influent
11
6
<1
5
22
er #2
Effluent
26
9
4.1
7
42
NOTE: 1. Influent and effluent samples taken at same time.
Analyses done by NUS Corp., Pittsburgh, Pa.
2. Granular carbon approximately 20 aonths in service,
97
-------
DISCUSSION
Following Monday Morning Session (April 30)
Q PROFESSOR McCARTY: Mr. Moser, I'd like to know more about
the one case in which you used spent activated carbon in the
filters for dechlorination. Do you feel it is a good idea to
use carbon loaded with organics for dechlorination? Did you
consider the organics problem at all? Did you investigate the
possible formation of chlorinated organics?
A MR. MOSER: In 1961 it seemed like a good idea, and I don't
mean to say we wouldn't do it again; however, I think we would
probably investigate some other factors first. We have virgin
carbon after the prefilters, so in a sense I don't see how that
differs from having carbon in one filter and expecting the
organics to be removed on the top. We really are relying on the
virgin material on the bottom to remove substances that might
tend to bleed through or slough off from the top portion of the
filter. The carbon in those primary filters has effectively
dechlorinated for up to 10 years without replacement. The
purpose of putting it there was to take advantage of its de-
chlorination and filtration properties, rather than for organics
removal. What was the second point of your question?
Q McCARTY: Have you studied whether organics are formed in
that process?
A MOSER: No, we have not. I think the best way to get an
answer to your question is to refer back to the person who was
instrumental in obtaining that data. Vern Snoeyink is here.
Vern, do you want to comment on chlorination of organics in the
carbon column or chlorination of the carbon itself?
SNOEYINK: Yes. We did one experiment I think that relates
quite nicely to Professor McCarty's question. We chlorinated a
carbon bed that was loaded with humic substances and measured
the amount of chlorinated organics formed. The principal halo-
genated compound found in the effluent was chloroform; there
was no bromide present, so we were not getting any of the
brominated compounds. When we analyzed the data carefully,
there was evidence that the chlorine was reacting preferentially
with carbon rather than with the humic substances. So although
chloroform was produced from the saturated carbon, not as much
98
-------
was produced as would have been if that compound had not been
attached to carbon. Chlorinated compounds are certainly formed,
but carbon destroys a lot of the chlorine before it can react to
form halogenated organics.
Q CHARLES BUSCHER, Continental Water in St. Louis: I'd like
to ask Mr. Moser a question concerning the criteria employed to
decide whether or not to use carbon in specific areas. You have
indicated that your company uses granular activated carbon at
only 19 of its 100 water treatment facilities.
A MOSER: To put it briefly, customer complaints. We use
carbon where we have chronic taste and odor problems. It doesn't
always work out that the granular carbon reduces treatment costs,
so we generally try to do what we can by using powdered activated
carbon, potassium permanganate, or other simple alternatives. We
avoid the expense of granular carbon installations unless we have
a problem that's so severe or so chronic that GAC is the only way
to handle it.
Q MR. BUSCHER: In the 19 plants that you have operated over
various periods, have you monitored TOC or trihalomethane con-
centrations to check the effectiveness of your particular GAC.
Can you control your GAC operation in that way?
A MOSER: We haven't tried to do that. If we still had the
Nitro facility operating, we could go back and check. Nitro is
the only plant at which we had a multiple, regenerated carbon.
We had some data on COD and CCE from Nitro, but nothing conclu-
sive. We haven't attempted to monitor for trace organic com-
pounds ourselves. We have participated in the ORSANCO study and
Dick Miltner will later in this conference present some data from
two of our plants, Pittsburgh and Huntington, on GAC's ability to
remove many compounds. Perhaps that will answer part of your
question.
CHARLES BUSCHER: In review of the American operations of
GAC to date, have you found anybody who has used any of the
parameters prescribed in the proposed regulations to measure the
efficiency of the GAC?
A MOSER: I believe there's some data available from the
Passaic Valley installation where they have an infrared furnace.
They are regenerating carbon in that furnace and, as part of
that study, are measuring volatile organics and TOC to try to
determine its life with respect to the criteria in the proposed
regulation.
I don't have that data, but to my recollection it was a couple
of days before one of the volatile organics having a concen-
tration greater than a half of a part per billion passed through
99
-------
the carbon. That's the only installation I know of where data
may be available.
COTRUVO: I believe there are data available from other
sources. Patrick Cairo from Philadelphia probably has some and
he might be able to mention it when his time comes.
Q ART BUSCH, North Texas State University: Dr. Suffet, I'd
like to continue the discussion of monitoring breakthrough using
a surrogate or a correlation. I share Dr. Cotruvo's position
I think, but I'm also a little bit confused. EPA has stipulated
a TOC standard that could be used, or considered, as a general
characterization, but it's several orders of magnitude higher
than their limit on THM's, for example. As long as the standards
are written in terms of specific compounds, it seems to me that
we have to address those compounds. However, we certainly don't
have any list at the present time. The question is, do we have
any general techniques that would approach the level of the THM
standards?
A DR. SUFFET: If specific standards are developed by EPA for
particular organic compounds of health concern, such as THM,
there are methods available to measure these at the levels of
concern. Screening procedures can and are being developed for
this purpose, but the expense is the key question. It would be
ideal if a surrogate parameter could be developed that correlated
with the compounds that need to be removed. Right now, there is
insufficient information on how these organics are simultaneously
affected by a treatment process to develop these correlations.
I'd like to see them developed. I think everybody would. It's
much easier to use surrogate than specific standards.
Q BUSCH: Perhaps so, but isn't there a legal problem in
imposing a surrogate standard for a specific compound? It
appears that, legally, we are going to have to identify specific
compounds.
A COTRUVO: I think I can answer that. This discussion falls
into EPA's province. It is possible for both to exist simultane-
ously. There can be standards for individual chemicals and there
also can be standards for surrogates. For example, coliform
bacteria are a surrogate standard which are intended to indicate
overall biological quality. The same kind of thing could exist
for organics. There could be a standard for trihalomethanes and,
at the same time, there also could be standards for TOC, TOCL,
or whatever. Of course, both specific and collective standards
would have to be satisfied.
MOSER: The development of surrogate parameters would
require collecting an immense data base to convince people
that they provided satisfactory correlation with specific
100
-------
contaminants. More typically, those types of correlations are
developed for a particular place with a fairly consistent raw
water situation. There are convenient ways for determining such
a relationship between parameters, and using that as an operating
criterion.
COTRUVO: Although we have some flexibility under the
statute when setting standards, we have a fairly strong directive
for setting a numerical standard when there is clearly an adverse
health risk, as in the case of carcinogens, and when the measure-
ment is manageable. That may not turn out to be the optimum
approach in situations where a large number of chemicals might
be present.
SUFFET: I think the screening methodology is an approach
that requires consideration of water quality fluctuations and
their implications for sample collection. Sample collection is
sometimes neglected in devising systems for collecting data.
Grab samples don't show you the whole story. You have to see
things over the longer span of time.
COTRUVO: By the way, we are not only concerned with tri-
halomethanes; we consider the trihalomethanes as both a chemical
MCL and as a surrogate because they represent, at least qualita-
tively, the formation of a whole range of chemicals during the
chlorination process.
Q MR. PANAGIOTAKOS: I'm from the Lowell Water Treatment
Facility. We are just installing a GAC unit. We have been
using PAC for about 15 years. I have a question concerning
chlorine residuals. In our State, we are required to maintain
a free chlorine residual in the effluent. With GAC we are going
to be removing the chlorine, and we are going to have to post-
chlorinate in order to satisfy the State's standard. What is
EPA's stand on this matter? Will' it be possible in the future
to lower or eliminate the free chlorine standard?
A COTRUVO: There are no EPA chlorination or disinfection
regulatory requirements. That is a State matter. Our intent is
to minimize the unnecessary introduction of synthetic chemicals
into drinking water while assuring protection from pathogens.
Disinfectants and their by-products are synthetic chemicals.
Nevertheless, there are definite benefits from disinfection as
well as from the maintenance of a disinfectant residual in the
distribution system. The appropriate combination of treatment
processes has to be worked out in each specific system.
Q PANAGIOTAKOS: What I was really trying to find out is,
if current studies show that we may cause more harm by post-
chlorinating after GAC than if we didn't.
101
-------
A COTRUVO: The point is, that after the GAC has been applied,
the total organic loading of the water is lower than it -was in
the influent; therefore, the amount of by-products that are
formed from post-chlorination are generally going to be less than
they would have been in the absence of GAC.
MOSER: Given that granular carbon is going to remove the
chlorine anyway, and that the pre-chlorination only makes your
trihalomethane problem worse, it would seem to make sense to
rely upon post-chlorination for much of our disinfection.
Q RHODES TRUSSEL, Montgomery Engineers: Mr. Moser, I'm
concerned about the implication that GAC may be effective for
removing trihalomethanes for periods as long as 9 months.
I noticed that data from two points in time at the Hopewell
Facility were presented. In the first instance, the trihalo-
methane levels were rather low and there was about a 50 percent
removal. Then, 9 months later, 78 percent removal was observed
with levels between 400 and 500 ppb coming in. Other studies
have generally shown that carbon is usually exhausted much
earlier than either of these two periods. I wonder if it
wouldn't be reasonable to suppose that the trihalomethane
concentration in the influent to your carbon is varying a great
deal, that the carbon isn't just serving to smooth out the
trihalomethane levels in the effluent, and the actual mass of
trihalomethane discharged from the carbons over a long period of
time probably equals the mass coming in.
A MR. MOSER: I admit that you certainly could be right. Only
after we collect more data from a greater number of samples and
average them out will we determine whether you're right or not.
MR. TRUSSEL: I'd like to see much more frequent monitoring
of that plant. It would be very interesting.
MR. MOSER: We're going to do so.
Q MOHSIN SIDDIQUE, the Washington Suburban Sanitary Commis-
sion: Mr. Moser, in your work on the regeneration of carbon,
did you analyze the gaseous products of regeneration?
A MR. MOSER: No. We weren't concerned with that since it
had not yet been shown to be a problem.
102
-------
LUNCHEON PRESENTATION
(Beginning of Luncheon, April 30)
DR. STEPHEN J. GAGE
I am very pleased to be here today for this luncheon speech.
First, I'd like to extend my congratulations to Professor
Sontheimer and Dr. Cotruvo for this very successful CCMS pilot
project and also for this very successful technical conference.
I have been involved wth CCMS for more than 6 years in one way or
another. I think that I can whole-heartedly say that this is by
far the most successful technical conference yet hosted by the
CCMS. I hope that the success of this conference is some reason-
able indication of the interest, in this country and abroad, in
the very important subject of protecting the health of our people
from contaminants in the drinking water supply.
I would like to make a few remarks about CCMS. The organization
was formed in 1969 as the civilian arm of NATO, the North
Atlantic Treaty Organization. At that time, considerable thought
was given to strengthening the role of NATO, especially with the
success of the economic restoration of the European countries.
It was felt that it would be most appropriate to establish a
civilian analogue to the military and technological activities
of NATO. The author of this idea, it's reputed, was Patrick
Moynihan, who was working in the White House at the time. Since
then, Mr. Moynihan has gone on to become our Ambassador to the
United Nations and more recently, a Senator from the State of
New York. He started this program and was the first American
chairman, and it has been running successfully ever since. In
fact, this Fall, here in Washington, we'll celebrate the tenth
anniversary of the Committee on the Challenges to Modern Society.
The U.S. has volunteered to host this meeting and all of the
other CCMS countries have heartily concurred. We will be holding
the tenth anniversary plenary session here in the latter part of
October. I hope some of you who are working on the CCMS projects
will be able to join us.
I understand that eleven of the NATO countries and three coun-
tries outside of NATO are participating in this pilot project.
That representation is probably typical for the projects that
103
-------
have developed in recent years. We have had some projects with
as many as 20 or 25 countries taking part, and other projects
with as few as three or four countries. All of them have had
the same basic goal, to improve the living conditions of the
people of the NATO countries and of the people of other countries
around the world. CCMS has emphasized concerns in health and
safety, environment, and energy that have strongly influenced our
public policies during the past decade.
All of the other CCMS Projects are run in essentially the same
way as this one. Typically, two to three countries agree that
they will sponsor the pilot project and thus become the project
leaders. In this case, the project leaders are the United
States, the Federal Republic of Germany, and the United Kingdom.
I think all of the NATO countries have sponsored a project at
one time or another. While the U.S. has played a major role in
supporting many of the projects, we have had excellent sponsor-
ship from throughout the North American and Western European
continents.
We have had some excellent projects, which include work on air
pollution surveys and control strategies, hazardous waste dis-
posal systems, and automotive safety. I might just mention that
I am packing my bags to attend the next CCMS plenary session in
Brussels, which opens Thursday morning. We are going to propose
three new pilot projects. One is on estuarine management, for
which the Greek government will be proposed as the project
leader. The United States will contribute with our massive study
of the Chesapeake Bay estuary system. That gives you some idea
of how closely such international projects are tied to our
interests. The other two projects range far afield in other
directions. One new project is on plastics recycling, which will
look at ways to reduce the amount of plastic waste products
burdening the environment and to reclaim more of those valuable
materials that can then be used as feed stocks for other plastic
processing operations. The third new project is related to the
protection of monuments from air pollution. Right now, inter-
estingly enough, the Greeks and the French are competing for
sponsorship of that project, both having expressed a great deal
of enthusiasm for it. Given the effect on the cathedrals and
other public buildings in France, or on the historic buildings
such as the Acropolis and Parthenon in Athens, this is a very
important new project.
The United States receives a great deal of benefit from these
CCMS projects. First, it always leads to an improved under-
standing among the various Western nations involved. In some
instances, other nations from around the world have been in-
cluded. Australia, New Zealand, Japan, and some of the African
countries have participated in some of these projects. So
in a sense, CCMS provides a forum for improved understanding
104
-------
throughout the world. It encourages the sharing of technical
information, as exemplified by this meeting/ and just as impor-
tantly, the sharing of information concerning institutional and
social development for improving health, safety, and the environ-
mental quality of the various countries.
Institutional and social concerns are probably the most sensitive
areas in the design of a new pilot study. The argument always
presented is that the work will interfere with the political
aspects of the individual countries or that working cooperatively
would be difficult. Frankly, the lessons we have learned in
these many projects—over 30 to date—cover the institutions and
the social structures of the other countries. In that way, we
have benefited greatly in the approach to using the technologies
that we have identified through the projects.
I would like to pass on to you a couple of anecdotes that I have
found quite interesting. One of the earliest projects started,
I believe in 1970, was on automotive safety. This was a project
in which several automobile companies were presented with the
challenge to design new safety features, such as side rails,
collapsible bumpers, and other devices that are now generally
accepted and are used in most automotive design. But as you will
recall, back in 1970, there was still a great deal of hostility
and skepticism toward those ideas in some quarters, stemming from
the belief that these innovations would make the cars too heavy
and/or that the benefits from such improved safety design would
not merit the additional cost and effort.
The United States wanted to be the major pilot on this project,
but the people who were organizing the project could not convince
any of the Detroit auto manufacturers to participate. The
Department of Transportation had to issue a competitive contract
and actually hire three companies with no previous automobile
manufacturing experience, to make modifications in automobiles
and to test the various types of innovative safety designs. The
automobile industries of Germany, Sweden, and Japan, on the other
hand, all took the project very seriously and divided the respon-
sibilities among themselves according to the size of the car.
One took a small car, another took an intermediate size car, and
the third took the larger car. They began designing and con-
structing automobiles and testing them, and committed a substan-
tial amount of resources to carry out destructive testing of a
full-scale automobile. About 1974, when the success of the
pilot project became obvious here in the United States, Detroit
suddenly became interested. I'm speaking of General Motors,
Ford, and Chrysler. They started their own independent work,
or at least made public some of the work that they were doing.
If I had a lesson to bring from that particular example, it
would be that the institutional changes that we helped bring
about in this country, as a result of our involvement in this
105
-------
international technology transfer project, were more important
than the technologies themselves. Although we did have safer
Volvos and Datsuns driving around on American streets, hopefully
we also have safer American automobiles.
I will mention another project, or a series of projects, in which
I first became involved with CCMS. After Mr. Moynihan left the
White House to become Ambassador to the United Nations, the re-
sponsibility for the CCMS chairmanship went to Mr. Russell Train,
who was then the Chairman on Environmental Quality and subse-
quently administrator of the Environmental Protection Agency.
I was working for Mr. Train in 1973, at the time of the Arab oil
embargo. Mr. Train asked me to put together some ideas for some
projects on energy that would not conflict with the interests
of other U.S. agencies, and that in fact, would help increase
their interest and attract interest from other countries. Three
ideas resulted. One was a project on solar energy, another on
geothermal energy, and another on energy conservation.
The first attempt to determine interest on the part of CCMS
countries resulted in a mixed response. In fact, we were dis-
couraged at that time. Basically, the countries said, well,
we're either too far north in Europe to use solar energy or
we're too far south in Europe, or we're too poor to be able to
invest enough in additional research to make a significant
contribution to solar energy development. We heard different
reservations for the geothermal and energy conservation projects.
As it turned out though, the world's perception of their use of
energy, or at least those of the western industrial nations,
changed so markedly during the intervening 3 years that when
these three pilot projects were reported in Brussels this past
fall, it was apparent that CCMS had actually fostered an over-
whelming success in international cooperation. Over 20 countries
took part in the solar energy project. Over 15 countries became
involved in the geothermal project, which entailed drilling
geothermal wells in Turkey, Greece, Portugal, as well as in the
United States. We were soon able to recruit New Zealand, Mexico,
and other countries for cooperation in the CCMS projects.
Probably the best story of all was that connected with energy
conservation .which was renamed, asRational Use of Energy. From
that project, we discovered that throughout the world some very
sophisticated research was being done on how best to use avail-
able energy for space conditioning, heating, or air conditioning.
That project, up until then, probably contributed the most to
the United States' National Energy Conservation Program, by
providing a good substantive technological base.
Recently, I was associated with the Flue Gas Desulfurization
Project. This was interesting for me in that it goes back to
106
-------
my first experience with international technology transfer. I
will conclude with this anecdote.
In the Spring and Summer of 1972, it became obvious that the
U. S. Environmental Protection Agency's support for using scrub-
bers, or flue gas desulfurization units, had aroused a great deal
of adverse reaction from the electric utility industry, which was
to be forced to use scrubbers in many of their installations.
The scrubber that had been used as the basis for the administra-
tive decision had been dismantled and moved from the site of the
electric power plant. Therefore, it was not feasible to return
to that same scrubber and operate it for longer periods of time
and under different operating conditions to show the utility
clearly that scrubbers could work. Consequently, a group of us
were asked to investigate the status of flue gas desulfurization
units in Japan. We spent 3 weeks there in the summer of 1972
and prepared an interagency report on our findings. The effec-
tiveness of the flue gas scrubbers in Japan supported the EPA
Administrator's decision and made it tenable. We found that
there were a number of scrubbers operating at full-scale there,
including one scrubber operating on medium sulphur, high-ash coal
at over 90 percent reliability without any serious chemical prob-
lems. We brought that experience back with all the documentation
that the Japanese company provided us, and that particular scrub-
ber became the most important scrubber in the world for the next
several years.
These anecdotes are an indication that we are willing in this
country to capitalize on the experience of the other advanced
technological nations in making economic decisions. To put it
another way, where we can find demonstrated technology in other
countries that can be transferred to this country, we will incor-
porate it into our environmental decisions. The reason why I
brought the scrubber case up now is that we are going back. In
fact, I started about 2 years ago to recap the Japanese experi-
ence. We also found operating scrubbers in Norway and in West
Germany to add to that bank of experience. At this time, the
Japanese now have over 200 operating scrubbers, nearly all of
which have had excellent success. They have increased the number
of coal-fired power plants using scrubbers from one to five, and
each of those installations are effective. In fact, the one
plant that was mentioned earlier has had a record of close to
100 percent reliability for the last 5 years. That's a tribute
to Japanese engineering, but also, it should be recognized that
when good technologies can be developed and dedicated, capable
people are there to employ them, some of the technological
problems that seem to pester us all the time can be mastered.
I will conclude by pointing out that we take the activities of
these CCMS pilot projects very seriously. The Congress of the
United States requires us to prepare a 5-year plan and submit
107
-------
annual updates of that plan to them. In our 5-year plan, our
research outlook for 1979 which was submitted to the Congress in
the early part of February, one paragraph from the chapter on
drinking water states that: "... water treatment practices in
other countries are similar to those used in the United States
with the important exception that greater use is made of granular
activated carbon and disinfectants other than chlorine. New
developments in the drinking water area are transmitted through
contacts among individual scientists, the NATO Committee on the
Challenges to Modern Society, the World Health Organization, and
other professional organizations." This shows that we have built
the CCMS project on Adsorption Techniques in Drinking Water
Treatment into our 5-year plan. We hope that this technical
conference, and the many exchanges that I know occur among our
scientists and the scientists from the various countries, will
enable this project to live up to our expectations as being one
of the most successful that we have yet sponsored under the Com-
mittee on the Challenges to Modern Society.
108
-------
EUROPEAN EXPERIENCE IN THE USE OF
ACTIVATED CARBON TREATMENT
109
-------
NATO CMM
•OTAN CCMS
NATO-CCMS
no
-------
AN EVOLUTIONARY APPROACH TO
ACTIVATED CARBON TREATMENT IN FRANCE
Pierre Schulhof
In France as in other industrialized countries, the treat-
ment of river waters to produce drinking water requires the
elimination of most of the organic substances.
The various types of organics present in raw water, as
categorized by engineers specializing in water treatment rather
than by chemists, have long been classified as
1. Organic materials with unpleasant taste
2. Organic materials without taste, but which acquire a
taste through treatment, notably chlorination
3. Organic materials without taste and unaffected by
treatment
4. Organic materials that interfere with treatment,
for example by causing foaming.
For years, because the pretreated raw waters were still
of adequate quality, this classification was the sole consider-
ation in water treatment. It is principally centered on taste
problems and more particularly on taste related to chlorine
treatment. The taste of chlorine has long been unacceptable
in France.
This aversion to chlorine taste has resulted in the use of
ozonation since the beginning of the century and of activated
carbon for the last twenty years.
Recently however, a new classification of organics has
been superimposed on the first one because of the decline in raw
water quality. This second list is not only linked to the
organoleptic quality of the treated water, but it also takes
into consideration possible long-term effects of some chemical
compounds. The second classification includes:
1. Organic materials of natural origin that are not
transformed by treatment (especially chlorination)
111
-------
2. Organic materials of natural origin that are trans-
formed by chlorination
3. Organic materials that are of synthetic origin.
The elimination of materials in the second category
requires treatment without chlorination; the elimination of
materials in the third category can be achieved by different
suitable means.
This being the case, a new treatment concept has evolved
in which the role of activated carbon remains very important but
is associated with a function quite different from that-for
which it originally had been provided. Today biochemistry
governs water treatment technology. Activated carbon no longer
acts only as an efficient adsorptive agent but also as a favored
site for biological deqradation.
Beginning in 1970 a few plants were constructed according
to this new concept and have been in operation since 1975.
Construction of other plants is in progress and some are ex-
pected to begin operation at the end of 1979. In several places
pilot plants have been operating for several years.
The role of activated carbon should not be considered
independently, however. Its combination with other treatment
processes, now available, should be emphasized.
EXPERIENCE WITH POWDERED ACTIVATED CARBON
In the past, activated carbon was used essentially to
eliminate tastes of natural origin from raw waters. These
tastes are very often associated with the presence of algae in
rivers. Indeed, in most cases they are not strong but are
exacerbated by the action of synthetic organics^ especially
those produced through chlorination.
Progressively increasing quantities of organic substances
and ammonia are contained in the rivers. The need for chlorine
addition to eliminate ammonia made the taste problem difficult
to solve, but fortunately activated carbon is very effective in
this respect.
In some cases when the water to be treated contains a par-
ticular synthetic organic that has a bad taste or interferes
with the treatment process, its elimination is obtained by
activated carbon. At first powdered activated carbon was used,
it being readily available in France at low cost and easy to
implement. (Table 1 lists principal plants using powdered
activated carbon.)
112
-------
Table 1
Principal Drinking Water Plants in France Using
Powdered Activated Carbon (Output >50 x 103 m3/day)
Plant
Output
m/day
Raw
Water
Origin
District
Supplied
Treatment
Date of
PAC
Going
Into
Service
Average
Treatment
Rate
tng/1
PAC
Function
Storage
and
Supply
Carbon
Origin
Choisy-le-Roi 800,000 Seine River
Ncuilly-sur- 600,000
Ma me
Mame River
Paris Chlorination; aluminum poly- 1964 0 to 30 tastes 2 silos (90 T coconut
southern chloride coagulation; pew- each); weigh-
suburb dered activated carljon ing systems
treatment; flocculation-
sedimentation; rapid sand
filtration, part on sand, if
part on activated carbon;
ozonation; addition of Cl~
Paris Chlorination; aluminum poly- 1968 0 to 30 tastes 1 silo (70 T) coconut
eastern chloride coagulation; pow- (plus con-
suburb dered activated carbon tainere);
treatment; flocculation- screw conveyor;
sedimentation; rapid sand proportioning
filtration; ozonation;
addition of C10
Orly
300,000 Seine River
Paris
southern
suburb
Chlorination; Al (SO.K plus
activated silica coagulation;
powdered activated carbon
treatment; flocculation-
sedimontation; rapid sand
filtration; ozonation;
addition of CIO,.
1970
0 to 20
tastes
3 silos; 2 gravi-
metric batchers
coconut
and peat
Mery-sur-Oi'f 200,000 Oise River
Paris Present treating line:
northern Chlorination; aluminum
suburb polychloride coagulation;
powdered activated carbon
treatment; flocculation-
sedimentation; rapid sand
filtration, part on sand,
part on activated carbon;
ozonation; addition ot CIO.,
1964
0 to 30
tastes;
organic
matter
Containers;
proport ion ing;
screw conveyors
-------
Table 1
Principal Drinking Water Plants in France Using
Powdered Activated Carbon (Output >50 x
(continued)
103 m3/day)
Croissy
120,000 Seine River
Aubergenville 120,000
Morsany
(first
sect ion)
Suresnes-
pont-
Valerien
75,000 Seine River
detergent
removal
(avoiding
foams in
sand pits)
containers ;
proportioning
coconut
and peat
Tastes (in Containers; coconut
connection proportioning and peat
with algae
and bacteria)
50,000
Seine River
Outer Coagulation; flocculation 1958 10
western (aluminum sulfate plus
suburbs activated silica); powdered
activated carbon; sedimentation
with powdered activated carbon;
aerated sand filtration; ground
level replenishing through
sand pits
Outer Coagulation; flocculation 1961 Varies
western (aluminum sulfate plus widely
suburbs activated silica); powdered
activated carbon; sedimenta-
tion with powdered activated
carbon; nitrification (bacteria
beds); sand filtration; ozona-
tion; addition o£ Cl^
Outer Prechlorination; coagulation 1969 15 tastes; 1 concrete peat and
southern (alumina sulfate plus activated organics silo (60 m ) anthracite
suburbs silica); powdered activated removal
carbon; flocculation-sedimen-
taticn with powdered activated
carbon; sand filtration; ozona-
tion; addition of Cl
Paris Chlorination; powdered acti- 1959 40 tastes containers; peat and
north- vated carbon; coagulation; proportioning anthracite
western flocculation-sedimentation;
suburbs rapid sand filtration; addition
of CIO,
-------
The activated carbon treatment technique has been steadily
refined. Activated carbon input locations were studied with
respect to the application points of other reagents. It was
found that the carbon should be introduced last, as far as
possible from the flocculating agent and chlorine or chlorine
dioxide. It was also found that adding the carbon in lots at
two different locations improved the economy of treatment.
Activated carbon in powdered form remains widely used in
France. It is very effective in eliminating taste from treated
waters containing low concentrations,of organic substances and
ammonia. It is also particularly well adapted to meet organic
matter peaks of short duration, without heavy capital costs. In
such cases the treatment consists of injecting large doses of
carbon, up to 100 mg/1, for a short period.
Thus while the effectiveness of powdered activated carbon
in eliminating organic substances economically is somewhat
limited, substantial removal can be achieved. Combined with a
well-designed and well-operated flocculation-sedimentation
process, a dose of 20 to 30 mg/1 can reduce the total of dis-
solved organic matter by about 60 percent. However, the con-
stant increase of chlorination doses accompanying the rise of
dissolved ammonia creates taste problems that powdered activated
carbon cannot eliminate. This is why, at the end of the 1960's,
French engineers responsible for water treatment looked to
granular activated carbon, not to replace the use of powdered
carbon but to extend its range when the yield was low. Ini-
tially this modification took place in the simple filtration
stage through the exchange of granular carbon for sand in the
filter beds.
GRANULAR ACTIVATED CARBON AS A FILTER BED
The concern for taste problems, especially chlorine taste,
contributed to the initial development of granular activated
carbon (GAC) filtration. However, it is not possible to stop
the reaction of organic substances with chlorine used at the
breakpoint rate to eliminate ammonia. This is evidenced by
current required chlorine doses of more than ten times the
amount of dissolved ammonia that is present, although a factor
of seven is sufficient to control the ammonia.
In addition, the competition between ammonia and organic
substances for chlorine consumption is sometimes difficult to
predict because the kinetics are very complex. This may lead
to an underestimation of the chlorine dose, resulting in a
partial oxidation of ammonia stopped at the chloramine level.
Chlorinated organic substances and chloramines give water
the improperly named "chlorine" taste to which consumers
object.
115
-------
Also, odor thresholds as high as 20 may be measured in raw
waters chlorinated to breakpoint, while they were measured at
4 before treatment.
Chlorine taste is difficult to remove with powdered
activated carbon. Organic materials should be eliminated before
they combine with chlorine. Very high doses of powdered acti-
vated carbon are necessary to obtain reliable taste removal.
The mechanism of granular activated carbon placed in a
filter bed is quite varied. First, it acts as a chlorine
removal agent and decomposes chloramine to regenerate ammonia,
which is eliminated fairly well by nitrifying bacteria contained
in the filter, yielding an ammonia-free water at the output.
Second, it adsorbs taste- and odor-causing organochlorinated
compounds which because of their slow kinetic reaction, remain
at low levels during the filtration process. Consequently they
can be effectively adsorbed by carbon for a long period.
Under such conditions it is not unusual for carbon filtra-
tion effectiveness on the taste of water to be maintained over
one or two years, whereas effectiveness in removing total dis-
solved organics lasts only a few months. This is illustrated
In Figures 1 and 2, which show the results of two years of
testing at the Choisy-le-Roi plant. Tests at this plant also
indicate that chlorine added to GAC-treated waters is consumed
quickly (Figure 3), indicating a new generation of organo-
chlorinated substances.
With some precautions, good results can be obtained in the
removal of chlorine taste in treated water through simple
filtration on activated carbon. A number of plants in France
have applied or are still applying the technique of breakpoint
chlorination followed by flocculation-sedimentation-activated
carbon filtration (Table 2). Disinfection is usually by ozona-
tion, and distribution network protection requires addition of
chlorine or chlorine dioxide. Carbon is regenerated every year
or every two or three years, depending on the case. Some carbon
beds have been regenerated up to five times without any diffi-
culty. Regeneration criteria are most often based on the taste
threshold test.
Filters are usually of the open type, made of concrete in
rectangular shapes with a surface area of up to 120 m . Carbon
layer depth and empty bed contact time vary from 0.70 to 1.0 m
and from 6 to 20 min, respectively. In most cases washing is
carried out with air and water.
An extended range of carbon grades is used, most of which
give satisfactory results with respect to the taste of treated
116
-------
70.
60.
C "
O 50.
-P
U 40.
"8 30]
10
OIOISY.lE.ROt plant
PtCACTlF ocivm«d ccrbon
2000 6000 WOOD 15000 20000
4OOO IOOO
Filtered Water Volume (m ) Per m
of Carbon
Figure 1. Reduction in Total Organic Carbon by
Single Filtration on Activated Carbon
Beyond That Achieved by Sand Filtration
to
M
-P
en
(0
4-1
1 2345678910
Duration of Test - Months
Figure 2. Taste Removal by Single Filtration
on Activated Carbon
117
-------
Cn
a
§1,6
-P
(0
4J
C
0)
O
C
O
u
0)
C
•H
activated carbon aged year
.^0,05
U
24 48 72
Duration of Test - Hours
Figure 3. Chlorine Consumption in Filtered Water After
Single Filtration on Activated Carbon
118
-------
Table 2
Principal Drinking Water Plants in France Using
Granular Activated Carbon for First Filtration (Output >50 x 103 m3/day)
Mery-sur-Oise 100,000 Oise Kiver
Viry-Chatillon 100,000 Seine River
upstream
of Paris
Paris Chlorination; aluminum poly- 1970 tastes;
northern chloride coagulation; floe- organic
suburbs culation-sedimentation; GAC matter
filtration; ozonation; chlori-
nation
Paris Prechlorination; aluminum 1973 tastes;
southern polychloride coagulation; organic
suburbs flocculation-sedimentation; matter
GAC filtration; C1O2 dis-
infection
108 15
12
15
9 at 40 10
6 at 32
30
1 filter
Picactif
i filters
Picafil
'i filters
Pic.iflo G
1 filter
F300
F 300 K;
Row 0.8
supra
vo
Saint-Charles 100,000 Moselle
Kiver
Vigneux-sur- 50,000 Seine River
Seine upstream
of Paris
Nancy
Outer
western
suburbs
Prechlorination; coagulation; 1974
5A plus alginate flocculation-
sedimentation; GAC filtration;
CIO disinfection
Prechlorination; aluminum; 1970
sulfate plus activated silica
coagulation; flocculation-
sedimentation; GAC filtration;
CIO., disinfection
tastes;
organic
matter
tastes ;
organic
matter
22
10
48 15
23 6
36
NCY 103
at F300
F300; F500;
Row 0.8
supra;
BS12
ilcuille 50,000 Houille
River
Choisy-le-Roi 30,000 Seine River
Annet-sur- 25,000 Mame River
Marne
Dunkerque Prechlorination; coagulation; 1973 organic
flocculation (Fe-CIO, plus matter
starch), sedimentation,in part;
flotation, in part; GAC fil-
tration; replenishment basins
Paris Chlorination; aluminum poly- 1978 tastes
southern chloride coagulation; floe-
suburbs culation-sedimentation; GAC
rapid filtration; ozonation;
chlorination.
Paris Aluminum polychloride coagu- 1971 tastes;
eastern lation; flocculation-sedi- organic
suburbs mentation; activated carbon matter
filtration; ozonation; chlori-
nation.
50 10
117 15
45 13
Row 0.8
supra
Not yet 1 filter
regen- Picaflo B
erated 1 filter
F300
36
1 filter
Picactif
1 filter
Picaflo
-------
water. The selection depends on performance and cost consider-
ations. In given circumstances, carbon grades may be mixed.
Reactivation is carried out by the manufacturers.
In addition to removing taste, activated carbon filtration
can be used when the water to be treated usually contains low
concentrations of synthetic organics, but is subject to occa-
sional contamination by industrial wastewaters that create
peaks of synthetic organics for short periods. Even when
already saturated with total organic carbon, activated carbon
adsorbs these peaks and releases less harmful organics. In
such cases, powdered activated carbon is also effective at high
doses, but a monitoring device is required to detect the need
to adjust the treatment,whereas a carbon filter is continuously
available. For this reason, granular activated carbon filtration
has sometimes been preferred over powdered carbon injection.
The experience acquired in France with simple filtration
has permitted carbon manufacturers and water treatment engineers
to master the technologies of carbon design (type and grain
size) applicable to water treatment, selection of bed depth and
contact time, filter backwash, and carbon regeneration tech-
niques. It has been confirmed that carbon is an excellent
filtering medium and that disinfection of filtered water is not
a problem in spite of the large numbers of microorganisms
contained in the filtering beds.
French experience also has shown that dissolved ammonia in
raw water can be partially removed through biological action.
The elimination of ammonia in this way, with a corresponding
reduction in chlorination of raw waters, constitutes a definite
move toward the elimination of chlorinated organics simply by
avoiding their generation in the course of treatment.
Finally, with respect to total organic carbon, complete
elimination tests have shown that a satisfactory result could be
obtained at the cost of reiterated regeneration. From Figure 1
and Table 2, it can be estimated that if the average filtration
velocity is 6 m/hr, regenerations at six-month intervals provide
a 20 percent reduction in organic carbon content, which combined
with flocculation-sedimentation-sand filtration gives a total
reduction of 60 percent of the raw water content. At present
regeneration costs, such a process stream is expensive.
NEW TECHNIQUES
The presence of substantial amounts of synthetic organics
in river water is a rather recent concern. This analysis of the
problem will essentially be centered on the Seine River flowing
through Paris, which supplies, directly or indirectly from
12U
-------
related underground sources, the city's five million inhabitants,
one tenth of the population of France (Figure 4).
Within ten years, micropollution of the river has increased
considerably even in the area upstream of Paris. In terms of
total organic carbon, the Seine River contains an average of
6 to 7 mg/1 of dissolved organic carbon, including at least
3 to 3.5 mg/1 of synthetic carbon. However, the proportion of
volatile organochlorinated material, particularly THWs, is very
small (< 10 yg/1).
As a result of increased amounts of ammonia and organic
matter, chlorine demand (breakpoint rate) has increased sub-
stantially, as has coagulant demand (jar test rate) which is
influenced by organic substances.
Despite the rise of total organic carbon content or chlorine
doses, trihalomethane levels have never been very high in the
treated waters, usually about 20 yg/1 and in exceptional cases
50 yg/1. This is perhaps due to the brief contact time between
chlorine and haloform precursors in the treatment plants.
Nevertheless, it is desirable to limit organochlorinated
substances as well as taste- and odor-causing substances in
supplied water. This implies severe restriction of the use of
chlorine and elimination of organic matter to the extent
possible.
The new techniques combine step-ozonation with sand and
activate carbon double filtration. The second filtration on
carbon is always preceded by ozonation, which efficiently pro-
motes the elimination of organic substances, as demonstrated
through tests begun in 1970 and subsequent research. This does
not mean that the use of powdered or granular activated carbon
without ozonation will be discontinued; they will still be used
where applicable.
To illustrate these new technologies while remaining in the
operational field of the Seine River and Paris district, data
obtained from a pilot installation in the 800 x 10 m /day
Choisy-le-Roi plant upstream of Paris will be presented in
detail, followed by reports on full-scale developments that
preceded or accompanied the pilot tests. These are the Morsang
plant (150 x 10 m /day), which treats water from 40 km upstream
of Paris; Rouen-la-Chapelle (50 x 10 m /day), which treats
groundwater affected by the Seine River downstream of Paris; and
Mery-sur-Oise (300 x 10 m /day), which treats water from the
Oise River, a tributary of the Seine.
121
-------
.ROUEN LA CHAPELLE
StOUEN
L'AUMONE- A MERY-SUR-OISE
CHOISY LE ROI
VIGNEUX-SUR-SEINE
PARIS AREA
LOCATION OF WATER
TREATMENT PLANTS
ULLY-SUR-MARNE
'SAINT MAUR
VIRY-CHATILLON<
Figure 4. Location of Water Treatment Plants
in the Paris Area
122
-------
THE CHOISY-LE-ROI PLANT
The former treatment train at Choisy-le-Roi consisted of
breakpoint chlorination (to an average of 6 mg/1); flocculation-
settling; sand filtration at a maximum velocity of 6 m/hr and
ozonation with 10-min contact time; and an ultimate chlorination
step with a dose of 0.8 mg/1 resulting in 0.4 mg/1 of chlorine
residual at the plant effluent.
The distribution network is long, and residence time is
currently 72 hr. In summer, water temperature is high, sometimes
exceeding 25°C.
The plant is equipped with powdered activated carbon devices
whose performance has already been described. As a test, sand
has been replaced by granular activated carbon in some of the
plant's 54 filters of 117 m surface area each. The supple-
mentary reduction in total organic carbon produced by the use
of activated carbon is plotted in Figure 1.
The results seeming insufficient, a pilot plant was in-
stalled which has been in operation for two years, permitting
comparison of alternate treatments. Two essential modifications
were tested in this pilot plant (Figure 5):
1. The suppression of prechlorination and its replacement
by step-ozonation of rather short duration (about
2 min)
2. A second filtration with activated carbon after initial
ozonation. Various carbon-water contact times were
tested.
It was found that using the same ozone dosage as in the
original installation (and thus with no extra cost) but dis-
tributing it differently—i.e., partially at the head of the
train and partially after sand filtration—resulted in the
total elimination of ammonia without the use of chlorine.
In addition, the combined effect of the first step-
ozonation and a properly selected coagulant dose has made it
possible to reach a 50 percent reduction in raw water total
organic carbon at the sand filter bed effluent.
Data from the second filtration on activated carbon, after
step-ozonation, are more difficult to analyze. As observed,
when carbon was used as a simple filter bed, the reduction across
the filter bed (between input and output) decreased to a_final
plateau after a throughput of 15-20 x 10 m water per m carbon
during which organics were effectively eliminated. Contrary
to the case of simple filtration, where the decrease in the
123
-------
M
O
CL
Chemical treatment Flocculalion Sand
WAC PAC sedimentation filtration
ClONa CIO
•»•»•» 2<*
Ozonatbn
Tank
§
1
Model
network
(3days)
Chemical
First treatment Floccuhtion Sand Carbon Second
ozonation WAC sedimentation filtration filtration ozonation
Tank
CIO,
Model
network
(3days)
(Step ozonation and BAC)
Figure 5. Choisy-le-Roi (800,000 m /day) General Flow
Scheme of Pilot Plant
-------
reduction nears zero, the final plateau shows a steady state
reduction of about 30 percent of the total organic carbon at
the influent (Figure 6). This plateau remains unchanged for
one or two years without the need for regeneration.
The reduction of organics through the complete treatment
process (step-ozonation, flocculation-sedimentation, double sand
and carbon filtration) has been between 65 and 85 percent during
tests carried on for more than a year. At the same contact time,
the second filtration on activated carbon without regeneration
resulted in a reduction of total organic carbon at least equiv-
alent to the reduction obtained with simple filtration with
activated carbon regenerated every four months.
Three empty bed contact times were tested with carbon:
10 min, 30 min, and 1 hr. No appreciable differences were
observed in the organics level of the reduction plateau. A
threshold was observed, however, in respect to filtration veloc-
ity. At 17 m/hr the reduction of detergents was limited to
50 percent, whereas the reduction was 100 percent at 12 m/hr.
Another interesting parameter is the chlorine demand of
filtered water. The dose of chlorine injected for network
protection depends strictly on the chlorine demand of the
treated water. Therefore, using the pilot network of treated
water, free chlorine is measured as a function of time, and the
chlorine demand of the treated water is given by the difference
between chlorine dosage and chlorine residual. Because this
demand varies with the temperature and the initial dose of
chlorine, these parameters must be well defined. (These tests
were conducted with one-year-old carbon; i.e., when the reduc-
tion plateau had been reached.)
The second filtration on activated carbon is very important
in stabilizing chlorine in the treated water. It has already
been indicated that organics reduction was substantial, though
limited. Significantly, the major portion of the organic _sub-
stances eliminated are those that react with chlorine to form
organochlorinated compounds. It is worthwhile to note that
preozonation, prior to carbon filtration, has considerable
influence on the rate of disappearance of chlorine (Figure 7),
while the influence of preozonation on total organic carbon
reduction is not so marked. Thus, ozonation acts preferentially
on organics that consume chlorine, which is the desired objec-
tive. Under such conditions the quantities of organochlorinated
compounds in the treated waters are extremely small. The
decrease in the dissolved oxygen shows the biological nature
of this phenomenon.
Many other studies are required to explain the basic
mechanisms operating during the second filtration on activated
125
-------
cT
o
«'-.
(1) 30.
<*° 10,
CHOiSY LE ROI pbnt
10000
20000
30000 40000 50000 60000
3 3
Volume (m ) of Filtered Water Per m Carbon
Figure 6. Reduction in TOC by Second Filtration on Activated
Carbon Versus First Filtration on Sand
U
Q2J
3
•d
•H
(0
-------
carbon after ozonation. With further research, optimum ozonation
contact times in each particular case will be better defined.
Above all, an activated carbon should be specially developed
for water treatment, to foster the effects of biological regen-
eration and take full advantage of this new technique. It is
unlikely that activated carbon standards, determined for other
applications, are suitable for specific water treatment operating
conditions. French manufacturers are marketing a wide range of
carbons and have agreed to cooperate in the development of spe-
cific carbons oriented to water treatment with a good quality-
cost ratio.
Choisy-le-Roi plant modifications. From the pilot tests,
what improvements are suggested for the Choisy-le-Roi plant?
At a rather reasonable cost, half of the 54 filters can
readily be filled with carbon and operated in series following
the other half of the filters, which would remain filled with
sand. With preliminary step-ozonation to reduce the chlorination
of raw water, the modified treatment would consist of sand and
carbon double filtration and a second step-ozonation.
The-Choisy-le-Roi plant was initially designed to process
800 x 10 m of water per day with a filtering velocity of 6 m/hr
through 54 filters. With the same filtering velocity, the new
output would be reduced to 400 x 10 m /day.
However, the process of double filtration employs an in-
creased velocity of 9 m/hr through the 54 filters arranged in
two filter series. The same turbidity level would be maintained
since two sequential filtrations at 9 m/hr give results equiv-
alent to a single filtration~at_6 m/hr. Thus, the expected
production would be 600 x 10 m /day. Output over this
maximum value would require that the two batteries of filters be
operated parallel and arranged in a single filtration mode.
3 -3
Modified to produce 600 x 10 m of water per day by a
double filtration process while retaining the possibility of
producing 800 x 10 m /day by simple filtration, the installation
at Choisy-le-Roi would remain flexible to handle peak demands.
Figure 8 shows organics reductions resulting from single
filtration and double sand-carbon filtration after ozonation as
a function of estimated additional cost, including capital and
operating costs and regeneration expenses. When activated carbon
is not used, organics reduction from initial concentrations in
the raw water amounts to 50 percent. Obviously the double fil-
tration after ozonation appears to be the most promising tech-
nique. It was on this basis that the decision was made to modify
the Choisy-le-Roi plant (Figure 9).
127
-------
5 0)
W to
o o
o
u
C
•H
0)
a
o
10 -
c 1
(0
(0
4J
(0
u
18 month regeneration period
50
60
70
80
Percent Reduction in Organics as Compared with Raw
Water
Figure 8. Organics Reduction Versus Capital and
Operating Cost
128
-------
800 000 m3 /day
PRESENT PLANT
treated
water
Sand
Flocculation Sedimentation filtration ozonation
Post
chlorination
600.000/800,000 m3/day
FUTURE PLANT
(Step ozonation and second stage
active carbon filtration)
ActfvatedLiJ
carbon Second
Treated
water
Sand carbon Second Post
Flocculation Sedimentation filtration filtration ozonation chlorination
Figure 9. Original and Modified Treatment
Processes at Choisy-le-Roi
129
-------
The regeneration criterion will be based on the chlorine
demand of treated water; a carbon regeneration period of eighteen
months is expected. Organics reduction is anticipated to be
70 percent. Including capital, operating and maintenance, and
regeneration costs, the added charge is estimated to be only
1.4 cents per cubic meter of water, corresponding to about one
more dollar per year per consumer. For this small extra cost, a
40 percent reduction of synthetic organics in the water supplied
will be gained, the improvement being mostly related to organo-
chlorinated substances. Higher reduction percentages are pos-
sible, but the cost of additional gains increases sharply.
THREE NEW TREATMENT PLANTS
The primary reason for deciding to modify the largest
water treatment plant in France was the confirmation of test
results from the pilot plant at Choisy-le-Roi. by experience
with otffier plants using step-ozonation followed by double
filtration which had been operatincr at full scale for four years_.
Three of these plants are descnoed below. Plant characteris-
tics and conditions of step-ozonation are summarized in Tables 3
and 4.
The Morsang plant. The 150 x 10 m /day capacity of the
Morsang plant is obtained from three parallel trains. The first
line uses powdered activated carbon; the second line, in oper-
ation since 1975, uses a second filtration on granular activated
carbon. Half of the second line constitutes a third train in
which an ozonation step is inserted before the carbon filtration
so that the full treatment line comprises flocculation, sedimen-
tation, sand filtration, ozonation, activated carbon filtration,
and chlorination. This line yields5 a 70 percent reduction in
organics, as measured by TOC.
The Rouen-la-Chapelle plant. "Following a pilot testing
program begun five years ago,the Rouen-la-Chapelle plant
(Figure 10) was put into service at the end of 1975. The treat-
ment sequence consists of a first ozonation with a short (3 min)
contact time, a sand filtration, an activated carbon filtration
with a contact time of 30 min, and finally a second ozonation,
also with a contact time of 10 min. Before pumping the treated
water to the distribution system, a 0.4 mg/1 chlorine dose is
added.
The first ozonation plays three roles:
1. It provides oxygen to water very depleted in dissolved
oxygen.
2. It oxidizes and precipitates iron and manganese.
130
-------
Table 3
Principal Drinking Water Plants Using Step-Ozonation
Before a Second Filtration on Granular Activated Carbon
Raw
Date of
GAC
Going Number
Average
Filter Bipty
\tolume Bed Regenera-
per Contact tion
Water
Plant Capacity Origin
Mery-sur-Oise 300 000 Oise River
Morsangf second 75 000 Seine River
section) upstream
f&ris
Rouen-la- 50 000 groundwater
Chape lie
District
Suppl ied
northern
suburbs
remote
southern
suburbs
Rouen
southern
suburbs
Into
Treatment Service
First ozonation; raw water 1979
storage; coagulation by alumi-
num chloride; flocculat ion-
sedimentation; sand filtration;
second ozonation; GAC filtra-
tion; chlorine disinfection;
decnlorination; C1O_
Prechlorination; coagulation; 1975
flocculation; sedimentation;
sand filtration; ozonation
(half the plant); GAC filtra-
tion; final disinfection
chlorination
First ozonation; sand f iltra- 1975
tion; activated carbon filtra-
tion; second ozonation;
chlorination.
GAC of GAC Unit
Function Filters m
organ ics 12 108
elimi-
nation;
chlorine
stability
& NH4
elimination
organ ics 4 90
elimination
micro- 6 75
pollution
& NH.
elimination
Time Period Carbon
min years Type
13 Not 2 Picactif
yet 2 Picafil
re- 4 PicafloG
gen- 2 F 300
erated
10 3 F400
13 1 5 Picactif
fil- 1 Picafil
ter
regen-
erated
after
4 yrs
testing
-------
Table 4
Conditions of Step-Qzonation
Plant
Mery-sur-Oise
Morsang
Rouen-la-
chapel le
First Ozonation
°3
Concentration
in the-Air
(g/mJ)
15 to 20
residual second
ozonation: 1.5 to 2;
with addition of new
03 12 to 15
Treatment
Rate
(g cynr5)
of water
0.8 to 1
1
Second Ozonation
Contact
Time in
Minutes
3
3
°3
Concentration
in the^Air
(g/tar)
12 to 18
12 to 18
12 to 15
Treatment
Rate
(g o3/mj)
of water
2 to 4
average 2.85
1
Contact
Time in
Minutes
10
10
10
U)
-------
ROUEN LA CHAPELLE
WATER TREATMENT PLANT
ACTIVATED
CARBON
FILTRATION
••BONO
OZONATION
Figure 10. Treatment Process at Rouen-la-Chapelle
(3OO.OOO m3/day)
Cj Flocculation
Sand Second
Sedimentation filtration ozonation
Treated
water
Activated
carbon Post
filtration chlorination
Figure 11. Treatment Process at Mery-sur-Oise
133
-------
3. It transforms the organic micropollutants into sub-
stances that can be adsorbed on the activated carbon
for a long period of time.
Ozonation coupled with activated carbon constitutes the
technique termed biological activated carbon (BAG). Ammonia is
biologically eliminated in the sand filter, for the most part,
while the activated carbon filter, through the BAG effect, com-
pletes the elimination of organics. The organic content in
treated water averages 25 percent of organic content of the raw
water.
The reduction rate of 75 percent, observed at Rouen-la-
Chapelle, is slightly higher than the 70 percent rate noted at
the Morsang and Choisy-le-Roi plants, but this difference may
be due to the nature of the dissolved organic content of the
source water. After three years, the carbon has not needed
regeneration.
At the Rouen-la-Chapelle plant, unlike the Morsang plant,
the last stage of ozonation is introduced at the end of the
treatment line, rather than before carbon filtration, and is
followed only by chlorination at low dosage to protect the
distribution system. Postozonation has sometimes been suspect-
ed of fostering regrowth of bacteria in the distribution net-
work. No evidence for this has been found at Rouen-la-Chapelle
or in other French water treatment systems using the same final
ozone-chlorine combination. Plants at Toulouse and Nice, in
fact, have been disinfecting with ozone only, without post-
chlorination, for many years with satisfactory results.
The Mery-sur-Oise plant. Another plant under conversion
by the end of 1979, will be the largest in France, and perhaps
in the world, using step-ozonation and BAG to eliminate organics
and limit the use of chlorine. This plant is located at Mery-
sur-Oise, 20 km north of Paris, and is supplied from the Oise
River, a Seine tributary severely polluted by industrial waste-
waters (Figure 11).
Mery-sur-Oise was originally a conventional plant using
simple filtration and ozonation, with a maximum output of
300 x 10 m /day. It is being modified by the addition of a
riverside storage reservoir with preozonation and by the separa-
tion of the conventional filters into two batteries, in series,
as at Choisy-le-Roi.
The raw water storage reservoir will hold 400 x 10 m of
water, corresponding to a little more than two days of average
output of the plant. Such facilities allow the smoothing out of
pollution peaks by pumping and storing water from the river when
its quality is higher. The storage reservoir also will improve
134
-------
the quality of the water through a natural purification process.
To promote this prepurification, an ozonation stage is placed
before the reservoir, eliminating some of the inhibitors of
natural purification as well as phosphates that could encourage
proliferation of algae in the reservoir.
After the storage basin the treatment line is comprised of
flocculation-sedimentation and sand filtration, at which level
ammonia is biologically eliminated; ozonation with a contact time
of 10 minutes; and filtration on activated carbon in twelve
filters of 100 m surface area, with an empty bed contact time
averaging 13 minutes. This stage eliminates organics as far as
possible before a final chlorination preceding pumping into the
distribution system. Only in this last operation is chlorine
used in the plant.
It can be noted that step-ozonation is carried out in these
three plants in three different ways: one ozonation between the
sand and activated carbon filtration at Morsang; a first ozona-
tion before sand and carbon filtration and a second after carbon
filtration at Rouen-la-Chapelle; and a first ozonation before
the sand filter and a second before the carbon filter at Mery-
sur-Oise. The particular quality of the raw water at each plant
leads to the particular treatment scheme. The common feature
is that in every case double filtration is used with at least
one ozonation step before the second filtration on activated
carbon.
Conclusions
Experience in France covers a full range of the possibili-
ties offered by activated carbon. During recent years, research
has been accelerated and new developments have brought about
continual improvements, with further progress in sight.
Two basic processes are used. The first, employed by six
treatment plants with a combined output of 500 x 10 m /day, is
adsorption—i.e., simple filtration on activated carbon be'ds.
This requires a rather high frequency of regeneration.
The second process, used by four plants with a total output
exceeding 400 x 10 m /day, is biologically enhanced adsorption
entailing double filtration on sand and carbon beds with step-
ozonation. Most conventional French treatment plants can be
converted to this process at reasonable cost. Regeneration is
still required, but less frequently.
Either of these processes can improve the quality of treated
water, but in each case there is an optimal choice between the
two. The choice depends both on the pollution level of the raw
135
-------
water and on the quality requirements for treated water. The
higher these are, the greater the inclination toward the double
filtration process.
It has proved possible to obtain a 70 percent removal of
organics and to control chlorination, at reasonable cost.
136
-------
OBJECTIVES AND PROCEDURES FOR GAC TREATMENT IN THE NETHERLANDS
Adriaan P. Meijers, Johan J. Rook, Bart Schultink,
Johan G.M.M. Smeenk, Johan van der Laan, and Cees L.M. Poels
The annual production of drinking water in the Netherlands
amounts to approximately 1 billion cubic meters, of which 40
percent is drawn from the Rhine and Meuse rivers, the Yssel
Lake, and the Haringvliet. The Rhine is seriously polluted
by many contaminants, while the other sources are polluted to a
lesser extent. Given the quality of the sources for the produc-
tion of drinking water, the Dutch waterworks are unanimous in
their opinion that at least a 1-month reserve supply should be
permanently available. Reserve water suplies are collected in
storage basins and by infiltration in the dunes. Selective in-
take of raw water prevents serious pollution from reaching the
purification plants. The collection of reserve supplies sub-
stantially improves and equalizes water quality, but pollution
is so high that treatment with activated carbon remains indis-
pensable. Powdered activated carbon is still widely used, but
many waterworks now are changing over to granular activated
carbon (GAC) treatment.
This paper describes several currently operating carbon
filter installations and pilot plants, focusing on the criteria
for contact time and operating time between regenerations,
referred to as running time.
ROTTERDAM WATERWORKS, KRALINGEN
Water Quality
Rotterdam obtains its water from the Biesbosch storage
basin where the water taken from the Meuse River has a resi-
dence time of approximately 3 months. The river is slightly
contaminated by micropollutants. Table 1 shows the water
quality of the Meuse at Keizersveer.
While the water is held in the storage basin, it satis-
factorily improves in quality as the concentration peaks of
pollutants taken in are attenuated. Algal growth in the deep
basin is suppressed by screens of rising air bubbles that ensure
the circulation of the water. The average chlorophyll content
is 5 yg/1 with occasional peaks of 40 yg/1.
137
-------
Table 1
Water Quality of the Meuse at Keizersveer, 1977
Parameter
Flow-m /s
Temperature-0 C
Suspended matter at
110°C-mg/l
PH
DOC-mg/1
Color-mgPt/1
Taste
Phenols- yg/1
Oil-mg/1
Pesticides-yg/1
Detergents-yg/1
Chlorophyll-a-yg/1
Ammonia-mq/1
Total phos.ph'a-te-ma/l
Minimum
23
2
7
7.5
3.0
9
1
1.5
<0.1
<0.1
40
1
0.1
0.7
Average
270
12
19
7.7
4.4
15
3
2.5
0.1
0.1
60
22
1.3
1.5
Maximum
1,015
22
214
8.1
5.8
25
6
5
0.9
0.14
130
86
4.7
2.2
DOC-Dissolved organic carbon
From this storage basin the water is transported 30 km to
the plants at Berenplaat and Kralingen. Transport chlorination
with 5 mg/1 chlorine takes place during the summer months; the
dose was reduced to 1 mg/1 chlorine in 1978.
Purification Processes
The plant at Berenplaat has been in operation since 1966
and has a capacity of 12,000 ir.Jhr. The storage basin, located
where water from the Oude Maas forrc erly was taken in, has a
residence time of 2 weeks. The water is purified successively
by breakpoint chlorination with 4 mg/1 chlorine, dosing with
powdered activated carbon, coagulation in a sludge blanket
flocculator, rapid sand filtration, and safety chlorination with
0.8 mg/1 chlorine.
In 1977 a new 5,000-m /hr plant was built at Kralingen.
The water is purified by coagulation and sedimentation in
lamella separators, ozonation with 3 mg/1 ozone, secondary iron
dosing, double-layer filtration, carbon filtration, and safety
chlorination with 0.8 mg/1 chlorine.
Both plants supply good quality water, although the taste
of the water from the Kralingen plant is said to be better.
138
-------
Objectives of Carbon Filtration
Carbon filtration is employed at Kralingen for protection
against occasional pollutants and toxic substances, for removal
of substances produced by chlorination, and for removal of
assimilable organic matter produced by ozonation. The addi-
tional precaution of carbon filtration along with ozonation is
deemed necessary because ozone fails to oxidize numerous sub-
stances including highly chlorinated organic compounds.
The Rotterdam waterworks is not primarily concerned with
trihalomethanes, but by the chlorinated substances having a
higher molecular weight that are represented by the parameter
AOC1 (adsorbable organic chlorine) (1). The AOC1 determination
method is still in the developmental stage, however, and cannot
yet be used as a criterion for contact times and running times
of filters. The same caveat applies to the toxicity tests of
effluents presently being developed by KIWA. The operating
conditions of the carbon filters at Kralingen are given in
Table 2.
Table 2
Conditions for the Carbon Filters at Kralingen
Filter Condition Value
Particle size-mm _ 0.8
Quantity of adsorbent per filter-m 116
Depth of adsorbent-m 4
Free board-m 3
Filter diameter-m 6^
Contact time-min 12
Adsorbent used was Norit ROW 0.8 Supra and N.
**
5 empty bed volumes/hr.
Results
During the summer months, transport chlorination was
maintained at 5 mg/1 chlorine, which led to the formation of
trihalomethanes (THM). The residence time in the pipeline to
Berenplaat was 8 hours, and to Kralingen it was 15 hours.
During the 1977-1978 winter season the transport chlorination
was discontinued; consequently, no THM formation occurred.
Table 3 shows the .THM concentrations in the influent at
Berenplaat and in the water supplied by the plant. Prechlori-
nation with 0.5 to 1 mg/1 chlorine and safety chlorination with
139
-------
0.8 mg/1 chlorine in summer causes an increase in THM concentra-
tions from 68 to 84 yg/1. The trihalomethane formation poten-
tial (THMFP) of the raw water after 48 hours at 20°C is 160
yg/1. During the winter period when only breakpoint chlori-
nation with 4 mg/1 chlorine and postchlorination with 0.8 mg/1
are applied, the THM concentration in the Berenplaat drinking
water is 38 yg/1.
Table 3
THM Concentrations (yg/1) in the Water of Berenplaat*
Season
and Type
of Water
CHC1.
CHCl2Br
CHClBr.
CHBr-
Total
THMs
Summer raw
water
Summer fin-
ished water
Winter raw
water
Winter fin-
ished water
30
40
<1
18
22
28
0
13
14
14.5
0
6
1.5
1.5
0
1
68
84
<1
38
The plant in Kralingen began operations in July 1977.
During the first month no carbon filtration was applied. Table
4 shows the THM concentrations during that period. The raw
water was subjected to transport chlorination with 5 mg/1
chlorine for 15 hours. The effect of safety chlorination with
0.8 mg/1 chlorine was measured after 6 hours. During the
coagulation process the THM concentration decreases slightly,
but this decrease is offset by the increase resulting from the
safety chlorination.
The effect of the removal of THM precursors can be seen in
Table 5. During the first 5 months of GAG operation the
influent contained approximately 100 yg/1 THM. In the first
month the THM concentration was reduced to 2 yg/1. The decrease
in the production of THMs by safety chlorination (by 16 yg/1 as
compared to 33 yg/1 without carbon filtration) indicates a
partial elimination of the precursors. After a few months the
trihalomethanes started to break through, with the brominated
compounds appearing in the lowest concentrations. In this
period 20 yg/1 THMs were generated by safety chlorination.
140
-------
Table 4
THM Concentrations (yg/1) at Kralingen
Without Carbon Filtration, July 1977
Type of
Water
Total
CHC13 OHCl2Br CHClBr2 CHBr3 THMs
Raw water
After coagu-
lation
After ozona-
tion
After safety
chlorination
50
32
32
49
32.5
22
23
35
18
13
12
2.5 103
1.5 68
69
16.5 1.5 102
Table 5
Running Time
of Carbon
THM Concentrations (yg/l)at Kralingen
With Carbon Filtration
CHC13 CHCl2Br CHClBr2
Total
THMs
1 month, after carbon
filtration
1 month,after safety
chlorination
2-5 months,after
carbon filtration
2-5 months,after
safety chlorination
2-10 months', after
carbon frltration
2-10 months', ^winter)
after satety chlori-
nation
1.2 0.8
3.2 3.7
17
18
11
12.5
5.4
6.5 6.5
13.5
1.5
9.5
5.7 18
32
8.5 52
9.5
6.5 29
141
-------
During the winter period, transport chlorinatibn was dis-
continued; consequently, no THMs were formed. However, after
carbon filtration small concentrations of THMs caused by
elution were still detected, while the removal of precursors
continued. This is illustrated by the constant increase of
20 yg/1 THM after safety chlorination, which yielded a propor-
tionately high concentration of brominated compounds. For
these reasons, the Rotterdam waterworks does not consider GAG
treatment for THM removal economically feasible.
This moderate reduction, especially for chloroform, is
illustrated by Figures 1 to 3. The breakthrough curves of the
DOC, UV extinction, and the reduction of detergents are shown in
Figures 4 to 6. At the end of 5,500 empty bed volumes (6 weeks)
the DOC had dropped from 2.4 to 1.0 mg/1, or by nearly 60
percent. After 6 weeks there is a rapid breakthrough, amounting
to 80 percent after 12,000 empty bed volumes.
A similar breakthrough occurs for compounds represented by
the UV extinction. Anionic detergents are also only partly
eliminated. The concentrations of phenols and pesticides after
carbon filtration are generally £l yg/1 and £0.04 yg/1, respec-
t ively.
The carbon filters are expected to be able to cope with a
sudden high influx of pollutants. A test filter that had been
running for 2 years and therefore was in equilibrium with the
existing pollutants, was dosed with 10 y.g/1 lindane for 7 months.
On the average, 0.6 yg/1 lindane was detected in the effluent,
indicating that the carbon filter functioned effectively as a
safety filter.
Eliminating chlorination before the coagulation process
does not lead to sliming of the installations. The water
evidently matures sufficiently in the Biesbosch. Since the
plant is indoors, weather conditions are not a factor.
At present, research is being conducted to determine the
most suitable type of adsorbent for a given set of analytical
and toxicological parameters.
PROVINCIAL WATERWORKS OF NORTH HOLLAND (PWN), ANDIJK
Water Quality
The PWN depends on the Yssel Lake for part of its water
supply. Table 6 shows the water quality of the Yssel Lake at
Andijk for 1977.
142
-------
chloroform
jug/I
40
30
20-
10
[Rotterdam Waterworks
lran»port_cnionnation oft_
_____
,' effluent
04-
0 2000 4000 6000 8000 10000 12000 14000 16000 16000 20000 22000 24000 26000
bedvolumes —•»•
Figure 1. Chloroform Content before and after Carbon Filtration
at Kralingen
t
CHCIjBr
/ug/l
Rotterdam Waterworks
25-
20-
15 1
transpon chtormjrjcm
— -
effluent
-Of
. X
i
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 22000 24000 26000
bedvolumes —•—
Figure 2. CHCl2Br Content Before and After Carbon Filtration
at Kralingen
143
-------
t
CHOBrj
I Rotterdam Waterworks |
transport chlonnation off
______
10-!,
S
B
I
--0
2000 4000 6000 8000 10000 12000 14000 16000 16000 20000 22000 24000 26000
bedvolumei »
Figure 3. CHCIBr- Content Before and After Carbon Filtration
at Kralingen
percent break through of
ooc
lOO-i
60-
60
40-1
Rotterdam Waterworks
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 22000 24000 26000
bedvolumet —*-
Figure 4. DOC Breakthrough Curves in Case of Carbon Filtration
at Kralingen.
2 to 4 mg/1.
144
-------
Rotterdam Waterworks
percent break through of
UVe»tmction
100-1
BO
60
70
60-
SO-
40-
30
20
10
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 22000 24000 26000
bedvolumes »
Figure 5. UV Extinction Breakthrough Curves in Case of Carbon
Filtration at Kralingen
detergents
>ug'l
BOH
60-
Rotterdam Waterworks
40-
20
influent
effluent
0 2000 4000 6000 8000 10000 12000 14000 160001 18000 20000 22000 24000 26000
bedvolumet —•—
Figure 6. Decrease of Anionic Detergents by Carbon Filtration
at Kralingen. Methylene Blue Method.
145
-------
Table 6
Water Quality of the Yssel Lake at Andijk, 1977
Parameter
Temperature-0 C
Suspended matter at
110°C-mg/l
PH
Chloride-mg/1
DOC-mg/1
Color-mgPt/1
Taste
Phenols-mg/1
Oil-mg/1
Pesticides-yg/1
Cholinesterase inhibi-
tors-yg/1
Anionic detergents-yg/1
Chlorophyll-a-yg/1
EOCl-yg/1
Ammonia-rog/1
Total phot* phate - mg/1
Minimum
1
2
7.7
164
6.2
12
1
<1
0.01
<0.01
0.2
<10
5
0.4
0
0.2
Average
11
18
8.6
223
8.6
21
2
7
0.07
0.02
1.4
62
78
3.4
0.3
0.7
Maximum
20
49
9.5
340
12.2
30
4
28
0.15
0.22
6.6
170
220
8.2
2.3
1.8
Usually the contamination by micropollutants is slight, but
higher concentrations of phenols, cholinesterase inhibitors,
extractable organic chlorine compounds (EOC1), and chlorophyll
(algal growth) may occur occasionally, so that a sophisticated
purification system is still necessary.
Purification Processes
The pumping station at Andijk has a maximum purification
capacity of 4,000 m /hr and was put into operation in 1967.
Water from the Yssel Lake is received in a basin with a storage
capacity of 3 weeks. Algae are removed by microsieves with a
mesh size of 35 ym. The water then is subjected to breakpoint
chlorination with an average of 9 mg/1 chlorine disinfection in
a detention basin with a 30-min, contact time, coagulation in
hydrotreaters, double-layer rapid sand filtration, carbon
filtration, and a second straining with microsieves to remove
large organisms. Finally, safety chlorination is applied with
0.03 mg/1 chlorine and chlorine dioxide. The dose can be kept
low because after breakpoint chlorination and carbon filtration
the chlorine demand of organic compounds is negligible. Thus
the concentration of trihalomethanes does not increase.
Recently, static mixers have been used in the coagulation
process, making it possible to reduce the coagulant dosage.
146
-------
However, elimination of organic -compounds has been less ef-
fective since this change.
Objectives of Carbon Filtration
The reasons for applying carbon filtration at Andijk are to
restrict bacterial growth in the distribution system; to protect
against occasional pollutants and toxic substances*; -and to
improve taste. The criterion for the restriction of bacterial
growth is that the water supplied be of sufficiently high
quality to preclude further oxygen consumption in a slow sand
filter. This criterion may be interpreted as a requirement for
the removal of the DOC.
Trihalomethanes were not recognized as a significant health
risk before the carbon filters were installed. Hence, their
elimination was not considered in establishing the original
criteria for the carbon filter running time.
Conditions for Carbon Filtration
Table 7 describes the operating conditions of the carbon
filters at Andijk. A 4,000-m /hr flow is maintained through
the carbon filters. Since only 2,000 m /hr are withdrawn on
the average, the water recirculates over the carbon filters from
1.7 to 2.5 times.
Every fortnight one carbon filter is regenerated by the
manufacturer; in the future this process will be done by the
waterworks. After the first carbon filtration step (12 filters)
the water is collected, pumped to the second carbon filtration
step, collected again, and 50 percent of it recirculated. The
filters have been operating in this way since the end of the
startup period in April)1978.
The adsorption loading of the GAC rises to between 20 and
25 percent weight during the course of a filter run. To achieve
comparable results with one carbon filtration step, the ran time
would have to be decreased by 50 percent and the regeneration
rate correspondingly increased. In spite of the additional
pumping step between the two carbon filters, this process proves
to be cheaper because of the lower consumption of activated
carbon. Preliminary research has revealed that for DOC reduc-
tion a residence time of 30 irin. yields the optimum cost-perform-
ance ratio (2).
The preliminary tests also proved that the performances
of the two carbon brands tested, type A* and type B** were
virtually identical. Lower regeneration costs determined the
choice of adsorbent; the initial cost of the carbon is of far
less importance.
* Norit N.V., Amersfoort, The Netherlands
** Chemviron S.A., subsidiary Calgon Corp., U.S.
147
-------
Table 7
Conditions for the Carbon Filters at Andijk*
Filter Condition Value
Particle size-mm 0.8
Quantity of adsorbent per filter-in 85
Depth of adsorbent-m 3
Free board-m _ to 3
Filter surface-m 27
Contact time-min 30**
Bed volumes before regeneration 16,000
(11 months)
Number of carbon filters 2 by 12
Type of carbon filters open,concrete
Backwash cycle*** once every
4 months
*Adsorbent used was Norit ROW 0.8 Supra, slightly smaller
pores than normal.
**The maximum permissible flow variation is 10 percent;
2 empty bed volumes/hr.
***Stratification is attempted by gradually reducing the
backwash velocity.
Results
The manner in which the filters are operated makes it
possible to obtain fairly constant effluent quality, as shown in
Figures 7 to 9, for UV extinction, DOC, and THM reduction.
Table 8 shows the average reduction of DOC, UV extinction, THMs,
and EOC1. The DOC and UV extinction decrease by a constant
percentage. The decrease in EOC1 is satisfactory during the
entire period; the average effluent concentration of EOCl is 1.0
yg/1. The reduction of trihalomethanes varies widely, reflect-
ing the fluctuations in the THM content of the influent (Figure
9). The high summer concentrations were reduced satisfactorily;
their removal led to a new criterion for the running time of the
filters at the plant. The regeneration cycle has to be adjusted
in such a way as to ensure that the effluent will not contain
more than 50 yg/1 THM.
Table 8
Average Percent Reduction of Some Collective Parameters
Parameter April-June July-December
DOC 35 31
UV extinction 55 56
THM 83 36
EOCl 79 68
148
-------
UV extinction/m
8 -
6 -
5 -
4 -
3 -
2 -
0
• influent
a effluent first filter
•A effluent second filter
\
Provincial Waterworks
of North-Holland
Apr. Mai June July Aug. Sept.Oct. Nov. Dec.
1978
Figure 7. Decrease of UV Extinction (254 nm) in GAC Treatment
at Andijk.
149
-------
DOC
mg/l
6 n
4 -
3 -
2 -
1 -
0
-• influent
-a effluent first filter
-A effluent second filter
a—
Provincial Waterworks
of North-Holland
5
8
Apr. Mai June July Aug. Sept.Oct. Nov. Dec.
1978
Figure 8. Decrease of DOC in GAC Treatment at Andijk
150
-------
trihalomethanes
influent
130^
120-
110-
100-
90-
80-
70-
60
50-
40-
30-
20-
10 -
0
a effluent first filter
-A effluent second filter
Provincial Waterworks
of North-Holland
o
8
1 I 1 1 1 1 1 ; 1 1
Apr. Mai June July Aug. Sept Oct. Nov. Dec.
1978
Figure 9. Decrease of Trihalomethanes in GAC Treatment
at Andijk
151
-------
After carbon filtration the bacterial counts measured at
22°C ranged from 800 to 10fOOO. Safety chlorination reduces
their numbers to between 10 and 100, and bacterial growth in the
distribution system is clearly diminished. However, carbon
filtration leads to a marked increase in the number of higher
organisms such as rotifers and worms. These organisms are
removed using 35-ym mesh microsieves. A striking phenomenon is
that when high bacterial counts occur, few higher organisms are
found after startup and backwashing, whereas bacterial counts
decrease when higher organisms occur in large numbers.
Taste-causing and harmful substances have been effectively
eliminated. The concentrations of pesticides and phenols lie
below the detection limit. Blue algae are removed by rapid sand
filtration and carbon filtration.
The oxygen consumption per filter is 1 to 2 mg/1. The
water is aerated between and after the two filters. In winter
virtually no oxygen consumption occurs.
AMSTERDAM MUNICIPAL WATERWORKS
Water Quality
The Amsterdam Municipal Waterworks obtain the major part of
their water from the Lek Canal near Nieuwegein. This canal is
fed indirectly by the Rhine River. The water quality character-
istics are shown in Table 9.
Table 9
Water Quality of the Lek Canal at Nieuwegein, 1977
Parameter
Minimum
Average
Maximum
Flow of the Rhine at
Lobith-mVs 1,050
Temperature-°C 1
Suspended matter—mg/1 12
pH 7.4
Chloride-mg/1 76
DOC-mg/1 2.9
Color-mgPt/1 7
Taste 31
Pesticides-yg/1 0.1
Detergents-yg/1 50
Chlorophyll-a-yg/1 0.1
Oil (Bergambacht)-mg/! <0.1
EOC1 (Ochten)-yg/1 2
Ammonia-mg/1 0.1
Total phosphate-mg/1 0.7
2,200
11
38
7.7
151
4.8
19
44
0.2
150
17
0.1
12
0.9
1.7
6,300
23
111
8.1
233
8.3
60
67
1.6
430
49
0.3
45
3.4
4.7
152
-------
In addition to the compounds named in Table 9, numerous
organic micropollutants are found in the untreated water,
including organic phosphates, chlorinated and nitro aromatic
compounds, chlorinated ethers, aromatic bases, and phthalates.
The concentrations measured in the Lek with the aid of gas
chromatography and mass spectrometry totalled 20 yg/1 on
October 30, 1977, and 30 yg/1 on December 6. These compounds
are assumed to be partly responsible for the taste of the water
and for its toxicity to fish.
After pretreatment the water from the Lek is infiltrated in
the dunes, satisfactorily eliminating these compounds and
equalizing the quality of the water.
During artificial recharge the volatile compounds virtually
disappear. The concentrations of the remaining micropollutants
are generally much lower than 1 yg/1, with the exception of the
phthalates. Nevertheless, the infiltrated water is not free
from taste and numerous pollutants from the Rhine still are
present, albeit in substantially reduced concentrations.
Purification Processes
At Nieuwegein water has been drawn from the Lek Canal since
1957. Coagulation and sedimentation take place in a basin,
followed by rapid sand filtration. After chlorination the water
is transported over a distance of 55 km to the dunes where it is
infiltrated. The water is aerated, dosed with 3 mg/1 powdered
activated carbon, and further purified by rapid sand filtration,
slow sand filtration, and safety chlorination. These operations
take place at Leiduin. When the temperature of the water from
the Lek falls below 10°C, the water to be transported is no
longer chlorinated. Above this temperature a dose of 2.4 mg/1
chlorine is added/ which leads to the formation of approximately
100 yg/1 trihalomethanes. The THMs are eliminated by infiltra-
tion in the dunes. At present the doses of chlorine vary, the
criterion being that after 20 min ,0.2 mg/1 chlorine should still
be present. The resulting concentration of THMs is 15 to 25
yg/i.
Objectives of Carbon Filtration
Since the results of dosing with powdered activated carbon
have not been entirely satisfactory, the municipal waterworks is
considering GAC treatment. This change is intended to eliminate
the remaining substances causing taste, to reduce bacterial
growth in the network system, and to remove unknown and possibly
toxic substances. GAC treatment will be placed after rapid sand
filtration in the treatment sequence. The results of the test
filters are being evaluated.
153
-------
Conditions for the Test Filters
In the test filters three types of adsorbents were tested
at different contact tiroes. Table 10 lists the test conditions.
Every 2 weeks the filters are backwashed, gradually reducing
the backwash velocity to maintain the stratification to the
extent possible.
Table 10
Conditions for Seven Test Filters at Leiduin
Type of Carbon
Filter
Condition Type A Type B Type C
Quantity of adsorbent
per filter-1 393 147 147 393 147 147 393
Depth of adsorbent-m 233 233 2
Filter diameter-m
Contact time-min
0.
40
5
0.
24
25
0.25
12
0.
40
5
0.25
24
0.25
12
0.
24
5
Results
Collective Parameters—
For the type A and type B carbons it was observed that with
a residence time of 12 win/ a constant TOC breakthrough of 75 to
80 percent occurs after 11,000 to 12,000 empty bed volumes,
while the permanganate consumption and the UV extinction reach
breakthrough levels of 80 and 70 percent, respectively. These
constant breakthrough levels last until 25,000 empty bed volumes
regardless of the hydraulic loading rate.
With a residence time of 24 mtn, constant breakthrough
levels are reached after 6,000 to 8,000 empty bed volumes. For
carbons A and B the breakthrough is then approximately 70 per-
cent for TOC and permanganate consumption, and 50 percent for UV
extinction.
Carbon type C* shows a higher breakthrough for these
parameters, from 75 to 80 percent. For these parameters no
difference in performance between the type A and type B carbons
was observed. Type C carbon clearly gives poorer results.
*Lurgi Apparate Technik GMBH, Frankfurt, Fed. Rep. Germany
154
-------
With a residence time of 40 min the constant breakthrough
is probably reached sooner (4,000 empty bed volumes) with lower
breakthrough levels. However, this residence time gives rise to
anaerobic condition? and its use in daily practice is therefore
not recommended.
Trihalomethanes are not detected after infiltration. Other
micropollutants, however, can still be discovered by measuring
the AOC1. The concentrations of AOC1 before carbon filtration
amount to 100 to 250 yg/1. For these parameters, too, the be-
havior of type A and type B carbons is identical. With a resi-
dence time of 24 min,, the AOC1 is reduced by 45 percent up to a
throughput of 11,000 bed volumes. After that, almost complete
breakthrough occurs. The type C carbon is also inferior in
removing these parameters. The initial reduction of 40 percent
soon drops to 20 percent. With shorter residence times the
reduction of AOC1 is even poorer, while longer residence times
bring no improvement. The adsorption of AOC1 is no better than
that of the TOC, which may be explained by the assumption that
in this case the AOC1 represents fairly polar chlorinated com-
pounds of little volatility. Furthermore, the safety chlori-
nation raises the AOC1 concentrations to their original levels.
During the-last stage of these experiments two EOC1 determina-
tions also were carried out. The concentrations were very low,
however,sranging between 0.2 and 0.6 yg/1 before carbon filtra-
tion and 0.1 and 0.2 yg/1 after carbon filtration. Subsequent
chlorination caused an increase to approximately 4 yg/1. No
difference was observed in the performance of the three types of
carbon.
Micropollutants—
In several samples, micropollutants were determined by gas
chromatography and mass spectrometry. The results are listed in
Table 11. The concentrations are very low and no clear distinc-
tion can be drawn in the performance of the three types of
carbon. The overall precision of this analytical method is
certainly no better than 0.1 yg/1.
Trihalomethanes—
Trihalomethanes are not present before carbon filtration,
and consequently they also are absent afterwards. However,
trihalomethanes are formed during safety chlorination with 0.7
mg/1 chlorine. After a 2-hr contact time the THM concentrations
average 30 yg/1, with slight differences associated with the
design of the preceding carbon filtration. In a purification
system without carbon filtration or with ozonation, safety
chlorination produces lower concentrations of 20 yg/1. When the
carbon filtration process is added with a contact time of 12
min«j a substantial increase to 40 to 50 yg/1 occurs during the
first few months; later this level falls to 20 to 30 yg/1. If
carbon filters with a residence time of 40 min. are installed, a
155
-------
Table 11A
Analysis of Micropollutants Before and After Carbon
Filtration at Leiduin on May 8, 1978 - Concentration in yg/1
Type of
Micropollutant
Effluents
Influent
Carbon
Type A
Carbon
Type B
Carbon
Type C
Acids
Phthalates
Organic phosphates
Dichloroisopropyl ether
Chlorinated aromatics
Polyaromatics
Trimethyloxindole
Aromatic bases
0.4
1.2
0.2
0.3
0.4
0.2
1.7
0.1
0.1
1.3
0.1
0.2
1.6
0.2
0.1
Table 11B
Analysis of Micropollutants Before and After Carbon
Filtration at Leiduin on July 12, 1978 - Concentration in yg/1
Type of
Micropollutant
Influent
Carbon
Type A
Effluents
Carbon
Type B
Carbon
Type C
Acids
Phthalates
Organic phosphates
Dichloroisopropyl ether
Chlorinated aromatics
Polyaromatics
Trimethyloxindole
Aromatic bases
0.1
1.2
1.1
:o.i
o.i
0.2
0.5
1.0
0.2
0.8
0.9
0.1
0.1
1.1
0.5
156
-------
slight reduction to 15 yg/1 takes place during the first few
months. However, after 3 to 6 months the THM content increases
again to 30 to 50 yg/1. Bromoform is primarily responsible for
these high concentrations.
Conclusive evidence indicated that installation of carbon
filters followed by chlorination with 0.7 mg/1 chlorine cause*!
an increase in the THM concentrations. The THM concentrations
can be reduced to 20 yg/1/ if the chlorine is dosed*after a 20-
min contact time the chlorine residual does not exceed 0.2 mg/1.
* (in an amount such that)
Evaluation of GAC Treatment
The Amsterdam Municipal Waterworks anticipates few problems
with regard to the elimination of unknown toxic substances as
well as those substances causing taste. It is also expected
that the bacterial growth in the distribution system will
decrease. During the testing of the carbon filters it was
found, however, that the concentrations of chlorinated products
are decreased little or not at all by carbon filtration. The
significance of the residual AOC1 and THM concentrations will
have to be demonstrated with the aid of toxicity tests. At the
present state of these investigations, however, the choice of
the optimal contact and running times is not conclusively
indicated.
MIDDEN NEDERLAND WATERWORKS PUMPING STATION AT ZEIST
WATER QUALITY
At Zeist very pure groundwater with extremely low levels of
organic substances is pumped. In recent years, however, pollu-
tion by trichloroethene has been observed in concentrations of
500 to 1,000 yg/1. Such pollutants have also been found in the
groundwater elsewhere in the Netherlands; for example, tri- and
tetrachloroethene at Hilversum, tribromoethene at Almelo, and
trichloroethene at Nijmegen, while chloroform also occurs -at
various locations.
PURIFICATION PROCESSES
The groundwater at Zeist is aerated and filtered over
CaCO^MgO filter material* which results in deacidification and
elimination of iron and manganese. A number of wells have been
equipped with carbon filters with a total capacity of 300 m /hr.
The entire pumping station has a capacity of 900 m /hr. The
quality of the treated water is shown in Table 12.
*Akdolite-Werk GMBH, Hochdahl, Fed. Rep. Germany
157
-------
Table 12
Quality of the Drinking Water at Zeist
Parameter Value
Temperature-°C 12
Color-mgPt/1 1
pH 8.2
Permanganate consumption - mn/1 <1
Total hardness-meq/1 2
GAC treatment is intended to reduce trichloroethene levels
to less than 1 yg/1. Table 13 describes the operating con-
ditions of the carbon filters at Zeist.
Table 13
Carbon Filter Characteristics at Zeist
Filter Condition Value
Particle size-mm 1
Quantity of adsorbent per filter-in 12
Depth of adsorbent-m 1.5
Free board-m 1
Filter diameter-m 3.3
Contact time-min, 12
Expected empty bed volumes before
regeneration >25,000
Number of carbon filters 3 by 2
Backwash cycle first filter only
RESULTS
In the course of the preliminary experiments it was found
that GAC treatment is most effective when inserted into the
sequence following filtration over CaC<50/t1gOCliter material.
This reduces the need for frequent bacfcvtasning and prevents
the risk of premature breakthrough.
158
-------
An attempt was made to fix the adsorbent within a packed
bed in a test filter. During backwashing the tension is re-
leased only in the uppermost 50 cm. This backwashing method and
a contact time of 8 min yielded a running time of 4 months. The
trichloroethene concentration front advanced regularly through
the GAC bed during the course of the test run, as expected
(Figure 10). At the equilibrium stage the load of the adsorbent
proved to be 50 g/kg carbon.
For regeneration all the carbon has to be removed from the
filter. If not, desorption and breakthrough occur immediately.
The operational filters have been in actual use since December
1, 1978. Two filters have been placed in series, with a total
contact time of 12 min. The running time is expected to be
7 to 8,months. The costs for this small capacity are hfl.
0.15/nr carbon ($0.075/m carbon at the May 1979 exchange
rate).
KIWA PILOT PLANT AT NIEUWEGEIN
The water quality of the KIWA pilot plant at Nieuwegein is
illustrated in Table 9.
Purification Processes
At Nieuwegein water is taken in from the Lek Canal. The
raw water is presettled in a basin and then coagulated and
filtered over a double-layer filter. This step is followed by
parallel ozonation, GAC treatment over a fresh carbon filter,
and GAC treatment over a filter that has been running for 7
months.
Objectives of Carbon Filtration
The most suitable criteria for contact time and running
time are TOC, UV extinction, EOC1 and AOC1, and taste. However,
the toxicological significance of the remaining concentration is
not known.
The operating conditions of the carbon filters used at
Nieuwegein are shown in Table 14. At the start of the experi-
ments on August 10, 1977, one carbon filter had been running for
2 weeks, whereas the other one had been in operation for 7
months. A great difference in the quality of the effluents from
these two filters was expected.
159
-------
26 May
0.1
hbed (cm)
r200
! Waterworks
Midden- Nederiand
2Juf»
Figure 10.
Breakthrough Curves of Trichloroethene by Carbon
Filtration at the Zeist Pilot Plant.
Contact Time 8 Minutes.
160
-------
Table 14
Conditions for the Carbon Filters at Nieuwegein
Filter Condition* Value
Particle size-mm 0.8
Quantity of adsorbent per filter-1 42.5
Depth of adsorbent-m 0.85
Free board-m 0.65
Filter diameter-m 0.25
Contact time-min 21
Backwash cycle every two weeks
at 25 m/hr
# '
Adsorbent used was Norit ROW 0.8 Supra.
Results
For 4 months, weekly analyses were made of DOC, UV ex-
tinction, and EOC1. On two occasions analyses were made using
gas chromatography and mass spectrometry. The toxicological
research was carried out on tropical tadpoles (Xenopus laevis)
raised in aquarium tanks with a constant water supply of varying
water qualities. During the larval development and the growth
period after the metamorphosis, it was not possible to find
criteria for judging the toxicology of the water. Two para-
meters eventually proved suitable, namely the mortality rate up
to the sixth day after egg deposition in the tanks, and the
enlargement of the liver in the tropical tadpoles after 4 months
at the end of the tests. The enlargement of the liver was
expressed as the liver somatic index (LSI); i.e., the weight of
the liver related to the total body weight of the tadpole. The
liver is the organ in which toxic substances (e.g., chlorinated
and aromatic hydrocarbons) are metabolized, causing liver
enlargements.
In order to find relations, the mean values of the DOC, UV
extinction, and EOC1 were computed on the 8th, 15th, and 22nd of
August and compared with the mortality rate of the tropical
tadpoles after 6 days. Table 15 and Figure 11 show the results.
The mortality rate of the tropical tadpoles proved to be
a less changeable criterion than the chemical parameters. This
cannot be a logical assumption, however, because of the totally
different nature of the toxicological parameters. In living
organisms there is often a threshold value that has to be
exceeded before toxicological effects become apparent. The
161
-------
• —-• mortality )• attar 6 day*
C—i DOC mti/l
A & UV attrnctjon/m
a—D EOCi.iJg/i
EOCI ./ug/l 6-| 12 UV
Lak RSF uona Uk RSF ti«hcarbon Lak
RSF utad carbon
Figure 11.
Relation Between the Mortality of Tropical Tadpoles
After Six Days and DOC, UV Extinction and EOCI
Lek = raw water; RSF = rapid sand filtrate
162
-------
Table 15
Comparison Between DOC, UV Extinction, and EOC1
and the Mortality Rate of Tropical Tadpoles After Six Days
Type of Water
Used as Habitat
Lek water
Rapid sand filtrate
Effluent ozone
Effluent carbon
Effluent carbon
Groundwater
(old)
(fresh)
DOC
mg/1
4.8
2.8
2.9
1.8
0.5
UV
Extinction
per m
11.3
8.0
3.5
4.1
0.3
EOC1
yg/i
5.1
3.1
1.4
0.3
0.2
0.1
Percentage
Mortality
After
Six Days
78
69
61
64
62
31
mortality rates of the tropical tadpoles in the effluents of
ozonation and of the fresh and old carbon filters hardly
differed. However, a vast difference exists in the mortality
rates between the effluent of a fresh carbon filter and ground-
water, although the analytical parameters of the effluent of the
fresh carbon filter yield favorable results. There seems to be
no linear relation between the mortality rate and the DOC, UV
extinction, and EOC1. All that can be deduced at present is
that a decrease in the value of the chemical parameters most
often goes hand in hand with a decrease in the mortality rate of
this test animal.
The chemical parameters have also been compared with the
LSI value. The mean values for the chemical parameters have
been computed over the entire test period. Table 16 and Figure
12 give the results for the comparison of the LSI values and
DOC, UV extinction, and EOC1. Table 17 and Figure 13 compare
LSI values with various groups of organic compounds.
The values of the groups of organic compounds have been
calculated by taking the average over the entire test period.
For the Lek water, five samples were taken at Bergambacht. The
other samples were taken weekly at Nieuwegein. The LSI as such
proves to be a sensitive parameter as compared with the differ-
ences between the chemical parameters. An accentuation of the
differences is obtained if the LSI value is related to the
decrease in the LSI of animals in Lek water and the control
animals by the formula:
A LSI percentage =
LSI - LSI (groundwater)
LSIfLek water) •* LSI(groundwater)
100
percent
163
-------
0 O DOC .119/1
*—t. uv •iiinciion/m
O a EOCI.,ug/l
.LSI % T ».
i
EOO./ug/l 6-1 ,-12 UV
DOC mg/i I I •»«.nttion/m
t J U ?
Ltk RSF ojon« Ltk RSF frnncvbon Lik RSF uMdorbon groundvviiir
Figure 12. Relation Between the Relative Value of the Liver
Somatic Index and DOC, UV Extinction and £OCI
O
•
O
1 nilfllld comatiCi
O cnlormiito •rOTKlict and llhtrt
• prtvnols
D IDCl
.LSI %
t
'00 •]
to
«0
40
20
conccntrition
organic ["
»ubtl§nct»
r«
EOCI
i
i
i
I
U
Itk RSF wont Ltk RSF Inthcirbon Lek RSF uuduroon grounavnitr
Figure 13. Relation Between the Liver Somatic Index and the
Concentration of Different Groups of Organic
Substances
164
-------
Table-16
Comparison of DOC, UV Extinction, and EOC1
with the LSI Percentage
Type of Water
Lek water
Rapid sand filtrate
Effluent ozone
Effluent carbon
Effluent carbon
(old)
(fresh)
DOC
mg/1
4.5
3.2
3.1
2.5
1.3
UV
Extinction
per m
12.1
9.9
4.8
5.9
2.4
ECC1
ug/1
10.1
2.7
1.1
0.3
0.1
A LSI
Percentage
100
78
62
60
36
LSI Increase
In Relation to
Control Animals
36.0
28.2
22.2
21.6
13.0
Table 17
Comparison Between Some Groups of Toxicologically
Suspect Compounds, EOC1, and LSI Percentage
Test
Parameter
Organic phosphates- yg/1
Nitrates aromatics-yg/1
Chlorinated aromatics
and ethers-yg/1
Phenols- yg/1
EOCl-yg/1
ALSI percentage
Lek
2.0
5.65
2.45
0.8
10.0
100
Rapid
Sand
Filtrate
0.5
3.6
1.15
0.5
2.7
78
Ozone
0.3
1.3
0.25
0.05
1.1
62
Old
Carbon
0.1
0.15
0.3
60
Fresh
Carbon
0.05
<0.1
0.1
36
The direct relation between the ALSI percentage and DOC and
UV extinction appears to be better than the corresponding
relation with the mortality rate. Each decrease in the value of
the chemical parameters, except for the DOC after ozonation,
parallels a decrease in the ALSI percentage. The difference in
LSI between the effluent of the fresh carbon filter and the
groundwater is smaller than the differences in the mortality
rate.
In this connection it should be noted that the carbon filter
had been running for 4 months at the end of the test period. The
relation between LSI and EOC1 is less favorable since the differ-
ences in EOC1 concentrations do not parallel the differences in
A LSI percentage.
165
-------
Figures 11 to 13 reveal that the'various groups of organic
micropollutants follow the trend indicated by the EOC1 fairly
closely. The relation between micropollutants and the A LSI per-
centage appears to be less clear, given the large disparity in
ALSI percentage between the two carbon filters and the slight
differences in the concentrations of the micropollutants. There-
fore, little purpose would be served by conducting a great deal
of GC-MS work in this type of research. At present it is suffi-
cient to observe the collective parameters, such as EOC1, and to
carry out occasional GC-MS research if the identity of the sub-
stances that constitute the collective parameters is unknown,
acute or highly toxic effects occur, or a sudden increase in the
value of the collective parameters occurs. Although the tropical
tadpole is apparently not the most suitable test animal for the
toxicological evaluation of water quality and the tests did not
lead to optimum results at certain points, some important find-
ings were obtained. It was determined that the toxicological
effects on the tropical tadpole decrease as the water quality
improves, and that given its biological effects on the tropical
tadpole, the Lek water quality—even after coagulation, rapid
sand filtration, and (fresh) carbon filtration—is inferior to
groundwater quality. This is not confirmed by the chemical
parameters, however.
Evaluation of Toxicological Parameters
From these tests, and from the research on full-grown trout
and trout eggs carried out some time ago at KIWA, it is known
that measuring toxicological parameters for the evaluation of
the water quality may yield useful results. Rapid measuring
methods need to be developed both for the toxicological and for
the chemical parameters. This type of research should be con-
tinued, but with parameters that can be measured in a relatively
short time, such as TOC, UV extinction and EOC1, the Ames test,
and short-term biolarval tests. If these rapid tests are con-
ducted at frequent intervals the results may be related to each
other more effectively than were the test data resulting from
this investigation.
ACKNOWLEDGEMENT
The author would like to thank J.J. Rook and A.J. van der
Veer of the Rotterdam Waterworks; L.J. Schultink and J. Zwaagstra
of the Provincial Waterworks of North Holland; J.G.M.M. Smeenk
and E.A.M. van Soest of the Amsterdam Municipal Waterworks; and
J. van der Laan of the Waterworks of Midden Nederland, who
generated most of the data described here. Thanks also are due
to my colleagues at KIWA, C.L.M. Poels, M.A. van der Gaag,
S.H.H. Olrichs, and R.Chr. van der Leer, for carrying out the
toxicological, analytical, and technological parts of the KIWA
investigations.
166
-------
REFERENCES
1. Kuhn, W. 1974. Vom Wasser, 43;327
2. Van Lier, W.C., et al. Experiences with Pilot Plant
Activated Carbon Filters in Dutch Waterworks. In trans-
lation of Reports on Special Problems of Water Technology;
Adsorption. EPA-600/9-76-030 (Dec. 1976).
167
-------
PRACTICAL APPLICATIONS OF
ADSORPTION TECHNIQUES IN DRINKING WATER
BELGIAN EXPERIENCES
Dr. W.J. Masschelein
INTRODUCTION
2
Belgium has a surface area of about 30,600 km and 10
million inhabitants. The average water consumption per capita is
220 liters per day, including industrial consumption.
The average rainfall is between 800 and 900 mm/year. The
theoretical potential.groundwater capacity amounts to 2.6 x 10
m of which 0.65 x 10 (+25 percent) m are effectively used.
9
About two-thirds of the reserves are situated in the South
of the country and one-third in the North. Fifty percent of the
reserves are^potentially exploitable, but only 900 x 10 to
1000 x 10 m are potentially usable groundwater because a con-
siderable amount in coastal areas or in coal mining zones is
brackish. Furthermore, the percolation yield (1 m /m/h for free
infiltration galleries) limits the exploitation in a given area
although the yield by pumping reaches 1 to 50 m /m/h. For this
reason, it is easily understandable that most of the water
(presently about 80 percent) originates from groundwater.
Well water has always occupied an important place in Belgian
water supply, as in Belgian art (e.g. in the Ghent alterpiece of
Van Eyck or in the well construction of Breughel). However, the
use of surface water is increasing and experience with advanced
techniques such as adsorption is expanding.
The available surface waters are those of the Yser River in
the coastal area, the Scheldt River in the northern and western
parts of the country, and, in the south-eastern part, the Meuse
River, including the North Albert canal fed with water from the
Meuse. At present, the two former lowland rivers are often
heavily polluted and their use necessitates bank-side storage.
JHence, the most used source of surface water is that of the
Meuse River. The Meuse is a "flashing river ",the flow being
influenced by rainfall and seasonal sources.
168
-------
PRINCIPLES OF THE USE OF ACTIVATED CARBON APPLIED IN BELGIUM
In the practical uses of activated carbon, already in oper-
ation or to be used in the future, there is a tendency to use
fixed beds., e.g., rapid filtration through granular activated
carbon (GAC), or pulsed beds,-e.g., powdered activated carbon
(PAC) in a flocculator. Expanded fluidized and moving bed tech-
niques have not yet found use.
Practical data on full scale experience with rapid filtra-
tion on GAC, at a surface loading of 4 to 6 m/hr in the treatment
of the lowland water stored at the Blankaart station, are indi-
cated in Table 1 [data given by the national "Institut d'Hygiene
et d'Epidemiologie" (I.H.E.)J.
Table 1. Typical Examples of THM Adsorption
in Lowland Waters
Sampling
Point
Intake
After chlo-
rination
After floc-
culation
After fil-
tration on
Nor it ROW
After fil-
tration on
Norit PKST
After fil-
tration on
Chemviron 40C
Concentrations
TOC,
g/mj
23-11
12-8
12
6
7
6
CHC1, CC1. CHCl_Br CHBr,,
342., 3
mg/m
3 0.1 0.6 0.3
40 3.2 5.6 12
40 1.4 14.6 14
35 1.2 14.5 2.6
32 1 14.5 2.6
3.5 (0.1) 0.3 0.3
THM
m Mol/m
(0.03)
0.74
0.54
0.40
0.36
0.03
In some instances the quality may become critical and
research is intended to diminish the residual THM content in the
treated water by the use of appropriate carbons. Significant
differences appear in the adsorption capacity for THM» even among
carbons with equal TOC removal capacities. Adequate treatment
169
-------
includes storage, pre-chlorination, coagulation-flocculation
(either with aluminum sulphate or ferric chloride), settling, and
filtration.
The Tailfer Plant
The Tailfer plant of the Brussels Intercommunal Waterboard
(CIBE) incorporates the most modern treatment technology in the
country. The station is located on the Meuse River, which has
its source in France and is a typical "rain-river" of the moder-
ate climatic zone. Upstream from Tailfer, the wastewater from
about five hundred thousand inhabitants (equivalents) is dis-
charged into the Meuse River. The natural tributaries to the
Meuse result from spongy peat-soil. In dry summer periods the
water is peaty. To enable navigation, the Meuse is forming a
cascade of locks and reaches. During the summer period, actino-
mycetes develop in the river. The organic content is increased
by decaying vegetation during the fall. Consequently, color,
taste and odor removal are among the objectives of the treatment.
The treatment sequence consists of pH adjustment, prechlo-
rination (can be by-passed), preoxidation with chlorine dioxide,
coagulation (aluminum sulphate + activated silica), floccula-
tion in a "pulsator", settling, filtration, ozonation, and post-
disinfection (C12 + NH3 and/or C102).
The powdered activated carbon in the sludge zone of the
pulsator acts as a coagulation-flocculation aid. At the normal
sedimentation rate, 0.011 m /s of sludge is withdrawn for a
water flow of 0.73 m /s, hence about 1.5 percent volume. The
concentration factor of the solids (carbon) is then 70. If 2 to
20 g/m PAC are injected, the concentration can attain 140 to
1400 g/m . This has a favorable effect on the density of the
floe and also improves the coagulation. The differences between
the injection, during or after the coagulation phase, are shown
in Figure 1.
The quality improvement in the water through use of PAC
must be considered in conjunction with coagulation-flocculation.
Typical data are shown in Table 2. The whole process of removing
dissolved compounds is kinetically controlled and the best im-
provements require longer contact times. This appears to more
favorably apply to the use of PAC, rather than GAC, with an
appropriate and specific technology. PAC has a higher surface
area per unit weight than GAC.
Except for the formation of small amounts of volatile bromi-
nated halomethanes, ozonization did not exert any significant
effect on preexisting trihalomethanes iTHM's) (Table 3).
Although the mixed waters from different origins, distrib-
uted by the Brussels' Intercommunal Waterboard, contain less than
170
-------
z
o
a:
UJ
II
O uj
U J/J
z z
u u
a. a.
2
O
111
Q
3 <
t/) tt
VI
u
Q.
Q
at
_i
i_
r^
t
H
UJ
™ fc
1 - 1
1
u
(Q CO
O T3
O -H
«3
CO
flj 'O
CJ 'D
< c
P-I S
y-i w
O 3
S w
< CO
r? CO CP
01 £.5
Q S3 >
^ LU ^ Q
Q 5 r1 4J S
UJ 5 H UK
_j Q- I- o
Hill 'II LJ
1_ W «H 5
r i "~: u w M-I
^J *f \ 1
a ^ a-
• _ ^
fc
171
-------
Table 2. Effect of PAC-Aided Coagulation/Flocculation
Parameter
TOC g/m3
Phenols mg/m3
MBAS mg/m3
COD g/m3
Carbon dose g/m
Raw
7
35
48
15
27
Water
± 2
+ 30
+ 32
± 4
.5 + 12.5
Water Treated with
PAC and Settled
2.5
•18
43
8
+ 1
± 13
± 31
+ 5
—
Table 3. Trihalomethane Concentrations at Tailfer
with Prechlorination
Raw Water
Prechlorinated
(2-5 g/m3) (10 min)
Chlor ine dioxide
(0.7 g/m3) (30 min)
After carbon &
settling (30 min)
2nd chlorination
& filtration
(1 g/m-3) (3 hrs)
After
postbzonization
CHC13 CC14 CHCl^r CHBr2Cl CHBr3
expressed in mg/m
2
15-20
20-22
10-15
15-17
17-20
-
—
-
-
M
-
-
5-7
8-11
8-11
8-11
10
-
0.5
1
1
1
2-3
1
-
-
-
_
0.5
Total
THM
m mol/m
-
0.2
0.26
0.20
0.20
0.20
172
-------
0.1 mg/m of total THM's, the laboratory will focus more atten-
tion on this question in the future. The possible formation of
haloalkanes of higher molecular weight, as well as other chlori-
nated compounds, requires very close attention.
From the data in Table 3, one can deduce that the formation
of THM essentially results from the prechlorination phase. Con-
sidering the possible destruction of the precursors of the meta-
cresol type, prechlorination was discontinued and the C1O2 dosage
and post chlorination were reinforced. Single data, comparative
to that in Table 3, is shown in Table 4.
Table 4. Trihalomethane Concentrations at Tailfer
without Prechlorination
Raw water
Chlorine-dioxide
(1.2 g/nT) (30 min)
After carbon and
settling (30 min)
Dost chlorination
(1.5 g/m3)
(3 hrs) + filtration
After post-
ozonation
CHC13 CC14 CHCl^r CHBr2Cl CHBr3
(mg/m3)
2
2
3
4-6
5
-
-
-
—
—
-
2-3
4
3-5
4
-
-
-
—
—
0.5-1
0.5
0.5
0.5
0.5
Total THM's
(mMol/W3)
0.02
0.035
0.04
0.07
0.08
It appears that the use of chlorine dioxide as a pre-
oxidation-predisinfecting agent largely reduces the formation of
THM's. This is probably caused by the destruction of precursors
which otherwise would form THM's during chlorination. Chlorine
dioxide is believed to be capable of oxidizing certain organic
structures which chlorine cannot. The present data are incon-
clusive; an investigation conducted over a whole year is needed
to indicate possible seasonal effects. Also the effect of CIO
treatment needs to be more closely studied.
The PAC also removes chlorine from water. This effect is
considered of secondary importance in Belgium since no objective
for research on dechlorination in treatment plants currently
173
-------
exists. From experience in the Tailfer plant, chlorine dioxide
appears much more stable than chlorine in the presence of PAC
under similar conditions. Laboratory experiments have indicated
that the reaction proceeds by at least two consecutive steps; the
integrated rate expression may be described by an exponential
function (1).
Percent Remaining
b x e
-mt
The half-life time of CIO. in the presence of PAC is about
ten times greater than that of (free) chlorine applied under the
same conditions.
PILOT EXPERIMENTS
There is no other experience at full scale with GAC or PAC
in Belgium. The method, however, receives considerable attention
at pilot-plant scale.
When considering overall quality improvement, the removal of
TOC constitutes one of the most critical criteria. At filtration
velocities of 15 to 20 m/hr, the thickness of the GAC layer must
be at least 2 to 3 m in order to obtain a satisfactory removal of
TOC.
Experiments with a column of 2.5 m Chemviron 400 carbon
(see also Table 1) indicated that for an extended service time of
several months, the level of abatement of TOC or THM remains
limited (2) (Table 5).
Table 5. TOC and THM Abatement
TOC Abatement
(mg/m )
1
1.25
1.5
1.75
2.0
—
—
_
—
—
THM Abatement
(m Mol/m3)
_
-
—
-
—
0.1
0.15
0.2
0.25
0.3
Service Time
(months )
8.6
7.2
6.0
4.8
2.9
6.2
4.1
3.1
2.5
2.1
174
-------
Pilot experiments have been carried out at the Havre treat-
ment works (also managed by CIBE) to help in the design of a
biologically regenerable GAC filter eliminating simple phenols
(including cresols) and monocyclic aromatic hydrocarbons (toluene
and xylene). These compounds had polluted a groundwater catch-
ment. The concentration of these organicsr which previously
ranged from 50 to 100 mg/m , has been lowered to less than 5
mg/m , usually 1 mg/m . The carbon used was Norit ROW 0.8 supra,
at a surface loading of 0.72 m/hr, during 6 to 10 months of
operation.
Other characteristics of the raw water were: oxygen, 1 g/m ;
iron, 150 to_200 g/m ; COD, 4 to 5 g/m ; TOC, 3 g/m ; and less
than 50 mg/m ammonia.
In-depth aeration was attempted by using hydrogen peroxide
in association with the GAC filter. Typical data for the oxygen
content (in g/m ) are given in Figure 2.
The final population of bacteria (on the surface of the
activated carbon) in the presence of-H^^ reaches a stationary
state at about one organism per 4 mm . Even at this rather low
count (populations up to one bacterium per 40 m have been
reported), the filter remained sufficiently active in biological
degradation.
Bacterial concentrations in the filtered water frcrc the pilot'- plant,
using adsorption and biological treatanent, have varied, from 50 to 45,000
total counts per ml. No systematic explanation has been given yet, hut the
point illustrates the need for an accurate post-disinfection after GAC filtration.
No treatment combining ozonation v?ith adsorption on GAC exists at
present in Belgium; however, the technique is considered to be promisina
for the future.
OTHER TECHNIQUES
The CIBE has two full-scale plants using diatomaceous earth
in occasional operation. One plant has a capacity of 0.2 m /sec
for final polishing filtration of water coming.from the River
Meuse. Another plant has a capacity of 0.12 m /sec for treatment
of raw water that is temporarily stored in open reservoirs to
meet potential peak consumption. In both cases, the adsorption
effects on organics (COD, TOC, detergents) are negligible.
COSTS
It appears that in BeIgiurn,under most conditions, the problems
related to organics can still be soTved by the use of PAC.
175
-------
10
g/m3 \t±
o
PSI
I
n
CN
o
0
-1
g/m" OUT
-6
100 —
100 —
L
M^NVM
J
BACTERIA/cm3 (OUT) AT PLANT
Figure 2. Bacteria Counts in the Effluent as a Function
of Oxygen Presence
176
-------
However, there is growing interest in the use of GAC, particu-
larly if, in the future, more polluted resources in central and
northern Belgium must be exploited.
On the Belgian market, prices are respectively 15 BF/kg for
PAC and 55 BF/kg for regenerable GAC (May 1979 exchange rate:
30 BF to 1 $U.S.).
The estimated regeneration loss is 10 percent per cycle,
with two regenerations per year in rapid filtration. The cost of
carbon losses is 11 BF/kg/year and the cost of regeneration
(thermal) is equal to 10 BF/kg/year. Total estimated regenera-
tion costs are 31 BF/kg/year.
Example: To treat 0.28 m /s (1000 m /h) at a filtration
velocity of 10 m/hr and through^a GAC bed depth of 2 m, the total
surface of the filters is 100 m . Hence about 80,000 kg GAC
could be necessary.
Annual charges are:
carbon 55 BF/kg x 80,000 kg x 0.1 = 440,000
(investment)
filters: (1500 BF/m3/d) 27,000,000 x 0.1 = 2,700,000
carbon losses and regeneration costs
(includes costs of emptying and
filling carbon beds) - 31 BF x 80,000 = 2,480,000
Total, BF/year (This includes power and main-
tenance costs for regeneration, but does not
include operational costs; i.e., backwashing
and power costs for filtration) = 5,620,000
At a net production of 8,500,000 m /yr, these estimated
costs amount to 0.66 BF/m or $0.022 US/m . If only one annual
regeneration is necessary, this production cost amounts to 0.52
BF/mJ or $0.017 US/m .
On a comparative basis, the cost of using powdered activated
carbon is less than 2,000,000 BF corresponding to a nominal pro-
duction cost of less.than 0.025 BF/m . The cost of using PAC
amounts to 0.15 BF/m or $.005 US/m at a dose of 10 g/m , and is
approximately proportional to the dose.
17?
-------
Various Impacts on Production Costs
Besides the above factors, other parameters have an effect
on the costs:
The average peak factor in Belgium, that is maximum day con-
sumption to average day, is on the order of 1.30 to 1.35. This
can involve, in local circumstances when only considering treated
surface water, an increase in the above mentioned production
costs of 30 percent,_or as in the example given, an additional
expense of 0.12 BF/nr or $.004 US/m .
Labor for PAC dosing in the Tailfer station, operated at
130,000 m /day,-amounts to less than 0.5 man/day/week or less
than 0.003 BF/m water produced. Estimates for the use of GAC,_
regenerated twice a year, indicate three working days/year/20 m
GAC. Assuming filtration at 10 m/hr, corresponding labor costs
amount to 0.016 BF/m or $.005 US/m .
The use of GAC necessitates a transfer procedure. During
these periods, the production must continue. Hence, a supple-
mentary filter or storage area may be necessary. The minimum
required volume is 20 m , or an additional investment of about
2,700,000 BF, or an annuity of 270,000 BF/yr to be imputed with
the production (respectively $90,000 and $9,000 US).
CONCLUSION
Belgium relies on its groundwater resources for about
80 percent of its water supply.
PAC is used in the Tailfer plant of the CIBE to improve
color, sometimes taste and odor,and, in conjunction with CIO ,
to reduce trihalomethane formation potential. The level of cri-
halomethanes found until now does not justify the introduction of
GAC in most plants; however, the technique is being evaluated
seriously.
Several pilot plants are operated to investigate the use
of GAC for abatement of TOC and THM, and as biologically active
filters. Examination of representative small scale columns is
necessary to shorten the delay until a conclusion can be reached.
On a comparative basis, estimated costs in Belgium are in
the ratio of 1/0.2 to 1/0.5 for regenerated GAC compared to PAC.
178
-------
REFERENCES
1. Masschelein, W.J., R. Goossens and G. Minon, Tijdschrift
BECEWA 49, 1 (1979).
2. Masschelein, W.J., R. Germonpre and F. van Menxel, Communi-
cation at the IWSA Symposium Brussels; Ed Pergamon Press
(1979).
179
-------
THE USE OF GAC FILTRATION TO ENSURE QUALITY
IN DRINKING WATER FROM SURFACE SOURCES- Swiss Experiences
Maarten Schalekamp
The lake waterworks in Switzerland use granular activated
carbon (GAC) filtration, Respite the relatively high quality of
their water sources, because of the proximity of oil pipelines to
drinking water storage basins and the occurrence of two phenol
spills in the St. Gallen and Zurich drinking water supplies,
respectively.
When the oil pipeline from Genoa to Ingolstadt was laid
adjacent to Lake Constance and its surroundings, the possibility
of oil leakage had to be taken into account. The consequences
of oil leakage into a drinking water storage basin could be
disastrous. As a result of the efforts of several waterworks,
the Energia Nationale Italiana (ENI) financed the installation
of ozone or GAC treatment facilities for all the waterworks in
the critical area. Thus, water contaminated by slight amounts of
oil can be purified for potable use.
The need for GAC filtration was further demonstrated by
the phenol spills that contaminated the water supplies of the
cities of St. Gallen in 1957 and Zurich in 1967. On both
occasions, the drinking water supplies were highly contaminated
and their tastes and odors were adversely affected.
These two events led to the installation of a GAC filtra-
tion system at the waterworks of St. Gallen by Lake Constance.
The phenol spill in Zurich had consequences so severe that it
provided sufficient impetus for the installation of a GAC
filtration step in the Lengg and Moos treatment plants. In
most of the other lake waterworks in Switzerland, the sudden
appearance of the floating mussel Dreissena polymorpha pallas
(DPP) made necessary chlorination and subsequent GAC filtration
for dechlorination.
The consumption of oil and gasoline has increased sevenfold
during the past few years-J therefore, the organic load of the
rivers also has increased-. It has been reported that many or-
ganic substances found in the lake waters are almost completely
absent from pure well water (Figure 1). Investigations carried
out in the laboratories of the Zurich waterworks confirmed the
presence of such substances in treated water drawn from Lake
180
-------
Zurich (Figure 2). Since each Swiss waterworks must ensure that
the quality of the drinking water obtained from surface water is
in no way inferior to the quality of water found in well and
groundwater, particularly with respect to the load of organic
substances, the need for purification of surface water by GAC
filtration is manifest.
Well Water from Bolder, July 5, 1973
L-.lJ IJ ^JlJLj * ... Jl- . .1 ^ l_
1
,
.23 21
26
Surface Water from Lake Zurich, July 7, 1973
Figure 1. Organic Substances in Well and Zurich Lake Water
(Each numbered peak represents an identified
organic substance.)
DECKLORINATION EFFICIENCY OF GAC
Several waterworks in Switzerland use chlorine to prevent
DPP from attacking raw water pipelines. The chlorine is then
removed through GAC filtration. Experiments have shown that
a concentration of 0.5 mg/1 chlorine remains after a dose of
1.0 mg/1 Cl has been added to the raw water catchment
installation, completely eliminating the DPP larvae within
14 days (Figure 3).
This amount of chlorine is sufficient since it has been
shown that the larvae remain in the rapid filter for up to
15 days. Because a residual chlorine concentration of only
0.05 mg/1 in drinking water is permitted in Switzerland, the
excess chlorine must be removed from the water.
Although the chlorine could be neutralized with thio-
sulfate, GAC was selected instead as a safety measure. The GAC
is applied either as a second upper layer in the rapid filters,
181
-------
0
200
I
5
o
200
9X>
17
fl
," •..
•Li *
33
26
26
40
. A
19
23
I.lie
33
53
"
"*
A
65
82
80
After Ozonation
60
65
82
80
(1 TO7L 7778808'8
•Mis?. S? ?..:i.e.2
Before Ozonation
33
26
26
3>
65
40
48
47
49
V
h
82
80
74
-73
75777?
83
64
tl
Figure 2. Modification of Organic Matter
During Water Treatment
182
-------
Figure 3. Elimination of Dreissena Polymorpha
Pallas Larvae in Raw Water
Figure 4. Dechlorination Half-Values of Various
Types of GAG at Different Filtration Rates
1 10
I
£
£
& 5
. GAC With Raw Water After Three Months Plus CI?
_ GAC With Rapid Filtrate
Plus Cl» - Oj S / v = 7 m hr
2 3
Months
Figure 5. Decrease in Declorination Efficiency
of Norit 07 GAC at Different Filtration Rates
183
-------
as a 5 to 10 cm thick layer above the sand in the slow filters,
or as a top layer in GAC filters.
To select the most suitable type of GAC, experiments were
conducted to test the different types of GAC for dechlorination
at the velocities of the slow as well as the rapid filters.
All the carbon types were suitable for the slow filters. For
the rapid filters, however, Hydraffin LW extra, Sutcliffe,
Norit 07, and Pittsburgh 200 were most suitable (Figure 4).
Since the effectiveness of dechlorination decreases with time,
different types of GAC were tested for several months at dif-
ferent velocities (Figure 5).
It was also discovered that the efficiency of dechlorina-
tion depends upon the grain size of the GAC: the smaller the
grain size, the higher the efficiency. GAC also was employed
to eliminate any ozone remaining in the treated water.
DOUBLE-LAYER RAPID AND SLOW FILTERS WITH GAC
Additional experiments were performed with one-layer and
double-layer rapid filters. The double-layer filters, consist-
ing of 60 cm quartz sand and 20 cm GAC, were compared with a
one-layer filter filled only with 80 cm quartz sand.
The operating time was five times longer for the
double-layer GAC filters compared to the single-layer filter
(Figure 6), with the cleaning efficiency being equal. The
pressure loss takes place in the GAC layer (Figure 7). The
longer or shorter operating time for the double-layer filter
is also dependent upon the grain size of the GAC. After
intensive air rinsing, it was possible to rigorously separate
the quartz sand and the GAC with a backwash water velocity of
only 15 m /m /hr.
Until the completion of the GAC filters, most of the lake
waterworks in Switzerland augmented their rapid filters with GAC
for dechlorination. In Zurich, dechlorination was accomplished
with the slow filter, with 10 cm of GAC as the top layer, until
the completion of the GAC filter. This modification resulted
in operating times at least five times better than those of the
single-layer slow filters (Figure 8).
A new cleaning process with cyclic and continuous systems
was developed (Figures 9, 10) eliminating the necessity of
removing the GAC during the backwashing process. The advantage
of this new process is that the slow filters no longer need to
be shut down.
Although the results of using the GAC layer on slow filters
were excellent, the recharge basins of the new Hardhof plant
were covered with 10 cm of GAC.
184
-------
V - 3.75 m/hr
Figure 6. Operating Times of Single- and
Double-Layer GAC Rapid Filters
28 30 Z 4 6 8 10 12 14 16
Q = 5000 m*/hr; V = 5.6 m/etay
Figure 7. Pressure Loss in Each Layer of a
Double-Layer GAC Rapid Filter
90 cm S« nd
90cmSind •••
cmNonlPKST
0 5 10 15 JO 25 30 35 40 «5
V = 75 m/day
Figure 8. Operating Times of Single- and
Double-Layer Slow Filters as a
Function of Loss of Pressure
185
-------
•_ BccKwishing Walar
-»• Sludge
Figure 9. Slow Filter Cleaning Instrument (Cyclic System)
Bachwashmg Wittr
To Canalisation
Figure 10. :Slow Filter Cleaning Instrument (Continuous System)
186
-------
THE REMOVAL OF ORGANIC SUBSTANCES BY THE GAC FILTER
Experiments in Zurich showed that, depending on the kind of
GAC, up to 75 g of organic substances can be removed from water
per kg of GAC. These substances are referred to as the sum of
dimethylformamide (DMF) plus dioxane extracts.
The removal of organic substances by several types of GAC
was investigated in Zurich. Table 1 shows the types of GAC
tested and the amount of organic substances removed from the
raw water of Lake Zurich. The best removal efficiencies were
achieved with Pittsburgh F400 and Lurgi LS-Supra.
Table 1. Cumulative Removal for Various Types
of GAC for Zurich Lake Raw Water
Carbon Type
Pituburflt FWO
Norit TOST
Uirft IS-Supra
Norn 07
Uiiji LW .
Pttuburih ROD
NoHlH
UcM
Experiment 1*
Cumulative
R.mov.U
«'k«
•1
41
X
21
a
Experiment It
Cumul>n»e
Removilt
«'k«
•1
41
«
2.
M
M
X
'Duration nine month*
tOwralion Ihrci month*)
tMtuund M turn of DMF plu* dioittnc t*tr«cit.
A new GAC, ROW 08 Supra from Norit, was tested and compared
with Pittsburgh F400. Experiments in larger plants had already
confirmed that the ROW 08 was equal to the Pittsburgh F400 after
1 month (Table 2). This was further substantiated by long-term
testing.
Table 2. Load Comparison of Various Types of GAC*
Carbon
Type
ROW 08 tupra
ROW 06 lupra M
F400
DMF Extract
g/kg CAC
Total
Load
5.5
4.1
6.1
New
GAC
1.8
1.6
3.7
Effective
Load
3.9
2.5
2.4
DMF Extract Extinction at
300 nm (Light Path - 1 cm)
Total
Load
0.062
0.093
0.121
New
GAC
0.044
0.044
0.109
Effective
Load
0.018
0049
0.012
•34 dayi: throughput-4.5 m'/kg CAC
Pittsburgh F400 carbon was used later in the GAC filtration
process of the Lengg waterworks. After a 2-month operating
period, a load of 37 g/kg in the upper GAC layer was attained.
The fresh carbon showed a load of 4 g/kg DMF + dioxane extract,
increasing after 3 months of operation to 63.5 g/kg, and
attaining the highest capacity of 68.5 g/kg after 7 months
(Figure 11).
The slow filter in the Moos waterworks was supplemented
with a 5-cm layer of PKST GAC. This new carbon initially
adsorbed an organic load of 5 g/kg; after 3 years the load
was 29.2 g/kg (Figure 12). The effectiveness of this carbon
187
-------
UV spectra of DMF extracts; DMF + dioxane
extract
Figure 11.
GAC Loads Over Time in the Upper
Layer of the Filter
300 350
Wave Length — nm
UV spectra of DMF extracts
Figure 12. GAC Loads Over Time in a Slow Filter
After GAC Step
After Ozone and GAC Steps
UV = 254 nm; light path = 1 cm
Figure 13.
UV Extinction Over Time After GAC
Filtration and After Ozone Treatment
Plus GAC Filtration
188
-------
in removing organic substances (i.e.f UV extinction) in the
treated water remained constant over a 3-year period.
Thus, the construction of a separate GAC filter for the
recharge of groundwater at the new Hardhof waterworks was not
necessary because the recharge basins performed with equal
efficiency when they were augmented with GAC. The operating
time of the recharge basins also increased by five times.
The operating time of GAC with Swiss lake water is only
6 months if the ozonation process is not used. When the ozona-
tion process is placed before GAC filtration, the GAC can be
used for about 3 years (Figure 13), thereby increasing the GAC
operating time sixfold at a significant cost savings since the
GAC does not have to be reactivated as often. If halogens are
present in the water, they will be more efficiently removed by
ozone than by GAC. Figure 14 shows an 80 percent decrease in
chloroform following ozone treatment and a 33 percent further
decrease after GAC.
Figure 15 illustrates the increase of the most important
aldehydes through ozonation. In plants without a GAC stage,
the continued oxidation of the aldehydes causes the formation of
acids that promote the reproduction of bacteria. To prevent
this problem, GAC filtration is used immediately after ozonation.
There are three situations when the GAC should be changed
or regenerated: first, when the halogen level in the drinking
water approaches the maximum level permitted, at 25>ug/l (Swiss
drinking water normally contains only 10 to 15>Ug/l halogens
(Figure 16)); second, when the UV extinction does not decrease
further after the GAC filtration step (Figure 17); and third,
when the load of DMF + dioxane extract in the upper layer of GAC
is very near the maximum possible and that in the bottom layer
is 70 percent of the maximum possible load.
Figure 18 shows that with the GAC ROW 08 Supra, the maximum
load is 40 g/kg. The bottom layer is regenerated at 28 g/kg,
or 70 percent. With the GAC Pittsburgh F400 the maximum load
is 75 g/kg, and regeneration is done if the bottom layer is
52 g/kg, or 70 percent.
TESTS AND MONITORING
In view of the danger to the consumer's health presented
by certain organic pollutants, it is imperative that water
suppliers know instantly when changes occur in the drinking
water, as monitored by biological and physio-chemical tests.
Nonspecific chemical tests such as oxidizability, TOC, DOC,
DOC1 (dissolved organic chlorine), and DOC1-N (dissolved organic
chlorine-nonpolar, measured in the liquid-liquid extract by
microcoulometric technique) are slow and laborious to carry
189
-------
§
•H
4J
R)
O
•i-l
U-l
•H
J-l
IV
•o
•o
I
u
.
baa
Figure 14.
Decrease of Halogens After Ozona-
tion and After GAC Filtraton
0
1000
i
o
I
i
Alter GAC Filtration
- •
Alter Ozonitton
4f
-
n
B
nndl till
40
i ,|
n
S3
?6 Heptinal
40Oc>anai
48 Nonanal
S3 Decanal
1 nntln
Bttote Ozonltion
.. /. .. _ nJHB _^lnB ^ninll
25 5
Ozone Level — mg/L
Figure 15. Aldehydes Before and After Ozonation
at Different Ozone Concentrations and
After GAC Filtration
190
-------
European limit—25
II
II
II
II
II
II
II
P!
jf' 'x
s^
'£•'&?•
>':'••£
'";-•''<}
;V:iS
III
II
II
II
II
II
II
II
II
II
M
M
S
9
O
Si
I!
Figure 16. Halogens in Raw and Finished Water
from Lake Zurich Supplies
Yearly average 1976-1979
-^«
;;•<»
:'";
>'tl!
^
1 i! 1
< >£ d
£ £o £
~1
s
!
K
55
1
r— i
«
S
1
Timfllo
£
>
1,
•»
2 :
on (1 m;2S4 ni
i
IU
- ?
ff
Figure 17. Variation of UV-Extinction Levels
During Water Treatment
191
-------
1.2-m GAC layer
I «
i
S 30
I "
I ,0
i •
—— Upper Lly«r GAC
Bottom Liver GAC
Marctl Mirctl Mlrch March
1»76 1177 1(78 1971
Figure 18. Loads of Upper and Bottom GAC
Layers as a Function of Time
10'
1 1C'
5
i '°'
I ">'
* 10'
Figure 19.
Bacterial Count After Three Days at
20°C in 1 nl of Chlorinated Rapid
Filtrate and Various GAC Filtrates
192
-------
out and are not continuous. Besides these tests, an instrument
for continuously monitoring organic matter in raw water, ground-
water, and finished water is employed. It measures UV extinc-
tion, which enables organic pollution to be detected immediately.
With this instrument, it is also possible to monitor the amounts
of organic matter during the treatment of drinking water. The
results obtained by UV extinction correlate satisfactorily with
those given by DOC or TOC, and they are available instantaneously
in the Hardhof control center.
BACTERIAL REGROWTH
Experiments in pilot plants in Zurich have proved that GAC
treatment leads to bacterial regrowth. Figure 19 shows that
the Norit PKST carbon initially had the highest bacterial count.
The bacterial count of the Pittsburgh F400 was about half that
of Norit PKST and the Lurgi LS-Supra count was a little lower
than the Pittsburgh F400. The Sutcliffe carbon showed the best
results. Nearly all the counts for the GAC types tested were
equally high, except for the Sutcliffe carbon. Its bacterial
count (343 colonies/ml) was only a third that of the others.
A close relationship exists between the reduction of organic
substances and bacterial regrowth, especially for PKST GAC.
Initially, PKST was the most effective GAC for removal of organic
substances, but its use also resulted in the highest regrowth
count. As shown by experiment 2 (Figure 20), with monthly back-
washing all carbon types already exhibited a 5-digit bacterial
count after 1 month of operation. The weekly backwashing of
Sutcliffe 206A and Norit PKST carbons reduced the bacteria to a
4-digit count, and after 5 weeks they had a 60 percent recur-
rence of bacteria. It must be concluded that the carbons have
a significantly lower reinfection when the GAC filter is more
regularly backwashed. Full-scale tests with rapid filters at
the Lengg waterworks confirmed this hypothesis.
In previous experiments, Lurgi LS-Supra and Norit PKST pro-
duced no regrowth; however, they were fed with chlorinated water.
The new GAC filter in Lengg, which is backwashed twice a week,
has never produced a count of more than 10 to 50 colonies/ml.
These are extraordinary results (Figure 21).
REACTIVATION
The reactivation is done by Norit's fluid-bed process.
The GAC must be either replaced or reactivated every 2 years.
The fluid-bed process was selected because it ensures the lowest
loss of carbon of all processes currently available. The volume
loss from filter to furnace and back to the filter totals only
3.5 percent (Figure 22).
193
-------
10'
Backwasning Montnly
, /\~^-/
2» 4
Mlrcn
11 14
Aoril
25 21 2
Mly
Figure 20.
Bacterial Count After Three Days at
20°C in 1 ml of Ozonized Slow Filtrate
and Various GAC Filtrates
Jul Aug S*f Oct
Figure 21. Bacteria] Count After Three Days at
20°C in 1 ™l of GAC Filtrate
194
-------
Nott Water Througt-put—271 m3/kg
N
-------
The author was present when Norit ROW 08 Supra from Duisburg
and PKST 1-3 from the Zurich waterworks were reactivated
(Table 3). Later, a low density GAG, Norit PKST 1-2, was reac-
tivated in the same furnace. Although this carbon is not usually
suitable for reactivation, the losses were only 50 percent of
those incurred with common processes. The effectiveness of the
reactivation was excellent. The PKST carbon had a small load of
only 28 g/kg (Figure 23). Whereas fresh carbon had a load of
5 g/kg, the value for the reactivated carbon was slightly higher
(8 g/kg).
Table 3. Reactivation Efficiency of Two Types of GAG*
Carbon
Type
ROW 08 supra
PKST 1-3
DMF Extract
g/kg CAC
Total
Load
28.5
20.9
New
GAC
1.8
2.2
Reactivated
GAC
3.1
4.5
DMF Extract Extinction at
300 nm (Light Path - 1 cm)
Total
Load
1.576
1.055
New
GAC
0.044
0045
Reactivated
GAC
0.066
0.057
•Reactivated in the Norit installation Mar. 26, 1974
fThroughput—3800 m'/kg GAC
This kind of reactivation plant was installed in the Lengg
waterworks and was put into operation at the beginning of 1976.
The reactivation of the GAC was excellent. The used Pittsburgh
F400 had a total load of 32.1 g/kg dioxane + DMF; the new GAC,
6.9 g/kg; and the reactivated GAC, 6.2 g/kg. The reactivated
GAC was better than the new GAC at a furnace capacity of
110 kg/hr. The ROW 08 had a total load of 22.1 gAg dioxane +
DMF; the new GAC, 2.3 g/kg; and the reactivated GAC, 5.8 gAg-
The reactivated ROW 08 GAC4was not as efficient as the new GAC,
but was as good as the F400. With a furnace capacity of 110
kg/hr the reactivated GAC was also better than the new GAC
(Table 4).
Table 4. Effectiveness of the Lengg Reactivation
Furnace for Two Types of GAC
Carbon
Type
F400'
KOWOBt
Oioxane Extract
g/kg CAC
Total
Load
7.8
5.4
New
GAC
2.9
1.0
Reactivated
CAC
1.8
2.7
DMF Extract
g/kg CAC
Total
Load
24.3
16.7
New
GAC
4.0
1.3
Reactivated
CAC
4.4
3.1
Total
g/kg GAC
Total
Load
32.1
22.1
New
CAC
6.9
2.3
Reacliv.itL-d
CAC
6.2
5.1)
•Throughpul-ial m'/kg GAC: furnace capacity-110 and 130 kg/hr
tThroughput-271 m'/kg CAC; furnace capacity-130 kg/hr
196
-------
CONSTRUCTION OF THE GAC FILTERS IN ZURICH
In Lengg, 12 GAC filters are in use, six on each side of the
pumping station. The filters are closed pressure filters of
smooth-faced concrete with an area of 44 m each (Figure 241.
The bottom of the filters contain 75 plastic nozzles per m .
The filter bed consists of a 50-cm layer of quartz sand, grain
size 0.7 to 1 mm, and a 1.2-m layer of Pittsburgh F400 GAC above.
Because this carbon exhibited a very high filtration resistance,
it was replaced during the reactivation by Norit ROW 08 Supra
which resulted in considerably lower pressure loss. Neverthe-
less, the F400 carbon has very good adsorptive characteristics
and is still used in the Moos waterworks where pressure losses
cause no problem.
The cleaning of the GAC filters is accomplished by back-
washing, first with air and then with the same type of treated
water as that used for the rapid filters. The backwashing
velocity, is 25 m /m /hr; if necessary, it can be increased to
50 m /m /hr. The normal velocity of the filter is 21.5 m /m /hr
when one of the 12 GAC filters is backwashing. For reactivation,
the GAC can be transported hydraulically to their respective
reactivation plant silos.
In the construction of the filters, measurements must be
taken to ensure that enough free space is available for the
expansion of the carbons, which is often greater than the esti-
mates of the supplier (Figure 25). The construction of recharge
basins as two-layer filters with 10 cm of GAC can be seen in
Figure 26.
THE COSTS OF GAC FILTRATION AND REACTIVATION
The construction costs for the 12 GAC filters with a total
area of 528 m and a daily output of 250,000 m are:*
Buildings $2,749,970
Technical installations $3,014,030
Consolidation of the filter $ 120,370
Total $5,884,370
The costs of the reactivation plant are not included in this
list. The capital costs per cubic meter daily capacity are
therefore $22. The whole lake water treatment in Lengg, includ-
ing the pumping station, will cost $220 per daily cubic meter.
Ten percent of the total capital costs must be spent on the GAC
filtration system.
*Based upon an exchange rate of $1 = 1.67 Swiss francs.
197
-------
Channel to GAC Filler Garner Water Cnai
Pressure Release Line
Rinsing Air Line
Channel From GAC Filter
Water
Channel
Nozzles Plate I
Quartz Sand —-*
Figure 24. Profile and Cross Section of GAC Filter
198
-------
I '.
tu
I
§
£
Multilayer Thickr
16 cm LW Extra
45cmPKST
Measured Expansion
Expansion Calculated From
Data Provided by Supplier
25 50
Flushing Rate — m/hr
Figure 25. Expansion of GAG Layers as a Function
of the Backwashing Rate
Figure 26.
Composition of the Recharge Basin
at Hardhof Plant
199
-------
The operating costs for interest and amortization (25 years
at 6 percent interest) are $560,000 per year. An additional
$80,000 should be included as annual maintenance costs. With an
annual output of 40 x 10 m , the capital investment costs are
$0.016/m . The operating costs for the annual reactivation are:*
Capital costs $106,670
Maintenance $ 26,670
Heating $ 21,330
Service $ 21,330
Loss GAC at 5 percent
(effectively only
3.5 percent) $ 37,340
Annual total operation costs $213,340
With an output of 40 x 10 m /year, the operation costs for the
reactivation, including the loss of GAC, are $0.0055/m . The
total costs will be $0.02l5/m . With an annual output of
40 x 10.? m , the treatment costs of the Lengg waterworks total
$0.17/m . Thus the costs for GAC filtration come to only
12.5 percent of the total operating costs.
To ensure that the quality of drinking water derived from
surface water is not inferior to that of well or groundwater,
GAC filtration is indispensable. At a cost of one eighth of the
total treatment costs, the use of GAC is also economically
feasible.
*Construction costs of the reactivation plant and transfer
installations = $560,000 (10 percent amortization, 6 percent
interest).
200
-------
CURRENT UNITED KINGDOM PRACTICE IN THE
USE OF GRANULAR ACTIVATED CARBON
IN DRINKING WATER TREATMENT
J.B. Goodall
and
R.A. Hyde
INTRODUCTION
Water suppliers in the United Kingdom are legally required
to supply water which is wholesome - this is generally taken to
mean water which is clear, palatable, and safe.
The facility for using activated carbon is available in
approximately 100 potable water treatment works throughout the
country. In most applications, activated carbon is used for
controlling unpleasant taste and odor, particularly the earthy/
musty taste and odor that occur as a result of the presence of
actinomycetes and algal matter in water. Because these taste and
odor problems are seasonal and often short lived, most water
undertakings find it more economical to use powdered rather than
granular activated carbon.
At present, the United Kingdom has four water treatment
works employing granular activated carbon, three of which use the
carbon after rapid filters and one where the carbon has partially
replaced the media in the rapid filters. One of the treatment
works employs granular activated carbon to remove "chemical"
taste and odor, the remaining three are for earthy/musty taste
and odor control. Only one of the treatment works has regenera-
tion facilities on site.
Research on the use of granular activated carbon for
synthetic organic control in drinking water treatment is being
undertaken by the Water Research Center and a number of water
suppliers.
Bough Beech (East Surrey Water Company)
Water from the unpolluted Eden River is pumped into a 9.1 x
10 m (2 x 10 gallons; 2.4 x 10 U.S. gallons) reservoir from
October to April each year. Water from the reservoir that feeds
the carbon filters is prechlorinated to maintain 1 g/m free
201
-------
chlorine in the water and then split into two parallel streams
(Figure 1); two thirds of the water is clarified by coagulation
with aluminum sulphate at pH 7.1 and floe blanket sedimentation.
The remaining one third of the water is softened and coagulated
by addition of sodium hydroxide and aluminum sulphate at pH 11
and clarified by floe blanket sedimentation. The two clarified
streams are combined, resulting in a blended pH of 8 to 8.3,
filtered through five rapid gravity sand filters, then pumped
through four horizontal pressure filters containing granular
activated carbon, rechlorinated, and stored prior to distribution.
During the summer months 27.3 x 10 m /d (6 mgd; 7.2 U.S. mgd)
of water is treated; in the winter this is reduced to 13.6 x 10
nr/cl.
The Eden River is the only surface water treated by the East
Surrey Water Company and treatment with activated carbon was con-
sidered necessary to avoid taste and odor complaints from custo-
mers accustomed to water supplied from deep borings in the chalk
and Lower Greensand. Consequently, it was decided that the
treatment works, commissioned in late 1970, should contain
granular activated carbon filters.
The activated carbon horizontal pressure filters are 2.44 m
(8 ft) in diameter by 9.14 m (30 ft) long and contain 0.91 m (3 ft)
of media. Initially, the bed was 0.69 m (2 ft 3 in) of 0.3 to
0.85 mm (18/52 BS mesh) granular activated carbon on 0.23 m (9
in) of sand and was later changed to 0.84 m (2 ft 9 in) of carbon
on 0.08 m (3 in) of sand. At present, the filters contain 0.76 m
(2 ft 6 in) of 0.3 to 0.85 mm (18/52 BS mesh) carbon on 0.15 m
(6 in) of 0.7 to 2.8 mm (6/22 BS mesh) carbon. The filters are
cleaned once every 2 days by air scour followed by backwashing.
Under summer treatment conditions, the carbon filters are
operated at a downflow velocity of 12.7 m/h (4.3 gal/min/ft :
5.2 U.S. gal/min/ft ) and an empty bed contact time of 4.3
minutes. In the winter the treatment rate is halved and conse-
quently the contact time is doubled.
To date, one coal-based and two different coconut-based
activated carbons have been used; the approximate t>ed life for
taste and odor control was 2 yeats for the coconut-based carbons
and somewhat less for the coal-based carbon. The granular acti-
vated carbon ceased to remove organics (monitored as permanganate
value) after only a few days of operation. Every spring,
exhausted activated carbon from two of the four adsorbers is re-
plenished with virgin and/or regenerated activated carbon.
During the last cycle of operation (2 years of treating water,
mechanical handling and carbon transfer, carbon regeneration) it
was estimated that 40 to 50 percent of the carbon was lost.
202
-------
Prcchlorination
(up to 6 mg/1)
Stored
River Eden water
pH 7,
Floe
Blanket
Clarification
Aluminium
Sulphate
pH 8 - 8.3
nH 1]
Sodivm Hydroxide
Softening &
Floe Blanket
Clarification
Ranid Gravity
Sand
Filtration
Horizontal Pressure
Granular Activated
Carbon Adsorption
Chlorination
Figure 1. Schematic Diagram of the Bough Beech
Treatment Works
203
-------
Church Wilne (Severn Trent Water Authority)
The Derwent River, a tributary of the Trent River, at Church
Wilne contains the industrial and municipal-effluents from the
city of Derby. At a flow rate of 22.7 x 10Jm /d (5 mgd;
6 U.S. mgd), stored Derwent River water is treated by prechlori-
nation (to maintain a residual of 0.05 to 0.1 g/m through the
carbon filters), coagulation with chlorinated copperas, floe
blanket clarification, rapid gravity filtration, granular acti-
vated carbon adsorption, and disinfection with chlorine dioxide
(Figure 2). This works has the only on-site regeneration furnace,
at a potable water treatment plant, in the United Kingdom.
Details of the design and operation of the granular activated
carbon adsorbers and the regeneration furnace are described
elsewhere (1, 2). The granular activated carbon filters are used
for controlling unpleasant taste and odor that results from
industrial and municipal effluent rontamination of the source
water. Prior to the commissioning»in 1975, of the granular
activafed carbon filters, up to 40 g/m of powdered activated
carbon w.eJ~_e needed for effective taste and odor control. Granular
activated carbon was selected as the most suitable method of
using activated carbon after pilot trials had shown that if
powdered carbon was.used, continuous dosing would be necessary.
A minimum of 20 g/m powdered carbon would be required and, under
the most severe conditions, doses -of 90 g/m were not adequate.
Trials also showed that oxidation using chlorine dioxide or ozone
was ineffective for taste and odor control and, in fact, the use
of potassium permanganate increased the problem.
The coal based granular activated'carbon is contained in
twelve 4.11 m (13 ft 6 in) diameter by 3.75 m (12 ft 3 in) high
vertical pressure filters operating in parallel. Initially, the
carbon bed depth was 1.53 m (5 ft); at a later stage this was
increased to 2.28 m (7 ft 6 in), but due to backwashing problems
(carbon loss at 50 percent bed expansion) this had to be reduced
to the present bed depth of 2 m (6 ft 6 in). -The carbon
adsorbers are operated at 6 m/h (2 gal/min/ft ; 2.4 U.S. gal/
min/ft ) down-flow velocity and have an empty bed contact time
of 20 minutes. The beds are backwashed daily, with water only,
to alleviate compaction of the bed.
The approximate bed life for controlling the "chemical"
taste and odor is 6 months. One adsorber is taken off line at a
time and the carbon transported as a slurry to the multi-hearth
furnace for regeneration. Regeneration of activated carbon
takes 1 week per adsorber, resulting in a 4 to 5 percent loss of
carbon, and achieves a 95 to 98 percent recovery of activity.
FOXCOTE (ANGLIAN WATER AUTHORITY)
This 9.1 x 10 m3/<3 (2 mgd; 2.4 U.S. mgd) treatment works
can receive water from three separate sources: a gravel pit;
204
-------
Prechlorinrihion
Stored
River Derwent water
Chlorinated
Copperas
Floe Blanket
Clarification
Rapid Gravity
Filtration
Granular
Activated Carbon
Adsorption
Chlorine Dioxide
Figure 2. Schematic Diagram of the Church Wilne
Treatment Works
205
-------
direct abstraction from the Great Ouse River; or from a 614 x
10 m reservoir. Treatment is breakpoint chlorination (to main-
tain 0.6 to 0.7 g/m free chlorine onto the carbon filters);
coagulation with ferric or aluminum sulphate; floe blanket clari-
fication; rapid gravity sand filtration; adsorption with granular
activated carbon; rechlorination; and ammoniation (Figure 3).
Prior to the installation of the granular activated carbon
adsorbers, consumer complaints were received about an unpleasant
earthy/musty taste in the water. Ozonation, permanganate dosing,
lime softening, and chlorination (both free and chloramine resid-
ual) all failed to control the taste problem. The use of chlor-
ine dioxide, though successful, was rejected on the grounds of
cost. Powdered activated carbon was found to be effective, but
at the doses required, up to 40 g/m , filtration difficulties
were encountered. Thus, after pilot trials in 1957 and 1958,
granular activated carbon filters were installed and operational
in 1960.
The 17.3 x 10 kg (17 tons; 19 U.S. tons) of coconiit-based
granular activated carbon is contained in eight 2.44 m <8 ft)
high, 2.44 m diameter, vertical rubber-lined steel pressure
vessels. The activated carbon beds, initially 0.91 m (3 ft) deep
but now 1m (3 ft 3 in), are supported on floors of porous stain-
less steel tiles. Water flows downwards through the_bed at a
rate of 10 m/h (3.5 gal/min/ft ; 4.1 U.S. gal/min/ft ) giving an
empty bed contact time of 6 minutes.
The carbon beds are backwashed twice per week; there is an
initial high rate water wash (31 m/h; 11 gal/min/ft ; 13 U.S.
gal/min/ft ) for 1.5 minutes, followed by 0.5 minutes for settl-
ing, and finally a 20 minute water wash at 15 m/h. Design and
operation details of this system have been described elsewhere
(3).
The useful bed life of the granular activated carbon for
taste control is 4 years, after which time the carbon is returned
to the supplier for regeneration. Approximately 25 percent of
the carbon is lost during the 4 years of operation, its transfer
from the filters to the furnace, and subsequent regeneration.
Although the carbon effectively controlled taste in the
drinking water for periods of up to 4 years, its capacity for
organics removal (measured as TOC and COD) was much more limited
(Figures 4 and 5). In an earlier paper about this plant (3), it
was reported that the effluent from the granular activated carbon
adsorbers had a substantially higher plate count (3 days at 22°C)
than the influent to the adsorbers. This observation has been
confirmed by more recent findings (Table 1).
206
-------
River Great Ouse
Gravel Pits
(River Great Ouse
water)
Foxcote Reservoir
(River Great Ouse water)
Prechlorlnation
Aluminium or ferric sulphate
Floe Blanket
Clarification
Rapid Gravity
Filtration
Granular
Activated Carbon
Adsorption
Chlorination and
Ammoniation
Figure 3. Schematic Diagram of the Foxcote Treatment Works
207
-------
Sand filtrate
Carbon treated water
!s
o
<12
HI
O
Z
UJ o
*rt **
X
O
0
6-
Downf low velocity 10 m/h
Empty bed contact time 5.4 minutes
O)
I4
CO
CC
o
o 3
Z
o
Figure 4,
50 100 200
DAYS OF OPERATION
Organics Removal on Foxcote No. 6 Filter
29 November 1972 to 12 March 1973
208
-------
to
O
VO
• Sand filtrate
Q Carbon treated water
Downflow velocity 10 m/h
Empty bed contact time 6 minutes
200 300
DAYS OP OPERATION
400
500
Figure 5. Organics Removal on Foxcote No. 7 Filter
21 October 1975 - 23 March 1977
-------
Table 1. 3 Day 20°C Plate Count
No of
samples
Sand filtrate
53
Colony
count/ml
0-10
11-100
Percent
sample
98.1
1.9
Carbon filtrate
83 0-10
11-100
100+
60.2
34.9
4.8
Final water
231
0-10
11-100
74.9
24.7
LLANDEILO-GRABAN (WELSH WATER AUTHORITY)
,33
Water from Llan Bwch-Llyn, at a rate of 1.1 - 2.3 x 10 m /d
(0.25 to 0.5 mgd; 0.3 to 0.6 U.S. mgd), is treated by direct in-
jection of aluminum sulphate and filtration through five 2.44m
(8 ft) diameter pressure filters.
A heavy bloom of blue-green algae gave rise to taste and
odor problems during 1978. As an emergency measure, at the end
of September 1978, the 0.3 m (1 ft) of anthracite was replaced
with 0.3 m of granular activated carbon. Since that time, no
taste and odor problems have been reported.
RESEARCH
The increasing need to use polluted surface waters to sat-
isfy the demand for potable water has led to increasing concern
about the presence of organic compounds in water supplies. These
compounds may constitute a health hazard or be esthetically
undesirable. Research work is being carried out by the Water
Research Center and a number of water suppliers on the use of
granular activated carbon for removal of trace organics in
potable water production.
In a program carried out by the Water Research Center,
jointly financed by the Severn Trent Water Authority and the
Directorate General Water Engineering of the Department of the
210
-------
Environment, pilot studies were undertaken to investigate the
treatability of the industrially polluted River Trent to a pota-
ble standard (5). Part of this program was to investigate the
efficiency of granular activated carbon to remove organics from
water. Results showed that activated carbon, even with an empty
bed contact time of up to 78 minutes, could not reduce the total
organic content to non-detectable levels. As an extension of
the River Trent studies, pilot plant trials are being undertaken
on River Thames water that has been pretreated (coagulation, floe
blanket clarification, rapid gravity filtration) before being
spiked with selected organic compounds. The spiked compounds
have been chosen to represent a wide range of molecular weights
and chemical types. The studies are to determine the efficiency
of commercially available adsorbents as well as the effect of bed
design and plant operation on the removal of these compounds.
Preliminary work has concentrated on the removal of six
chlorinated compounds; chloroform, bromodichloromethane, a and Y
isomers of hexachlorocyclohexane ( otand Y HCH also known as a
and Y benzene hexachloride), dieldrin, and 1,2-dichlorobenzene.
Initially, these were dosed into treated River Thames water via a
methanol stock solution (6). The use of methanol was necessary
to assist in getting the chlorinated organics into solution in
the aqueous phase. It was desirable to keep the methanol concen-
tration in the spiked water as low as possible in order to mini-
mize the associated biological activity. In later work (7) it
was possible to eliminate the use of methanol by dissolving the
solid organics ( ctHCH, Y HCH and dieldrin) into the liquids
(chloroform, bromodichloromethane and 1,2-dichlorobenzene) and
adding these at the required flow rate (approximately 0.3 ml/h)
using a syringe pump into 1 m /h of treated River Thames water.
Results have shown that 1,2-dichlorobenzene was the most
readily removed of the six compounds; in general, pesticides are
more readily adsorbed than haloforms and bromodichloromethane is
better removed than chloroform. There was also a tendency for
the carbons that were best for haloform removal to be worst for
pesticides, and for the worst for haloforms to be the best"for
pesticides removal (Figures 6 and 7). Over the range investi-
gated (2, 5 and 10 m/h), downflow velocity did not appear to
influence the performance of granular activated carbon filters.
Similarly, increasing bed diameter from 0.15 to 0.3 m (0.5 to 1
ft), while maintaining the same downflow velocity and contact
time, did not influence performance.
THE FUTURE
It is unlikely that the use of granular activated carbon
for potable water treatment in the United Kingdom will greatly
increase unless legislation makes this necessary or there is
organic contamination of source waters where no alternative
source is available.
211
-------
to
Spiked water
75
100
DAYS INTO RUN
Figure 6. Comparison of Granular Activated Carbons
for Bromodichloromethane Removal
-------
to
I-1
u>
• Spiked water*
A LSS
D Haycarb
O WVG
CC1230H
25
50
DAYS INTO RUN
75
Figure 7. Comparison of Granular Activated Carbons
for Hexachlorocyclohexane ( HCH) Removal
-------
The Water Research Center is embarking on a study to assess
the implications on the United Kingdom water industry if legisla-
tion, such as that proposed by the United States Environmental
Protection Agency (8), is brought into force.
Research work is continuing within the water supply industry
and by the Water Research Center where comparisons of a wider
range of adsorbents will be made. The removal of other groups
of compounds such as aliphatic, aromatic and polyaromatic
hydrocarbons, phenols, and chlorinated phenols will be studied.
ACKNOWLEDGEMENTS
The authors wish to thank the Director for permission to
publish this paper. They would also like to thank all those
members of staff from the Anglian, Severn Trent, and Welsh Water
Authorities, the East Surrey Water Company, and the Water Re-
search Center who helped in collecting the information reported.
REFERENCES
1. Osborne, D. Experience at Wilne Treatment Works. Presented
at the "Adsorption Techniques Conference," Washington,
April/May 1979.
2. Osborne, D. Experience with multihearth furnace at Wilne
Treatment Works. Ibid.
3. Ford, D. The use of granular activated carbon for taste
and odor control. Proc. of the Water Research Association
Conf. "Activated Carbon in Water Treatment" Paper 12,
Medmenham, England (1974).
4. Clark, R.G. Private communication.
5. Melbourne, J.D. "River Trent Treatment" Technical Report
74. Water Research Center, Medmenham, England (1978).
6. Hyde, R.A. Removal of haloforms and pesticides by granular
activated carbon. Activated Carbon Adsorption of Organics
from the Aqueous Phase (Ed. McGuire and Suffet) Ann Arbor
Science, Ann Arbor, Michigan (in press).
7. Hyde, R.A. Pilot plant investigations on haloform and
pesticide removal by activated carbon. To be presented at
IWSA Specialized Conference "The Use of Activated Carbon in
Water Treatment" Brussels (May 1979).
8. U.S. EPA. Interim Primary Drinking Water Regulations.
Federal Register (Thursday, February 9, 1978) Part II.
214
-------
DESIGN CRITERIA AND PROCESS SCHEMES FOR GAC FILTERS-
GERMAN EXPERIENCES
Prof. Dr. Heinrich Sontheimer
Since 1975, when a special conference was held in Karlsruhe,
Germany,* on practical experiences with the use of activated car-
bon in drinking water treatment in Germany, research has contin-
ued at an accelerated pace. The German Ministry of Research and
Development has subsidized investigations on the optimization of
carbon use and regeneration, on biological treatment in activated
carbon filters, and on macroreticular resins for humic acid
removal, to name just a few recent projects. Many of these
studies were conducted on-site at water treatment plants focusing
on the practical aspects of activated carbon use. More special-
ized studies on process design parameters for sorption processes
sponsored by the Deutsche Forschungsgemeinschaft, were undertaken
at the University of Karlsruhe.
To provide a better understanding of the problems connected
with activated carbon treatment in Germany, this paper describes
its historical development and the resulting design criteria and
process schemes for GAC filters. Defining the purpose and aim
of the carbon treatment in itself poses a problem. It is said
that different waters need diffprent types of treatment, and this
means different control methods a/id design criteria for GAC as
well. In addition, many open-«nded questions on activated carbon
treatment still await furthe-r research.
GAC TREATMENT IN GERMANY
In surveying the development of the use of activated carbon
in Germany, it must first be noted that powdered carbon is used
infrequently. Many waterworks store this type of carbon for use
in the event of environmental accidents, but fortunately such
accidents are rare. By contrast, granular activated carbon
filters have been used in water treatment for the past 50 years.
Table 1 shows the development of the use of activated carbon
filters in Germany. Carbon filters have been used for dechlori-
nation since 1930, especially where high chlorine dosages are
*The proceedings of this conference were published by the
Engler-Bunte Institute, available in English as EPA-600/9-76-030
(Dec. 1976).
215
-------
are needed for ammonia removal and where carbon treatment is
required before infiltration into the ground. In Germany, taste
and odor problems are largely associated with the use of Rhine
River water. The most extensive studies on the removal of taste
and odor with activated carbon filters were made in 1955, in
Dusseldorf. One of the reasons for this study was that the
waterworks was reluctant to install an additional pumping station
that would have been needed in order to use powdered carbon.
When it was later discovered that with ozone pretreatment, less
carbon was needed for taste and odor removal, the additional
pumping station was still needed, but this application was felt
to be more efficient.
TABLE 1
Development of the Use of
Activated Carbon Filters in Germany
APPLICATION DATE
Dechlorination 1930
Taste and Odor Removal 1955
Organics removal 1970
Biological activated carbon 1976
The grades of carbon used for taste and odor removal had
to have good phenol values, but required no high activity for
other substances. Therefore, the carbons had only small pores
and a small surface area. When in 1970 the removal of organics
became increasingly important, new types of carbon with a higher
activity and surface area had be to used. For these more expen-
sive carbons, regeneration was an economic necessity. Since
on-site regeneration, if done properly, is cheaper in most
instances than off-site regeneration at the carbon-producing
company, the first carbon regeneration installations at the
waterworks were constructed and are now in operation.
The use of a higher quality carbon led to observations on
biological activity in carbon filters, especially at the water-
works on the Rhine River where preozonation is used. From these
findings, data were obtained in 1976 on the practical use of this
process, which combines the economy of biological oxidation with
the safe operation of granular activated carbon treatment.
216
-------
DESIGN CRITERIA FOR GAC FILTERS
Table 2 describes the design criteria of dechlorination,
taste and odor removal, organics removal, and biological acti-
vated carbon. The development of these treatment processes
reduced high filtration velocities to fairly low velocities and,
at the same time, increased the height of the filter beds. Thus
the empty bed retention times increased considerably, as did the
investment costs for activated carbon filters. Concurrently,
on-site regeneration provided a better margin of safety for the
removal of organics and also reduced operating costs, as did
biological activity in the filters.
TABLE 2
Design Criteria for GAC Filters in Germany
Dechlorination
Taste and odor removal
Organics removal
Biol. activated carbon
Filtration
velocity
m/h
25-35
20-30
10-15
8-12
Bed
height
m
2
2-3
2-3
2-4
Empty bed
retention
time
min.
2- 4
6-10
8-15
15-25
Throughput
ratio before
regeneration
m3/m3
> 1 000 000
- 100000
- 25 000
- 100000
Today, biological oxidation is critical to the economical
operation of the carbon filters. Table 3 describes the median
reduction of removal efficiency in carbon filters, using data
from the Mulheim treatment plant for filters 2.5m in height and
217
-------
lOm/hr in filtration velocity, having a 15-min empty bed reten-
tion time. The decline in DOC removal with time, resulting from
the reduction in biological oxidation and from increased loading
onto the carbon of nonbiodegradable organics, indicates that the
oxidation process used here was not sufficiently effective to
convert more substances into biodegradable matter. However, the
median data shown in the tables are still fairly good.
TABLE 3
Reduction of Removal Efficiency
in Carbon Filters (RWW Dohne)
Throughput
m3/m3GAC
0—18000
18000—39000
39000—59000
0—59000
DOC-
Remova!
9 DOC/m3d
135
77
26
78
Biological
DOC-Oxidation
g DOC/m3d
101
64
22
61
Biologically
Oxidized
%
71
80
82
78
Loading of
Carbon as DOC
g/kg
39
58
61
61
PROCESS SCHEMES FOR GAC FILTERS
The data on the Miilheim treatment plant emphasize the impor-
tance of using the correct means of oxidation and the optimum
process scheme. The most important types of treatment processes
are shown in Figure 1. The optimum process for treating river
water usually includes two-stage ozonation. The first dose is
applied before flocculation. After sedimentation and before
filtration, a second dose is applied. This type of treatment
enhances efficient flocculation, and if the correct dosage is
used, the conversion of nonbiodegradable into biodegradable sub-
stances is also aided. In this scheme, carbon filters are used
for the adsorption of hazardous substances and provide a site for
biological oxidation of organics.
218
-------
Stepwise Ozonation Stepwise Chlorination Ozone - Floccuiation
pre-ozonation pre-chlorination ozonation
I I <
flocculation flocculation addition of flocculants
I 1
sedimentation sedimentation
1 t
main - ozonation main - Chlorination
j or ozonation
filtration filtration filtration
I I I
GAC-filters G AC-filters G AC-fitters
I i
(ground passage) (ground passage)
J I
safety Chlorination safety Chlorination safety Chlorination
Figure 1. Oxidation and GAC Filters Process Schemes
This combined treatment scheme extends the operating time
period between GAC regenerations by three to ten times. If
suspended or colloidal substances are present in low concentra-
tions, flocculation or sedimentation are not needed, and ozone-
induced flocculation in the filters is all that is necessary.
These modifications do not, however, negate the general process
principles. These remain in effect even if chlorine is applied
in small dosages instead of ozone.
Thus sufficient experience has been obtained for this
combined biological adsorptive treatment and also enough design
criteria to decide on the empty bed retention time and the type
of pretreatment for each type of water. How to operate a pilot
plant to gather more data on the special design of treatment
facilities for a particular type of water also is known. How-
ever, it is often difficult to choose among the many available
alternatives for control of activated carbon filters. Figure 2
lists a number of the possible control methods for activated
carbon filters that are currently in use.
219
-------
1) Carbon loading on the top and the bottom
layer of the filter
a) DMF-extract
b) Dioxane-extract
c) Loading with organic chlorine
d) Loading with nonpolar chlorine
e) Loading with specific compounds
2) O2 consumption for biodegradation
Measurement with two or more filter velocities
3) Breakthrough behaviour:
a) For sum-and group parameters of organic
substances:
e.g. DOC, UV, DOCI, DOCIN
Taste and Odor, Aromatic Amines etc.
b) For defined substances:
e.g. CHCI3, C2HCI3, Chlorobenzenes etc.
Figure 2. Possibilities for Activated Carbon Filter Control
METHODS OF CONTROLLING ACTIVATED CARBON FILTERS
Control of the filter efficiency and the carbon activity is
achieved by three processes. The first process was developed to
measure the different carbon loadings, either by extraction using
dimethylformaraide to obtain all the organics, or by using dioxane
or similar solvents to measure nonpolar organics. When extrac-
tion is employed, it is important to look for group parameters
such as nonpolar organic chlorine compounds, or for specific sub-
stances like Chlorobenzenes that are especially significant for
monitoring the carbon filter efficiency for Rhine River water.
Besides extraction, group parameters such as organic chlo-
rine can be measured through direct pyrolysis of the carbon,
saving time and permitting easier control. However, these
methods have the disadvantage that loading in itself does not
provide specific data on the breakthrough behavior of a filter.
This is true to soma extent for the measurement of oxygen con-
sumption. The data in Figure 3 show that oxygen consumption
declines with time. Additional information is needed on the
filter breakthrough behavior and on the significance of these
measurements.
220
-------
2000 4000 6000 8000 10000
Throughput in m3 Water / m3 AC
12000 14000
Figure 3. Oxygen Consumption in an Activated Carbon Filter
as a Function of Water Throughput (Dusseldorf)
The problems related to breakthrough measurements are well
illustrated by measurements for four different parameters on two
large filters in the Dusseldorf waterworks. At Dusseldorf, two
different types of carbon have been used along with the same
pretreatment. The breakthrough behavior for the different water
quality parameters is shown in Figure 4. The dimensionless con-
centrations (c/c ) are given as a function of the throughput
ratio in volume Per volume for the last half of the operational
period and for one of the two filters. A UV reduction of 50 per-
cent was achieved for this filter after a throughput of 8000 m
of water per m of activated carbon in the filter. Although the
DOC reduction is not as good, the two curves run parallel and
therefore the UV measurements can be used for information on
the DOC values. The data show that the DOC1 reduction lies in
between the values for DOC and for UV removal. The problem
requires further study. More data are needed for organic sulfur,
organic acids, and other group parameters like the corrosion
inhibition factor, which may become more significant in the
future.
221
-------
CHOOSING THE CORRECT CONTROL PARAMETER AND TYPE OF CARBON
However, the more parameters and criteria are measured,
the more difficulties will be encountered in choosing the correct
parameter for optimum control. For example, in Figure 4 the
breakthrough behavior of chloroform shows that higher concen-
trations are found in the effluent than in the influent to the
filter. On the other hand, a very good filter efficiency was
achieved for all the other volatile nonpolar organic chlorine
compounds, tri- and tetrachloroethylene being the most important.
Although the influent concentrations of all substances were very
low in this test, which was performed at a high water flow in
the river, the difference in the behavior of the organic chloro-
compounds remains very high. Thus the question as to which of
the two parameters should be used here to control the filter
efficiency is unresolved.
The same difficulty arises when two different carbons are
compared. One possibilty would be to examine the breakthrough
behavior for the UV-absorbing organics, an example of which is
shown in Figure 5. Filter 8 has a lower efficiency at first,
than the other carbon in filter 4. This remains true for UV-
absorbing organics, but for organic chlorine compounds there is
little difference, although filter 8 tends to be more efficient.
Isotherm data can provide simlar results, but the test method
used in Figure 5 is preferred for assessing activated carbon
quality.
Figure 6 shows the isotherms measured by using the same
water that was to be treated with the activated carbon filters.
The isotherms indicate that the activated carbon in filter 4 has
a somewhat higher capacity than that in filter 8 with respect to
overall orqanic substances in water (OSW), as measured here by
UV absorbance. 'Moreover the same effect occurs for a sub-
stance like p-nitrophenol (PNP), and similar effects can be
expected for substances such as trichloroethylene. Nevertheless,
the differences remain quite minimal in this test, the results of
which correlate quite well with the breakthrough behavior in the
full scale filter plant.
However, the close comparison between the equilibrium
test and the large filter breakthrough behavior does not really
indicate which parameter should be used for control and which
method might be preferable. For this information the many dif-
ferent possibilities for controlling activated carbon filters
have to be reexamined (Figure 2).
In Germany the breakthrough behavior for a certain organic
pollutant will seldom supply sufficient information to control
activated carbon filter efficiency. Only for some groundwaters
with tri- and tetrachloroethylene, can this type of control be
used advantageously.
222
-------
1.4
1.3
1.2
1.1
1.0
0.9
0.8
,0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Chloroform
C0-1.68(19/1
• UV c0- 1.98m"1 ,
other volatile Chtoro-Compounds
c<,-2.72ugll
.... . J. . A * A A
~ ~ ~ » • ~ A* ~ * A
5 6 7 8 9 10 11
Throughput 1o'm3 Water/m3 Carbon
Figure 4. Breakthrough Curves for Four Parameters
1.0
0.5
5 Throughput in 103m3/m310
15
Figure 5. Breakthrough Behavior of UV-Absorbing Substances
for Filters with Different Carbon Grades
223
-------
O)
100
50
10
0.1
100
50
10
0.1
100
« 50
3
c
10
OSW
OSW + (PNP)
PNP
0.01
1.0
1.0
0.1
c in g/m3
10.0
10.0
1.0
Figure 6. Comparison of Equilibrium Data for
Two Different Activated Carbons
224
-------
When surface water is being treated, especially the water
from the heavily polluted Rhine River, it is usually not known
in advance which type of chemical pollutant will be most serious
at any given time. This determination can be attempted by
special analytical methods, but these methods usually require so
much time that the particular type of pollution may have changed
in the meantime. Thus it is generally not practical to use the
breakthrough behavior of specific organics as a means of control.
In practice, the waterworks has to be certain at all times
that the carbon filter has enough capacity for sudden high con-
centrations of unknown pollutants. This can be accomplished
either by studying the carbon loading at different filter heights
or by observing the overall removal efficiency for organics. UV
absorbance measurements are taken for the latter purpose since
these can be obtained easily and correspond to DOC measurements.
When UV absorbance measurements indicate the need for paral-
lel tests, results very similar to those from carbon loadings can
be obtained by controlling the oxygen consumption of biological
regeneration of the carbon in the filter. If parallel tests are
made for some time, and for different process conditions, compar-
ing the proposed simple testing methods (e.g., UV absorbance)
with more sophisticated analytical measurements will provide
enough information to decide when regeneration should be done.
In Dusseldorf, for example, it is known from practical ex-
perience that taste and odor problems will occur in the farthest
parts of the distribution system if a 50 percent reduction in UV
absorbance is not achieved. Since chromatographic enrichment of
chloroform is going on at the same time, some other organic
substances responsible for the renewed formation of taste and
odor may also be breaking through.
Thus, many problems in the use of carbon filters remain to
be solved. Further research is needed, particularly on sorption
analysis for water with unknown organics; on the prediction of
breakthrough behavior in sorption filters; on the mechanism of
biological treatment in carbon filters; on the combination" of
flocculation, filtration, and adsorption; and on control means
for carbon filters. In the future, waterworks should be able to
predict carbon filter performance by laboratory tests because
this will be the only way to optimize the production and regener-
ation of carbon for this particular purpose. Better information
will be needed on the mechanisms influencing the efficiency of
biological treatment in carbon filters, and other combined treat-
ment effects will have to be studied intensively.
225
-------
SUMMARY AND RECOMMENDATIONS
The use of activated carbon filters for drinking water
treatment will become increasingly widespread in the future.
Carbon filters will be primarily a safety factor similar to
safety chlorination. Most chemical pollutants that are hazardous
to human health can be removed with a carbon filter if the right
design is selected and the adsorption efficiency is control led.
Only with this treatment will the problems, arising from unknown
organics in river water and in many groundwaters as well, be
resolved.
If, on the other hand, carbon filters are used chiefly for
safety purposes, they can still be applied for the removal of
other organics. This application will permit an easier control
because the overall adsorption capacity can be monitored and
thereby the safety factor will be maintained for unknown sub-
stances as well.
The use of activated carbon filters for safety purposes
implies that the adsorptive quality be adapted to the change
in the possible pollutants; for example, by using sorbent
mixtures.
Some questions still remain regarding the application
of activated carbon treatment to drinking water. Exchange
of information on an international level can ensure progress
toward safer, cleaner drinking water.
226
-------
DISCUSSION
MONDAY, APRIli 30
AFTERNOON SESSION
Q UNIDENTIFIED QUESTIONER: I wonder whether you could
comment on the extent to which you get removal of heavy trace
metals.
A PROF. DR. SONTHEIMER: I must say we are very happy in our
country that we do not have problems with heavy metals. All
along the Rhine there is a low heavy metal concentration, even
lower than drinking water standards permit. Furthermore, we
achieve good removal in treatment. Removal efficiency depends
on the concentration of heavy metals, but generally lies in the
range of 50 to 90 percent. The removal efficiency varies
widely, but we are not concerned because the concentrations in
the Rhine River are low. The concentrations in solution are
especially low, so heavy metal pollution does not pose a real
problem.
Q DR. ANDREW BENEDEK, McMaster University: Professor
Sontheimer, I am interested in your data on oxygen uptake. As
you know, oxygen is taken up by fresh activated carbon and you
showed a decline. Is that over and beyond the chemical uptake,
or do you feel that the chemical uptake is negligible in that
particular case?
A PROF. DR. SONTHEIMER: We do not distinguish between the
two: chemical and biological. We know that in the first 2 days
chemical uptake is appreciable, but I can't see any reason why
chemical uptake would continue for months and there are no data
to support that hypothesis. Why should the activated carbon
surface become available for chemical oxidation after the surface
is loaded with adsorbed material?
Q DR. ANDREW BENEDEK: Have you attempted to find out what
types of organics are being oxidized, particularly by measuring
changes in ultraviolet spectra?
A PROF. DR. SONTHEIMER: Yes, I think it is the same in all
countries. The bulk of the organic material consists of humic
substances. And most of the organics oxidized are humics. We
have data on metals, change of polarity, and change of the
molecular weights of the humic materials during oxidation; those
227
-------
data correlate exactly with the amount of oxidation and the
effect of oxidation within the carbon. Hence, we are certain
that the changes we measure are mostly governed by the oxidation
of the organics. It's a very important problem, because on one
hand we shouldn't have too high of an organic content, because
then we need too much chlorine for disinfection purposes. It is
my opinion that we should reduce the total organic carbon concen-
tration to approximately 1 mg per liter or less before chlorina-
tion; less than 10 jag/1 of THM's should then be produced by
safety chlorination.
Q DR. COTRUVO, EPA: I would like to ask a practical question
for the engineers and water plant operators. The necessity of
reactivating carbon is an additional workload placed upon the
plant operation. Undoubtedly, there would be some practical
limits to reactivation. In your country, what do you feel would
be a practical minimum time interval between reactivations for
carbon?
A PROF. DR. SONTHEIMER: I would say it should be in the
range of 3 months. If you are using the right combination of
pretreatment in the right way, then you will be able in most
cases to reprocess in 3 months. We are now happy about our
situation in the Rhine river because it is no longer heavily
polluted. We reactivate every 4 or 5 months, but I think we
should be able to reduce that to 3 months.
Q PAUL ROBERTS, Stanford University: The U. S. EPA has pro-
posed a carbon treatment standard that would entail 50 percent
removal of total organic carbon and allow a maximum rise of
1/2 mg/1 of Total Organic Carbon (TOC). Would it be possible to
cite experience from Europe that would show conditions under
which that standard could be satisfied, and if so, what would be
the combination of empty bed contact times and reactivation
frequencies that would permit that standard to be fulfilled.
A PROF. DR. SONTHEIMER: I shouldn't comment on EPA's recom-
mended standards, but I must say I don't agree entirely. That's
my personal opinion. However, I think the general idea behind
it is excellent. A collective parameter must be chosen. There's
no doubt TOC is very easy to measure and a parameter that might
be considered. We use ultraviolet absorption, yet I couldn't
say if UV is possible everywhere. We measure TOC as well, and
usually find a relatively constant relation between UV and TOC,
but we think it's more important to remove the most dangerous
substances. I would never propose to use activated carbon after
breakpoint chlorination because I think that it is a waste of
money. Breakpoint chlorination can be applied, but after 2-3 or
6 weeks of operation, you achieve the same results without it.
Omitting breakpoint chlorination results in lower chemical costs
and gives you the same, or much better, water quality.
228
-------
I must say I honor the general idea behind these regulations,
but I do reserve comment because Germany-is a country small
enough not to need such regulations. I have a close working
relationship with most of our waterworks people and we can
discuss our problems in person. We can discuss the best way for
each water utility to operate. I'm very satisfied with this
relationship, because this gives us a possibility of finding
economic solutions to achieve good drinking water quality.
I also agree with Dr. Schalekamp's statement that we really
should attain the same water quality from surface water supplies
that we are accustomed to in the very best ground water. This
is achievable and economically feasible. I am convinced that,
in treating roost surface water supplies to assure good finished
water quality, we need activated carbon for most cases. But it
is impossible for me to establish a general figure or set
standards which must be met. I must say that I am not familiar
with all the factors in the U.S.A., so I cannot comment on this.
In Germany, we are glad that we do not need formal standards.
Q PROF. MICHAEL SEMMENS, University of Minnesota: I have two
questions for Dr. Sontheimer or anyone else who would like to
respond. The first is concerned with the ozonation process used
as a pretreatment step before coagulation-flocculation. Could
you give me some idea of the dosages of ozone that you use in
that process?
A PROF. DR. SONTHEIMER: Yes, the ozone dose is in the range
of 0.1 to 0.4 mg/1 per liter; in other words, very low dosages,
but nonetheless very beneficial. There is, normally, a very
well-defined optimum in the range of 0.2 to 0.3 mg/1 per liter.
Q PROF. SEMMENS: If you exceed that range, do you see a
deterioration in the effectiveness of coagulation?
A PROF. DR. SONTHEIMER: No, you see it in the removal of
organics. You must add some ozone to get better flocculation,
but not too much or you will reduce its effectiveness.
Q PROF. SEMMENS: The second question relates to the coagula-
tion process itself. Historically, coagulation and flocculation
have been used to reduce the turbidity of the water prior to
filtration. If your objective is to reduce the organic content
of the water, do you still use turbidity removal as the control
or means of determining the best coagulant dose? Or do you use
organics removal?
229
-------
A PROF. DR. SONTHEIMER: We use both. I think we have to use
both. Turbidity removal is very important, especially for virus
removal and for other reasons related to disinfection. So, we
should really have low turbidity levels. I tell the German
water supply specialists that they should achieve 0.1 turbidity
units, but the other important objective is to remove the
organics. This is one problem that remains. We save money and
minimize sludge problems with activated carbon, but it also
takes out organics that may be corrosion inhibitors. So I think
a very important research project for the future will be to
optimize flocculation, not only for turbidity removal, but also
for organic removal. I think that optimized flocculation, in
combination with some preoxidation has a real future.
Q BRIAN TOMLINSON, Fluor Northwest, Fairbanks, Alaska: We
have problems in the north where as much as 6 to 8 feet of ice
accumulates on our lakes. There is no dissolved oxygen in the
water. You indicated 78 percent average reduction of organics
by aerobic biological oxidation. What has been your experience
on anaerobiosis and how do you treat it, by pretreatment or
scavenger chemical or something like that?
A PROF. DR. SONTHEIMER: Biological treatment has been widely
used for hundreds of years in Germany and in most European
countries. We do not have any German waterworks for surface
water that do not use biological treatment. The effect of ground
filtration is achieved quite easily. The filters contain acti-
vated carbon and, if properly designed, are very similar. We use
activated carbon instead of sand filters or ground infiltration.
It's only a matter of money and space that determines the prac-
ticality of using activated carbon. It's not a general proposal,
you can obtain the same data by ground filtration, so we don't
make an issue of these things. But I think biological treatment
is so inexpensive and so easy to use in combination with the
right pretreatment, that I recommend it.
Q MR. TOMLINSON: Maybe I misunderstand your presentation.
Are you performing biological treatment within the activated
carbon column?
PROF. DR. SONTHEIMER: Yes sir.
MR. TOMLINSON: And it's strictly aerobic?
Q
A PROF. DR. SONTHEIMER: It is strictly aerobic. We are very
sure about this. We must have greater than 5 mg/1 of oxygen
outside the carbon. The oxidation kinetics and the diffusion
coefficient shows that the oxygen concentration inside the carbon
is higher than zero.
230
-------
Q MR. TOMLINSON: Therefore, you have had no experience with
anaerobic biological reduction of organics?
A PROF. DR. SONTHEIMER: Yes we have, with nitrates and so
on, but it wasn't a special study. I should say we have received
unexpected experience.
Q GERALD SPEITEL, University of North Carolina: I wanted to
follow up on the question about pretreatment organics removal.
What kind of experiences do you have in terms of percentage
removal of TOC?
A PROF. DR. SONTHEIMER: The removal of TOC in pretreatment
is 35 to 40 percent. Although not very high, it is still
s ignif icant .
Q MR. SPEITEL: Do you feel that coagulants or pH ranges
greatly influence the pretreatment conditions?
A PROF. DR. SONTHEIMER: Oh yes, we pay attention to that,
but on the other hand, it would be uneconomic to optimize floc-
culation by means of two or more stages.
MR. SPEITEL: Does the optimum for organics removal differ
greatly from the turbidity optimum?
A PROF. DR. SONTHEIMER: To some extent, yes. We studied
this and found there are waters that should be flocculated at
different pH values for optimum organic removal and optimum
turbidity removal.
231
-------
vvEPA
NATO CMM
•OTAN CCMS
NATO-CCMS
232
-------
DESIGN CRITERIA FOR ACTIVATED CARBON ADSORBERS
233
-------
&EPA
NATO CMM
-OTAN CCMS
NATO-CCMS
234
-------
PREDICTION OF MULTICOMPONENT ADSORPTION BEHAVIOR
IN ACTIVATED CARBON ADSORBERS: KINETIC ASPECTS
Werner Merk
INTRODUCTION
A major problem in the design of adsorption beds for water
treatment is the prediction of breakthrough behavior in multi-
component mixtures. The objective of this research is to
formulate the kinetic models needed and to verify the results
with experimental data.
Experimental data of three multi-component systems are
shown. In the first bi-solute system, p-nitrophenol/p-chloro-
phenol, both solutes have a similarly high affinity for acti-
vated carbon; in the second bi-solute system, p-nitrophenol/
phenol, both solutes have different equilibrium behavior, and in
the three-solute system, p-nitrophenol/p-chlorophenol/propionic
acid, propionic acid has a much lower affinity to activated
carbon than p-nitrophenol or o~chlorophenol(1). In all exper-
iments, the activated carbon B 10 II* was used (apparent particle
density pR=0.59 g/cm , equivalent particle radius R=0.66
mm. ).
Design of fixed-bed adsorbers requires, in addition to rate
parameters, information about adsorption equilibria. Adsorption
equilibria of bi-solute systems are predicted from single solute
data using the ideal-adsorbed-solution (IAS) theory (2,3), while
the single-solute equilibria are represented by Freundlich
isotherms (1,4,5).
EXPERIMENTAL
To investigate adsorption kinetics, two different experi-
mental methods were used (Figure 1). In a batch reactor test
(left part of Figure 1), at time t«0, a known quantity of carbon
granules S was added to the aqueous solution with the initial
concentration c of solute i. The concentration profile,
c.(t) decreasing with the time t, was measured.
•Manufacturer:Lurgi, Frankfurt, Germany.
235
-------
batch reactor
tb i S
fixed-bed adsorber
concentration -
lime profile
lilz.ll t breakthrough curves
c,.
Figure 1. Concentration Profiles in an Activated
Carbon Particle "Film-Diffusion Model"
In a column experiment (right part of Figure 1), the
aqueous solution with the influent concentration c._ flows
from the bottom to the top of the bed. The variation of the
concentration c. in the water phase with time t was measured
at different bed lengths z.
Both concentration-time profiles from batch measurements
and breakthrough curves from column experiments were used for
testing model predictions. Moreover, from the batch measure-
ments, the required rate parameters for design of multi-solute
fixed bed systems were determined.
KINETIC MODELS
Adsorption kinetic models require an assumption about rate
controlling steps. In the systems investigated in this work,
the adsorption kinetics may be controlled by: diffusion of
solute molecules through the external film, in the following
called film diffusion; and by diffusion within the particle, in
the following called internal diffusion.
Multi-solute adsorption is more complicated than single-
solute adsorption, owing to solute-solute interactions during
the diffusion process and to competition between the different
species for available sites on the surface of the adsorbent.
As a first approximation, it is assumed that the several
solutes diffuse independently of one another. Thus competitive
effects are taken into account only through the adsorption
isotherm.
236
-------
FILM-DIFFUSION MODEL
If mass transfer is governed only by external resistance,
there are no concentration gradients within the particle of
diameter 2R (Figure 2).
Figure 2. Concentration Profiles in an Activated
Carbon Particle "Film-Diffusion Model"
The diffusion of solute molecules through the external
film may be expressed by the following rate equation for each
component:
Bi,F
(1)
where c. is the bulk liquid concentration; c- R is the liquid
concentration at the outer surface of the particle; and 3 * p
is the external mass transfer coefficient for component i. It is
assumed that equilibrium is achieved at the outer surface of the
particle, between the concentration in the liquid c. _ and the
concentration in the adsorbed phase q. » q.. The16oncentration
gradient n. , which is the driving fitce fo£ mass transfer, is
in the liquid phase.
B. F can be predicted by the general correlation for heat
and mass transfer in fixed beds (1,6). The Sherwood-number,
Sh., of component i, is given by:
Sh
i,F
2R
(2)
237
-------
where D. L is the diffusion coefficient in the bulk liquid,
Sh. can be calculated using the following equations:
Shi,Packed Bed = U+1.5(l-e) ]Sh^ sinqle Particle (3)
Shi,Single Particle " min Shi + F ^^(Re)1/2 (4)
min Sh. =2 for spheres
0.664
0.0557Re
°'3
1/2
(5)
H-2.44(Sci2/3-l)Re~°'1
where Re, Sc. ?and Pe. are the Reynolds-,Schmidt-, and Peclet-
numbers (1), respectively.
These equations can be used, if Sc. > 0.6 and Pe. =
Re • Sc/> 1000. 1 x
The diffusion of solute molecules through the external film
is the only rate determining step for low concentrations and for
solutes having a very favorable isotherm (1,4,7,8). In Figure 3,
results are shown for the bi-solute system p-nitrophenol/p-
chlorophenol. The influent concentration of both solutes was
0.1 mmol/1. The fluid phase concentration x.f expressed
relative to the influent concentration of p-chlorophenoJ in the
liquid phase, can in this case increase up to 5t3 percent above
the influent concentration. The enrichment of the weaker
adsorber, p-chlorophenol, in the liquid phase can be explained
by displacement effects.
The experimental breakthrough curves of both solutes are
accurately described by the "film-diffusion model.11 The agree-
ment is remarkable, considering that no adjustable parameter is
used in the calculation. The only information used from experi-
mental results is the single solute equilibrium for each compo-
nent. The assumption of equilibrium at the outer surface of the
carbon particle and the hypothesis that there is no interaction
between the diffusing molecules in the liquid phase seems to be
confirmed, or at least cannot be refuted.
At higher influent concentrations, deviations are observed
between measured breakthrough curves and those calculated from
the "film-diffusion model." The deviations are caused by the
internal mass-transfer resistance which becomes significant at
higher concentrations.
Internal diffusion can occur either by pore or by surface
diffusion, or by both simultaneously. The increasing influence
238
-------
100
200
300
tOO 500
tth]
— film diffusion
^00 500
t[h]
Figure 3. Calculated and Measured Breakthrough Curves of the
Bi-Solute System p-nitrophenol/p-chlorophenol
(bed density PB = 0.39 g/cm , interstitial fluid
velocity u = 5.65 m/h, external mass transfer
coefficient 6,p=B2F=1.7 10~ m/s, as
in all column Experiments presented in this paper.)
Equilibrium represented by lAS-method.
239
-------
of internal mass transfer resistance, observed in this work,
indicates that diffusion within the carbon particle occurs
predominantly in the adsorbed phase. Recent literature data (9,
10, 11, 12, 13) also favor the "surface diffusion model." With
the assumption that accumulation of adsorbate within the pores
can be neglected, the frequently used "film-surface diffusion
model" gives the same result as the "film-homogeneous diffusion
model."
Film-Homogeneous Diffusion Model
The "film-homogeneous diffusion model" assumes adsorption of
solute molecules at the outer surface, which can be described by
an equilibrium model, followed by diffusion within the particle
(Figure 4). The flux n. „ is described by Equation (1) and the
flux n. „ of each component within the particle can be described
by FicK'S first law.
-5F
"I.*'''
n.
Figure 4. Concentration Profiles in an Activated Carbon
Particle: "Film-Homogeneous Diffusion Model"
The diffusion coefficient, D. , which describes the
mobility of solute i in the partiile, is obtained from single-
solute batch reactor tests by comparing experimental and theo-
retical data at q./q- = 2/3.
i if*
Figure 5 shows the estimated diffusion coefficients D
of the single solutes p-nitrophenol, p-chlorophenol, phenol'and
propionic acid. Although the diffusion coefficients D
240
-------
U1.K
[1012m2ls]
5 10
- c/0 fmmolllj
'1.K
HOmls]
p - Chlorophenol
LIS
fllg}
0.5
A
7.0
•
2.0
V
F//-
ter
O
10
15
C1Q [mmolllJ
20
Figure 5. Diffusion Coefficients of the "Film-Homogeneous
Diffusion Model" D, K versus initial (influent)
Concentration '
241
-------
depend on the initial concentration, they are in good agreement
with those estimated from equivalent column experiments. This
leads to the following conclusion for the prediction of break-
through curves:
The average diffusion coefficient D. R can be estimated
from a batch experiment with an initial concentration identical
to the influent concentration of the column experiment.
These coefficients can be used to predict breakthrough
curves of bi-solute systems, if one assumes that there is no
mutual interaction between the diffusing molecules.
Figure 6 shows measured breakthrough curves of the bisolute
system p-nitrophenol/phenol. The influent concentration c,Q of
p-nitrophenol is 1 mmol/1 and the influent concentration c2Q of
phenol is 5 mmol/1. Good agreement for all column lengths Is ob-
tained between measured breakthrough curves and those calculated
from the "film-homogeneous diffusion model." The calculations
were performed using only single solute equilibrium data and rate
parameters, determined from batch experiments with the single-
solute systems. However, the "film-homogeneous diffusion model,"
using only single solute data, has limitations, as indicated in
Figure 7. This figure shows experimental results for the same
mixture as in Figure 6, but now the influent concentrations of
both components are 5 mmol/1. An increasing deviation can be
seen as bed length increases. The breakthrough curves predicted
by the "film-homogeneous diffusion model" show a later break-
through for the stronger adsorber p-nitrophenol, an earlier one
for the weaker adsorber phenol, and for both components a faster
approach to equilibrium than that observed. These deviations
occur in all experiments with mixed solutes if there is a domi-
nating internal resistance and marked displacement inside the
carbon particle.
A better understanding of the deviations described above is
shown in Figure 8. This figure illustrates the fluxes of both
solutes on a single particle, depending on time and bed length,
according to run 12/3 (Figure 7). The length of. the arrows is a
measure of the fluxes and the area of the segments is a measure
of the amount adsorbed. For solute 1, the flux only changes its
quantity with the time, whereas for solute 2, there is a change
in quantity and direction.
As Figure 8 indicates, only a little phenol will be adsorbed
in the presence of p-nitrophenol, the component with the higher
affinity to activated carbon. Therefore, for short operating
times, p-nitrophenol will be predominantly adsorbed at the begin-
ning of the column and phenol will be adsorbed further along the
bed length. After longer operating times, as p-nitrophenol
migrates down the bed, phenol is displaced by p-nitrophenol.
With displacement occurring, phenol must diffuse out of the
242
-------
p-NHrophenol (1)
film - homogeneous diffusion
Figure 6. Calculated and Measured Breakthrough Curves of
the Bi-Solute System p-nitrophenol/phenol.
Equilibrium represented by lAS-Method.
243
-------
lilm-homogfrtfous diffusion
Figure 7. Calculated and Measured Breakthrough Curves of
the Bi-Solute System p-nitrophenol/phenol.
Equilibrium represented by lAS-Method.
244
-------
oo
in
R
o
u
•l-l • —
4J
u w
< £
4J
O en
4J C
c a)
•H ^H
T3
cn d)
QJ en
x
3 T3
rH C
Eh (0
00
0)
^
3
245
-------
carbon while p-nitrophenol diffuses in. This counter-diffusion
could explain the deviations between calculated and experimental
results in this experiment.
For further investigation of this possibility, experiments
with preloaded columns were performed. Results of one experi-
mental run are shown in Figure 9. Prior to the experiment, the
bed was in equilibrium with a solution of 5 mmol/1 phenol. The
experiment was carried out with a bi-solute solution of p-nitro-
phenol and phenol with equal influent concentrations of 5 mmol/1.
After the bi-solute system is introduced, there is continuous
displacement of phenol by p-nitrophenol. Therefore all phenol
concentrations, determined experimentally, exceed the influent
concentration. During the entire experiment, there is counter-
diffusion inside the carbon particles. Large deviations can be
seen between measured breakthrough curves and those calculated
from the "film-homogeneous diffusion model" using only single
solute data. The deviations are caused by a much higher mass-
transfer resistance in this experiment than expected from single
solute data.
Extended Film-Homogeneous Diffusion Model
In order to include diffusional interference, the "film-
homogeneous diffusion model" has to be extended:
n2,K = D21,K K 3 r + D22fR PR 3 r (8)
Thus, the flux of solute 1, fci K, depends on the driving
forces of solute 1 and solute 2. Tfte same is valid for the flux
of solute 2, n, . That means that for a bi-solute system,
four coefficients must be determined from mixture data. These
coefficients can also be determined in batch reactor tests. But
now, two batch reactor tests with the bi-solute system are neces-
sary, the first with fresh carbon and the second with preloaded
carbon (1,8). Coefficients from these batch tests were used for
predicting breakthrough curves in Figure 10, which are compared
with experimental curves already shown in Figure 7. It can be
seen that for both solutes, the agreement between measured and
predicted values is much better than before, when mutual inter-
ference was neglected.
246
-------
^» **
2.0<
f2
7.5
7n
,o
0.5
0
I n°°o
/ "
'o
P/I
^ * 8 cm
DNn
•not (2)
un tmmo°lll)
77/5 5.02
^•a
— film - homo
<^°n
D D o 0 0
fmmoUtl I mmollg} 1 mmollg I
5.03 0 2.99
g»n»ous diffut
iion
O O O D O
0 5 10 15 20 2i
• trni
Figure 9. Calculated and Measured Breakthrough Curves of
the Bi-Solute System p-nitrophenol/phenol.
Activated Carbon Initially Preloaded with Phenol
247
-------
p • Nitroph»nol HI
Figure 10. Predicted and Measured Breakthrough Curves of
the Bi-Solute System p-nitrophenol/phenol
(D
D
21,K
• 2.5 • 10~2m2/s;
1.0 • 10~2 m2/s;
'12,K
1.5
m2/s;
2.0
10"12 m2/s
Equilibrium Represented by lAS-Method
248
-------
SIMPLIFIED CALCULATION OF BREAKTHROUGH CURVES
To simulate a more practical case, the problem now is how to
handle the calculation of breakthrough curves for a mixture in
which one solute has a much higher concentration than the others.
This is illustrated with the bi-solute system p-nitrophenol/p-
chlorophenol in Figure 11. The bed was saturated with p-chloro-
phenol after 20 hours. During this time, p-nitrophenol occupies
only a negligible capacity of the bed because of its compara-
tively low concentration. Thus, p-chlorophenol is adsorbed
nearly undisturbed by p-nitrophenol.
Although the presence of p-chlorophenol has a strong effect
on the adsorption of p-nitrophenol, its breakthrough curves can
also be calculated like a single-solute system. Actually, for
this case, p-nitrophenol diffuses into the carbon which is al-
ready in equilibrium with an aqueous solution containing 5 mmol/1
of p-chlorophenol. In order to calculate the breakthrough curves
of p-nitrophenol, assuming a single-solute system, equilibrium
data and rate parameters have to be estimated from batch experi-
ments using carbon which was already in equilibrium with an
aqueous solution containing 5 mmol/1 of p-chlorophenol.
The breakthrough curves of p-nitrophenol and p-chlorophenol,
in the three-solute system p-nitrophenol/p-chlorophenol/propionic
acid, are predicted using the same simplification that was
pointed out in the previous example. P-nitrophenol and p-chloro-
phenol are adsorbed on activated carbon, which is already in
equilibrium with the water/propionic acid solution. Thus, equi-
librium data and rate parameters of p-nitrophenol and p-chloro-
phenol have to be estimated from batch tests using activated
carbon already in equilibrium with the water/propionic acid
solution. With these data, breakthrough curves of p-nitrophenol
and p-chlorophenol in the three-solute system are calculated like
a bi-solute system. Results are shown in Figure 12. The agree-
ment between measured breakthrough curves and those calculated
from the "film-homogeneous diffusion model" is good.
The method used in these two examples can be generalized.
If one knows which solute should be removed from water, two
estimations have to be made:
- Solutes which are present in very small concentrations
compared to the solute to be removed, and/or which have
a small affinity to activated carbon, do not disturb
the adsorption of the specific solute one is interested
in and thus may be neglected.
- Solutes which are present in very high concentrations
compared to the solute to be removed, and/or which have
a high internal diffusion resistance, may be added to
the water phase.
249
-------
ff- Nitrophenol (1)
A—4 A. A 4.
p-Chiorophenol (2)
run
1312
«»
ImmollU
0.704
'20
Immollll
5.0
I
film-homogeneous diffusion
79.21
60
80
100
tlhJ
Figure 11.
Calculated and Measured Breakthrough Curves of
the Bi-Solute System p-nitrophenol/p-chlorophenol.
The Curves of Both Solutes Are Calculated Assuming,
in Each Case, a Single Solute System.
250
-------
p-Nitrophenol (1)
p-Chlorophenol (2)
Prop ionic acid (3)
100
200 300
— t[h]
400
-o-
p -Nitrophenot (1)
p-Chlorophenol (2)
Propionic acid (3)
' C20 C30
Immolll]
0.103
0.
100
film- homogeneous diffus.
100
79.22
200 300
— tfh]
Figure 12.
Calculated and Measured Breakthrough Curves of
p-nitrophenol/p-chlorophenol in a Water/Propionic
Acid Solution
251
-------
SUMMARY
Although the concentrations of solutes are very low in water
treatment, it cannot be expected that the "film-diffusion model"
predicts breakthrough behavior satisfactorily because, in
general:
most of the solutes have less favorable isotherms than
p-nitrophenol or p-chlorophenol; many solutes are much
larger molecules and hence have a much higher internal
diffusion resistance than p-nitrophenol or p-chlorophenol;
displacement effects cause a much higher internal resis-
tance than expected from single-solute data.
Therefore, internal diffusion resistance has to be incor-
porated into the "film-homogeneous diffusion model." All param-
eters needed in this model can be obtained from batch reactor
tests.
Multi-solute systems may be regarded as a single-solute
system or a bi-solute system, respectively, if all solutes except
one or two are present in very different concentrations; and/or
have a comparatively small affinity to activated carbon; and/or
have a comparatively small internal diffusion coefficient.
252
-------
NOTATION
c fluid-phase concentration
D diffusivity
L volume
1 total length of packed bed
n flux
q concentration of adsorbate
R radius of particle
r radial coordinate
S mass of carbon
t time
u interstitial fluid velocity in packed beds
x normalized fluid phase concentration (=C/CQ)
y normalized concentration of adsorbate
z axial distance coordinate in column
mmol 1
m2 s-1
liter
cm
mmol s~ cm~
mmol g
mm
mm
g
s
-1
m h
-1
cm
GREEK SYMBOLS
6 mass transfer coefficient
e void fraction i(=i-vg/v )
v kinematic viscosity
P.. density of dry particle
ms
-1
g cm
-3
253
-------
NOTATION
Pe = Peclet-number (=Re»Sc)
Re - Reynolds-number (u2R/v )
Xi
Sc = Schmidt-number («vr/DT)
jj LI
Sh * Sherwood-number ( = 3.,2R/DT)
r Li
SUBSCRIPTS
F in the film
i solute i
K in the particle
L in the bulk liquid
R at the outer surface of the particle
s solid
t total
V preloaded
0 initial value
00 final value
254
-------
REFERENCES
1. Merk, W. Thesis, University of Karlsruhe, 1978.
2. Radke, C.J., and J.M. Prausnitz, 1972. J. AICHE-. 18, 761,
3. Jossens, L., J.M. Prausnitz, W. Fritz, E.U. Schlunder, and
A.L. Myers. 1978. Chem. Eng. Sci., J3, 1097.
4. Fritz, W. Thesis, University of Karlsruhe, 1978.
5. Fritz, W. and E.U. Schlunder. Competitive Adsorption of
Two Dissolved Organics onto Activated Carbon. Part I:
Adsorption Equilibria, Chem. Eng. Sci. (submitted for
publication).
6. Gnielinski, V.f 1978. vt Verfahrenstechnik,363,12.
7. Fritz, W., W. Merk, and E.U. Schlunder. Competitive
Adsorption of Two Dissolved Organics onto Activated
Carbon. Part II: Adsorption Kinetics in Batch Reactors.
Chem. Eng. Sci. (submitted for publication).
8. Merk, W., W. Fritz, and E.U. Schlunder. Competitive
Adsorption of Two Dissolved Organics onto Activated
Carbon. Part III: Adsorption Kinetics in Fixed Beds,
Chem. Eng. Sci. (submitted for publication).
9. Crittenden, J. Thesis, University of Michigan, 1976.
10. Crittenden, J. and W.J. Weber, Jr. 1978. J. of the Envir.
Eng. Div., 185.
11. Crittenden, J. and W.J. Weber, Jr. 1978. J. of the Envir,
Eng. Div., 433.
12. Peel, R. and A. Benedek. The Modelling of Activated
Carbon Adsorbers in the Presence of Bio-Oxidation. Paper
presented at AICHE 68th annual meeting, Los Angeles,
Calif. Nov. 16. / 20. 1975.
13. Benedek, A. A Study of the Removal of Synthetic Organics
from Drinking Water by Activated Carbon. Health and
Welfare, Canada, Dec. 1977.
255
-------
PREDICTION OF MULTICOflPONENT ADSORPTION BEHAVIOR:
EQUILBRIUMl ASPECTS
Bernd Frick
INTRODUCTION
The knowledge of adsorption equilibria, and the ability
to describe them, is essential to determining the breakthrough
behavior of organics in GAC filters. When dealing with multi-
solute systems, for instance in water treatment, it is necessary
to use an adsorption model which can predict competitive adsorp-
tion. In many cases, competitive adsorption will be very impor-
tant in the treatment of water supplies containing anthropogenic
organics in addition to large amounts of natural substances.
For the prediction method to be useful, the mathematical
structure of the equilibrium model has to be simple and
only single-solute data should be necessary.
A number of competitive adsorption models more or less
fulfill these provisions (1,2), but they all have advantages
as well as disadvantages. For the prediction of competitive
adsorption equilibria, the author used the Simplified Com-
petitive Adsorption Model (SCAM) proposed by Baldauf, et al.
(3,4).
All these models can be applied successfully only to
defined systems with a limited number of components. However,
the waters used in practice are mixtures of thousands of compon-
ents, of which only a few percent can be identified. Because of
the lack of information on the composition of water, it seems
impossible to predict the adsorption behavior of such an unde-
fined system.
It is the aim of this paper to show that, despite this
fact, there are possibilities to use such a model for solving
the practical problems of activated carbon adsorption. Based on
the discussion of the adsorption behavior of multi-solute
mixtures and on the explanation of the observed effects by means
of model calculations, a method will be proposed which allows
the prediction of adsorption equilibria in unknown mixtures.
Results obtained by this method will be shown for a mixture of
practical interest.
256
-------
ADSORPTION BEHAVIOR OF MULTI-SOLUTE MIXTURES
The collection of equilibrium data is an important part
in the design of a GAC adsorber. This is done by measuring
adsorption isotherms. In most cases, the organics are measured
collectively by group parameters such as DOC or UV absorbance.
Therefore, the resulting isotherms represent the adsorption
behavior of all adsorbable and non-adsorbable substances. DOC
seems to be the most suitable because it covers all the organ-
ics. Figure 1 shows DOC isotherms of two actual waters and of
two single solutes. It is striking that the isotherms of the
river water and of the humic acid are relatively steep while the
single-solute isotherms of p-nitrophenol and phenol-4-sulfonic
acid are flat. All the isotherms can be fitted by the Freund-
lich equation. This is appropriate for the two single solutes,
but it is not so with the others because they are multi-solute
mixtures. By treating them as single-solutes, valuable informa-
tion will be lost, especially with regard to the competitive
adsorption behavior. From Figure 2 it can be seen that the
description of a mixture by a single-solute equation is a gross
simplification. The data in Figure 2 are for the same humic
acid as shown in Figure 1. By increasing the carbon dosage,
however, lower concentrations can be achieved; in this range,
the isotherm shows a totally different shape. This indicates
that we have to deal with a multi-solute mixture.
In most cases, smaller carbon dosages are used and only the
steep part of such isotherms can be observed. It is therefore
interesting to look for the reasons for such steep isotherms.
There are two main questions:
1. What are the conditions for steep isotherms?
2. Is it possible to describe the adsorption behavior
of such mixtures by a model?
For this purpose, a number of model calculations have been
carried out and some examples of these calculations are shown in
Figure 3. It can be seen that the slope of the isotherms is a
function of the single-solute data of the components and of
their initial concentrations. In the case of very different
single-solute behavior, the isotherm slope increases when the
concentration of the weaker-adsorbed component is increased.
(Figure 3, left-hand side). For constant initial concentrations
of the two components the isotherm slope is a function of their
adsorbability. For similar single-solute data, the isotherm
slope of the mixture is almost identical to that of the single
solutes. But when there is a large difference in single-solute
behavior, the isotherm slope reaches values which are four- to
six times larger than those of the single-solute isotherms.
257
-------
150
100
• so
p-NHioptxnct
humfci
Pfwnof ~ sulfonfc
to
c In mg DOC/1
SO
100
Figure 1. DOC-Isotherms of Single Solutes
and Multi-Component Mixtures
Hunicteid from fltor Ncdur
C*rt»on: LSS-2
WX>
Figure 2
40 ao no
DOC-Isotherms of a Chlorinated
Humic Acid (after Sander)
258
-------
® fe-ttLC,0-*
200
SO
8.0/2.0
qa-BO-c,"
2OOI
«O
Ci_. In
6
mg'l
50
8 « 2
6
In mg/l
Figure 3. Influence of Single Solute Data and Concentration
on the Total Adsorption Behavior
These findings were confirmed experimentally with a de-
fined ternary system. This mixture consists of phenol-4-sul-
fonic acid, 4-hydroxybenzoic acid (HBA), and 4-nitrophenol
(PNP). The sulfonic acid is the weakest-adsorbed substance,
while HBA and PNP are strongly-adsorbed compounds. Figure 4
shows the DOC isotherms of this mixture with different initial
concentrations of the three components. The greater the amount
of the weaker-adsorbed component, the greater is the isotherm
slope of the overall isotherm.
To show that the competitive adsorption model used in
this report can predict the adsorption of multi-solute systems,
Figure 5 presents a comparison between experimental and predic-
ted results for the mixture containing the largest amount of
phenol-4-sulfonic acid.
It can be summarized that the isotherms of mixtures
strongly depend on the single-solute behavior of the components
and on their concentration distribution. When this interaction
is recognized, it is possible to predict the behavior of the
mixture using the single-solute data of the components. None-
theless, the description of such a mixture as a pseudo-single
solute is an oversimplification. It may be adequate for a
qualitative evaluation, but it is quite insufficient for the
prediction of competitive adsorption effects.
To overcome the drawback of missing information on the
composition of the mixture, research was directed towards the
development of a combined laboratory- and calculation procedure
259
-------
® Phenol-4-Sulfonic Acid (5) Hydroxybenzoic Acid (§) 4-Nitrophenol
I
£
80
60
40
2.48/1.22/1.04
2.89/1.22/0.52
Cw-3.31 / Cao-0.61 / 030-0.52
1.0
2.0 4.0
CT in mg DOC/I
6.0 8.0
Figure 4. Influence of Initial Concentrations on the
DOC-Isotherms of a Three-Component Mixture
so
Phenol-4-Sutfonic Acid
C01 • 3.31 mgDOC/l
Hydroxy-Benzotc-Acid
Cos • 0.61 mgDOC/l
4-Nitro-Ph«nol
Cog- 0.52 mgDOC/l
cOTam* 4.44 mg DOC/I
JL
2 3
CT in mg DOC/I
Figure 5. DOC-Isotherm of a Three-Component Mixture
260
-------
which allows the prediction of mixture adsorption by reducing
the mixture to a binary or ternary system.
PREDICTION OF MULTI-SOLUTE ADSORPTION OF UNDEFINED MIXTURES
From a chemical point of view, raw water is a mixture
of many different substances with specific structures affecting
the quality of the final water. From the engineering point of
view, the raw water is a mixture of compounds with different
adsorbability. In contrast to the chemical viewpoint, it can be
assumed that fewer components need to be considered if only the
adsorption behavior is regarded, because it is possible that in
spite of the different chemical structures, there are similari-
ties in adsorption behavior. It will be necessary, however, to
reduce the number of components by introducing key components
which represent a specific range of adsorption behavior. This
is supported by the following assumption:
The overall adsorption is determined by the organic back-
ground of the fulvic , humic , and lignin-sulfonic acids. This
behavior can be described by regarding the unknown mixture as a
system of two or chree key components with different adsorb-
abilities. The effects of competitive adsorption with regard to
the behavior of trace organics can be predicted by using the
single-solute data of the key components and the data of those
organics which are of interest.
CHARACTERIZATION OF ADSORPTION BY SPLITTING INTO KEY COMPONENTS
The splitting procedure is based on the competitive
adsorption model which needs only single-solute data for the
prediction of competition. This model, strictly valid only for
defined systems, can be applied to unknown mixtures when conver-
ting this mixture into a semi-defined system. This can be
achieved by adding a specific amount of a defined substance to
the mixture. By watching the adsorption behavior of the added
compound (tracer substance) in the presence of the unknown
organics, it is possible to obtain information on the single-
solute data and on the concentrations of the key components.
Figure 6 shows the flow sheet of the procedure.
For this purpose the SCAM model was modified for applica-
tion to three and four component mixtures, respectively.
When considering the unknown mixture as a system of two
key components, there are six independent variable parameters
to describe the adsorption behavior of the unknown organics and
their influence on the adsorption behaviour of the tracer sub-
stance. The parameters are the initial concentrations of the
two key components C io and the constants K. and n^ des-
cribing the single-soiute behavior of the key components, ex-
pressed by the Freundlich equation a. = K. C. ni
261
-------
UNKNOWN ( <
MIXTURE "" \
V
A
( i
v
~
1
^s
2
J KNOWN
^ COMPOUND
1
3/
\^
N
T
2 )
/
ADSORPTION EXPERIMENT
1
\
\
«3
\
CALCULATION (SCAM)
|
tf **1
Hl*l\^ * C-|
\
<*•«* c?
Ml
u
cnn
u
«
X
n
9 1(79
Figure 6. Flowsheet of Splitting Procedure
200
» 150
o>
i 100
'i
9
SO
20
AC:LSS
cOPNp'Sm«"
0.1
1X)
5.0 10.0
Figure 7. PNP-Isotherms in the Presence of Two
Different Lignin Sulfonic Acids
262
-------
The number of parameters has to be reduced due to a simpli-
fication of the fitting procedure. The initial concentrations
of the key components can be coupled with the total initial
concentration of the mixture
'TO
= C
10
+ C
20
The adsorption behavior of the single solutes can be
formulated arbitrarily or by using the adsorption behavior of
defined compounds representing good or poor adsorbability.
A more precise way of making inferences regarding the
single-solute behavior is to interpret the adsorption behavior
of the tracer substance and of the overall isotherm of the
mixture. The adsorption behavior of the tracer component gives
valuable information on the strongly-adsorbed compounds in the
mixture when using a tracer substance known as a well-adsorbing
substance.
Figure 7 shows the PNP isotherms in two different mix-
tures of lignin sulfonic acids. While the lower isotherm is
shown as a straight line in the log-log plot, the other isotherm
shows curvature. This is the result of a better adsorption be-
havior of a certain part of the lignin sulfonic acid. When
observing such an effect, we can better infer the adsorption be-
havior of the key component representing the strongly-adsorbed
part of the unknown mixture (5,6).
Information on the adsorption of the weaker-adsorbed
component can be obtained from the overall isotherms for lower
concentrations. There, the isotherm is almost identical with
the single-solute data of the weakest-adsorbed compound in the
mixture. As an example, Figure 8 shows the total isotherm in a
log-log plot of the mixture described in Figures 4 and 5. In
the concentration range where the overall isotherm is nearly
identical with that of phenol-4-sulfonic acid, the only compound
remaining in the solution is the sulfonic acid.
100
i
r
s
50
10
i PhMMl-4-Sulfonic Acid (2) Hydrexybenzofc Acid © 4-Nttroph*no<
smgto sokit* teothwm of
PfwnoM-SuLAckJ
0.01
0.05 0.1 OS 1.0
CT in mg DOC/I
5.0 10.0
Figure 8. DOC-Isotherm of a Three-Component Mixture
263
-------
The evaluation of the single-solute behavior by means of
overall isotherms and the shape of the tracer isotherm is an
important step for calculating the concentration distribution of
the key components. The definition of fixed adsorption behavior
seems to be advantageous, especially if the characterization of
the adsorption is carried out with more than two components.
The whole range of adsorbability can then be divided into inter-
vals with defined adsorption data.
In the case of characterization by two key components it
is advantageous to define the adsorption data of one component
only and to vary the other data in order to get a better des-
cription of the adsorption behavior. This way it is possible
to obtain a number of solutions; i.e., combinations of initial
concentrations and single-solute adsorption data. All these
combinations can predict the competition of the key components
with the tracer substance.
To be able to decide which of the combinations will produce
the best description of the adsorption behavior of the mixture,
an additional criterion is necessary. There are two criteria
which can be used for this purpose. One is the DOC isotherm of
the unknown mixture without the tracer substance, and the other
the DOC isotherm in the presence of the tracer substance. That
combination which agrees best with the two criteria should be
used for the characterization of the mixture.
RESULTS OF THE CHARACTERIZATION METHOD
For demonstrating the proposed method, a lignin sulfonic
acid (LSA) solution was used. After flocculation of a commer-
cial product to separate high-molecular-weight compounds, the
overall adsorption behavior can be compared to that of an actual
raw water of the Rhine River (5). The aim of the characteri-
zation method is the description of the entire adsorption by two
key components. For this purpose, three adsorption experiments
are necessary to evaluate the following:
1. The single-solute isotherm of the tracer substance
(p-nitrophenol = PNP);
2. The overall isotherm of the LSA;
3. The isotherms of the lignin sulfonic acid as well
as those of the PNP in the mixture. The concen-
tration of PNP can be measured independently of DOC
by UV absorbance at 317 nm.
Figure 9 shows the experimental results. Curve 1 repre-
sents the LSA isotherm and curve 2 the LSA isotherm in the
presence of 1 mg/1 PNP. The curvature of the second isotherm is
264
-------
a good indication for the presence of a larger amount of weakly-
adsorbed substances. The isotherm of PNP in the presence of LSA
shows no curvature.
100
80
>
,60
40
100
i
50
1mg/l PNP
1.0
cinmgDOC/l
6.0 8X) tOX)
10
PNP
single
solute
0.52 mg PNP (DOC)
+ 4.7mg LSA (DOC)
0.01
0.1
cf>NPin mg DOC/I
1.0
Figure 9. LSA and PNP Isotherms as Single
Solutes and in Admixture
Therefore, the single-solute isotherm of the component
representing the strongly-adsorbed substances was assumed to be
only 90 percent of that of the tracer substance. The single-
solute data of the second component were varied. By means of
the proposed equilibrium model, a number of combinations were
calculated and compared to the two LSA isotherms.
Figure 10 gives a comparison between experiment and cal-
culation for the best fitting combination. There is reasonable
agreement between the forms of the measured and the predicted
LSA isotherm. The single-solute data used for the calculation
265
-------
are shown at the top of the figure. Curve number 3 is the
measured LSA isotherm in the presence of PNP;curve number 4
the corresponding isotherm, predicted with the same data as
for curve number 2. In this case the agreement is less satis-
factory, but the effects of strong competition are evident.
single solute data for the prediction of (2) + @
c-n • 3.82 mg DOC /I c20 = 0.96 mg DOC/1
* 22.7 -c?4
q2 -69.6-C2'2
80
§
I
£
8
LSA+PNP
AC:LSS
I
i
1.0
2.0 4.0
c in mg DOC/I
6.0
8.0
Figure 10. Comparison of Measured and Predicted DOC-Isotherms
for Lignin Sulfonic Acid (LSA)
Similar results were obtained by using another type of
activated carbon with the same LSA (Figure 11). The prediction
is not met as well as it is in Figure 10.
One of the reasons for the deviations between experi-
mental and predicted curves may be the simplification or re-
duction to a two-component system. It can be assumed that the
characterization by three or four components improves the
quality of description. The other reason is an experimental
one. Unfortunately, DOC measurement is less exact than 5
percent. This can result in deviations as large as 30 percent
in the carbon loading, and therefore to an apparent larger
deviation between calculation and experiment.
266
-------
single solute data for the prediction of (5) + ®
c10 - 4.25 mg DOC l\ cx • 0.47 mg DOC/1
q1 » 26.3-c?4
q2 -81.3-of2
100
80
60
40
20
LSA
J LSA+PNP
AC: ROW
1.0
2.0 4.0
c in mg DOC /1
6.0
8.0
Figure 11.
Comparison of Measured and Predicted DOC-Isotherms
for Lignin Sulfonic Acid (LSA)
Nevertheless, this kind of treatment for an unknown mixture
will improve the prediction of adsorption behavior, especially
in regard to competitive adsorption. While it is possible to
describe the competition between the key components and organics
of interest by the proposed adsorption model, it has proved im-
possible when regarding the unknown mixture as a single solute.
This is shown byBaldauf (3).usingthe proposed characterization
method for the prediction of GAC flter performance.
CONCLUSIONS
It has been shown that the adsorption behavior of actual
waters or similar solutes is determined by the adsorption
behavior of the single solutes and their concentrations. The
main effects of mixture adsorption can be predicted using a
competitive adsorption model.
267
-------
The application of such a model to unknown systems re-
quires knowledge of the single-solute behavior. This was
investigated by using a combined experimental- and calculation
procedure, splitting up the unknown mixture into two key compo-
nents of defined quasi-single-solute behavior. With the aid
of these two hypothetical components, it is possible to de-
scribe the adsorption behavior of the mixture as well as the
competitive effects when additional solutes are present.
268
-------
REFERENCES
1. Jain, J.S. and V.L. Snoeyink. 1973. Adsorption from
Bisolute Systems on Active Carbon. WPCF 12, 2463-2497.
2. Jossens, L. et al. 1978. Thermodynamics of Multi-Solute
Adsorption from Dilute Aqueous Solutions. Chem. Eng. Sci.
33, 1097-1106.
3. Baldauf, G. and B. Frick and Sontheimer, H. 1977.
Berechnung des Sorptionsverhaltens von Gemischen.
Vom Wasser 4_9, 313-329.
4. DiGiano, F.A., G. Baldauf, B. Frick, and Sontheimer, H.
1978. A Simplified Competitive Equilibrium Adsorption
Model. Chem. Enq. Sci. 33, 1667-1673.
5. Frick, B., R. Bartz, H. Sontheimer, and DiGiano, F.
A. Predicting Competitive Adsorption Effects in Granular
Activated Carbon Filters. ACS Meeting, Miami Beach (1978)
6 Martin, R.I and K.S Al-Bahrani, 1977. Adsorption Studies Using
Gas Liquid Chromat ography - II. Competitive Adsorption.
Water Research 11, 991-999.
269
-------
LABORATORY ACTIVATED CARBON TEST METHODS FOR WATER
UTILITIES
Gerd Holzel
Many different types of activated carbon are offered for
water-treatment purposes by various producers. In order to
find the most suitable and most economic product for a particu-
lar application (1,2), it is necessary to have specific test
methods.
OVERVIEW OF TEST METHODS
Table 1 gives a summary of tests which, in our opinion, are
helpful for evaluating activated carbon products for use in
water utilities. As a first step, it is necessary to get a
representative sample of the carbon. Because this is often
difficult, it is important to come to an agreement with the
producer over sampling.
The mechanical test methods yield information on apparent
density, necessary to determine carbon quantities, and on
particle size distribution. A good evaluation of the sieve
analysis is the method accepted by the German Materials Stand-
ards Institute (DIN) (3). This analysis gives two parameters:
a characteristic particle diameter, and the particle size dis-
tribution.
Another important mechanical property of granular activated
carbon is abrasion resistance. A weak carbon may produce large
losses in the filter operation, and the fine particles may
cause complications in the distribution system.
The physical and chemical parameters are easy to guarantee
by the producer. In the case of impurities, this guarantee is
of hygienic and toxicologic importance; but we never have
measured significant concentrations of those substances in
activated carbons used by the German waterworks.
The pore volume of the carbon, especially the pore
volume distribution [4,11] give information for equilibrium and
kinetic properties. However, it seems to be more convenient for
270
-------
Table 1
Test Methods for Activated Carbons
1. SAMPLING
2. MECHANICAL TEST METHODS
2.1 Apparent Density
2.2 Particle Size Distribution
2.3 Abrasion Resistance
3. PHYSICAL AND CHEMICAL PARAMETERS
3.1 Moisture
3.2 Impurities
4. PORE VOLUME
4.1 Densities
4.2 Pore Size Distribution
5. COMPETITIVE ADSORPTION TEST
6. KINETIC TEST METHODS
6.1 External Mass Transfer
6.2 Effective Diffusion Coefficients
7. THERMOANALYTICAL TEST METHODS FOR
SPENT CARBONS
271
-------
the user to measure both properties directly by sorption tests
(i.e., the competitive adsorption test and the kinetic test).
These test methods will be discussed in detail later in this
paper.
Finally, to characterize spent carbons, we find it suitable
to use thermoanalytical test methods to determine significant
parameters, not only for the characterization of the loading of
spent carbons, but also for a subsequent regeneration.
For the adsorption behavior, several test methods were
evaluated. Some of them, such as iodine number or molasses
number, are described in the AWWA [5], ASTM [6], or the German
DIN [7] standards. But, to.,get direct information about the
adsorption capacity (e.g. m water treatable with 1 m of
granular activated carbon), it was necessary to develop new
methods beyond the known empirical ones. This new concept of
testing granular activated carbon is presented in this paper.
TESTING ADSORPTION BEHAVIOR
From filter measurements, it is well known that weakly
adsorbed substances are replaced by more strongly adsorbable
substances. This effect was considered in the proposed method
for a competitive adsorption test. Figure 1 shows the experi-
mental results of such a test. Here, the loading of the carbons
is plotted vs. the equilibrium concentration. The dashed lines
represent the Freundlich isotherms of organic substances in
natural water which enters the GAC filters. The carbon manu-
factured by Norit, indicated by the thick dashed line, has clear
advantages over carbon NK 12, represented by the thinner dashed
1 i ne.
The competitive conditions in a real filter are simulated
by adding 1 mg/1 p-nitrophenol (PNP) to the water under in-
vestigation. This admixture causes an evident change: the
loading of organic substances in water (OSW) on Norit carbon
decreases considerably, as shown by the thick solid line that
represents the isotherm of OSW with the admixture of PNP,
whereas the loading of OSW on NK 12 changes to a much lesser
extent, as shown by the thin solid line. The adsorption of PNP
from the same solution mixture is also different for both carbons
the NK 12 shows better adsorption characteristics than the Norit
carbon.
Frick (9) explained how these equilibrium isotherms could
be used to describe the adsorption behavior of multicomponent
mixtures. Now it will be shown how the same data can be used
to get representative information on the adsorption capacity of
the carbons.
272
-------
6
4
0>
0) 2
£
cr
i '
8
q = kfcn CosvsnrwaU)
fiosw • = 1 Tig m / 1
OSW OSW « PNP '
.__ ..__ KinD|T i
NIK t? /
MI\ if. /
7 ^
7
_— " ^- Ay/
/
/
PNP
OSW
< 6 ,Q-1 2 4 6 1Q0
concentration c in mg /1
< 6
FIGURE 1
Comparison of the Equilibrium Isotherms of Two Different
Activated Carbons.
PNP = p-nitrophenol, OSW = organic substance in water.
273
-------
A practical evaluation method was developed by Martin,
et al. (10). In this method, filter breakthrough behavior is
predicted based on a model assumed of a total mixed reactor;
therefore, only the filter influent concentration and two
parameters of the Freundlich isotherm, K_ and n, are needed.
r
In Figure 2, the dimensionless concentration, c, is divided
by the constant influent concentration, c , and is plotted vs .
the throughput V , expressed as 10 m wa?er treated by
one m GAG. For a given throughput, an actual filter effluent
concentration, as indicated by the upper line in the figure, is
obtained. In treatment plants, filters with different carbon
qualities, different operation times, and therefore different
effluent concentrations normally work in parallel. The effluent
of all these filters then is blended. To characterize the
carbon under investigation for such a filter plant it is assumed
that all the effluent from that given carbon filter is collected
in the same tank. The concentration in this tank, which is the
integral filter effluent concentration, is given by the lower
curve. For comparison of different carbons, it is helpful to
define a criterion for terminating a filter run; for example, a
throughput where the integral filter effluent reaches a certain
concentration. This parameter may differ for different treat-
ment plants. Here, a purification of 75 percent, or a remaining
concentration of 25 percent of the influent concentration is
assumed as the significant test value, V *.
1.0
(V0.536 g/cm3
carbon : SF300,
i 2.1mg/l
: 61 *>'3g/g
n : 0.56
actual concentration of
filter effluent
integral filter effluent
itratkMi
0.25
40 50
in103m3/m3
60
FIGURE 2
Determination of the Competitive Adsorption Test Data
274
-------
The test method not only allows the comparison of virgin
granular activated carbons, but also reactivated ones.
In Table 2, there are collected different test values for
the OSW and PNP equilibrium data obtained by the competitive
adsoL-ption test and by the former explained procedure. In the
first section of Table 2, four carbons of the same type, but
reactivated in different reactivation plants, are compared with
SF 300. There are no significant differences in the V*-OSW
throughput values, but there are in the V*-PNP data, it seems
that the V*-PNP test values are of great importance for re-
act ivationv control. This will be illustrated later in this
paper.
To make the competitive adsorption test useful for the
producer and to make comparison possible for water utilities,
we have tried to find a substance that represents the organics
in the waters to be treated. After a long series of experiments
we obtained good results with flocculated lignin sulfonic acid
(LSA). In the second section of Table 2, where those values are
listed, slight differences are found between the V*-values for
LSA but significant differences are found in the values for PNP.
In the last section of Table 2 the test values of various
virgin GAC's are given in order to demonstrate the possible
range of the test results. Those values differ for LSA between
45 x 10"* m^ treated water per m GAG for the best carbon, down
to 11 x 10 m /m for a less effective one. It is also
possible to distinguish between the different selectivities of
the carbons. The LSS 1.5 seems to be especially effective for
removing PNP.
It is obvious that good reactivation is possible. The data
from the reactivated carbons in the second section of Table 2 do
not differ significantly from those of the virgin carbons in the
last section.
As mentioned earlier, a smaller V*-PNP value implies a
lesser suitability for water-treatment purposes. For more in-
formation about this effect, we have analyzed the pore structure
of various types of GAC. The Freundlich coefficient, K of PNP,
the KF of crystal violet (KV) dye stuff, and the iodine number
were correlated with the microporous volumes of those carbons.
These microporous volumes are defined by an upper pore radius.
The results are shown in Figure 3, in which the K values and
the iodine numbers are plotted -vs. the pore volume expressed in
mm per g GAC. It is obvious that a small K or a small
iodine number is due to a small por«e volume. Hence, it follows
that it is important in the activation of virgin carbons and the
reactivation of spent carbons to choose the conditions for
reactivation in such a way that the microporous volume is in the
desired range.
275
-------
Table 2
Competitive Adsorption Test Data
A. CARBON
SF300
R6
R 7
R8
R16
SF300
R 6
R 7
R 8
R 16
LSS 3
LSS 1.5
NORIT SM
NK12
CHEM 5
WATER
osw-
BENRATH
PNP
LSA
+
PNP
LSA
+
PNP
VvQSW
103m3/m3
21
19
19
16
26
16
21
17
18
17
11
13
21
20
45
VCPNP
103m3/m3
54
24
22
26
39
52
23
11
23
45
25
60
22
34
30
276
-------
IN
- f (VP)r
• Iodine Number Pore Volume up to r- 1 nm
oRNP r-0.7nm
• KV r-1.0nm
0.5-
:0.4
Oi3
[0.2
0.1
1500
- Z1000
500
I
100 200 3OO
PoraVWunw Vp inmm3/g
FIGURE 3
Correlation Between Micropore Volume and Freundlich
Coefficient or Iodine Number
400
cooling
water
activated
carbon
test-cell
pump
FIGURE 4
Apparatus for Kinetic Tests
277
-------
In many cases, not only the equilibrium characteristics of
a carbon, but also the kinetic behavior are of considerable im-
portance. The apparatus shown in Figure 4 was developed for a
kinetic test which would supply data that can be used for filter
breakthrough behavior prediction. Therefore, the test solution
is pumped from a double-walled tempered breaker through the
activated carbon test cell, in which the thin layer of carbon
is fixed to avoid abrasion.
For kinetic measurements, it is possible to distinguish
between two rate-limiting steps in GAC treatment, namely:
1. The transport of organics from the bulk solution
through a boundary layer to the carbon particle (ex-
ternal mass transfer), and
2. The diffusion of adsorbed substances within the carbon.
We have studied the external mass transfer resistance for
various types of GAC and different substances. Figure 5 pre-
sents the results for PNP. Here, the measured mass transfer
O -1«tS (1-
Sh - f(R«,Sc)
2-
vp'0.29 I A SNK12
!*-0.39
• SF300
D Merit Row Oj8
: A BKA
! O L»v 35
V L«v25
10'
-i-.. U. i
I
i i
10U
_
ntynoftdi At * y
FIGURE 5
Sherwood Numbers for Different Carbons Compared with
Gnielinski's Correlation
278
-------
coefficientsr designated as B.r are expressed by dimensionless
Sherwood nuntoers, and the fluid velocity within the test cell is
expressed by Reynolds numbers. The dashed and solid lines are
calculated for different bed porosities from Gnielinski's (14)
correlation. This correlation for heat and mass transfer in
packed beds has been verified for many similar chemical engi-
neering problems. It is obvious that the Sherwood numbers for
our experiments are higher than the values predicted by the
correlation. This leads us to the idea to take into considera-
tion an additional surface which originates from the outer pores
of the activated carbon particle and to correct the measured
mass transfer coefficients to account for these areas.
The effective surface area can be determined by an inde-
pendent experiment based on mercury porosimetry or by a rapid
kinetic test with a substance of known diffusivity in the bulk
liquid. Thus, it becomes possible to predict the mass transfer
coefficients for different fluid velocities and different GAG
types.
Besides the external mass transfer, the transport within
the carbon particle is also of importance for the adsorption
kinetics as shown by Baldauf (13). This overall mass transfer
can be described by the homogeneous diffusion model as explained
by Merk (8). In this model, there is only one adjustable
parameter, the effective diffusion coefficient, Deff This
effective coefficient is another significant kinetic test value
in this concept.
In Figure 6, the dimensionless concentration is plotted vs.
time. The points are the measured concentrations, and the solid
line was calculated with the homogeneous diffusion model. The
effective diffusion coefficient was evaluated for the best fit.
In the tabulation included in Figure 6, the diffusion coeffi-
cients, as determined by this kinetic test, are given for four
different carbons. Three of them were reactivated samples. The
obtained values vary from 3.3 x 10 m /sec to 6.1 x 10
m /sec.
With these three test methods, it is possible to predict
filter breakthrough curves for substances if their diffusivities
in the bulk liquid are known; it is important to have this capa-
bility. In practice, however, we are not concerned with single-
solute solutions, of which we know the bulk liquid diffusivity,
but rather with unknown multicomponent systems of indefinite
composition. Before further calculations, are possible, the
effective diffusivities of the mixture components must be known
or a mean diffusion coefficient has to be estimated. This
279
-------
1.0
o
o
O 0.5
Effective Diffusion Coefficients (D^) of Different Carbons
Measured with OSW(Holthausen)
Activated
Carbon
0*r1O"*nPlme
Norit
ROW 03
3.3
Reac.8
2.6
Reac.13
4.1
Reac.21
6.1
Reac. 13
OSWHott.
D-f-4.1-10"15m2/sec
10
Time t in mina5 20
30
FIGURE 6
Determination of the Surface Diffusion Coefficient
becomes possible together with the above mentioned correlation
of Gnielinski, if carbon properties are known. Table 3^illus-
trates that a given water/ analyzed with different CiACs, is
characterized in all tests by the same bulk diffusivity, D , for
the organics. The mean value was in this case 1.0 x 10
m /sec. On the other hand, we measured different diffusion
characteristics for different waters using the same carbon.
The analyzed waters were taken from carbon filter influents from
Langenau, Mulheim, and Stuttgart.
With this method, it also becomes possible to determine the
influence of water treatment on the mean bulk diffusion coeffi-
cient of the remaining organics. In the lower part of Table 3,
it is shown that significant differences exist between the raw%
water and the water after ozonization and filtration. There is
a considerable improvement in view of the adsorption kinetics
expressed by the bulk diffusivity of the organics in treated
waters from different Dusseldorf treatment plants. The bulk
diffusion coefficient increases from 1.0 x 10
to 5.1 x
10~iu
m /sec at the Flehe treatment plant, and from 0.9 x
1,0 iu to 3.9 x 10~iu m /sec for the Holthausen plant.
280
-------
TABLE 3
Diffusion Coefficient (D ) of Different Organics
in Water
Water
Holt.
Activated Carbon
Norit ROW Reac.8
1.0 1.2
Reac.21
0.9
Reac.13
0.9
Mean
Value
1.0
Lang.
Miil.
Stutt.
2.9
0.6
0.5
INFLUENCE OF TREATMENT ON THE DIFFUSION
COEFFICIENT
Water Raw Water
[DL] - 10'10m2/sec
after Ozonation and Filtration
Rene
Holt.
1.0
0.9
5.1
3.6
281
-------
All the significant parameters which are necessary to
describe the adsorption behavior of granular activated carbons
have been discussed. Figure 7 is a schematic diagram of the
proposed test concept.
TEST
METHOD
INPUT
COMPETITIVE
ADSORPTION TEST
FILM DIFFUSION
TEST
SURFACE
DIFFUSION TEST
OUTPUT
OPERATIONAL
TEST
PARAMETERS
FILTER
PREDICTION
FIGURE 7
Test of Adsorption Behavior of Activated Carbon
282
-------
First, we have to evaluate the equilibrium parameters, Kp
and n. Frick (9) has shown how to determine those parameters
for multicomponent solutions. For comparison and classification
of different carbons, w,e can calculate the Y$-OSW and the V^-PNP
valuescorrespondingito m^ water treated.per m GAC.
Secondly, we can determine a mass transfer coefficient with the
film diffusion test and evaluate an effective external surfacearea
A ff Finally, we can measure an effective diffusion coeffi-
cient as a characteristic value for the internal kinetic be-
havior of the carbon under investigation.
With the results of the granular activated carbon tests
and the diffusion characteristics of the organics in the bulk
liquid, it is possible to predict filter breakthrough curves
of unknown multicomponent solutions, as discussed by Baldauf
(13).
ACKNOWLEDGEMENT
I would like to express my thanks to the waterworks of
Dusseldorf, Wuppertal, and Cologne, and to the Ministry of
Research and Technology of the German Federal Republic for their
support in this work.
REFERENCES
1. Sontheimer, H. and G. Holzel, Grundlagen der Aktivkohlead-
sorption zur Aufbereitung von Trinkwasser. Haus der
Technik, Essen, Germany, Vortragsveroffentlichungen 404
(1977).
2. Adsorption, Translation of Reports on Special Problems of
Water Technology, Vol. 9, edited by H. Sontheimer, Univer-
Sitat Karlsruhe, EPA-600/9-76-030, December 1976; US-Envi-
ronmental Protection Agency, Cincinnati, Ohio (1976).
3. Darstellung von Korngrossenverteilungen im RRSB-Netz DIN
66145, Beuth-Verlag, Gerlin.
4. Holzel, G. and H. Sontheimer, Methoden und Modelie zur
Strukturierung des Porenvolumens von Wasserreinigungsaktiv-
kohlen, Vom Wasser, Bd. 52, Verlag Chemie, 1979.
5. AWWA-Standard for Granular Activated Carbon (AWWA B 604-74)
J. AWWA (1974), 672.
6. ASTM-Committee D-28 on Activated Carbon (1974).
7. Aktivkohlen zur Wasseraufbereitung, DIN 19603 (1969)
BeuthVerlag, Berlin.
283
-------
8> Merk,- W. Prediction of Multicomponent Adsorption Behavior
in Activated Carbon Adsorbers: Kinetic Aspects, pp. 234-
255, in this Symposium.
9. Frick^B. Prediction of Multicomponent Adsorption Behavior
in Activated Carbon Filters: Equilibrium Aspects. pp 256-
269 in this Symposium
10. Martin, Holzel, Sontheimer, Eine Methode zur Bewertung der
Gleichgewichtseigenschaf ten von Aktivkohlen zur Trinkwasse-
raufbereitung (in preparation).
11. Holzel, Lentz, A Mercury Porosimeter for Pressures up to
1.5 kbar Accepted by High Temperatures - High Pressures.
12. Fritz/ Konkurrierende Adsorption von zwei organischen
Wasserinhaltsstof f en an Aktivkohlekornern. Dissertation,
Universitat Karlsruhe, 1978.
13. Baldauf, G. , Prediction of Breakthrough Patterns for
Organics from Laboratory Tests, in this Symposium.
14. Gnielinski, Gleichungen zur Berechnung des Warme- und
Stof f austausches in durchstromten ruhenden Kugelschuttungen
bei mittleren und grossen Pecletzahlen. Verf ahrenstechnik,
12, 363, 1978.
284
-------
PREDICTION OF BREAKTHROUGH PATTERNS OF ORGANICS
FROM LABORATORY TESTS
Dr. Gunther Baldauf
The papers by Merk, Frick, and Holzel presented different
aspects of dealing with the problem of multicomponent mixtures
of organics in waters to be treated for drinking water purposes.
The following views are widely accepted:
1. The efficiency of granular activated carbon filters can
be predicted for mixtures containing two or three single
components, using equilibrium and kinetic data for the
single components only.
2. Multicomponent mixtures of organics as they are found in
our natural waters can be described as a mixture of two or
three single groups of organics, each group representing a
range of adsorption behavior.
3. While this division into groups is possible for the
equilibrium data, we have also been able to determine
adsorption kinetic data for multicomponent mixtures.
By using this knowledge, we can predict the breakthrough
behavior of waterworks filters.
LABORATORY TESTS
To study these problems, special filter tests have been
made in Karlsruhe. They allow a comparison between experimental
data and precalculated concentration profiles, thus showing
whether the methods discussed so far are worthwhile v;heru
applied in practice. Two different types of activated carbon
were used for the experiments. The adsorption isotherms are
shown on a log-log scale in Figure 1.
285
-------
»or
50 -
f
s
I
Lignin SUfonic Acid
K •O2-10"1zm2/«
.02-10~0m2/«
to
2JO 4.0 6jO B.O
concentration c in mg DOC /1
Figure 1. DOC-Isotherms of LSA for Two Activated Carbons
A lignin sulfonic acid solution (LSA) from which the high-
molecular components were removed by flocculation was used in
these experiments. As indicated by Holzel (I), this mixture of
lignin derivatives behaves similarly to organic substances in
Rhine-water during adsorption. Two types of activated carbon
were studied. One can see from Figure 1 that of the two carbon
types, the adsorption capacity of the carbon R is considerably
better. Therefore, this carbon should produce the better
results in an activated carbon filter. If one compares the two
internal diffusion coefficients, which are also shown in Figure
1, one can see that the carbon L,which adsorbs less strongly as
evidenced by the equilibrium isfcfeherms, has a much better
kinetic behavior than the carbon R. Exactly the same tendencies
are found if UV absorption is used as a measure for concentra-
tion instead of dissolved organic carbon (DOC). As a conse-
quence, the isotherms are less steep, and according to Frick
(2), mainly those substances which behave as a single solute
will be covered by UV absorption, as far as adsorption is
concerned. To learn how the two activated carbons behave in an
activated carbon filter, experiments were made with a pilot-
plant filter.
A plexiglass column with an inner diameter of 52 mm and a
height of 1.50 m was used. The depth of the carbon layer was
1 m. Sampling tubes were installed at intervals of 10, 30, 50,
70 and 90 cm. The filter velocity was 10 m/h. A lignin
286
-------
sulfonic acid solution containing 4 mg DOC/1 was fed to the
filter. The concentration of lignin sulfonic acid was analyzed
spectrophotometrically at 280 nm. Parallel to this, the DOC
value of all samples was measured.
In Figure 2 the experimental UV breakthrough curves of the
two activated carbon filters are shown. In the ordinate of this
diagram the UV values for two bed heights related to the UV
values of the influent are plotted above the period of operation
of the filter. One can see that the UV-active substances in the
filter containing the carbon with poorer kinetics but a better
equilibrium behavior will produce an earlier breakthrough. This
example shows clearly that it is important to observe the
adsorption kinetics for the adsorption in the fixed-bed
adsorbers. For a comparison of carbons of different kinetic
behavior, it is not enough to use equilibrium data only for the
characterization of activated-carbon quality. To check whether
the adsorption process in the column is film-diffusion control-
led, the breakthrough curves were calculated using the film
diffusion model. This model was used because Merk (3) had shown
that the adsorption of low-molecular weight substances is
film-diffusion controlled.
so
100
150 200
tint* in hours
250
3OO
350
Figure 2. UV-Breakthrough Curves for Two Activated Carbons
In Figure 3 the calculated concentration profiles at
different bed heights are plotted. Beside the calculated
curves, the measured points are also shown. Apart from the
initial period, the UV-active substances in the filter have a
quicker breakthrough than would be expected from the
287
-------
10
05-
>z>10cm
o o
SO
SO 100 150 200 250 300 350 £00
tminhoiira
Figure 3. UV-Breakthrough Curves for Lignin-Sulfonic-Acid
(Activated Carbon R). Experimental data are
compared with curves calculated by the film
diffusion model [3]. The parameter—10,30, and
50 cm—refers to the sampling depth in the
activated carbon bed.
calculation. This earlier breakthrough can be explained by an
additional solute transport resistance within the carbon grains.
Whereas with the simple film diffusion model it is assumed
that the entire transport resistance takes place within the
film covering the carbon particle, the film homogeneous diffu-
sion model takes into consideration an additional transport
resistance within the grain. Despite the low overall concentra-
tion, the transport resistance has to be taken into account
because of the relatively high molecular weight of the sorbing
substances. For the calculation of the breakthrough curves
according to the film homogeneous diffusion model the diffusion
coefficient DR within the carbon particle must be known. This
can be obtained simply from the reaction- and time-curve.
In Figure 4, the curves calculated with this model are
plotted in addition to the measured points. Taking into account
the internal mass transport resistance within the carbon par-
ticle, a good description of the experimental results can be
obtained.
So far, the lignin sulfonic acid has been regarded as a
single-solute substance. This simplifying assumption is
288
-------
1.0-
0.5-x
T
50
100
150
200
time in hours
300
350
600
Figure 4. UV-Breakthrough Curves for Lignin-Sulfonic-Acid in a Pilot Plant
(Activated Carbon R). The paramster—10, 30, 50, 70, and 90 can—
refers to the sampling depth in the activated carbon bed. Experi-
mental data are compared with curves calculated by the film-
homogeneous diffusion model (3).
i.o-
0.5-
z >10cm
viffiMion
50 100 150 200 250
thm in hours
300
350
i.00
Figure 5. DOC-Breakthrcugh Curves for Lignin-Sulfonic-Acid in a Pilot Plant
(Activated Carbon R). The parameter—10, 30, and 50 cm—refers to
the sampling depth in the activated carbon bed.
289
-------
justified here since a mere fraction of the total substance can
be detected spectrophotometrically. These compounds differ only
slightly from each other. If the DOC value instead of UV adsorp-
tion is used as a measure for the concentration, systematic
deviations between measured values and those calculated from the
film homogeneous diffusion model will result. This can be seen
in Figure 5.
The DOC breakthrough predicted by the film homogeneous
diffusion model are plotted together with the experimental
data. The earlier breakthrough in the pilot plant filter
may be explained by assuming that only part of the components
that are detected by DOC measurement are easily adsorbed,
whereas compounds more difficult to absorb cannot be retained in
the filter. For this reason, the lignin sulfonic acid was
divided into fractions which are easy or difficult to absorb, in
accordance with the method suggested by Frick [2]. Figure 6
indicates the breakthrough curves calculated by using the film
homogeneous diffusion model, assuming such a two-component
mixture.
The good agreement with the measured points in Figure 6
shows that the description of the DOC breakthrough curves is
also possible for complex mixtures of substances if they are
divided into two groups.
Apart from the high-molecular-weight compounds, represented
here in the form of lignin sulfonic acid, the undesired and in
some cases toxic low-molecular-weight compounds must also be
removed in an activated carbon filter. The following is a
demonstration of how these species, mostly present in low
concentrations only, behave. P-nitrophenol was selected as a
model substance for these low-molecular compounds because it
can be analyzed quickly and exactly, and because its properties
are similar to the aromatic chlorinated compounds which are of
particular importance in the Rhine River.
Filter tests were made in the pilot plant with a mixture
consisting of lignin sulfonic acid with a DOC concentration of
4 mg DOC/1 and 1 mg/1 of p-nitrophenol. The DOC concentration
and the UV absorption were measured at the inlet and at several
sampling points along the column.
With these data the concentration of both substances can be
calculated. The breakthrough patterns of p-nitrophenol are
plotted in Figure 7. Assuming that there is no mutual influence
between the two substances in the adsorbed phase, the break-
through curve No. 1 is calculated. A comparison with the
measured points shows that this simplified assumption does not
hold, because the p-nitrophenol breaks through much earlier. On
the other hand, by regarding p-nitrophenol as a single-solute
substance, which is adsorbed from an aqueous lignin sulfonic
290
-------
1.0
7T0.5
X X X
,10cm
X „ X
predicted curves
(Film-Homog«n*ou«-Diffu*ion)
50
100
150 200
time in hours
250
300
350
Figure 6. DOC-Breakthrough Curves for Lignin-Sulfonic-Acid in a Pilot Plant
(Activated Carbon R).
1.0 -
XX)
200
300 4OO
Km* in hour*
500
Figure 7. Breakthrough Curves for p-nitrophenol (PNP) in Presence of Lignin-
Sulfonic-Acid (LSA) By Activated Carbon R . Closed circles are
experimsntal data. Calculated curves: 1) no interaction between
solutes, 2) regarding PNP as a single solute in aqueous LSA solu-
tion, 3) considering conpetition between PNP and LSA, 4) consider-
ing conpetition between PNP and the better adsorbable portion of
LNA, and 5) considering conpetition between PNP and both portions
of LSA.
291
-------
acid solution, curve No. 2 will result. Nevertheless, this
curve also shows a large deviation from the data. The third
possibility is to calculate the concentration profiles of the
lignin sulfonic acid and of p-nitrophenol as a two-component
mixture with the film homogeneous diffusion model, taking into
account a mutual competition in the activated carbon filter,
with the result shown in curve No. 3; the results obtained are
still unsatisfactory. The reason why it was not possible in
cases 1, 2, and 3 to describe the measured concentration
profiles is that the lignin sulfonic acid was treated as a
single-solute substance. However, the fact that this simplifi-
cation is not permissible has also been shown by the results
obtained with the pure lignin sulfonic acid. For this reason
the lignin sulfonic acid will subsequently be treated as a
two-component system, consisting of a well- and a poorly-
adsorbable portion.
If the better-adsorbable component of the lignin sulfonic
acid only determines the breakthrough behavior of p-nitrophenol,
then curve No. 4 in Figure 7 will result. It is shown that the
better-adsorbable part of the lignin sulfonic acid, amounting to
about 10 percent of the total substance, contributes decisively
to the early breakthrough of p-nitrophenol. The deviations,
arising even here, seem to indicate that the less-adsorbable
part of the lignin sulfonic acid, amounting to about 90 percent,
also competes with the p-nitrophenol. A precalculation of the
measured concentration profiles will be possible only if the
less-adsorbable component is also taken into account.
The precalculation of three-component mixtures is rela-
tively complicated. Therefore, the attempt was made to regard,
mathematically, the three-component- as a two-component mixture.
As the less-adsorbable portion of the lignin sulfonic acid is
insufficiently retained in the filter, it can be assumed that
for p-nitrophenol and for the better adsorbable component of the
lignin sulfonic acid, the activated carbon will be in
equilibrium with the concentration of the poorly-adsorbable
component of the lignin sulfonic acid. A comparison of the
calculated curve No. 5 with the measured points shows that with
this kind of calculation an earlier breakthrough is achieved.
In this connection it must be mentioned that in calculating the
p-nitrophenol breakthrough shown in Figure 7 the very low
diffusion coefficient of the lignin sulfonic acid was used. On
the other hand, by applying the diffusion coefficient of
p-nitrophenol in the presence of lignin sulfonic acid for the
calculation of the concentration profiles, a better adjustment
of the experimental data can be obtained.
The UV breakthrough curves of the lignin sulfonic acid in
the presence of p-nitrophenol, shown in Figure 8, were calcu-
lated by using the single-solute isotherms only. The good
agreement with the measured points can be explained by the fact
292
-------
that the lignin sulfonic acid is only slightly displaced by
p-nitrophenol.
Film - Homogeneous- Diffusion
I I
100
150
time in hours
200
250
300
Figure 8. UV-Breakthrough Curves for Lignin-Sulfonic-Acid
in Presence of p-nitrophenol (Activated Carbon R).
SUMMARY AND CONCLUSIONS
The investigations have shown that it is possible to
predict the breakthrough behavior of complex mixtures in a
pilot-plant filter. The aim of future studies will be to
predict the breakthrough behavior of waterworks filters in the
same way as discussed here. This seems to be fairly simple if
the filters are working with adsorption only. If, in addition,
biological processes occur in a filter (as is generally
observed), these processes have to be taken into account because
an additional mechanism exists for removal of organic substances
in the activated carbon filter. Investigations in a new re-
search program will also need to consider the influence of
ozone on biodegradation.
Use of the methods already discussed, can help to optimize
existing plants, as well as a proper control of the efficiency
of the regeneration processes.
293
-------
REFERENCES
1 • Holzel, G., Laboratory Activated Carbon Test Methods for
Water Utilities, pp. 270-284 in this Symposium.
Frick, B., Prediction of Multicomponent Adsorption
2. Behavior; Equilibrium Aspects, pp. 256-269 in this
Symposium.
Merk,.W., Prediction of Multicomponent Adsorption Behavior
3. in Activated Carbon Adsorbers: Kinetic Aspects, po. 234-
255 in this Symposium.
294
-------
PREDICTION OF FULL SCALE PLANT
PERFORMANCE FROM PILOT COLUMNS
Jack DeMarco
and
Noel Brodtniann
INTRODUCTION
The purpose of this paper is to compare pilot granular
activated carbon (GAC) column data with full scale activated
carbon bed data from a field investigation sponsored by the
U.S. Environmental Protection Agency. The literature does not
contain reports of a very large data base that shows the re-
sults of parallel operation of pilot and full scale activated
carbon beds for removal of a variety of organic substances.
This paper describes an effort to compare pilot and full scale
activated carbon beds for removing a variety of organics with
the objective of providing the needed data for ultimate full
scale design.
ANALYSES AND SYSTEM DESCRIPTIONS
Analyses were performed for purgeable halogenated organics
collected by purge and trap techniques and for substances
described by general parameters such as Total Organic carbon
(TOC) ana T-rihalomethane Vlbrmation. Potential CTHMFP) tD. For
research purposes, several criteria for comparison were selected
that compare the effluent quality of the activated carbon
systems at different points of the resultant breakthrough
curves. Data presented include concentrations of:
1. Total Organic Carbon (TOC)
2. Trihalomethane Formation Potential (THMFP)
3. Chloroform
t
4. 1,2-Dichloroethane
295
-------
These were the substances that occurred most frequently
and at the highest concentrations. The granular activated
carbon units, located at Jefferson Parish, Louisiana, are being
studied by personnel at the water treatment plant..
Figure 1 is a flow diagram showing the full scale unit
processes at the study site. The water source is the Missis-
sippi River. The raw water is pumped to the plant site where
potassium permanganate is added for taste and odor control.
Next a cationic polymer is added as the primary coagulant. As
the water enters the flash mixing chamber, a low dosage of lime
is added for pH adjustment and a ferrous salt used at times for
alkalinity control. The water then passes through an upflow
precipitation process for removal of floe and lime precipitated
hardness. The water is then disinfected during transport to
the sand filters by addition of chlorine and ammonia as shown.
Only chloramine disinfection is practiced, which accounts for
the generally low trihalomethane (THM) concentrations present
in the finished water. Sand filtered water passes through the
clearwell and is either pumped directly to the distribution
system or to on-site reservoirs.
The 30 in (76 cm) of filter sand was removed from one
existing sand filter and replaced with 30 in (76 cm) of WVG 12
x 40 mesh--grariular-activated carbon (GP£) ",. The bed received
coagulated, settled; alnd disinfected water and is referred
to in this paper-as-a filter/adsorber (sand replacement system)
because it performs both the filtration and adsorption process.
Two other existing sand filters were connected in series
operation. The 30 in (76 cm) of-filter sand was removed from
the second bed in series andthfe sand was replaced with 30 in
(76 cm) of WVG 12 x 40 mesh GAC. This activated carbon bed
received water that was coagulated, settled, disinf ected\ a'nd
sand filtered. The activated carbon bed is referred to In this
paper as an adsorber (post-filter adsorber) because it primarily
performs the adsorption function and does not serve the purpose
of filtration. Samples for the full scale beds were routinely
analyzed at the sample points shown in Figure 1.
Figure 2 shows a schematic configuration of the pilot
granular activated carbon column system and full scale acti-
vated carbon beds studied during the phase of the project
completed in January, 1979. The pilot adsorber and pilot
filter/adsorber received the same influent water as their full
scale counterparts. The activated carbon bed depths were the
same as the full scale systems on initiation of the study. The
objective was to attempt to match the empty bed contact time
296
-------
>
= GC
U. UJ
o £
5 o
SoS
OC W H
O O -I
Z < u.
II II U
(A < <
°
7
g
g
V
u
3
297
-------
NJ
VD
OC
FULL SCALE
SAND
FILTER
FULL SCALE
ADSORBER
PRECIPITATOR
FULL SCALE
FILTER
ADSORBER
t
PILOT
ADSORBER
PILOT-2.J
FILTER
ADSORBER
L
PILOT CONTACTORS
Figure 2. Schematic of Carbon Units
-------
(EBCT)r hydraulic loading rates, and operation of each set of
pilot columns with their respective full scale system.
A set of pilot columns connected in series was also oper-
ated to provide a range of empty bed contact times and show the
influence of contact time on effluent quality for various
parameters.
Table 1 includes background data on the activated carbon
systems used. The full scale carbon beds started with a 2.5 ft
(0.8 meter) activated carbon depth. Carbon losses during the
180 days of operation, however, resulted in an average bed ctepth
of 2.4 ft (0.7 meter) for the adsorber and 2.1 ft (0.6 meter)
for the filter/adsorber. The average contact time for the
full scale units are shown in Table 1. The pilot adsorber and
filter/adsorber are 3 in (8 cm) diameter glass columns and the
pilot contactors are 4 in (10 cm) in diameter.
Table 1
Background Data
Svstcr.
Adsorber
Filter Adsorber
Pilot Adsorber
Pilot Filter Adsorber
Pilot Contactors
Col. 1
Col. 2
Col. 3
Col. 4
Area
f:
381.
381.
0.
0.
0.
0.
0.
0.
-
2
2
05-
05
09
09
09
09
n"
35.
35.
0.
0.
0.
0.
0.
0.
Ave . Carbon
4
4
005
005
008
008
008
008
ft
2.4
2.1
2.5
2.5
3
3
3
3
n
0.
0.
0.
0.
0.
0.
0.
0.
Depth Ave.
I
7
6
8
3
9
9
9
9
gpd
0.6xl06
0.64sl06
70.5
E0.3
271
271
271
271
Flcv
n ,'dav
2271
2422
0.
0.
1.
1.
1.
1.
A'.
*
267
3C7
027
027
027
027
.•=. 11.^7
'ir.-ics
17.5
13.6
21. i
16.!
10.9 •
10.9
10.9
10.9
299
-------
RESULTS
Contact Time
The contact time for the pilot adsorber was approximately
22 jpercent Iqpger than the contact time in the full scale
adsorber that it was to simulate. Also the contact time for
the pilot Filter adsorber was about 24 percent longer .than
the contact time in the full scale filter/adsorber that ifc
paralleled. This must be kept in mind when comparing the
results of the pilot systems to full scale bed comparisons. The
longer contact times in the pilot systems would theoretically be
expected to produce better effluent quality results if removal
of the given substance increases with increasing contact time.
The results in Figures 3 to 6 show that the pilot contactor data
generally sjupport the theoretical expectations for TOC, THMFP,
chloroforjJiidnd 1,2-dichloroethane. The rate of rise of _ effluent
concentration is lower for longer contact times (EBCT). In
Figures 3 to 6, the values of EBCT are estimated by summing the
values for the individual contactors in series; the correspond-
ing values are 10.9, 21.8, 32.7, and 43.6 minutes for contactors
1, 2, 3, and 4, respectively.
The TOC data shown on Figure 3 indicate the same pattern
reported by other investigators (2). The TOC data appear to
be less affected by contact time beyond about 21.8 minutes
(contactor 2 effluent). Also the lack of equilibrium conditions
(influent concentration equals effluent concentration) for TOC
has been frequently observed in many investigations and reported
as the effects of biodegradation. Although this is likely true,
the continued difference between the influent and effluent
concentrations is also likely to be caused by the removal of
a component of. the TOC that is comprised of strongly adsorbed
substances that are totally removed for very long time periods.
The pilot contactor data for THMFP was the most variable,
which may partly be explained by the analytical procedures used
to quantify this organic fraction. Further studies will be
conducted to investigate the high THMFP variation during the
first 60 days of data collection. The TOC and specific organics
data show the influence of contact time more clearly.
300
-------
9.0
in
re-
1.0
INFLUENT (S. T. IFF)
CONTACTOR «1
CONTACTOR «»
CONTACTOR »3
CONTACTOR »4
to
I
20
40 90 M
TO M> I
DAYS OF MM
110 12O 130 14O ISO 1*O
Figure 3. Effect of Contact Timeon TOC
J5O-
300
190
;>oo
I ISO
INFLUENT (S.F. EW.)
---CONTACTOK *\
CONTACTOW «J
- - CONTACTOR «3
CONTACTOW »4
\ f \
x >.;' •> •
M> 20 30 40
90 M TO 10 tO 100 110 120 130 140 190 1*0
DAYS OF HUN
Figure 4. Effect of Contact Time on Tf^FP Panoval
301
-------
•JO
V>
ia
•x
is*
J a*
2-0-
- MFUJCNT (». t. 0F.)
-CONTACTOR n
CONTACTOR «
CONTACTOR *3
-CONTACTOR**
«-•
13.2
1J&
(
J\
M^iJ-'
) 10
v\
• ^^.^
so
(•
+.
f f
'\ •'
1 » /
'-<
'xvr
30
J
40 SO
u
u
^^ >
SO
>
//
'/
70
«
^^_-r
SO
DAYS OF
*„*'*
^f
•0
RUN
A
/ * x
/ ' x"
v ' vl_-r'"_ ---_-_.-'
1OO 110 120 13O MO ISO 1«
Figure 5. Effect of Contact Time (on Chloroform Removal
Mn.UEWT (8. F. IFF.)
CONTACTOK W1
CONTACTOR «2
- - CONTACTOR «3
CONTACTOR »4
X) «O SO M TO «0 W 1OO
120 130 14O ISO 1(0
Figure 6. Effect of Contact Time on. 1,2 Dichloroethan^
302
-------
Pilot Adsorber (Post-Filter Adsorber) and Full Scale Adsorber
(Post-Filter Adsorber) Comparisons~
Total Organic Carbon (TOC)-«*
The pilot system appears to produce a better quality
effluent than the full scale system using TOC data (Figure 7).
As cited previously, the pilot carbon adsorber column had about
22 percent longer contact time than did the full scale carbon
adsorber. Using an effluent TOC criterion of £ 1 mg/1 in the
effluent wduld result in an observed allowable operating "time
of 128 a ays ijp the full scale system compared to 53 days in the
pilot system or about 89 percent more service time. The result-
ing effect wOfcld be that about 25 days less service time would
be achieved in the full scale system than would be predicted by
the pilot system.
An effluent criterion of 2 mg/1 TOC in the effluent would
not be reached for about 88 days in the full-scale system and
95 days in the pilot system. This results in 7 days less
service time for the adsorber than would be predicted by the
pilot adsorber. Thus, the comparisons of differences in ser-
vice time varied between 7 to 25 days depending on the criterion
selected. Further study will include attempts to reduce the
difference in empty bed contact time to evaluate whether this
caused the differences in predicted service time in this in-
stance.
Trihalomethane Formation Potential (THMFP) —
A consistent pattern of improvement of effluent quality
did not occur in the pilot system that had the longer contact
time, mentioned previously. Detailed evaluation of the erratic
results of Figure 4 may indicate that small changes in contact
time are not as significant for removal of THMFP beyond about
20 min EBCT as compared to the other parameters tested for
these waters.. This may partially explain the results shown
in Figure 8. Repeat testing is planned to further evaluate
the preliminary results obtained.
Chloroform and 1,2-Dichloroethane
The results shown in Figures 9 and 10 present a different
pattern than indicated for TOC in Figure 7. Figures 9 and 10
indicate that no practical differences occurred between the
effluent quality of the full scale adsorber and pilot adsorber,
even though a longer contact time existed in the pilot system.
An effluent quality £ 1 yg/1 chloroform would be maintained
for about 45 days and £ 2 yg/1 chloroform would be maintained
for about 62 days in both systems. Equilibrium between the
effluent and influent chloroform concentrations occurred in
approximately 110 days.
303
-------
14
12
»O
1.
IT
MFLUENT <«. f. CFF.)
AOSOMCK fff.
«LOT AMOMKft «W.
100 HO 120 130 140 1*0
DAYS V NUN
Figure 7. Adsorber and Pilot Adsorber Comparison—TOC
MO
300
a so
poo
I iso
100
• MFVUCNT ($. r
• ADSORBER CTf.
• M.OT AOSOMEM IFF.
1O 20 W 40
TO W K
DAYS OF Km
MO 120 130 140 100 WO
Figure 8. Adsorber and Pilot Adsorber Comparison—THMFP
304
-------
4B.S
IB-
M
12
(NFUKHT (S. F. iW.)
ADSORBER EFT
MLOT ADSORBER or
W MX) 110 120 130 14O ISO 160
Figure 9. Adsorber and Pilot Adsorber Comparison—Chloroform Removal
INFLUENT
-------
Similar results were achieved for I,2-dichloroethane.
Using an effluent criterion of 1 yg/1 would result in a service
time of about 80 days for the full-scale and pilot adsorber.
Also equilibrium with the influent 1,2-dichloroethane concentra-
tion occurred in the effluent betwen 90 and 110 days. Both
figures show the variable influent conditions that existed in
the influent water and the ability of the granular activated
carbon systems to reduce the concentrations of organics even
when considerably higher influent conditions were present.
Pilot Filter/Adsorber (Sand Replacement) and Full Scale
Filter/Adsorber (Sand Replacement) Comparisons
Total Organic Carbon-B-
using TOC data, the pilot system seems to provide a better
effluent quality than the full scale system (Figure 11). For
these systems the pilot filter/adsorber has about 24 percent
longer contact time than the full scale filter/adsorber. An
effluent concentration £ 1 mg/1 TOC was maintained for only
3 days in the full scale system compared to about 25 days in the
pilot system. This results in about 22 days less service time
for the full scale bed than predicted by the pilot system.
Using an effluent TOC criterion of £ 2 mg/1 resulted in a
difference of about 23 days between the full scale and pilot
system.
If the contact times had been equal, the breakthrough
curves and predicted service times for TOC might have been more
equal. These TOC results show a similar trend to those results
observed for the adsorber system data shown in Figure 7.
Trihalomethane Formation Potential (THMFP)«
Figure 12 shows erratic results similar to Figure 4 and
Figure 8 for precursor (THMFP) removal and further evaluation
of these results will be performed through repeat testing and
investigation of the analytic techniques. The lack of a clear-
cut difference in quality of effluent may be caused in part by
the lack of a large impact in contact time beyond 20 minutes,
EBCT, see Figure 4.
Chloroform and 1,2-Dichloroethane
The results shown in Figure 13 and 14 for the filter/
adsorber systems indicated that a slightly better effluent was
consistently achieved by the pilot filter/adsorber system. The
difference in quality between the pilot and full scale adsorber
effluents is more pronounced for the chloroform data than for
the 1,2-dichloroethane data. The respective average influent
concentrations were about 5 yg/1 for chloroform and about 2 yg/1
for 1,2-dichloroethane. Equilibrium with the respective influ-
ent concentrations occurred at about 110 days for chloroform and
88 days for 1,2-dichloroethane. Without the occurrence of
306
-------
II
•TLUEHT T FK.TEM ADSOMEN I
! \ 1 \\ A \
//U-'\/\\
/
0 W » 10 40 M »0 TO M «0 100 110 1» 110 140 1*0 1M
OATS OF RUN
Figure 12. Filter Adsorber and Pilot Filter
— HFP Rernoval
Adsorber Comparison— THMFP
307
> ,!»**.«
-------
WFLUENT (PTT Iff.)
FILTER ADSOHKK EFF.
PILOT FILTER AOSOMEA
14
10 ao ao 40
•O TO K> K 100 110 120 110 MO «O MO
DAYS Of RUN
Figure 13. Filter Adsorber and Pilot Filter
Corsparison—Chloroform^enoval
INFLUENT (PTT EFF.)
FILTER ADSORBED EFF
PILOT FILTER AOSOftKR tff.
ISO UO 140 ISO IK)
Figure 14. Filter Adsorber and Pilot Filter
Adsorber Comparison—1,2 Dichloroethane Rerroval
308
-------
periodic high influent concentrations the systems could also be
judged to be at equilibrium as early as 70 days for chloroform
and about 85 days for I,2-dichloroethane. The plateaus for the
breakthrough curves appear to start at about this time period.
Using a criterion of £ 1 yg/1 of chloroform in the efflu-
ents would result in a predicted service time of 32 days for the
full scale system compared to 38 days for the pilot system data
shown in Figure 13. Figure 14, however, shows that an effluent
concentration of £ 1 yg/1 of 1,2-dichloroethane would be main-
tained for about 74 days by both the pilot and full scale
filter/adsorber.
DISCUSSION
The series pilot contactor system data (Figures 3 to
6) show that increased empty bed contact time would generally
increase the service time allowed using a selected criterion.
Comparisons of the chloroform results for the adsorber and
filter/adsorber systems shown in Table 2 do not consistently
agree with the series pilot contactor results.
Table 2
Comparisons of Adsorber and Filter
Adsorber Systems for Chloroform
System Ave. EBCT Service Time (days)
(minutes) At < 1 yg/L At < 2 yg/L
Filter/Adsorber 13.6 32 45
Pilot Filt./Adsorber 16.8 38 60
Adsorber 17.5 45 62
Pilot Adsorber 21.4 45 62
The systems in Table 2 have been listed by increasing
average empty bed contact times. The data show that the filter/
adsorber and its pilot system have different service times for
each criterion whereas the adsorber systems do not. The contact
times in each pilot system were longer than the respective full
scale systems. Although the filter/adsorber systems show a ,
relationship with increased contact time, the adsorber systems,
which were backwashed less frequently and also received a
different quality water, did not.
309
-------
The pilot filter/adsorber aud full scale filter/adsorber
comparisons were more consistent; for example both the total
organic carbon data (Fig. 11) and the chloroform data (Fig. 13)
showed increased service times for the pilot system. This is
consistent with the general expectation of longer service time
with increased contact time. This was not as apparent, however,
when 1,2-dichloroethane data (Fig. 14) were used to evaluate the
pilot and full scale filter/adsorbers.
The pilot adsorber and full scale adsorber showed expected
longer service times only for the general parameter of TOC. An
increase in service time with increased contact time did not
occur when the pilot and full scale adsorber were compared using
chloroform and 1,2-dichloroethane.
The results are unexplained to date (May, 1979) but further
evaluation of all data will be performed.
Statistical Comparisons
Judgement of whether a difference did or did not exist
in the effluent quality of each pilot adsorber column and its
full scale counterpart is sometimes made subjectively. Little
debate usually exists for breakthrough curves that are "widely
separated"" and judged different, even if some scatter exists.
Likewise oarves that are "very close" to one another and judged
to show no difference are not generally debated. Evaluations
that r£iy on these subjective approaches leave room for improve-
ment. Applying effluent quality criteria such as was previously
done is a method of providing practical evaluations of the
significance of any difference.
To improve this situation a statistical aid to judging
differences in the data has been tested. Efforts to date
consist of developing an equation or model for each breakthrough
curve, and then using a log transformation of the data to
develop a best fit least squares regression line. This straight
line plot is then used to determine the slopes and intercepts of
each line.
The hypotheses that the slopes are or are not equal is
then tested at the 0.05 significance level and the same method
is used to test the intercepts. The concept is similar to that
suggested by Cairo and coworkers (3). Chloroform and 1,2--
dichloroe thane data for the pilot and full scale adsorber have
been used with both the percent breakthrough approach described
in their paper and our model, which used influent and effluent
concentrations.
310
-------
Both approaches indicate that no statistical difference
existed in either the slopes or intercepts of the transformed
lines. Thusr these results agree with a visual subjective
evaluation and the effluent quality criteria tests apoJied.
Additional work will be done with breakthrough curve/.'having
varying degrees of separation,to assess the overall Usefulness
of the statistical tools.
Summary
The first attempt to provide a data base for comparing
a pilot and full scale system for various organics produced
results that require further investigation.
Although both of the pilot systems had longer average
contact times than their respective full scale systems, a
uniform pattern of longer service time for the different sub-
stances evaluated did not occur. These results may not be
consistent with the data collected using pilot contactors in
series to observe the effect of contact time.
The series pilot contactor system provided data that
show the effect of varying empty bed contact time on the removal
of the organic substances studied. Improved effluent quality
was generally achieved with increased contact time. A pilot
series system similar to the one used is considered essential
in collecting and interpreting design data.
A method of statistical interpretation of the results
was developed that shows promise for use in evaluating pilot
column results.
Further evaluation is required to assess that all pertinent
factors were properly controlled during the investigation. Alsrt,
repeat testing will be conducted, attempting to control more
closely the contact times between the pilot and full scale
systems. The comparative granular activated carbon particle
sizes also will be more closely examined.
Although pilot to full-scale comparisons have been success-
fully performed for single parameters, the results for multiple
organics have not been previously reported. Such a data base,
if properly collected, will provide invaluable background in
assessing what types of design safety factors are required when
using pilot column data to predict full scale results for a
naturally occurring mixture of organic substances.
ACKNOWLEDGEMENTS
The co-authors wish to express their thanks for the
assistance provided for the development of statistical tests
311
-------
by Michael Pence and Pat Cairo of the Philadelphia Water
Department.and to Robert Clark and Michael Laugle of the
Drinking Water Research Division of the U.S. Environmental
Protection Agency. Also Wayne Koffskey of the Jefferson Parish
Water Department provided assistance in the analytical data
evaluations used in this paper.
REFERENCES
1. Stevens, A.A. and J.M. Symons. 1977. "Measurement of
Trihalomethane and Precursor Concentration Changes," JAWWA,
69, (10) 546-554.
2. Wood, Paul R. and J. DeMarco. "Effectiveness of Various
Adsorbents in Removing Organic Compounds for Water - Part
II Removing Total Organic Carbon and Trihalomethane
Precursor Substances," Presented at American Chemical
Society Symposium on Activated Carbon Adsorption of
Organics from the Aqueous Phase, Annual ACS Meeting, Miami
Beach, Florida, Sept. 10-15, 1978.
3. Cairo, Patrick R., et al., "Development of Criteria for
the Design of Full Scale Carbon Adsorption Systems,"
Proceedings AWWA Water Quality Technology Conference,
Louisville, Kentucky, Dec. 3-6, 1978, Paper 4-4, pages
1-25, 1979.
312
-------
THE MULHEIM PROCESS FOR
TREATING RUHR RIVER WATER
Ewald Heilker
Granular activated carbon filters are widely used in
Germany. The process design criteria differ depending on
whether the carbon filters are used for chlofine removal or in
a combined adsorption and biological treatment. Besides the
process design criteria, many other design problems occur in
practice as demonstrated by the history of the use of granular
activated carbon filters for treating the Ruhr River water at
the Rheinisch-Westfalische Waterworks Association at Millheim.
In order to meet the peak demand for water in the Ruhr area,
the waterworks has depended on the Ruhr River surface water since
the beginning of industrial development because only a small
amount of groundwater occurs in the gravel layers of the river
valleys. The treatment of the Ruhr water in slow sand filter
basins and ground passage has been the classical and most favored
treatment in nearly all the waterworks along the Ruhr for many
decades. After a 1- to 3-day retention time in the aquifer
(which is about 12- to 20-m thick), the water is collected in a
series of wells parallel to the slow sand filter basins, dis-
infected, and pumped into the distribution system.
The Rheinisch-WestfSlische Waterworks Association supplies
about 1 million people in the western part of the Ruhr area, as
well as the Ruhr industries, with water from three rivers. It
has had difficulty in meeting the demand in the drought years of
1959 -and 1964. With an increasing amount of nutrient material
in the Ruhr, algae blooms have occurred in early summer and
autumn, reducing the hydraulic capacity of the sand filters.
Geohydrological considerations and the density of popula-
tion in the nearby Ruhr valley make it impossible to augment
the,water supply by building new filter basins or wells. The
first approach to the problem was to install rapid filters in
two waterworks. However, all the colloidal dissolved iron in
the Ruhr water could not be satisfactorily removed despite the
use of different granular media combinations in the filters.
The dosage of aluminum sulfate before filtering caused post-
flocculation at the basin surface and in the injection wells,
313
-------
fading to a considerable decrease in the infiltration rate
within 3 months.
After pilot plant tests for about ,3ne year with several
pollution loads of the JRuhr water, flocculators were built in
two of the waterwork^,-^jid. the water was treated in a conven-
tional way with superthlorination, flocculation, sedimentation,
and filtration.
The ammonia content of the raw water made a 10-15 mg/1
dosage of chlorine necessary for breakpoint chlorination. The
filter effluent was seeped into the slow sand filter basins and
injected into the ground as before, producing high infiltration
rates of 3 m /m /day and slow sand filter runs of 6 months. As
a result of the relatively high chlorine excess of 1 mg/1, algae
did not develop significantly on the basins. However, the bio-
logical decomposition of taste- and odor-causing substances
decreased considerably because of the chlorine in the slow sand
filters and in the ground passage. Therefore, filtration with
activated carbon was initiated, primarily in order to eliminate
the excess chlorine, but also to reduce the organic load*.
After the startup of the activated carbon filters, the
water quality greatly improved. The taste of the carbon filtrate
was no longer impaired and the organic load at the beginning of
the filter run was reduced. It was believed that the ground
passage step could be eliminated and the carbon filtrate could
be pumped directly into the distribution system. Virological
tests demonstrated that the flocculator outlet was virus-fre^
despite the high virus counts in the Ruhr water. Because of the
organic load, however, consumers could not be provided with water
that had not passed through the ground. The breakthrough of
taste and odor-substances indicated that the activated carbon
was exhausted after a 4 month run.
The effectiveness of dechlorination with carbon made it
possible to obtain a satisfactory final purification of the water
by biological processes in the slow sand filters and in the
ground passage. Considering the anticipated future demand, a
treatment process without ground passage had to be developed.
THE MULHEIM PROCESS
The effectiveness of the different treatment processes is
shown in Table 1. The data were obtained by the MUlheim Water-
works and by the Engler-Bunte-Institut of the University of
Karlsruhe. The high prechlorination caused the formation of
organic chlorinated compounds which could not be reduced to a
tolerable degree either by activated carbon or by ground passage.
At the time the data were compiled the activated carbon was
highly loaded and had spent most of its adsorptive capacity.
Because these results were confirmed by tests performed in a
314
-------
Table 1. Organochloro Compounds after Breakpcint
Chlorination Treatment
DOC1* DOC1 IT*
Source of Sample _ /«g/l _ .Ag/1
Raw Water (Ruhr) 17 5
After Breakpoint Chlorination
flocculation and filtration 203 30
After GAC-f liters 151 17
After Ground Passage and
Safety Chlorination 92 18
signifies dissolved jornarlc chlorine; N- signifies nonnolar
pilot plant with different kinds of activated carbon, it was
decided to abandon breakpoint Chlorination in the development
of a new treatment system. The removal of ammonia was to be
achieved by biological oxidation. This approach had been tested
at other plants, which found that nitrification was effective in
removing ammonia and that the negative effects of breakpoint
Chlorination could thus be avoided.
Figure 1 shows the general Mulheim process scheme and
Figure 2 illustrates the composition of the filter media. The
raw water is taken out of the impounded Ruhr navigation canal and
pumped into the mixing basin (1.8 by 1.8 by 5.2 m) where, instead
of chlorine, off-gases from the ozonation basin are drawn back
into the flash mixing basin by a high speed aspirating turbine
and injected into the raw water. Additional fresh ozone can then
be added directly to the mixing chamber. The amount of ozone
added, averaging 1-1.5 mg/1, depends mainly on the degree of
turbidity of the raw water. The water is dosed with 4-6 mg/1
poly aluminum chloride and mixed thoroughly by the turbine.
The pretreated water is pumped into the flocculator through
perforated pipes on the bottom of the pulsator basin. After
90 minutes of clarification, nearly all the particulate materials
are settled, and water with a turbidity less than 1 FTU flows
into the ozonation basin. Fresh ozone is injected through porous
ceramic pipes. Residual small floes are subsequently removed in
the rapid sand filters. Figures 3 through 9 show construction
details of the GAC treatment installation.
Since a biological process requires a longer retention time
than a purely adsorptive process, the GAC bed depth was increased
from 2 to 4 m, increasing the empty bed contact time from 5 to
10 min.
In the activated carbon filters, a temperature-dependent
interaction was observed between adsorption and biological oxida-
tion behavior. At low temperatures, non -decomposable, as well
as decomposable, substances are concentrated by adsorption. At
315
-------
increasing temperatures, an efficient microbial population de-
velops and the previously adsorbed biodegradable compounds are
biologically oxidized, increasing the filter rate. At the same
time, the nitrification activity in the activated carbon filters
led to a nearly complete oxidation of the ammonia to nitrate,
despite the extremely low winter temperatures (0.5°C). Pure
oxygen is injected before the filter pumps so that a sufficient
amount of oxygen (7-8 mg/1 ) is present in the carbon filtrate.
The bacterial investigations of the carbon filtrate
produced satisfactory data, although in the summer of 197*8 a
population explosion of nematodes was observed in the raptel sand
and carbon filters. Clearly, nematode development was caused
by long backwashing intervals of more than a week. The nematodes
disappeared when these intervals were reduced to 3 days (the
reproduction cycle of this genus of nematodes).
Backwashing in the gravel filters is accomplished by a
mixture of air and water at a rate of 25 m/hr. The GAC con-
tactor columns are backwashed, first with air to loosen the
material, and then water is pumped through at a rate of about
40 to 50 m/hr.
Table 2 compares the process data of the conventional treat-
ment (with breakpoint chlorination) with the new treatment in-
cluding adsorptive and biological granular activated carbon
filters. In essence, the breakpoint chlorination was replaced
by a two-stage ozonation and the retention time of the carbon
filters was lengthened, resulting in an increase of the bio-
logical capacity of the filters.
The treatment plants in the Dohne waterworks have been oper-
ating for more, than 1-1/2 years using the revised process. The
drinking water quality has significantly improved without incur-
ring higher treatment costs. The Dohne plant is less susceptible
to disturbance and can thus be operated with half of the former
staff size. The activated carbon filter runs are three to five
times longer than before. The other waterworks of the Rheinisch-
Westfalische Waterworks Association are being rebuilt to accom-
modate the Mulheim process.
FILTER CONSTRUCTION AND CORROSION INHIBITION
Some aspects of the construction and technical equipment of
filters have been particularly important.
Steel filters v:ere selected because they were better suited
for the overall process design and easier to work with in modi-
fying existing treatment plants. However, it was found that
corrosion problems caused high repair costs.
316
-------
Table 2. Changes in Water Treatment Processes
at the Dohne Plant
Treatment
Step
Classic
Treatment
New
Treatment
Preoxidation
Dosing
Power Input
10-50 mg/1 Cl
4-6 mg/l-Al,0,
0.1 kW/mJ
Retention time
0. 5 min.
1 mg/1
4-6
2.5 kW/m
Retention time
0.5 min
Flocculation
Sedimentation
5-15 mg/1 Ca(OH)
Retention time
1.5 hr.
5-15 mg/1 Ca(OH)2
Retention time
1.5 hr.
Ozonation
2 mg/1 O
Retention time
5 min.
Filtration
Filter Velocity
v « 10.7 m/hr
Filter Velocity
v = 10.7 m/hr
Activated
carbon filter
Bed height
22 m/hr
2 m
22-11 m/hr
h = 4 m
Ground Passage
Retention time
12-50 hr.
Retention time
12-50 hr.
Safety
Chlorination
0.4-0.8 mg/1 Cl.
0.2-0.3 mg/1 C12
After 10 years of testing different filters having addi-
tional cathodic corrosion protection, a method has been found
that offers nearly comple'te protection. Twelve double-stage
filters are being modified to prevent corrosion: all interior
welded seams are being straightened and ground; rust is being
removed with sand blasters; condensate is being prevented by
aeration during coating; solvent-free epoxy resin, which is
resistant to abrasion and ozone, is being used; a 450 to SOO-^um
poreless coating is being applied and is being tested with i-,-5
to 2 kV.
Further experiments have determined that the interior filter
surface and all installations can be completely protected if
combined cathodic protection with sacrificial anodes is used and
the electric power supply is transmitted over platinum-covered
titanium anodes.
The costs for all corrosion protection devices being applied
to the carbon filters amount to approximately 10 percent of the
total plant modification.
317
-------
Mtting B*S'n Pu'SKO1
"or C* OH :
Figure 1. Process Scheme at Mulheim
318
-------
•^JU»«- • w»i[&£,&£&»
A-Hahia - O.A.O.
I
Figure 2. Composition of Filter Media
The double-layer filter and activated carbon filter, shown
separately in the schematic diagram (Fig. 1), are combined
in a single vessel as shown here.
319
-------
Figure 3. Cross Section through a Double-Stage-Filter/
Upper Stage with Multi-Layer-Filter, below
the GAC-Filter- Dimensions are in urc.
The upper filter bottom is to avoid clogging of the
filter nozzles during backwash.
320
-------
Figure 4. Nozzle Bottom with Filterpipe as Protection
for the Anode Wires
Figure 5. Detail of the Previous Picture
321
-------
Figure 6. View of the Nozzle Bottom from Below
Figure 7. Details of the Nozzles; nozzles with
long shafts are used for the backwash
with air and water.
322
-------
Figure 8. Filter Hall
323
-------
Figure 9. Raw Water Inlet, Also Used as Backmsh T'Tater Outlet
324
-------
DISCUSSION
TUESDAY, MAY 1
MORNING SESSION
Q HAROLD PEARSON, Metropolitan Water District, Southern
California: I'd like to ask about the relationship between water
temperature and your biologically activated carbon operations.
I noticed your rates of flow were from 22 down to 11 nMi. Do
you correlate that with the temperature and the biological
activity so that you get maximum efficiency in the removal of
the organics?
A PROF. DR. SONTHEIMER: No, not exactly. This can be easily
explained. Normally in winter, water use declines, enabling us
to reduce the throughput. If the temperature increases, we have
more biological activity and more biological regeneration of the
carbon. It is very clear that biological activity in activated
carbon depends on temperature; you must pay careful attention to
this, especially if there are very low temperatures in winter, as
in Germany. This past winter we had 6 or 8 weeks with tempera-
tures below 1°C.
Q UNIDENTIFIED QUESTIONER: Mr. DeMarco, you indicated that
everyone had different criteria for regenerating carbon, and
depending upon the criteria, different regeneration times would
apply. The regeneration costs and facilities are very important.
One criterion in the proposed regulations states that when the
effluent COD or TOC increases 0.5 mg/1 above that for freshly
regenerated carbon, then one should regenerate. In looking at
your data, I had difficulty telling when that might be; depending
upon how you look at it, it may be from 1 to 20 days. I wonder,
as a representative of EPA, if you would interpret that data in
view of that criterion.
A MR. DeMARCO: All you really need to know would basically
be what the breakthrough concentration was at startup. About
0.5 mg/1 came through very quickly, and this was not based on a
3-day running average. What I attempted was to interpret the
unadjusted data and say when it exceeded a given criterion: in
a state above that criterion, that's what I did my comparison on.
You could probably get a rough estimate of when you might have
to regenerate, according to the proposed regulations, by looking
at the 1 mg/1 effluent concentrations that I talked about for
TOC.
325
-------
Q PROF. PERRY McCARTY: Using 1 mg/1? Would you suggest
using a 3-day running average? Are you just picking that
number?
A PROF. PAUL ROBERTS: I have a suggestion along that line.
It seems to me, if you recognize that in most large installations
there will be many of these adsorbers operated in parallel, and
that the concentration of interest is really an average over
those many filters, then what you ought to look at would be a
cumulative average of the effluent concentration rather than
the instantaneous value. One of the speakers from Karlsruhe
presented a graph in which he showed a curve of effluent concen-
trations and another that he called "integrated effluent Concen-
tration." I suggest that the integrated effluent concentration
would be appropriate for a multi-adsorber installation. That
would do two things for you: one, it would be approached more
slowly; and secondly, it would tend to iron out a lot of the
daily variability in the data.
A MR. DeMARCO: That's a good comment. I think it ties in,
somewhat, with some of the comments made the other day about
compositing, which I firmly believe in. I think the answer is
a combination of compositing and integrating for our research
work. What we intend to do at one of our full-scale locations
where we have more discreet contactors, such as in Cincinnati, is
to study the combined effluent of a series that will be a staged
operation, just as you would do in a full-scale plant. We
couldn't do it in the original test, because we didn't have a
staged operation in the plant. When we do have that staged
operation, we will look at the composite quality that comes out
of the parallel operation of three beds at different stages of
operation. In addition, we will be able to compare the effluent
quality of the staged operation with one unit to determine the
differences. I do think that we will see some smoothing out, as
you suggest.
Q RICHARD WOODHALL, Connecticut State Health Department. I
would like to ask the gentleman from Germany exactly what is
meant by soil passage treatment. What kind of soil and how many
feet of it?
A PROF. DR. SONTHEIMER: Soil passage means that we have a
slow sand filter. We introduce the water into the ground, which
is mostly sand, approximately 12 meters thick. There are wells
approximately 50 meters from the infiltration basin, and the
water passes through the ground to the wells. This varies from
one waterworks to another, since some have sufficient space and
some have not. In some cases, the underground residence time is
only 10 or 12 hours, whereas in other treatment works operated
by the same companies the residence time in the ground is 50
days, so it depends on local conditions. Ground passage means
326
-------
introducing the water into the ground, where it is mixed with
natural ground water as it moves to the wells and is removed
after some time.
Q STAN SMITH, Georgia Tech: Mention was made in the last
paper about a pre-activated carbon, which is used as filter
material. I'm wondering about the fate of that. Is that mate-
rial actually taken out and activated to useful carbon, or is it
perhaps added as a makeup to the regeneration cycles?
A PROF. DR. SONTHEIMER: No, it's made from pyrolysis of
carbon. It's made out of coal; the coal is ground and coke is
produced. It must have a very rough surface; this is the most
important factor for filtration efficiency. It's like anthracite
coal, but it's a type of coke.
Q MR. SMITH: I am curious to know if it was actually removed
and made into activated carbon later.
A PROF. DR. SONTHEIMER: No, it's actually made from coal and
it's an intermediate step in the fabrication of activated carbon.
g
MR. SMITH: It's used as a matter of convenience?
A PROF. DR. SONTHEIMER: No, it's a very good filter mate-
rial. It's sold as a filter material, like other available
filter materials.
Q MR. SMITH: I had another question on an earlier paper about
a parameter that was used as a test criterion, called OSW. Is
there a speaker still here who could explain this?
A PROF. DR. SONTHEIMER: Yes, OSW means Organic Substances in
the Water. It-'s only an acronym, nothing else.
DR. SYMONS: You see, they're learning from us to confuse
people with initials.
Q DR. RIP RICE, Jacobs Engineering: For Herr Heilker please;
you have been operating your new process with the 4 m GAC column
since November 1977, a period of a year and a half. Have you
regenerated the carbon yet? And if you do, or when you do, what
will be the regeneration criteria?
A PROF. DR. SONTHEIMER: That is a difficult question that
I tried to answer yesterday. I think it is very clear in this
case. The carbon has to be regenerated immediately because
oxygen uptake is very low.
327
-------
0
PROF. DAVE HENDRICKS, Colorado State University: A question
that's been bothering me for several years has been referred to
here by a number of people, concerning the fortuitous benefit of
biological treatment in conjunction with activated carbon adsorp-
tion. It seems that there's a trade-off between the lessening
of the adsorption capacity because of the bacteria occupying the
adsorption area, and also, possibly, the lowering of the transfer
coefficients because the access to the carbon pores may be
blocked. Logically, there appears to be a trade-off, but I've
never seen this articulated. I wonder if it's been considered
by people who observe this and by those who wish to promote the
synergistic features of biodegradation and adsorption.
A PROF. SONTHEIMER: I understand your concern, but our data
show that there is no real difference in adsorption capacity.
That means that the bacteria on the carbon do not have any sig-
nificant influence on the adsorption capacity of the carbon.
Our data show that the mass transfer rates are not really changed
by the presence of bacteria, but this has only been done with
very low concentrations of organics. Mass transfer considera-
tions may limit the amount of bacteria desired on the carbon.
There are some data, I believe, that show some changes in the
mass transfer criteria. This may occur, but we didn't find it
under the conditions that we investigated.
328
-------
OPERATIONAL EXPERIENCES WITH ACTIVATED CARBON ADSORBERS
329
-------
&EPA
NATO CMM
•OTAN CCMS
NATO-CCMS
330
-------
OPERATIONAL EXPERIENCES WITH THE CARBON ADSORPTION PLANT AT
CHURCH WILNE
D.J. Osborne and C.A. Kennett
The Church Wilne Works are of a conventional clarification
filtration type, taking wat^r from the River Derwent and pro-
ducing a potable supply foripart of the City of Nottingham.
They were commissioned in 1^70 with a flow of 45,500 m /day
and have now been doubled in size as part of a possible final
extension to 180,000 m /day). The subject of this paper is the
granular-activated carbon absorption plant of 91,000 m /day
(24 x 10 US gal/day) capacity built as an extension to the
works in 1973.
The works intake is located 16 km (10 miles) downstream of
the large industrial town o:: Derby. The combined industrial and
municipal sewage of Derby is treated to a conventional standard
in a large modern sewage works and discharged to the river,
where it is diluted betweenj4 and 20 times, depending on the
river flow. The sewage undergoes further self-purification
before a portion is removed for treatment at Church Wilne
(Figure 1).
It was known that the duality of this water was near the
limit of that considered .acceptable as a potable supply source;
consequently, a pilot plant study on treatment methods was
undertaken for several years prior to and during construction of
the main plant.
The treatment process decided upon was as follows:
(i) Long reservoir storage, minimum 30 days
(ii) Flocculation and clarification using chlorinated fer-
rous sulphate. The chlorine added is controlled to
give a level of 0.8 mg/1 free chlorine at entry to the
final contact tanX.
(iii) Powdered activated carbon dosing.
(iv) Rapid gravity sand filtration.
331
-------
Planned NewIntake
Figure 1. Plant Location
332
-------
(v) Residual chlorine adjustment with sulphur dioxide to
give 0.3 mg/1 going into supply after 12 hours
residence in the contact tank.
PILOT PLANT
The bankside pilot plant (Figure 2) was a replica of the
proposed works.
In the first year's operation of the pilot plant, taste and
odor control became difficult. Powdered carbon dosages for
complete taste and odor removal averaged 40 mg/1 and peaked at
90 mg/1. At this time, studies with potassium permanganate,
ozone, chlorine dioxide, and superchlorination all produced no
noticeable odor removal. A 30 cm (1 ft.) dia. single stage
granular activated carbon column experiment did, however, give
complete odor removal. At this stage, Humphreys and Glasgow,
Ltd., and Chemviron (Calgon Corp. in the USA) suggested a trial
in the summer of 1970 using a test rig comprised of four 10 cm
(4 inch) diameter columns with F200 carbon. Contact times per
column, which were placed in series, were 1.5, 1.8, 1.8, and
1.8 minutes. Taste and odor results from this unit are shown in
Figure 3. Measurements were also made of phenol values, anionic
detergents, and *TOC using an early-model instrument with a
detection limit of approximately 2 mg/1 (Figure 4). By that
autumn, sufficient data had been accumulated to enable inquiry
documents for a granular activated carbon plant to be prepared.
While this pilot plant work was going on, the main works
had started up and a high, and consequently expensive, usage of
powdered carbon had become a regular necessity. The most
effective way pf dosing the powdered carbon was found to be a
special air-mixed precontact tank with a 20 minute residence
time ahead of the clarifiers. Some air stripping of volatiles
probably occurred in this tank. Moreover, the carbon was not
influenced by coagulation chemicals, as it was when added to the
clarifiers. The high dosage rate precluded addition solely at
a point before the filters, since otherwise unacceptably short
filter runs would have resulted.
BID SPECIFICATION AND CONTRACT
The design requirement was that the equivalent carbon
dose rate would be 20 mg/1. This was based on the assumption
that if the powder dose rate were 40 mg/1, as in the summers of
1969 and 1970, then the granular dose rate would be 20 mg/1, and
that the reservoir storage of 4 to 8 weeks would not give any
improvement or deterioration in water quality. The contractor
was given freedom of choice over the bed contact time to use in
333
-------
u>
1
1
-
n
__. —
LL
r~5~~|
- —
4
4 ]
In
' n
i n
1
D
n
^; r
D '
n
L
11
i
12
1
1
J
1
h
( J
5
r
- i
I
/^
r
v
I v
r
6
\
V-
' N
14
_^
• \
>
13
- S
n
RIVER
1
\
J
^j
'
r i
7
/
I
8
-*
._.
c
\
^.
.1
0
23
-""-*.
17
- — -*
-" — -*,
16
-*
f -
L"
N
S
N
"
1
\
/
/
\
}
J
2<
1
L
/
i
j
i
/
P
k.
r
-*
/
X
22
21
L»*
Jb
1
J
—k
•^w
27
1.
2.
3.
4.
6.
•j
g
10.
11.
12.
15.
ir,.
17.
18.
J9.
20.
21.
22.
23.
24.
25.
i?(» .
27.
Control Panel
Carbon
Lime
Soda
Coayulant
Sil ica
Sod. Chlorite
Aerator
Tank
Dlond Chamber
1 i llel •;
Carbon Filter
Ozon iser
rinnl Tank
Sludge Pit
Decnnt Pit
I.agoon
Cliloi i lie Piox ide
Chloi iiuilor >••
Figure 2. Bankside Pilot Plant
-------
CP
sn
20
10
-
i
\
L
fl
/ \
r' \ '
i '
.j \
. »..
u
1970 F200
I-'IJUIV. DDSK KAIf AT BRhAKIHHUUbH
2.S4 kg.hr
Col 1
284 I,in x 227 litres
2.54 x 3.34
1156 x 227
- :J9 mg/l
- 22 tng/1
Raw Water
(no reservoir storage)
Col. 1
1.5 mln. contact time
wt/cartjon 2.54 Kg.
Col. ?
3.34 mln. contact time
wt/uarbor 5./ Kg.
^^ Col. 3
5.15 mln. contact time
Flow Rate = 227 lltraa/hr.
100
*f
moo
T- Running Time, Hours •
June ' ' July Aug
Figure 3. kiver Bank Pilot Plant
1400 1500
Note: Contact times are cumulative from each preceding column in series;
e.g., the contact time for column 3 represents the contact time
for columns 1,2, and 3 in series. Column 4 results were below
detection limits.
-------
18
to
0%
16
S 12
e
D
U
Q 1U
«—
I
m
Ml
£ «
1.
O ion
- 1
ionn
Raw Water
{no reaorvolr s* ni'ii'i-'
CoJ. 1
1.5 mln. contact tlrm:
Col. 4
7.0 mln. nontrt.:t
How Rate - '>}.! iJt.res/t r.
15RO
RUNNING TIME, HOURS
Figure 4. River Bank Pilot Plant
F200
-------
his bid. The specification also called for the following
guarantees, regarding:
complete taste and odour removal at the full scale plant
when operating under raw water quality conditions not worse
than those encountered by the water treated by the pilot
plant; maximum total quantity of carbon required to be
regenerated for any particular annual throughput; and the
costs of on-site regeneration per ton of carbon regenerated
in terms of annual costs. These guarantees were for a
two year period. The contractor was required to pay for
any losses above 7 percent of the regeneration furnace
throughput as part of this guarantee.
The specification did not state that regeneration of the
carbon had to be on site; it could be transported to the manu-
facturer's works for regeneration if this were economic.
CARBON ADSORBERS
The overall flow scheme for adsorption, regeneration,
and carbon transfer is shown in Figure 5. The 12 carbon
adsorber vessels are arranged in a rectangle of 3 rows of 4
vessels. There are 2 main inlet and outlet headers running
between rows 1 and 2 and beyond row 3. The first serves the
vessels of rows 1 and 2, the second the 4 vessels of row 2.
There are similar headers for inlet and dirty backwash water.
Each carbon vessel is 4.1 m dia by 3.75 m on the straight
side. There is an allowance for a carbon bed depth of 2.3 m
supported on a 457 mm gravel layer. The carbon depth of the
initial charge was 1.6 m, allowing a 4 min contract time.
Water is distributed across the top surface by a cross-shaped
distributor and collected within the gravel bed by a special
slotted pipe/nozzle system. The vessels are painted internally
with a 7 mm thick Wailes Dove Bituros enamel paint, applied
hot in the vessel fabricator's works. The carbon vessels
were built taller than necessary with sufficient sidewall for
an extra 763 mm depth of carbon to be added at a later date
if required. All pressure loss and other calculations were
made assuming the extra carbon depth was in place.
The plant hydraulic design assumption allow for flow
surges up to 119,000 m3 per day at the sand filter plant
upstream. This consideration also allows daily output to be
maintained while one carbon adsorber is temporarily out of service
for bed replacement and another being backwashed.
Carbon is transported throughout the plant as a water
slurry of about 1 kg of carbon per 7 litres of water. It is
important for minimum attrition of carbon particles that they
experience no high shear forces. In the wet carbon system, no
337
-------
00
CLARIFIED SAND
FILTERED /C\
WATER
SUPPLY
BACKWASH
REJECT •
rr~r
Mf*Ml L
OTHER ADSORBERS
REGENERATION
DEWATERINC FURNACE
SOFTENER
TOWNS WATER
FUEL
COMBUSTION —fl
AIR W
M.W.
FINAL
CHLORINATION
J_U U
TO
CHLORINE
CONTACT
TANK
MAKEUP CARBON
FROM BAGS
OR TOTE BINS
M.W.
M.W.— MOTIVE WATER
Figure 5. GAC Regeneration System
-------
pumps are used, only water operated jet eductors. There is one
large 3-inch eductor per pair of adsorbers, six in all, three
smaller 1-inch and 1.5-inch eductors associated with the carbon
regeneration furnance and one 3-inch eductor at the clean carbon
hopper.
The plant has two large open-top carbon storage hoppers for
storage of spent ("foul") and regenerated ("clean") carbon
(Figure 5). When a carbon adsorber charge requires regeneration
it is transferred to the foul carbon hopper and a fresh charge
of regenerated carbon is transferred into the adsorber from the
clean carbon hopper. Each operation can be easily completed in
two eight-hour shifts. The regeneration of the carbon is then
carri-ed out at a slow, controlled rate over the next several
days, carbon being taken from the foul hopper, regenerated, and
returned to the clean hopper.
COMMISSIONING - ADSORBER SECTION
Construction of the adsorber section was completed in
August 197$, and loading of the gravel layers and carbon was
completed that September. The initial carbon charge was sup-
plied in 30 leg bags. These were loaded into the carbon makeup
hopper. This was designed to take later deliveries of carbon
in standard 1 ton tote-bins. For initial loading, a temporary
plastic-sided structure and platform with water sprays was
erected to keep dust from dry carbon to a minimum. The carbon
was educted from the makeup hopper to the clean carbon hopper
to ensure good wetting of the particles and from there to each
adsorber in turn. In all, 135 tons of carbon (13-1/4 adsorber
charges) was loaded in two weeks. This left 1-1/4 charges in
the clean hopper after all adsorbers were filled. Each adsorber
was backwashed for about 45 minutes to stratify the carbon,
leaving smaller granules on top, larger below, and to thoroughly
wash fines from the beds. Some difficulty was encountered with
air removal from the pipework around the plant. This resulted
in unpredictable flows until the full pressure of the low lift
pumps was supplied by fully closing the interrupter/bypass valve
at the plant inlet/outlet. The predicted plant pressure drops
were borne out in practice to within 100 mm (Hg), measured at
the top liquid level in the adsorbers.
The second phase of commissioning was a 1-month training
period for the operators, during which charges of carbon were
moved out of each adsorber into the foul carbon hopper and
replaced from the clean hopper. A temporary connection was made
from the foul hopper to the clean hopper to bypass the furnance.
Each operator shift had about three carbon transfers to handle.
All 12 adsorbers were emptied at some tine. The "emptying tech-
nique for the flat-bottom adsorber beds was developed by trial-
and-error applications of backwash water, while educting the
carbon from an exit point at one side of the adsorber. It is
339
-------
not possible to empty the adsorbers fully? some 100 mm to 150 mm
of carbon always remain above the gravel. There were some fears
that this may, at some future date, need to be removed if the
same carbon remains at each regeneration; as yet, no problem
appears to have been attributed to this. Manual entry with a
portable eductor would be necessary to remove this carbon. At
the older plant at Gothenburg (converted from an open top gravity
sand filter), a rotating gravel-removal screen was installed at
the foul carbon hopper when this emptying method was used, to
remove any small gravel sucked up with the carbon.
WATER QUALITY
Taste and Odour
Experience with taste and odour removal has been excellent
(Figure 6). Before the granular carbon plant was completed and
only the powdered system was in use, the complaint level from
80,000 consAfiners ran at around 10 to 12 per week, but this
dropped to 1 per month when the adsorbers were put on line.
One supply area, previously on an ultra-pure sandstone bore-
hole water, and then switched to the river supply, did produce a
number of complaints at a time when the carbon adsorbers were
nearing exhaustion. Since final disinfection was changed to
chlorine dioxide in 1977/ no complaint of taste and odour from
consumers has been made.
Results from the interlaboratory studies of Taste and Odour
Numbers vary widely. One sample of very poor raw water was
split four ways and sent to different labs; each recorded intra-
laboratory repeatable result is indicated in Table 1.
Table 1
Interlaboratory Comparison of Taste and Odour Numbers
Taste and Odour Numbers
Works Laboratory 90
Local Industrial Laboratory 80
Trent River Authority 420
W.R.C. Medmenham 600
The last two numbers were determined under very strict air
cleanliness research conditions using a panel of samplers and
are much higher than the results determined under routine
methods in the Works Laboratory.
340
-------
70 ,
00
SO
40
U>
t,
S
S
•o
•
•*
:
30
30
10
River
--- Reservoir Outlet
•MO Carbon Plant Outlet
Reservoir storage averaged 80 days.
Taste and Odour Binlailsed by
selective abstraction.
Carbon Plant Product T.O.N. always 1.0
Dee Jan Feb Mar Apr May Jun Jul Aug >ep Oct Nov Dec Jan Feb
Figure 6. Taste and Odor Removal at
Church >Wilne Works
-------
Microbiological
One problem mentioned by other carbon bed users is a pro-
liferation of microbiological growth within the bed. This has
never been a problem at Church Wilne. The water at the plant
inlet contains 0.8 mg/1 of free chlorine and the dechlonnation
reaction within the beds has never taken the concentration below
0.05 mg/1 (normally 0.1 mg/1) at the outlet, even at a 15 ^nute
contact time. One UK plant reported a severe outbreak of Nais
worms when dechlorinated water was fed to carbon beds.
At one period in the pilot plant study an outbreak of the
algae anabaena was incompletely stopped by the sand filters and
decomposition of these within the carbon bed produced a strong
"chloroform" type odor at the plant outlet. This occurred to a
lesser extent a£ the main plant, but a more rigorous back-
washing, particularly of any idle adsorbers, cured the problem.
Total Organic Carbon
Removal of TOC was not a criterion for the plant design
using F200 carbon. In a pilot column test a contact time of 4
minutes with new carbon gave a reduction to 40 percent of the
influent TOC level of about 3 to 5 mg/1 for 5 to 6 days. This
then rapidly rose to 95 percent of the raw water level. In fact,
TOC's higher than the influent levels are often recorded. This
is probably due to a selective leaching effect or possibly some
sinking movement of heavier exhausted grains to the bottom of
the bed.
The pilot four column unit was initially used in parallel
with the main plant for guarantee purposes. Later, when these
initial objectives had been met, it was used for research on
TOC removals at long contact times using F400 carbon. At this
time, accurate TOC measurements were possible at low levels
using a CO,/CHA converting, GLC type, TOC instrument
(Figure 7)f *
If a 1 mg/1 TOC standard were set, a carbon bed with 12 and
18 minute contact times would be exhausted in 17 days and 24
days, with equivalent dose rates of 196 mg/1 and 218 mg/1,
respectively.
HALOFORMS
Haloform levels produced by chlorination are not signifi-
cantly reduced by activated carbon beds when free chlorine is
present throughout the bed (Table 2). Levels are reduced if
chlorine is absent, but other operational problems of a micro-
biological nature can result. One advantage of the use of
ClO^ for the distribution residual is that no further increase
of naloforms is found in the supply network.
342
-------
to
3.5
3.0
2.5
D
i,
m
o 2.0
C.
o
M
t,
o
1.5
1.0
0.5
'Paw* clarified
sand flit&r from
Church Wi l'ie raa.
Col. 1
3 mln contact time
Col. 2
6 miti contact time
»Col. 3
12 inln.conl.dL:. tlmu
Col. 4
"""IB min.c;ontae:t time
Flow Rate = 136 lltres/tir
-t.
1 4 fl 12 16 20 24
Days •» Period Ian. - March
.12 36 4C 44 48 52 5G 6fl
Running Time, Days
Figure 7. TOC Removal in a Column Test
With P400
-------
Table 2
•3
Typical Haloform Concentrations, mg/m
CHC13 CHCl2Br CHClBr2 CHBr3
River Water 3.0 0.3 0.5 2.4
Before Carbon Plant 14.8 20.2 19.6 2.4
After Carbon Plant 10.4 14.4 9.8 2.4
Detergents
Removal of anionic detergents, although not expected to
be significant at the low contact times with F200, was in
fact quite good. The low influent levels of 0.08 to 0.10 mg/1
were reduced by 90 to 100 percent for 34 days, followed by a
gradual rise to the input level.
OPERATION
Some difficulties occurred in the summer during plant
restart after shut-down periods. In normal operation, no growth
of organisms in the beds is apparent. During one period, after
a shut-down of only 4 hours, a bloom of a bright green discoidal
diatom ("Cyclotella") caused clogging of the beds and a long
backwash was necessary to restore forward flow. The first back-
wash water contained thousands of cells per ml.
It is now normal practice to give the adsorbers a short,
easing backwash before restart after any shutdown. Otherwise,
problems such as air blinding, gas bubble generation within the
beds, bed compaction, or biological growths invariably cause high
bed pressure drops.
Winterization
The plant is one of the first in the UK designed to
conform to normal outside petrochemical plant practice, but
operation in severe winter conditions of -1 to -10*C has still
caused operational difficulties. A fairly extensive program of
preventive measures against freezing was devised, but has proven
too onerous on the operators, especially when the furnace is not
operating. The furnace area has now been enclosed in a light
weatherproof building and some of the adsorber valves, notably
the backwash drain valves where the butterfly discs were often
frozen to their seats, have been trace heated.
344
-------
Normal Backwashing
The adsorber valves are all electrically operated
butterfly types. They can be set to any position between fully
open and fully closed, either locally by hand, by local control
of the motors, or automatically in the control room. This
facility, coupled with the display of flowrate from magnetic
flow meters on each adsorber, enables the total flow to be
equally divided among the adsorbers in service.
The backwash velocity for a given bed expansion is a func-
tion of the density difference between the bed media and water at
the operating temperature (Figure 8). This density difference is
smaller for GAC/water than sand/water. This means that in GAG
adsorbers, a greater expansion can be produced by a given flow
than in similar sand filters. However, with the adsorber, the
expansion, and hence cleaning and stratification, efficiency is
influenced to a greater extent by water temperature and a higher
backwash rate is needed in summer than in winter. The tempera-
ture of the water is displayed on the control panel to allow an
appropriate rate to be set. Some difficulty arose when two
adsorbers were loaded with a 2.3 m (7.5 ft) bed depth, as some
carbon was generally carried over with the backwash water. This
may have been due to poor adherence to the precise instructions
or possibly to a slight biological film on the top of the bed
making clumps of carbon granules less dense than normal granules,
thus enabling them to be washed out during backwash.
LATEST OPERATION
In 1976-7, new analytical procedures allowed the detection
of a series of Organic compounds of industrial origin in the
final water at trace levels of 20 microgram/litre (yg/1). These
were not causing taste and odor problems, but Severn-Trent
decided to use the plant for their removal, if possible. Thus,
all the adsorbers were put on line even though the daily flow
was below half the design flow at approximately 35-^45,000
m /day. This results in a contact time of 8-10 rain.as opposed
to the original 4 min. The compounds in question are removed to
below 1 yg/1 and this level has been set as a standard for this
water.
At one period, 2 years ago, two adsorbers were filled with
15 tons of F200 carbon and 15 tons of F400 carbon using the
extra bed depth available, creating a 10 itfin, contact time. A TOG
reduction from the average 3-4 mg/1 inlet level to 1.5 mg/1 at
the outlet was seen for the F400. The F400 was no better than
the F200 in removing the trace organics mentioned above.
345
-------
14 T
13
12
11
10--
a -
80%
45%
40%
35%
20%
10 TCMP. °c
15
20
Figure 8. Bed Expansion on Backwashing
-------
SUMMARY
The Church Wilne Works now has a carbon treatment facility
that allows water from a problematic source to be supplied to
part of the City of Nottingham. The difficulties with odour,
from warm water, have been overcome and the originally unsus-
pected problem of trace organic contamination has been satis-
factorily resolved. The Works has its own self-contained carbon
regeneration facility, which is covered in a separate paper at
this conference.
FUTURE OPERATION
The long term continued operation of the Church Wilne carbon
plant is in some doubt. Since the UK water industry reorganiza-
tion, a more rational use of water supplies in the region has
become possible. Already, a short pipeline has been completed
to a treated water aqueduct running down the Derwent Valley to
Leicester.
The quantity that can be taken from this line is limited
but it allows the shortfall at Church Wilne, due to the increased
contact time, to be made up. A blended water is currently sup-
plied to Nottingham.
Within the next four years, a new raw water intake upstream
of Derby is due to be constructed. The adsorbers may then be
allowed to fall into disuse or be retained in a minor taste and
odour control role, but continued upkeep of the regeneration
facilities will be in doubt.
ACKNOWLEDGEMENTS
This paper is given by kind permission of W.H. Richardson,
C Eng, MICE, MI. Mun. E, MRTPI, Dip. TP, Divisional Manager
Lower Trent Division of the Severn-Trent Water Authority and of
Humphreys and Glasgow Limited (H&G). The authors thank the
operations and scientific staff of Severn-Trent and commission-
ing staff of H&G, on whom the successful running and careful
monitoring of the plant depended and whose work is reported in
this paper. The comments made and conclusions drawn are those
of the authors alone and not those of Severn-Trent W.A. or H&G.
347
-------
TREATMENT OF GROUNEWATER WITH
GRANULAR ACTIVATED CARBON
Paul R. Wood
and
Jack DeMarco
Introduction
Three years of research were conducted at the 23 X I03m3/
day (60-mgd) Preston water treatment plant in Hialeah, Fla.
Groundwater obtained from wells 18 to 30 m (60 to 100 ft) deep,
in the porous limestone aquifer, is the sole source of drinking
water in southern Florida. The groundwater contains 10 mg/1
total organic carbon (TOC) and the trihalomethane formation
potential (THMFP) ranges from 650 to 950 ug/1. After lime
softening, approximately 18 mg/1 of chlorine is added for
breakpoint chlorination. The water passes through a chlorine
contact basin, a rapid sand filter [2 mm/s (3 gpm/sq ft)] and
then into the clearwell. Finished water entering the distri-
bution system contains 3 mg/1 of free chlorine. Free chlorine
drops to zero in 24 to 48 hr resulting in water at the con-
sumer's tap containing trihalomethanes in the 300-yg/l range.
The presence of other purgeable halogenated organic compounds
(HOC) increases the total HOC level to the 330-yg/l range.
The results of the first 2 years of research on adsorption
of HOC, TOC, and THMFP substances by granular activated carbon
(GAC) and synthetic adsorbents from raw, lime softened, and
finished water, have been presented in an earlier work (1).
Results of the first year of research on the 3-year EPA pilot
plant project for removing organic substances from drinking
water were presented in periodic progress reports to EPA.
This paper, covering a portion of both projects, describes
the effectiveness of four types of GAC in removing organic
substances from raw, lime softened, and finished water. Carbon
1* was tested in beds 0.75, 1.5, 2.25, and 3 m (2.5, 5, 7.5,
and 10 ft) deep with empty bed contact times (EBCT) of 6.2,
*Filtrasorb 400, 12 by 40 mesh, Calgon Corp., Pittsburgh, Pa.
348
-------
12.4, 18.6, and 24.8 min. Carbons 2**, 3***, and 4**** were
tested in beds 0.75 m (2.5 ft) deep with an EBCT of 6.2 min,
each.
DISCUSSION
It was found that GAC was the best adsorbent for the
water tesjted because of its ability to remove both precursor
substance's, as measured by TOC and THMFP, and volatile HOC.
Therefor?* GAG was extensively studied, as represented by the
composite "flow diagram in Figure 1, for all phases of the 3
years of study. The bench scale adsorption unit located in the
laboratory at the Preston plant receives raw, lime softened, and
finished water directly from the plant. The glass columns mea-
sured 1.5 m (5 ft) in height by 2.54 cm (1-in) inner diameter.
A flow rate of 2 mm/s (3 gpm/sq ft) was maintained by rotameters.
The columns were backflushed as required to maintain a given
range of inlet pressure.
PURGEABLE HALOGENATED ORGANIC COMPOUNDS
The average concentration of 19 specific purgeable HOCs
for the 3-year study period for raw, lime softened, and finished
water at the Preston plant is shown in Table 1. The raw ground-
water had a very low level of high molecular weight extractable
compounds (industrial and agricultural pollutants) and a con-
siderable concentration of purgeable HOC averaging 48 ug/1. The
three major compounds in order of concentration were cis-1,2-
dichloroethene, vinyl chloride, and trans-1,2-dichloroethene.
The generally lower concentrations of HOC in the lime softened
water versus that in raw water can be explained by the loss of
compounds through volatility and adsorption on the precipitated
sludge. The increase in the THM levels can be attributed to a
small amount of chlorine, which was added before the lime to
activate the sodium silicate used as a flocculation aid. The
concentration of HOC in finished water was less for some com-
pounds and greater for others. Finished water from the clear-
well contained approximately 187 pg/1 of HOC. Detailed results
of changes in HOC concentrations in the overall treatment plant
profile, and adsorption by adsorbents were presented in an
earlier work (1). The only other observation drawn from
the earlier paper is that despite a reduction in TOC from
10 mg/1 in raw water to 5.4 mg/1 in finished water, the adsorp-
tive capacity of carbon 1 for cis-1,2-dichloroethene was 30
**Nuchar WVG, 12 by 40 mesh, Westvaco Corp., Covington, Va.
***Hydrodarco 1030, 10 by 30 mesh, ICI Americas, Inc.,
Wilmington, Del.
****Witcarb Grade 950, 12 by 30 mesh, Witco Chemical Corp.,
New York, N.Y.
349
-------
PRESTON PLANT
C.B.
S.F.
C.W.
Distribution System
BENCH SCALE
ADSORPTION UNIT
Spike
i
ILL
TTTT
C - Granular Activated Carbon
S - Sand
NOTE: All carbon depths are 2.5 feet unless otherwise noted.
Figure 1. Composite Flow Diagram For All Phases of
GAG Studies Over A Three Year Period
-------
Table 1
Average Concentration of Nineteen Specific Purgaable
Halogenated Organic Compounds in Raw, Line Softened (L.S.),
and Finished Water (F.W.) at the Preston Plant
Average Consentration
ug
VINYL CHLORIDE
METHYLENE CHLORIDE
TRANS 1,2-DICHLOROETHENE
1,1-DICHLOROETHANE
CIS 1,2-DICHLOROETHENE
CHLOROFORM
1,1,1-TRICHLOROETHANE
1,2-DICHLOROETHANE
CARBON TETRACHLORIDE
TRICHLOROETHYLENE
BROMODICHLOROMETHANE
TETRACHLOROETHYLENE
CHLORODIBROMOMETHANE
CHLOROBENZENE
BROMOFORM
p-CHLOROTOLUENE
m-DICHLOROBENZENE
p-DICHLOROBENZENE
0-DICHLOROBENZENE
N=Nil
ND=Not Determined
RAW
12.8
0.45
1.5
0.58
25.6
.08
0.1
0.34
N
0.003
N
1.1
N
0.02
N
0.51
0.17
L.S.
9.7
ND
0.95
0.45
22.3
1.2
0.12
0.33
0.35
0.004
0.13
0.72
0.007
0.03
N
0.28
0.16
F.W.
6.2
ND
0.77
0.4
19.9
67.3
7.7
0.68
47
0.003
33.6
0.86
2.6
0.1
N
0.21
0.14
TOTAL
43.25
36.73
187.36
351
-------
percent greater in raw water than in finished water. As ob-
served in Table 1, the level of cis-1,2-dichloroethene compared
to the total level of all 19 HOCs is 59 percent in raw water
and 11 percent in finished water. The greater adsorptive
capacity of GAC for cis-1,2-dichloroethene in raw water is
attributed to less competitive adsorption by other HOCs, despite
the larger amount of TOC present. The adsorptive capacity of
GAC increases with the influent concentration of a substance.
This was included in the calculations (1) and the 30 percent
increase was based on equal influent concentrations.
ORGANIC REMOVAL BY FOUR TYPES OF GAC
The bench scale adsorption unit was set up, as shown in
Figure 2, to study the effectiveness of four types of GAC to
adsorb TOC, THMFP, and HOC from finished water. The volume
and EBCT (6.2 min) were constant for all carbons.
HOC Removal
The effectiveness of the four types of GAC in removing HOC
is discussed briefly; emphasis is placed on the removal of THM
precursor substances. Detailed HOC removal data based on both
equal volume and equal weight of GAC are reported elsewhere (2).
Chloroform breakthrough curves are presented in Figure 3.
These curves are typical of the relative performance of each GAC
for all the HOC monitored. The results of integrating these
breakthrough curves are shown in Table 2.
Table 2
Percent of Chloroform Removal By Four Types of GAC
% Chloroform Removal
Type of Weight of Carbon % Chloroform Removal By Equal Weight of GAC
GAC in Grams By Equal \folume of GAC (172.4 Grams)
WVG
ICI
F400
W950
195.6
144.1
163
172.4
40
48
53
78
36
57
56
78
352
-------
i:xE
SOFT
ENED
J_
MM^
W
V
MMM
I
c
CLE/
WEI
±
4
a
,R
W
9
5
3
DIET.
SYS.
1
2.5'
T T T T
Figure 2. Bench Scale Adsorption Unit Flow Diagram for the
Study of Four Types of GAC on Finished Water
100-
131
Finished Water Entering Each Column —Q—
Effluent from 2.5' WVG — A—
Effluent from 2.5' ICI --O—
Effluent from 2.5' F400 _.Q._
Effluent from 2.5' W950 Q
Days
14
28
42
59
Figure 3. Chloroform Breakthrough Curves
Through Four Types of GAC Columns
353
-------
On an equal volume basis, W950 had the largest adsorptive
capacity with 78 percent removal of chloroform, followed by
carbons F400, ICI, and WVG with 53, 48, and 40 percent, res-
pectively. The last column of data in Table 2 shows the per-
centage removal at equal carbon weights of 172.4 g, the weight
of W950 used. Once again, W950 is superior with 78 percent
removal. F400 and ICI are very similar wi*h 56 and 57 percent
removal, respectively. The removal of chloroform by WVG was 36
percent.
In Figure 3, the scale of the "Y" axis does not show
that W950 allowed a continuous low level passage of chloroform
before the major breakthrough at 28 days. This continuous low
level passage averaging 1 yg/1 is shown in Figure 4. Since the
surface of this GAC (made from a hydrocarbon resin) is hydro-
phobic and there is a higher percentage of surface area in the
small diameter pore range than in the other types of GAC, the
manufacturer recommends boiling the carbon in water to deaerate
it. This was not done for the initial test. Toward the end of
the 59-day test period another column was placed in service
using boiled W950 GAC. Low level passage of chloroform through
this column is shown in Figure 4. The average passage was again
1 yg/1 indicating that deaeration did not affect the level.
Further study with greater bed depths of this carbon will be
necessary to determine the mass transfer zone.
GAC Bed Life
In this water system, it was discovered that the most
economical use of GAC is on finished water. Finished water
normally leaves the plant with 3 mg/1 of free chlorine and a
total HOC level in the 185 yg/1 range. The free chlorine reacts
with precursor substances still present in the water after it
leaves the plant, dropping to zero in less than 24 hr. Occa-
sionally the decline in the amount of free chlorine will take
up to 2 days. The growth of additional THM in the distribution
system raises the level of total HOC at the consumer's tap to
more than 300 yg/1.
If GAC were used on finished water from the clearwell
before the water leaves the plant, in order to remove HOC and
THMFP substances and limit the level of total THM at the con-
sumer's tap to not more than 100 yg/1, several problems would be
encountered. Carbon removes all free chlorine so that the
effluent from a GAC column would have to be rechlorinated to
3 mg/1 of free chlorine to meet current state requirements.*
Rechlorination is also necessary because of the high bacterial
'Free chlorine levels in finished water vary according to
individual state requirements.
354
-------
Unbollu-d W950
Boiled W950
Days g
Figure 4. Low Level Passage of Chloroform Through GAC
W950 Before Major Breakthrough at 28 Days
200-
100-
Day*
-•— Finished Water
_Q— Finished Water Aged 2 Days
.£_ Distribution System
A 7.5' of F400
-O—10' of F4oa
A
28
42
*1
147
70
81
122
Figure 5. Bed Life of 7.5 and 10 Feet Deep Filtrasorb
400 Adsorbers
3S5
-------
count (1) in the GAC effluent. The GAG bed life for this system
was determined by using the criterion that bed exhaustion occurs
when a sample of the bed effluent, rechlorinated to 3 mg/1 of
free chlorine and held for 2 days, reaches a THM level of 100
yg/1. Bed life for 2.25- and 3-m (7'. 5 and 10-ft) deep columns
of carbon 1 are shown in Figure 5. Th€ influent concentration
of total THM in yg/1 in the finished water entering the dis-
tribution system is shown. Values showing the total THM in
finished water aged 2 days in the laboratory and having 2 days'
residence in the distribution system are also shown. Rechlori-
nated 2-day, bottle-aged sampling was initiated on day 45 of the
122-day test period for the 2.25-m (7.5-ft) deep column. For a
7.5-ft GAC column, the 100 yg/1 line was crossed at 65 days.
For a 3-m (10-ft) deep column, the bed life was 81 days, which
represents a carbon use rate of 10 kg/ml (80 Ib/mil gal).
During most of the 81 days, the level of total THM at the
consumer's tap would exceed 50 yg/1.
By plotting THM breakthrough in the 2.25- and 3-m (7.5 and
10-ft) deep carbon 1 beds, as shown in Figure 6, it can be seen
that even if all the THMFP were removed from finished water by
the carbon, bed failure would occur (i.e., breakthrough would
reach the 100 yg/1 level) in 94 and 119 days, respectively. If
vertical lines are drawn from the X axis at 65 and 81 days
(Figure 6), which are the bed life failure days for 2.25 and
3 m (7.5 and 10 ft) of GAC in Figure 5, it is observed that at
65 and 81 days, respectively, one half and one fifth of the 100
yg/1 level were a result of THM breakthrough.
In a separate phase of the work, bed life based on the
above criterion was determined for a bed of carbon F400 that was
0.75 m (2.5 ft) deep (Figure 7). Curves are also presented for
the concentration of total THM in the finished water entering
the column, the total THM in 2-day, bottle-aged finished water
(representing distribution system growth), THM breakthrough in
the column effluent, and THM growth in rechlorinated 2-day aged
column effluent. If the carbon removed all the THMFP, bed life
would be 35 days because of THM breakthrough. However, actual
bed life was only 9 to 10 days and the failure was solely a
result of THMFP breakthrough.
In Figure 8, data for the THMFP in finished water and
removal by 0.75 m (2.5 ft) of carbon F400 are plotted. A
vertical line from 9 to 10 days meets the THMFP breakthrough
curve at approximately 200 yg/1 THMFP. Therefore a GAC bed 0.75
m (2.5 ft) deep fails (based on the criterion chosen above) when
the THMFP breakthrough level reaches 200 yg/1. When the break-
through curve for TOC was plotted, the value of TOC breakthrough
at failure (9 to 10 days) was 3 mg/1.
356
-------
en
200
150 -
a:
in
C.
100 _
50 -
o e
Days 0
J /'
FINISHED WATER
M '\ A A l\
i \ ' * I ' r^ t \ ' ^
/ u • •" '
w
\
i
Day Day
65 81
14 28
42
7.3 FI.
GAG
<
o
X
O
/
^O
0.
0, /
f' ^
-rwr^0"
^•&-ev-a-ip v i | |
50 70
f-ox
i
1
x .' .^
'* \
0 ,' i
1
r]
^^J-J
TT
ill!
84
>Vx:
0-°^°^
P' 10 FT.
(' GAC
<
II M II
98
1
•
1
122
Figure 6. Trihalomethane Breakthrough Curves for
7.5 and 10 Feet of Filtrasorb 400
-------
300 -
Av. 225 Finished Water
2-Day Bottle-Aged
Av. 120 Finished Water
O 2.5 ft. GAC Effluent
THM Breakthrough
D 2.5 ft. GAC Effluent
Rechlorinated - 2-Day Aged
Days 0
14
28
42 53
Figure 7. Bed Life of 2.5 Feet of Filtrasorb 400 and
THM Breakthrough
600 -
400 -
a.
EL.
200
Days 0
Av. 434
98
122
Figure 8. THMFP In Finished Water and Removal by 2.5
Feet of Filtrasorb 400
358
-------
THMFP and TOC Removal
In Figure 9 the THMFP concentration in finished water
entering columns containing four types of GAC and the THMFP
breakthrough curves for each carbon are plotted. Adsorption by
the four carbons was compared at two points. The curves were
integrated at the end of the 59-day test period and at a THMFP
breakthrough level of 200 ug/1, the point of bed failure for
0.75 m (2.5 ft) of GAC. At 200 yg/1 breakthrough (Figure 9) the
bed life for carbon F400 was 10 days, for carbon WVG, 7 days,
for carbon ICI, 3 days, and for carbon W950, approximately 1-1/2
days. Although carbon 4 w&.s the best carbon for HOC removal, it
was the poorest for THMFP removal. Similar results are observed
with TOC data (Figure 10) when compared at the TOC breakthrough
of 3 mg/1.
Table 3 shows the amount of THMFP adsorbed by each carbon
column in 59 days. This compares the carbons at equal volume
and the adsorption on an equal weight basis, compared to the
163-g weight of carbon F400.
Table 3
THMFP Adsorbed by Each Carbon in 59 Days
THMFP Entering in 59 Days = 2002 milligrams (mg)
Equal Vol.
F400
WVG
ICI
W950
Grams
(gm)
163
195.6
144.1
172.4
Adsorbed
mg
524
545
314
160
% TfMFP
Removed by
Equal
Volume of
GAC
26
27
16
8
Wt. to Adsorb
as much as
163 gm F400
188 gm
240 gm
565 gm
% THMFP
Removed
Equal Weight
of GAC
(163 grams)
26
23
18
8
By the 59th day, 2,002 mg of THMFP had entered each
column. The weight of each carbon is shown in Table 3. F400
(163 g) adsorbed 524 mg, WVG (195.6 g) 545 mg, ICI (144.1 g) 314
mg, and W950 (172.4 g) 160 mg. The last column of data in Table
3 shows the number of grams of each carbon required to adsorb as
much as 163 g of F400. The required weight of WVG was 188 g
(163 g of WVG would adsorb 454 mg). The required weights for
ICI and W950 were 240 and 565 g, respectively.
359
-------
-•— Finished Water
_0- 2.5' GAC WVG
-^y 2.5' GAC ICI
_o_ 2.5' GAC F400
-Q-2.S' GAC W950
Days 03 7 10 14 17 21 24 28 31 35 38 42 45 49 52 56 59
Figure 9. THMFP In Finished Water and In the Effluen*
From Four Types of GAC
8-1
Days 0 3
Q Finished Water
O — Finished Water thru 2.5' F400
D— Finished Water thru 2.5' W950
A— Finished Water thru 2.5' HVG
Finished Water thru 2.5' ICI
Figure 10. TOC in Finished Water and Removal by Four
Types of GAC
360
-------
An adsorption comparison of each carbon at approximately
10 days is made in Table 4. At this comparison point, F400
appears to adsorb better than at the 59-day comparison point.
Similar data for TOC adsorption were found for the two
comparison points.
Table 4
T5MFP Adsorbed by Each Carbon in Approximately 10 Days
TJMFP Entering in Approximately 10 Days * 172 mg
Equal VD!. Grans Adsorbed % TtMFP Wt. to Adsorb
(gm) mg Removed by as much as
Equal 163 gm F400
Afolume of
GAC
% TfMFP
Removed by
Equal Weight
of GAC
(163 grams)
F400
WVG
ICI
W950
163
195.6
144.1
172.4
92.6
60.3
23.7
15.1
54
35
14
9
300 gm
563 gm
1057 gm
54
29
16
8
COMPARISON OF GAC ON LIME SOFTENED AND FINISHED WATER
In another phase of the study, F400 was evaluated on the
lime softened water. The flow diagram of the bench scale
adsorption unit for this phase is shown in Figure 11. In
section A, 2.25 m (7.5 ft) of GAC was evaluated to compare the
removal of organics with a similar column on finished water
evaluated in a previous phase. In section B, 1.13 m (3.75 ft)
of GAC was evaluated for removal of organics found in the
water and for removal of high molecular weight extractable
compounds (industrial and agricultural chemicals) added in the
spiking unit. In this paper only two points concerning this
phase of the study will be discussed: (1) THMFP breakthrough
in section A, and (2) removal of spiked compounds in section B.
THMFP and TOC Breakthrough
The influent concentration of THMFP to the 2.25-m (7.5-ft)
GAC column is shown in Figure 12. The THMFP breakthrough for
the 2.25-m (7.5 ft) of GAC on lime softened water in this phase
and for finished water from a previous phase is also shown. The
average influent THMFP concentration on lime softened and
361
-------
Dist. System Points
ter
2.5'
S
A
N
D
«-co2
Bench Scale Adsorption Unit
| 1
vrt
r—
•
F
4
0
0
1 '
F
4
0
0
3.75'
2.5'
1
S
A
N
D
Spiking Unit
1 ' 1 ' 1 ' 1 '
3.75'
4
0
0
V
SECTION A
7.5' GAC
SECTION B
3.75' GAC
Figure 11.
Flow Diagram for Evaluation of
Filtrasorb 400 on Lime Softened Water
362
-------
la
O
US
(T5
U-l U
o
m ^
• tn
r- -r-i
c
T5
m c
JJ 03
(0
Q T3
0)
J= C
U> (D
D JJ
O u-i
U O
OJ -H
CQ
C
Cu O
t
s: o
g5
OJ
1J
3
en
363
-------
finished water feeds were 463 and 434 yg/1, respectively.
Corresponding partial saturation plateau levels (1) were 265 and
250 pg/1. With the average influent concentration of THMFP in
the lime softened (463 yg/1) and finished water (434 yg/1)
phases so close to each other, it is surprising that the satura-
tion time needed to achieve almost equal partial saturation
plateau levels is so differeAt*-«-afrproximately 19 and 39 days,
respectively. The explanation Ttiay be found in a plot of TOC
data for the two phases in Figure 13. The average influent
concentrations of TOC in the lime softened and finished water
test periods were 6.5 and 5.4 mg/1. This is a difference of
17 percent (5.4/6.5 X 100) compared to a 6 percent difference
(434/463 X 100) for the THMFP levels in Figure 12. In
Figure 13 the 59-day test period is too short to determine the
partial saturation plateau level. The average plateau level
shown for finished water was achieved in 122 days of testing.
However, if the two TOC breakthrough curves are considered
at a common level of 3 mg/1, it is seen that the time to reach
this level is 21 days for lime softened water and 39 days for
finished water. The higher TOC level on the lime softened run
resulting in earlier saturation might explain the similar
results found with THMFP data despite the closeness of the
THMFP.
Removal of Spiked High Molecular Weight Extractables
Since the raw groundwater entering the Preston plant is
low in higher molecular weight extractable compounds from
industrial and agricultural sources, section B of the bench
scale adsorption unit (Figure 11) was used to spike some of
these compounds. Figures 14, 15, and 16 show the results with
chlorinated pesticides. The spiked influent levels are shown.
In all six examples some breakthrough occurs even on initial
startup. The zero-day sample was taken 30 min after water flow
was begun. The percent breakthrough at each sample point is
given. Overall the carbon removed more than 90 percent of the
high molecular weight extractables. The results with industrial
chemicals are shown in Figures 17, 18, and 19. In Figure 17
breakthrough was observed at 14 days for 2-nitrotoluene. Ttffs
was followed by a period of complete removal and further break-
through. Eth^yl benzoate was completely removed up to day 42.
In Figure 18 $11 of the naphthalene was removed except for
some breaktr.tough on day 3, all of the l-chloro-2-nitrobenzene
was removed up to day 42. In Figure 19 all the dimethylphtha-
late was removed until test day 10. After test day 10 varying
amounts of this compound were observed in the blank and it was
necessary to abandon tracing of this compound. Di-isobutyl-
phthalate showed early breakthrough with high levels of break-
through occurring after day 14. This was followed by a period
of low level breakthrough. The results of the studies on the
removal of spiked high molecular weight compounds are in the
364
-------
U)
Oi
V)
8 -
o< ..
E 6
U
O
EH
4-
2 -
Effluent from 7.5 feet F400 on Finished Water
Inlet concentration (lime softened water)
Effluent from 7.5 feet F400(lime softened water)
•—& Av. 6.5 Inlet concentration
on lime softened '
water
Av. Inlet concen-
"• tration on
5.4 Finished Water
Av. 3.75
Days o 3 710 14 17 21 24 28 31 35 J8 42 45 49 52 56 59 63
70
77
84
Figure 13. TOC Breakthrough Data for 7.5 Feet of Filtrasorb
400 on Lime Softened and Finished Water
-------
«* UIIC
.012-,
.010
U>
.012 —
.010 —
.008 —
.006 —
J3
Q.
Q.
O
g -004-
.002 —
t BIIC
--•- IN
-o- OUT
Days 0
Days
-f 1 » 1 » » ?
7 10 14 17 21 24 28
Figure 14. Concentration of Spiked Compounds
In and Out of 3.75 Feet of GAC
-------
U)
a\
•vl
.006
.005-
.004-
.00>
Xl
a
a
0)
C -002H
R)
•O
M
O
U
.001-
o
•-I
o
o
Chlordane
_ IN
-OUT
o
OUP
Ou>
0-
Days 0
of
u>
o
7
U,
1^
10
JP
GO
O
14
OP
«-H
O
17
*>
o
UP
n IM
in, °~~.
T 1
^1 24
UP
0
o
— - o
2U
.0010-
.0000
XI
a
a
C
•H
M
(U
•H
Q
.0006
.0004-
.0002-
Dieldrin
IN
OUT
Days 0
10 14 17 21 24 28
Figure 15.
Concentration of Spiked Compounds
In and Out of 3.75 Feet of GAC
-------
.020-,
00
.020 _
.016 -
•a
0)
c
m
•d
O
>i
X
o
.012-1
.008-
. 004-
Day s
Days
Figure 16. Concentration of Spiked Compounds
In and Out of 3.75 Feet of GAC
-------
.12 -
U)
Ol
VO
a
o o.o
N
c
u
JO
J3 °'04-
U
O.U2 .309 I172
. 0.1*1 •
7 10 14 17 21 24 28 31 35 38 42 45 49 52
0 't -» -i—f V—• *—»—
lyi 03 7 10 14 17 21 24
Day
28 31 35 31 42 45 49
Figure 17. Concentration of Spiked Compounds In and Out of 3.75 Feet of GAC
-------
1.24
0.12
IN
OUT
.84
u»
-J
O
.2-
y
Days
3 7 10 14 17 21 24 28 31 35 38 42 45 49 52
IN
OUT
Days 03 7 1Q 14 17 21 24 28 31 35 38 42 45 49 52
Figure 18. Concentration of Spiked Compounds In and Out of 3.75 Feet of GAC
-------
u>
-o
0.60-t
0.50-
0.40-
x:
4>
&
a,
-• 0.30
S
-*4
Q
0.20-
0.10-
IN
OUT
••» f » 1 1 ' ' 1-
Days 03 7 10 14 17 21 24 28
0.30-
0.20-
a.
3
a
o
I
0 0.10.
.05-
Days 02 7 10 14 17 21 24 2B 31 35 38 42 45 49 52
Figure 19. Concentration of Spiked Compounds In and Out of 3.75 Feet of GAC
-------
preliminary stage. Other compounds will be studied for longer
periods of time and desorption rates will be studied after
deliberate spikes in influent levels are created.
CONCLUSIONS
A study of chloroform removal from finished water showed
that equal volumes of four types of GAC removed 78 (W950),
53 (F400), 48 (ICI), and 40 (WVG) percent of the chloroform
entering the water during a 59-day test period. On an equal
weight basis, W950, ICI, F400, and WVG removed 78, 57, 56,
and 36 percent, respectively. The removal data for 18 other
purgeable halogenated organic compounds rated the carbons
in a similar order.
Removal of THMFP substances from finished water showed
that, during a 59-day test period, an equal volume of four types
of GAC removed 27 (WVG), 26 (F400), 16 (ICI), and 8 (W950)
percent of the total THMFP entering the water. On an equal
weight basis, carbons F400, WVG, ICI, and W950 removed 26, 23,
18, and 8 percent, respectively.
In the Hialeah, Fla., water system, a sample of water from
a GAC column rechlorinated to 3 mg/1 of free chlorine and
bottle-aged 2 days simulates the level of total trihalomethanes
that will reach a consumer's tap. If a maximum level of tri-
halomethanes of 100 ug/1 is chosen, the bed life of 0.75, 2.25,
and 3 m (2.5, 7.5, and 10 ft) of carbon 1 is approximately 10,
65, and 81 days. The carbon use rate based on 3m (10 ft) of
F400 is 10 kg/ml (80 Ib/mil gal).
The four types of GAC were evaluated at equal bed depths
of 0.75 m (2.5 ft). Since bed life for this bed depth (based
on the above criterion) is only 10 days, when the THMFP break-
through curves were integrated at this time period, at an equal
volume, carbons F400, WVG, ICI, and W950 removed 54, 35, 14, and
9 percent of the entering THMFP substances. At an equal weight,
F400, WVG, ICI, and W950 removed 54, 29, 16, and 8 percent,
respectively. The removal data for total organic carbon rated
the carbons in a similar order.
A method of spiking water with high molecular weight
industrial and agricultural chemicals was devloped and pre-
liminary adsorption data are presented. Chlorinated pesticides
exhibited early low level breakthrough. Overall, more than 80
percent of the spiked high molecular weight compounds were
removed and more than 90 percent removal usually was obtained.
Removal of 2-nitrobenzene, ethyl benzoate, naphthalene, 1-
chloro-2-nitrobenzene, dimethylphthalate, and di-isobutylph-
thalate ranged from complete to poor. A wide range of chemicals
will be tested for longer periods and desorption rates after
periods of higher influent levels will be examined.
372
-------
ACKNOWLEDGEMENT
This research was the joint effort of the U.S. Environ-
mental Protection Agency, Cincinnati, Ohio, and of the Dade
County Health Department, Miami-Dade Water and Sewer Authority,
and the Drinking Water Research Center of Florida International
University, all of which are located in Miami, Florida.
REFERENCES
1. Wood, P.R., et al. Removing Potential Organic Carcinogens
and Precursors from Drinking Water. U.S. EPA, Cincinnati,
Ohio (in press).
373
-------
TREATMENT OF OHIO RIVER WATER
Richard Miller
The problems of trihalomethane formation during the
treatment process and trace synthetic organic chemical concen-
trations in the source waters are of concern to many water
utility managers. The Cincinnati Water Works has initiated both
unilateral and multi-agency research to determine some of the
causes, the scope, and possible solutions to these problems.
CINCINNATI WATER WORKS LOCATION AND WATER TREATMENT SYSTEM
Our primary source of water is the Ohio River at a point
463 miles (745 kilometers) downstream from its headwaters at the
confluence of the Allegheny and Monongahela Rivers. Six tribu-
taries enter the Ohio River in this 463 mile (745 kilometer)
reach, but perhaps the most important is the Kanawha River. The
Kanawha River valley accomodates most of the giant chemical cor-
porations in this country and is the location of many points of
possible toxic discharge. Also located along this reach of the
Ohio's main stem are approximately 270 point sources of discharge
from both municipal and industrial facilities (Figure 1). How-
ever, we are indeed fortunate that the 110 mile (177 kilometer)
stretch immediately upstream of Cincinnati has very few major
discharge points. This is the situation regarding our primary
source of water.
Our raw water is obtained via intake structures on the
Kentucky side of the river and flows through tunnels which termi-
nate at our raw water pumping stations. Two stations with a
combined capacity of 340 mgd (1.3 million cubic meters per day)
lift the water to twin pre-settling basins having a combined
capacity of 372 million gallons (mg) (1.4 million cubic meters).
To my knowledge such basins are unique to only the Cincinnati and
Louisville water systems in our area of the country. They are a
definite asset to a system located on an industrial stream as
they provide several days supply if intakes are closed due to a
chemical spill.
From the pre-settling basins the water flows by gravity
through coagulation and clarifier basins, then into our filter
gallery. The filter gallery contains 47 rapid sand filters.
374
-------
KENTUCKY
Figure 1. Known Industrial and Municipal Wastewater
Discharges to the Ohio River. Lines Indi-
cate Industrial Discharges. Solid "T's"
Indicate Municipal Wastewater Discharges.
375
-------
Each filter has a rated capacity of 5 mgd (19,000 cubic meters
per day) at a loading rate of 2-1/2 gallons per minute per square
foot (6 m/h). After filtration, the water collects in two clear-
wells with a combined capacity of 28 rag (106,000 cubic meters).
A schematic of the treatment facilities is illustrated in
Figure 2.
:HE.MICAL
HOUSE
WEST
4
Figure 2. Schematic of Treatment Facilities
Facilities are available to apply chlorine, aluminum sul-
fate, and powdered activated carbon in the raw water lines
upstream of the pre-settling basins. An additional facility
provides for the -capability of adding chlorine, ferric sulfate,
lime, soda ash, and powdered activated carbon between the pre-
settling basins and the coagulation basins. Recent improvements
provided the capability of chlorinating downstream of the clari-
fier basins immediately ahead of the filters and increased the
capacity of our post chlorinating facilities. Currently, all
disinfection is by chlorination at these two points.
The relocation of the chlorine feed resulted from a uni-
lateral research project conducted in 1975 and 1976. As a result
of this relocation, we have reduced the formation of trihalo-
methanes approximately 80 percent. As an associated benefit of
this change, we have been able to reduce our chlorine feed rates
from 12 milligrams per liter (mg/1) to approximately 3.6 mg/1.
U.S. EPA CARBON ADSORPTION AND REGENERATION GRANT
Early in 1977, we applied to the United States Environmental
Protection Agency (USEPA) for a research grant for a project
entitled "Feasibility Study of Granular Activated Carbon Adsorp-
tion (GAG) and On-Site Regeneration." A grant award was made in
August of 1977. This project will investigate both the cost and
effectiveness of converting sand filters to GAC-filter-adsorbers
in comparison to deep bed, post-filter carbon contactors.
376
-------
A second phase, currently in progress, involves two pilot
column studies. One study is investigating the effect of regen-
eration on the carbon's adsorptive capabilities (Figure 3). The
other study is comparing bituminous and lignite based carbon in
deep bed pilot columns (Figure 4). This latter study may also
be used to investigate the reliability of deep bed, pilot carbon
columns to predict performance of the plant-scale post filter
carbon contactors. The final phase of this project will evaluate
both the effectiveness and feasibility of on-site GAC regenera-
tion.
Much of the analytical work for this project is being
conducted in-house. A former lecture room has been converted
into an acceptable temporary organics laboratory. It currently
houses two gas chromatographs and all of the necessary ancillary
equipment. A low-level module was added to our Total Organic
Carbon (TOC) analyzer and a gas chromatograph mass spectrometer
will soon be added to our equipment inventory.
In the first phase of our project we converted three sand
filters, each rated at 5 mgd (19,000 cubic meters per day) to
GAC filter adsorbers. In one filter, identified as 19A, 18
inches (0.46 meters) of sand was replaced with 18 inches (0.46
meters) of 12 x 40 mesh size GAC. This gives an empty bed
contact time of 4-1/2 minutes. A second filter, identified as
21A, had all 30 inches (0.76 meters) of sand replaced with 30
inches (0.76 meters) of 12 x 40 mesh size GAC. A third filter,
identified as 23A, had all 30 inches (0.762 meters) of sand re-
placed with 30 inches (0.76 meters) of 20 x 50 mesh size GAC.
Filters 21A and 23A both have an empty bed contact time
of 7-1/2 minutes. All carbon is bituminous base and is from a
single manufacturer. In all, 266,700 pounds (121,200 kilograms)
of carbon were purchased at a cost of $101,500.
An effluent sample point and several intermediate depth
sample points were installed in all filters. The intermediate
depth samples were not reliable indicators of water from their
respective depths due to the rapid development of air pockets
around the sample intakes. It was found that the water quality
improved at increasing depths.
The objectives of this phase of the study were:
• To determine if current water quality standards
can be met and maintained without adhering to the
Ten State Standard requirement of 12 inches (0.305
meters) of sand under the carbon media.
377
-------
Figure 3. Pilot Column Study - Effect of Regeneration
on Adsorption Capacity of Carbon
378
-------
Figure 4. Bituminous and Lignite Based Carbon Pilot Columns
379
-------
o To compare the carbon filters to the rapid sand
filters with regard to physical, chemical, and
bacterial data and to compare effectiveness in
removing specific organics.
o To determine the life expectancy of the GAC filter
based on TOC and trihalomethane (THM) data.
Analytical Results
Following the start-up (February 14, 1978), data were
collected at least weekly from the three filters for six
months, at which time the filters were well exhausted.
The physical-chemical data collected to compare the GAC
filter to the sand filters were similar. The effluent from
the carbon filters was, for the most part, equal in clarity
to that of the sand filters. Three brief episodes of moderate
turbidity occurred with filter 19A which were of little or no
significance. Apparently the 12-inches (0.305 meters) of
fine sand in filter 19A does not improve effluent clarity.
As expected, the carbon filters produced an effluent
that was either odorless or had a barely discernible odor.
The filter also removed all but marginal traces of free
chlorine and moderate levels of chloramines.
Figure 5 shows a plot of the concentration of TOC in the
influent and the effluents of the three GAC filters. It is
apparent that, initially, the GAC filters are very effective
in the removal of TOC. The proposed EPA regulations on
organic contaminants were used to determine when the carbon
filters would have needed to be replaced or regenerated. The
criterion states that, when the effluent TOC rises by 0.5
mg/1 above the baseline concentration in three successive
weekly effluent samples, the carbon must be regenerated or
replaced. From the figure it can be seen that filters 19A
and 23A would have required regeneration 28 days after startup
while filter 21A would not have required regeneration until
42 days after start-up. It is interesting to note that after
six months the carbon filters are still removing TOC.
All three GAC filters were found to have high total
bacteria plate counts (Table 1) and total coliform counts
(Table 2). These findings have been reported in the literature
and were, therefore, expected. The counts varied so much that
it is difficult to draw any firm conclusion from the results;
but it does appear that the counts in the effluent are higher
than in the influent.
380
-------
to
00
LEGEND!
INFLUENT
19 A EFF
21 A EFF
I4FEB78
60
I9APR78
120
I8JUN78
180
I7AUG78
Figure 5. Total Organic Carbon Concentrations
-------
Table 1. Total Bacteria Plate Counts
Plate Count
Day
18APR78
20JUN78
11JUL78
Project
Day
59 -
122
143
Raw
6,100
7,200
19,000
Per ml
Influent
21
3
7
21A-
EFF
200
1,100
6,800
Table 2. Total Coliform Counts
Total Coliform Per
Day
18APR78
20JUN78
11JUL78
Project
Day
59
122
143
Raw
2,700
32,000
27,000
100 ml
Influent
1
1
1
21A-
EFF
1
28
300
Before discussing the effectiveness of the GAG filters in
removing the trihalomethanes (THM), some terminology must first
be explained:
- Trihalomethane, as referred to in this presentation is the
summation of chloroform, bromoform, dibromochloromethane,
and bromodichloromethane.
Instantaneous Trihalomethanes are the THM already formed
at the time of sampling.
Trihalomethane Formation Potential (THMFP) is an indirect
analysis of precursors available for chlorination under the
test conditions and thus, a measure of removal of the same
precursors by the GAC. In our case, the test conditions were
set beyond those ever likely to exist in our water treatment
system; they were as follows: to a 325 ml sample bottle,
add 15 ml of buffer to obtain a pH of 9.5; then ad£ chlorine
solution to obtain 15 mg/1 free chlorine residual, fill with
sample, and incubate at 85° for 7 days.
382
-------
- Seven-Day Simulated Distribution THM was intended to predict
the effect of various water treatment measures on the qual-
ity of finished water at remote points in the distribution
system. Tests were conducted as follows: to a 325 ml
sample bottle, add 15 ml of buffer to obtain pH of 8.2, add
sufficient chlorine solution to result in 2.5 mg/1 free
chlorine above what was in water at that time, and incubate
for 7 days at finished water temperature using flow through
water bath. The analysis is performed by analyzing for
total THM after a water sample has been exposed to the same
pH and chlorine dose used in the treatment plant at the pre-
vailing temperature of the distribution system, for a length
of time originally estimated to equal transit time to the
more remote points of the distribution system. We recently
found that these results are not truly indicative of our
system; a very recent test, utilizing fluoride as an
indicator, shows approximately three days travel time to
the most remote points of our distribution system. There-
fore, the values obtained in our research project would be
greater than expected in the actual system.
The effectiveness of the GAC filters in removing Instantan-
eous THM can be seen in Figure 6. The section of dotted line on
the influent data represents the THM data from a sampling point
in the distribution system, which is water mainly composed of
sand filter effluent since only a small fraction of the filters
were converted to GAC. These somewhat higher THM values were
used because of spurious influent samples caused by improper
quenching of the chlorine.
From the figure, it is apparent that filters 21A and 23A
were very similar in the amount of Instantaneous THM they re-
moved. This result was expected as both filters have the same
bed depth of carbon. Filter 19A was not as effective due to the
fact that it contains less carbon.
f
If the criterion for filter exhaustion is defined as the
point at which the filter effluent concentration passes 50 per-
cent of the Instantaneous THM influent concentration, then filter
19A was exhausted on day 31, filter 21A on day 80, and filter 23A
on day 87. This time span is longer than that expected for true
values because of the sampling problem.
The effectiveness of the GAC filters in removing THMFP can
be seen in Figure 7. The dotted section of the influent data
corresponds to THMFP'* determined from the same sampling point in
the distribution system as mentioned above in regard to Figure 6.
These approximations cause a decrease in apparent THMFP in the
influent, while the effluent values are correct*. The net result
is that the adsorption effectiveness shown is Inferior to actual
performance.
383
-------
to
00
-p;"::TT
LEGEND
o INFLUENT
• ISA, EFFJ : i
'ii-j ' i -
I!'!!]!
:.L.u.LI i..i
I -ji
&n.H;!H
r; : : : i i i !
j ; : . . : .
.1 j i = : ' 1 ! i.
' t
^J
. :r ^v
• • i ; ' 1 •
•J
I:
i i <
I4FEB78
I9APR78
80 t DAYS! 100
120
I8JUN78
140
160
180
I7AUG78
aoo
Removal of Trihalomethanes in GAC Filters
Note:
Figure 6.
The dashed vertical lines labeled "XXX Regenerated" denote the
point at which the removal efficiency has declined to 50%.
-------
I mWHA-OMETHANE FORMATION POTENTIAL)
d ^"iii-JJtiiriKfcgAC!.FILTERS
|VV
:6A.C.F1LTE
^i;-;!^:;:-'-;
(• I9A EFF.
o 2IA EFF.
b23A EFF/
IAFES78
•• iftni at
ISJUN7B
I7AUG78
Figure 7. Removal of Precursors
(Trihalomethane Formation Potential)
in GAG Filters
Note: The dashed vertical lines labeled "XXX Regenerated"
denote the point at which the removal in the respective
column has declined to 50%.
385
-------
The gradual loss in the effectiveness of the filters in
removing THMFP can be seen by the gradual rise of the effluent
concentrations to levels approaching those of the influent.
If the criterion for filter exhaustion is defined as the
point at which the filter exceeds 50 percent of the influent
THMFP concentrations, then filters 19A and 23A were exhausted on
day 10 and filter 21A on day 45.
The effectiveness of the GAC filters in removing the Seven
Day Simulated Distribution THM can be seen in Figure 8. Again,
it is apparent that filter 21A is outperforming both filters 19A
and 23A.
If the criterion for filter exhaustion is defined as the
point at which the filter passes 50 percent of the Simulated
Distribution THM, then filter 19A was exhausted on day 17, filter
21A on day 38 and filter 23A on day 24. However, if exhaustion
is defined as the point at which the Simulated Distribution THM
of the filter passes 0.1 mg/1 (100 ppb), then filter 19A is not
exhausted until day 102, filter 21A on day 116 and filter 23A on
day 110. It should/be noted that the influent exceeded the
Maximum Contaminant Level (MCL) of 0.1 mg/1 (100 ppb) on day 8.
Contract laboratories have been analyzing samples for purge-
able nonhalogenated organics and base neutral extractables. The
purgeable nonhalogenated organics of main concern are benzene,
hexane, toluene, ethylbenzene, o-xylene, and tetralin. Base neu-
tral extractables, specifically noted in the contract, are iden-
tified in Table 3. So far, nothing of significance has been
found in either fraction.
Table 3
Base Neutral Extractables
Isophorone
Naphthalene
Benzidine
Dimethyl Phthalate
Nitrobenzene
Diethyl Phthalate
Di-N-Octyl Phthalate
Bis (2-Ethyl Hexyl Phthalate)
2,4-Dinitrotoluene
2,6-Dinitrotoluene
2,4-Dimethyl Pyridine
1,3-Dichlorobenzene
Butyl Benzyl Phthalate
Diethylphtr.alate
2-Picoline
Aniline
Pyridine
0-Naphthylamine
Di-n-Butyl Phthalate
Butyl Isobutyl Phthalate
Di-Isobutyl Phthalate
386
-------
u>
00
LEGEND'"
-fINFLUEN
;--tl9 A EFff
*T
MAXIMUM CONTAMINANT
140
160
I4FEB78
I9APR78
I8JUN78
180
I7AUG78
200
Figure 8. Removal of 7-Day Simulated Distribution System THM in GAC Filters
The dashed lines labeled "XXX Regenerated" denote the point at which the given
column would have to be regenerated to meet a given standard: 50% removal
(left), or MCL = 100/fg/l (right).
-------
As mentioned earlier, we are currently performing pilot
column studies to determine the effects of carbon regeneration.
Virgin carbon was in service for six months and was then regener-
ated. The regenerated carbon was compared to virgin carbon.
Figure 9 shows that regeneration had no apparent effect on the
carbon's adsorption capability.
We are also currently conducting pilot column studies to
compare the adsorption capabilities between bituminous- (Figure
10) and lignite-based GAG (Figure 11). As of May 1979, the data
are still being evaluated.
Contracts have been awarded and construction is currently
underway that will enable us to begin the third phase of our
research. An initial contract for site preparation, sewer
laterals, and floor slab to accommodate the post filter carbon
contactors has been completed. The site is adjacent to our
filter building. A connecting multi-purpose employee passage-
way-utility tunnel was also constructed as a part of this initial
contract.
A second contract, currently very active, calls for furnish-
ing and installing 4* deep bed carbon contactors (Figure 12) with
allied controls, pumps, utilities And a protective building.
Completion is scheduled for early 1979. The bid price for the
contactors was $989,000.
Upon completion of the contactors, we will install carbon in
at least one additional filter and begin parallel runs of both
the filters and contactors. Please remember that these are full
plant size units, not pilot scale or test tube models. Data so
garnered will be difficult to refute.
Evaluation of the data generated by these parallel runs will
enable us to make an intelligent decision as to the cost effec-
tiveness of both methods with respect to the proposed U.S. EPA
regulations. Naturally, both capital and operating costs will be
considered in this decision.
Carbon Regeneration
The last phase of our project calls for construction of an
on-site regeneration or reactivation furnace of limited capacity
(Figure 13). The contract for this furnace has been awarded to
the Westvaco Company; on a 500 pound per hour (227 kilograms per
hour) fluidized bed model, at a cost of $604,130. Our project is
the only one, to my knowledge, that invited competitive bids for
furnaces with differing operating principles. Both fluidized bed
and infrared models were evaluated with respect to initial
capital costs, operating costs, and life expectancy.
388
-------
OJ
00
VO
24OCT78
23DEC78
21FEB79
22APR79
Figure 9. Total Organic Carbon (TOC) Concentrations in Westvaco WV-G
Filter/Adsorber columns
-------
6,
—]
u>
vc
o
APPLICATION
WESt!FLUME
_j._ _; iLfiKEito'jJ
-! I i •
'nun1.! i .
irt.iri.1 !
Vf' • * V •• UvUCILj
(|7,ft.; depth)
:(hl!ft!.; depth
:Asift.!|!epth)
: l:-i i ' • t r •
10/24
11/13 12/3
(1978)
ill i •! I i :i^^^_j,-
3/13 4/2 4/22
(1979)
5/12
Figure 10. Total Organic Carbon (TOC) Concentrations in "Bituminous"
Contactor Columns
-------
vo
M- •- -i----; •
il CHLORINE APPLICATION
'I MOVED TO WEST FLUME
, .ON DECEMBER 4, ;M76i
PPA-1 (
'lant Efflucht)
I'M I , I ', !:(•'•
[ \ ' ! ; CONTACT I TIME
•••••• PRA-6 (B ft. depth),
—t- PFA-7 (p;ft. depth)1.
PI-A-8 (
11 ft. depthk • H.7
PFA-9 (15 ft.idepthhi . , 16.0G'min.!
i I . . . . : . ,1,1.
3.23 ihin.,
7.43 min.
'•' min.
LOADING ;RATp: j |7 qpm/sa. ft.
10/2)
ll/:.7.
5/12
Figure 11. Total Organic Carbon (TOC) Concentrations in "Lignite1
Contactor Columns
-------
Figure 12. Deep Bed Carbon Contactors
392
-------
Figure 13. Regeneration Furnace - How it Operates
393
-------
Completion of the furnace is scheduled for the fall of 1979.
At that time we will reactivate the spent carbon from the origi-
nal carbon filters. Spent carbon from the second carbon filters
and the carbon contactors will be reactivated as needed.
Reactivation efficiencies and carbon attrition rates will be
evaluated. Figure 14 is a schematic of the regeneration cycle.
Information obtained from this project will be valuable in
determining design criteria for in-plant carbon transportation
systems. This information will also be necessary for the eco-
nomic evaluation of the conversion of existing sand filters
versus construction of new post filter carbon contactors.
CONCLUSION
Upon completion of the project and evaluation of all data,
an engineering review will determine the most cost effective mode
of operation for the entire treatment plant. A recommendation
will then be made to our City Council who will probably make a
final decision. Naturally, there will be citizen input.
In conclusion, I think that Cincinnati has taken a reason-
able approach to develop possible solutions to problems of
concern. It is hoped that the lessons we learn may be applicable
to other water systems, particularly those on the Ohio River.
However, we should all remember that rectifying a problem at the
ultimate end, which is the water treatment plant, is far less
desirable than preventing the problem at the original point of
pollution discharge.
394
-------
Spent Carbon
Storage
Treated
Effluent
vo
Influent
I
Carbon
Metering and
Dewatering
Off Gas
Treatment
Drying
Granular
Carbon
Adsorption
1
Wet
Spent
Carbon
Incineration
Air —
Regenerated
Carbon
Storage
Carbon
Cooling
Regeneration
Regenerated
Carbon
Fuel
t
Steam
Figure 14.
Westvaco Fluid Bed Regenerator in a Typical
Adsorption System
-------
RESULTS FROM OHIO RIVER VALLEY WATER SANITATION
COMMISSION STUDIES
Richard J. Miltner
INTRODUCTION
In November 1976, the Ohio River Valley Water Sanitation
Commission initiated a project, funded by 11 water utilities
in the Ohio River Valley and by a United States Environmental
Protection Agency research grant, to investigate trihalomethane
formation and control at those utilities and to investigate the
levels of other organic compounds in raw and treated waters.
Three of the utilities employed granular activated carbon (GAC).
They were the Huntington Water Corporation, West Virginia, the
Beaver Falls Authority, Pennsylvania, and the Western Pennsyl-
vania Water Company, Pennsylvania. Treatment at these utilities
is described and major results of project studies examining GAC
adsorption/filtration are discussed.
METHODS
Instantaneous and terminal level samples for analyses were
collected and handled as suggested by Stevens and Symons (1).
Terminal level samples were buffered to clear well pH, chlori-
nated sufficiently to satisfy a 7-day demand, and stored 7 days
at clear well temperature. Samples were analyzed by a contract
laboratory for trihalomethanes and carbon tetrachloride by
purge, trap and desorption followed by gas chromatography and
halogen-specific detection (2). Analyses for all other para-
meters were conducted by the utilities in accordance with
Standard Methods (3). Discussion of organic analytical proce-
dures and quality control is detailed elsewhere (2).
GAC OPERATION
GAC use at these three utilities was similar. At each,
GAC was employed in the filtration/adsorption mode in beds
originally designed for rapid sand filtration. The filter/
adsorbers received chlorinated, settled water and preceded
post-chlorination at the clear well.
396
-------
Huntington Water Corporation
Westvaco WVW 14 x 40 GAG was evaluated at Huntington. The
selection of GAG was based on its history of taste and odor con-
trol at the utility. The virgin GAG replaced taste-and-odor-
exhausted GAG. No previous pilot scale studies had been con-
ducted to determine optimum selection of GAG or bed depth for
organics control. The bed was placed with 76 cm (30 inches) of
GAG on top of 30 cm (12 inches) of sand and gravel. Hydraulic
data demonstrated a mean loading rate of 6.1 m/hr (2.6 gpm/ft )
and a mean empty bed contact time (EBCT) of 7.2 minutes. Treat-
ment is illustrated in Figure 1. Water quality data for the
utility during the time of the study are given in Table 1. The
virgin GAG bed represented only 8 percent of the utility's total
flow. Periodically, influent and effluent waters for older WVW
14 x 40 GAG beds were sampled in order to evaluate performance
after long periods of time in operation.
Beaver Falls Authority
Three virgin GAG beds were evaluated. The utility had
previously conducted pilot column studies with several GACs for
taste and odor control. Those studies did not determine optimum
selection of GAG or bed depth for organics control. The three
beds were geometrically identical. One bed was placed with 61
cm (24 inches) of Calgon Filtrasorb 400 on top of 30 cm of sand
and gravel. A second bed was placed with 61 cm of Calgon
Filtrasorb C on top of 30 cm of sand and gravel. A third bed
was placed with 61 cm of ICI Hydrodarco 8 x 16 on top of 30 cm
of sand. Treatment is illustrated in Figure 2. All three beds
were placed in service simultaneously. Hydraulic data collected
during the study indicated mean loading rates of 3.1 - 3.5 m/hr
(1.3 - 1.5 gpm/ft ) and mean empty bed contact times of 10 to
11 minutes. Water quality data for the utility are presented in
Tables 2, 3, and 4.
Western Pennsylvania Water Company
Treatment at this utility included filtration/adsorption
with 2-1/2 year old Calgon Filtrasorb 400 GAC. Placement and
operation of these beds were discussed by Ross (4). Treatment
is illustrated in Figure 3.
FINDINGS FROM GAC STUDIES
Adsorption of Trihalomethanes
The GACs studied were exhausted for the removal of total
trihalomethane (TTHM) and the trihalomethane precursors [as
measured by trihalomethane formation potential, or THMFP (1)]
397
-------
Ul
vo
00
IU
I
o
T
33%
T«O HR5
T= I MR
MIX
% LIME CHLORINE
Fe504
(PAC) (PAvC)
(POLYMER) (KMr>04)
(CuS04) (POLYMER)
LEGEND
SAMPLE POIWT
(OPTIONAL FEED)
HYDRO
TREATER
OLD
WVW 14*40
GAC
FILTERS
SETTLING
(A
VIRGIN
WVW I4«4O
GAC
FILTER
OLD
WVW 14x4-0
GAC
PI LT E RS
I CLEAR
WELL
=O=
T- 5-25 MRS
CHLORIME
8%
7-3.5 MRS
Figure 1. Treatment o£ Huntington Water Corporation
64,000 m3/Day (17 MGD)
July 1977 - March 1978
-------
10
VD
T«OHR
COAG
(KMp.O4)
(CHLORIKIE)
LEGEMD
Q * SAMPLE POINT
(OPTIONAL FEED)
SETTLE t=O=!l
"s 9 HR5
LIME (RAG)
CHLORIME
T«4HRS
SAND
FILTERS
7O%
FILT
4OO
6 AC
FILT C
C?AC
FILTER
HD8*lb
GAC
FILTER
CHLORINE
CIO-
T« 12 MRS
Figure 2. water Treatment Scheme
Beaver Falls Authority
Eastvale Plant
17,000 m3/Day (4.5 MGD)
-------
o
o
(CHLORINE DIOXIDE)
12 MOD
ir°f
< 22 MGO
2\(POLYMER)
O
2
r \
(CHLORINE) LIME
u==t
ALUM
|5E7TL£K>
LEGEND
uuu
GAC
FILTER
°r
CHLOF
FLUOG
L.Ut/SK.
WELL
MNE
!lDE
SAMPLE POINT
(OPTIONAL FEED)
II
Figure 3. Treatment at Western Pennsylvania Water Company
129,000 m3/Day (34 MGD)
-------
Table 1
Water Quality Data (Mean Values)
Huntington Water Corporation
July 1977 - March 1978
Week of
Virgin CAC
Operation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
20
22
23
25
27
31
Raw
Mean
Temp, °C
27
28
28
28
28
27
27
27
27
27
26
24
19
19
14
15
15
11
8
5
3
2
2
Water
pH « 7,
Turbb
14
21
26
13
15
80
37
34
17
18
25
24
47
98
34
22
18
42
240
160
24
30
34
,5
TCC
1,600
1,200
910
870
1,500
3,000
5,300
2,300
1,400
970
1,100
1,700
3,100
4,300
3.900
2,600
2,800
3,900
1,400
26,000
2,800
5,900
610
C,AC Influent (Settled)
Mean j>H •= 8.9
Turbb
2.0
4.6
4.9
4.4
6.5
5.8
3.8
5.9
3.3
4.6
7.9
4.4
8.7
4.3
16
9.1
10
5.5
9.8
8.0
7.0
9.0
14
Chlorine3
Free
0.8
0.3
1.8
0.5
0.6
0.5
0.4
0.3
0.5
0.5
0.4
0.5
0.5
0.5
0.5
0.5
0.4
0.6
0.2
0.5
ND
0.9
0.3
Total TC
1.4 <1
0.4 <1
3.7 <1
0.7 <1
0.9 <1
0.8 <1
0.7 <1
0.6 <1
0.7 <1
0.7 <1
0.6 <1
0.7 <1
0.8 <1
0.9 <1
0.7 <1
0.9 <1
0.9 <1
0.8 <1
0.3 <1
0.6 <1
0.7 <1
1.1 <1
0.9 <1
SPCd
4
52
42
7
18
28
17
22
24
26
28
28
31
—
34
39
18
200
55
36
30
—
—
Virgin
Mean
Turbb
0.2
1.5
1.4
1.7
.1.8
1.1
0.5
1.5
0.3
3.2
1.7
0.4
0.4
0.4
0.4
1.0
0.5
0.7
12
0.8
0.5
0.2
0.3
GAC Effluent
pH = 8.7
Chlorines
Free
ND
ND
0.1
TR
0.4
0.1
TR
0.2
TR
TR
TR
TR
TR
0.1
TR
TR
0.3
0.2
0.2
ND
0.3
0.4
Total TC
ND <1
TR <1
0.3 6
0.3 8
0.6 5
0.2 <1
0.1 <1
0.4 2
0.2 2
0.2 <1
0.2 <1
TR 1
0.2 <1
0.4 <1
<1
0.5 <1
0.6 <1
0.4 <1
0.4 <1
0.3 <1
0.5 <1
0.5 <1
0.8 <1
SPCd
100
53
12
41
18
13
3
25
46
140
23
12
30
~
2
10
2
4
11
3
<1
—
— -
.Chlorine, mg/1
Turbidity, NTU
jTotal coliforra/100 ml
Standard plate count/ml
-------
O
K)
Table 2
Influent Water Quality Data (Mean Values)
Beaver Falls Authority
September 1977 - April 1973
Kaw
Mean pH - 7.2
Week of
Operation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
17
18
21
23
25
27
29
32
Temp, °C
21
21
15
11
16
16
16
10
10
8
6
3
4
2
1
1
1
1
4
4
7
10
11
Turbidityb
44
28
22
9.5
7.5
9
10
9
16
10
14
10
22
10
10
12
8
14
10
150
12
8
6
TCC x 103
98
71
140
150
39
190
80
98
220
120
120
69
89
75
65
48
27
6
23
84
13
24
8.4
GAC Influent (Settled)
Mean pH - 7.4
Chlorine3
Free
2.0
1.7
1.3
1.1
1.2
1.4
1.1
1.0
1.3
1.0
1.4
1.0
1.2
1.3
1.0
1.4
1.0
0.4
0.3
TR
—
0.2
0.2
Total
•••
1.7
1.4
1.3
1.4
1.6
1.2
1.0
1.6
1.3
1.6
1.1
1.7
1.5
1.2
1.7
1.1
1.6
1.6
1.4
1.4
1.1
1.6
Turbidityb
5.6
4.8
2.3
2.9
2.5
3.3
3.6
3.2
4.6
4.5
3.7
5.9
4.6
6.6
4.8
5.9
5.5
6.4
5.8
6.6
6.3
1.7
1.9
TCC SPCd
<1
<1
<1
<1 100
<1 800
<1 350
<1 10
<1 42
2 110
<1 33
<1 95
1 360
<1 660
1 200
<1 120
<1 150
<1 33
<1 30
<1 24
<1 38
<1 58
<1 33
<1 17
*Chlorine, mg/1
Turbidity, NTU
^Total coliform/100 ml
Standard plate count/ml
TR = trace
-------
Table 3
Effluent Water Quality Data (Mean Values)
Beaver Falls Authority
September 1977 - April 1978
o
u>
Week of
Operation
Raw
Water
Teap.'C
Filtrasorb
Mean pH -
Chlorine*
Free
Total
Turbb
400
7.3
TCC SPC"1
GAC Effluent
Filtrasorb C
Mean pK - 7.3
Chlorine8
Free Total Turbb TCC
HD 8x16
Mean pH - 7
SPC"1
Chlorines
Free Total
Turbb
.3
TCC
SPC*1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
17
18
21
23
25
27
29
32
21
21
15
11
16
16
16
10
10
a
6
3
4
2
1
1
1
1
4
4
7
10
11
ND
ND
ND
ND
ND
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
__
—
--
ND
ND
ND
ND
ND
<0.1
^0.1
<0.1
TR
<0.1
<0.1
<0.1
<0.1
TR
<0.1
<0. 1
^0.1
0.3
0.4
0.2
0.1
0.1
0.3
0.4
0.4
0.3
0.3
0.4
0.6
0.3
0.4
0.5
0.3
0.6
0.6
0.4
0.1
0.4
0.4
0.5
0.6
0.6
0.6
0.3
0.3
0.3
64 —
75 ~
98 1,000
45 1.400
34 25.000
42 2.000
28 20,000
22 29,000
13 6,500
12 1,600
2 960
1 270
<1 480
1 440
<1 44
<1 50
<1 21
<1 30
<1 13
<1 21
<1 41
<1 31
<1 69
ND
ND
ND
ND
ND
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
__
—
™
ND
ND
ND
ND
ND
-------
Table 4
Water Quality Data (Mean Values)
Beaver Falls Authority
September 1978 - December 1978
*••
o
*»
GAG Influent
Week of
Operation
53
54
55
56
57
58
59
60
61
62
63
64
Raw
Temp, °C
26
23
22
19
14
12
14
13
11
9
8
6
(Settled) Filtrasorb
TCa
18,000
10,000
22,000
9,200
31,000
10,000
8,700
19,000
5,000
12,000
82,000
8,000
Free
Chlorine
1.4
1.2
1.6
1.6
1.4
1.1
1.4
1.3
1.5
0.8
1.2
1.0
Free ,
TCa Chlorine
<1 TR
<1 TR
<1 TR
<1 TR
<1 TR
<1 TR
<1 TR
<1 TR
<1 TR
<1 TR
<1 TR
<1 TR
400
TC3
100
120
230
470
62
44
30
8
—
1
<1
<1
GAG Effluent
Filtrasorb C
Free
Chlorine
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TR
TCa
64
25
21
5
9
10
3
<1
2
<1
<1
-------
after 7 to 15 weeks of operation. Figure 4 illustrates trihalo-
methane control at Huntington. WVW 14 x 40 was exhausted for
the removal of TTHM after 7 to 8 weeks of operation. Prior to
that time, influent concentrations exceeded effluent concen-
trations. Following that time, effluent concentrations exceeded
influent concentrations, or effluent and influent concentrations
could not be differentiated as determined by the precision of
project trihalomethane data (2). WVW 14 x 40 was exhausted for
the removal of THMFP after 7 to 10 weeks of operation as illu-
strated by Figure 4.
Figure 5 illustrates trihalomethane control by one of the
GACs at Beaver Falls. Filtrasorb 400 was exhausted for the
removal of TTHM after 9 to 10 weeks and for the removal of THMFP
after approximately 11 weeks of operation.
GACs studied were exhausted for the removal of chloroform
after 7 to 15 weeks of operation. Chloroform control by Fil-
trasorb C at Beaver Falls is illustrated in Figure 6. After 12
to 15 weeks of operation, effluent and influent chloroform
concentrations could not be differentiated as determined by the
precision of project chloroform data (2), indicating that ex-
haustion had occurred. WVW 14 x 40 was exhausted for the re-
moval of chloroform after 7 to 8 weeks of operation as illu-
strated by Figure 7.
GACs studied were exhausted for the removal of bromodich-
loromethane and dibromochloromethane after 8 to 15 weeks of
operation. Figures 6 and 7 illustrate exhaustion for the re-
moval of dibromochloromethane after 13 to 15 weeks of operation
by Filtrasorb C at Beaver Falls and exhaustion for the removal
of bromodichloromethane after 11 to 14 weeks of operation by WVW
14 x 40 at Huntington, respectively.
Evaluation of the GACs for the control of bromoform and
dichloroiodomethane was complicated by the fact that at the
concentrations typically present (i.e., below 0.5 yg/liter),
the precision of the data may be ^100 percent (2). This was
generally found to be the case for other volatile halocarbons.
GACs at Beaver Falls could not be evaluated because, when
/dia,tec.ted, concentrations of these trihalome thanes were below
0.5 jag/liter. However, bromoform and dichloroiodomethane
Concentrations at Huntington are given in Table 5. These data
indicate that adsorption of bromoform occurred during week 12,
but that influent and effluent bromoform concentrations could
not be differentiated beyond this week.
Thus, the virgin GACs at these two utilities were an
effective trihalomethane control for 2 to 3 months when
operating in the filtration/adsorption mode, in beds designed
405
-------
30O-
2OO-
IOO-
INFLUENT
THMFP
EFFLUENT THMFP
10
15
2O
•25
30 35
40
HUNTINGTON WATER CORP.
GAG * WVW I4K4O
DEPTH » 7fc CM (3O INCHES) GAC
LOADING RATE • fe.l M/HR(2.6»
C&CT * l.t Ml MUTES
\NJFLUeWT TTHM
EFFUUEWT TTHM
10 IS 20 25 30 35 4O 45
TIME IN OPERATIOM, WEEKS
28 28 21 14. II 3 2 4
TEMPERATURE, °C
13
14
Figure 4. Trihalomethane Removal by Granular Activated Carbon
406
-------
3OO-
20O-
-------
40 .
20
_J
\
-------
(50
IOO
50
J
IT)
Z
o
h
INFLUENT
CHC13
40 -
h
z
o
o
5 10 15 20 25 3O 35
INFLUENT
EFFLUENT
CHBrClz
IO
15
20
—i—
25
—i—
30
—i—
35
—i—
4O
^EFFLUENT CHCI3
HUNT1NGTON WATER CORP.
GAC * WV W I4-X4-O
DEPTH - 16 CM (3O INCHES) GAG
LOAOIMG RATE - 1.1 M/HR(2-fe
E8CT s -?.l MINUTES
45
45
30-
10 -
INFLUENT
CHBr2Cl
10 15 20 25 30
TiME IN OPERAvTlOW.
27
14
II
TEMPERATURE. °C
13
14-
Figure 7. Trihalomethane Removal by Granular Activated Carbon
409
-------
Table 5
Removal of Trihalomethanes By
Granular Activated Carbon
Huntington Water Corporation
July 1977 - May 1978
Concentration,0 yg/1
Week of
Operation
1
2
3
4
5
6
7
8
9
10
11
12
15
16
17
19
21
22
35
39
42
45
Bromoforra
Influent
1.6
4.4
1.2
0.3
0.2
0.2
0.1
0.5
0.6
1.5
1.6
1.9
<0.1
0.1
0.1
0.2
ND
ND
ND
ND
ND
0.5
Effluent
ND
<0.1
<0.1
ND
<0.1
ND
ND
<0.1
0.1
0.2
0.2
0.2
0.1
<0.1
0.1
<0.1
ND
<0.1
ND
<0.1
ND
0.2
Dichloroiodome thane
Influent
<0.1
0.1
0.3
0.2
0.2
0.4
0.7
0.6
0.4
0.2
0.2
0.1
<0.1
<0.1
0.1
<0.1
ND
ND
ND
0.1
<0.1
0.2
Effluent
ND
ND
ND
<0.1
<0.1
<0.1
<0.1
<0.1
ND
<0.1
<0.1
0.1
<0.1
<0.1
<0.1
ND
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
aGAC - WVW 14x40
Bed depth - 76 cm (30 inches) GAG
Loading rate - 6.1 m/hr (2.6 Rpm/ft2)
EBCT - 7.2 minutes
°GC/Hall detector, approximate lower
detection level -0.1 ug/1
ND - not detected
410
-------
for sand filtration, receiving chlorinated water, anC. when
placed in operation during summer months when temperatures were
high. Figures 4 and 7 illustrate higher influent trihalomethane
concentrations when temperatures were high and lower influent
trihalomethane concentrations when temperatures were low,
suggesting that time to exhaustion might have exceeded 2 to 3
months had the beds been placed in operation during colder
months.
Adsorption of Carbon Tetrachloride
Organic compounds other than trihalomethanes were not
frequently found at detection levels of 0.1 yg/liter, and when
detected, were generally at concentrations below 0.5 yg/liter
where influent and effluent concentrations often could not
be differentiated because the precision of the data below 0.5
yg/liter may be +1QQ percent (2). Thus, it was difficult to
evaluate the adsorption capacity of the GACs for other com-
pounds .
Carbon tetrachloride was frequently found at Huntington,
however. These data are presented in Table 6. Breakthrough was
not observed until the 9th week of operation. GAC was an
effective barrier for relatively high carbon tetrachloride
loading, as demonstrated during the 10th week of operation when
13 yg/liter influent and 0.4 yg/liter effluent concentrations
were observed. After 11 weeks of operation, however, the
compound was always detected in GAC effluent waters. Although
influent and effluent concentrations were low after 11 weeks of
operation and oftentimes could not be differentiated, these data
suggest that GAC was not an effective barrier for routine, low
level carbon tetrachloride loading after 3 months of operation.
Desorption from GAC
Desorption from GAC was observed whenever GAC influent
loading was significantly lowered. This was observed at each
of the three utilities.
At the Western Pennsylvania Water Company, chlorine dioxide
was substituted for chlorine as the raw water disinfectant, see
Figure 8. When chlorinating raw water, no difference was
observed in TTHM concentration influent to (35 yg/liter) and
effluent from (35 yg/liter) the 2-1/2 year old Filtrasorb 400.
With chlorine dioxide applied and little or no free chlorine
present, the GAC influent concentration was significantly
lowered (to 2.4 yg/liter) and the difference between TTHM
influent (2.4 yg/liter) and effluent concentration (12 yg/liter)
was attributed to desorption of trihalomethane from GAC.
411
-------
Ul
*»
ro
O
u
22
\\X\\\\\\V
\\
35
35
42
ZO
RAW
RA.W
Figure 8. Trihalomethane Formation (Mean Values)
Western Pennsylvania Water Company
-------
Table 6
Removal of Carbon Tetrachloride By
Virgin Granular Activated Carbona
Huntington Water Corporation
July 1977 - May 1978
Week of
Operation
1
2
3
4
5
6
7
8
9
10
11
Concentration,13 U8/1
Influent
<0.1C
ND
NFB
0.4C
0.6C
O.lc
0.1
O.lc
0.3
13+
0.4
Effluent
ND
ND
NFB
ND
ND
NFB
NFB
NFB
<0.1
0.4+
0.1
Week of
Operation
12
14
15
16
18
22
35
39
42
46
Concentration, fe US/1
Influent
0.5
0.2
0.2
0.3
0.2
<0.1
<0.1
<0.1
0.1
0.3
Effluent
0.1
<0.1
0.1
0.3
<0.1
<0.1
0.1
0.1
0.2
0.2
Table 7
Removal of Carbon Tetrachloride By
Older Granular Activated Carbona
Huntington Water Corporation
GAC placed October
Month of
Operation
9
94
11
13
14
1976
Concentration,* Ug/1
Influent
ND
ND
0.5
0.2
0.1
Effluent
2.2C
2.3°
1.0
0.4
0.3
GAC
Month of
Operation
27
274
28
29
31
32
placed April
Concent rat
Influent
ND
ND
0.1
0.2
0.2
0.1
1975
ion,b yg/1
Effluent
1.5C .
0.7C'+
0.6
0.7
0.3
0.2
GAC - WVW 14x40
Bed depth « 76 cm (30 inches) GAC 2
Loading rate - approximately 6.1 m/hr (2.6 gpm/ft )
EBCT - approximately 7.2 minutes
GC/Hall detector, approximate lower detection level
°Co-elution with 1,1,1-trichloroethane
•*• " GC/MS confirmed as carbon tetrachloride.
ND - Not detected
NFB = Not found after blank correction.
0.1 yg/1
413
-------
Beaver Falls halted the practice of breakpoint chlorination
after 21 weeks of operation, see Figures 5 and 6. With little
or no free chlorine present, GAC influent trihalomethane con-
centrations were significantly lowered and desorption of tri-
halomethanes resulted. For example, in Figure 5 at week 21,
influent and effluent TTHM concentrations of 37 yg/liter and 41
yg/liter, respectively, could not be differentiated. After
breakpoint chlorination was halted, however, the effluent TTHM
concentration of 15 yg/liter was differentiable from the influ-
ent TTHM concentration of 3 yg/liter and was attributed to
desorption of trihalomethane from GAC.
Desorption of carbon tetrachloride was observed from older
GAC beds at Jkmtington. These older beds were in operation in
February 1977 when GAC influent concentrations of carbon tetra-
chloride exceeded 100 yg/liter during a spill condition. When
sampled from July 1977 to December 1977, carbon tetrachloride
influent concentrations were below 0.5 yg/liter (i.e., signifi-
cantly lower than 100 yg/liter), but effluent concentrations
reach 2.3 yg/ liter indicating desorption of carbon tetrachlo-
ride occurring several months after the spill, see Table 7.
GAC Bacterial Densities
GAC effluent bacterial densities oftentimes exceeded GAC
influent bacterial densities at each of the three utilities
during warmer months.
Table 1 presents bacterial densities in the influent to and
effluent from the virgin GAC bed at Huntington. Throughout the
study, influent total coliform (TC) counts were less than one
per 100 milliliter. At lower temperatures effluent TC counts
were less than one per 100 milliliter, but at higher tempera-
tures effluent TC counts reached eight per 100 milliliter.
Tables 2 and 3 present bacterial counts in the influent to
and effluent from the three virgin GAC beds at Beaver Falls.
These data demonstrate significantly higher GAC effluent TC and
standard plate counts (SPC) than the corresponding influent
values, when temperatures exceed 10°C. Note that effluent TC
counts reached 130 per 100 milliliter and SPC values reached
92,000 per milliliter. Influent and effluent values generally
could not be differentiated when temperatures dropped below
10°C. This phenomenon was observed again one year later and
is seen in data presented in Table 4. Again, at higher temper-
atures effluent TC counts were significantly higher than influ-
ent values, but as the temperature dropped, influent and
effluent TC counts could not be differentiated.
414
-------
Similar phenomena were observed in a 2-1/2 year old GAC
bed at the Western Pennsylvania Water Company when disinfectant
studies were conducted during a period when temperatures exceed-
ed 20°C. However, at each of these three utilities, post-
chlorination at the clear well assured the bacterial integrity
of the finished water.
These data indicate that pre-chlorination does not stop
bacterial growth on GAC during warmer months of operation and
that utilities employing GAC as filter/adsorbers must take care
to monitor the bacterial quality of their water.
REFERENCES
1. Stevens, A.A. and J.M. Symons. 1977. Trihalomethane and
Precursor Concentrations Occurring During Water Treatment
and Distribution. Jour. Amer. Water Works Assoc. 69(10):
546.
2. Water Treatment Process Modifications for Trihalomethane
Control and Organic Substances in the Ohio River, USEPA
Grant No. R-804615, in preparation.
3. APHS, AWWA, WPCF. Standard Methods for the Examination of
Water and Wastewater. 14th ed. (1976).
4. Ross, R.M. 1976. Conversion of Rapid Sand Filters to
Granular Carbon Filters, Jour. Amer. Water Works Assoc.
68(12): 663.
415
-------
DESIGN AND OPERATIONAL EXPERIENCES
WITH ACTIVATED CARBON ADSORBERS:
TREATMENT OF DELAWARE RIVER WATER, U.S.A.
Patrick R. Cairo
and
Irwin H. Suffet
Removal of trace organics at a drinking water treatment
plant is one of the most difficult problems facing U.S. water
utilities today. The Environmental Protecton Agency is consider-
ing regulatory alternatives either to specify treatment technol-
ogies to reduce organic levels or to impose specific limitations
(MCLs) on certain organic compounds that are shown to be signifi-
cant human health risks (1). Regardless of the initial approach
taken, it is expected that drinking water standards will be
imposed for specific organics on the basis of health effects
studies currently underway. It is also very likely that both the
drinking water limits and the types of compounds regulated will
be more stringently controlled while new treatment processes are
being constructed and placed in operation. The expected evolu-
tionary nature of the regulations presents a unique challenge
both to scientists and engineers performing the studies that will
determine design and operating characteristics for organic
removal processes. Research protocols in analytical chemistry,
aquatic microbiology, and process engineering must be employed to
ensure proper consideration of all factors.
This paper discusses design and operational experiences with
pilot plant testing of carbon adsorption systems used in the
treatment of Delaware River water at Philadelphia. This experi-
ence is presented in the context of the chemical and biological
considerations that must be merged in an engineering evaluation
to design and operate a GAC facility. Because specific process
design objectives are expected to evolve during the operating
lifetime of the facility, a framework for in-process evaluation
of the chemical and biological data also is developed.
The significant elements that must be integrated in the
design and subsequent operation of the granular activated carbon
facility are shown in Figure 1. These begin with biological and
chemical considerations measured by specific and nonspecific
analytical tests as well as screening protocols utilizing iso-
lation methods such as macroreticular resin (MRR), continuous
liquid-liquid extraction (CLLE), purge-and-trap (P&T), and
416
-------
BIOLOOICAL CONSIDERATIONS
SPECIFIC ANALYSIS
NON SPECIFIC ANALYSIS
CHEMICAL CONSIDERATIONS
SPECIFIC ANALYSIS
SCREENING PROTOCOL SCREENING PROTOCOL
........
1' t I'.J*
NON SPECIFIC ANALYSIS
1 It M
DISTRIBUTION PROBLEMS
TIME SERIES PLOTS AND BREAKTHROUGH CURVES
I 1
HEALTH EFFECTS PRIORITIZATION (CROUP I-V)
I - ' *
J (^"^A- TRANSIENT
| ' I 1
CC COMPUTER
PLOT PROGRAM
I
MASS LOADING M
STATISTICAL ANALYSIS
ENGINEERING EVALUATION
I
ECONOMIC ANALYSIS (CAPITAL AND OPERATING COSTS)
Figure 1. Design Elements for Adsorption Process
-------
bacterial analysis by coliform and standard plate count (SPC)
tests. The data generated must next be evaluated in the context
of potential health effect prioritization. This would include
both direct impacts to human health and indirect impacts affect-
ing the operating performance of the GAC process or esthetic
problems arising in the treated water during its transmission
in the distribution system.
Through chemical and microbiological testing the engineering
aspects of alternative GAC configurations are evaluated. This
work can be performed using bench, pilot, and full scale systems.
Each scale presents certain advantages and disadvantages that
must be evaluated carefully. Bench scale experiments permit
maximum flexibility and can be performed quickly and economi-
cally. However, the direct application of these results to full
scale installations is not recommended since the experiments
usually do not represent yg/1 concentration levels or the types
of hydraulic and water quality problems encountered when large
volumes of river water are being treated. Pilot scale testing
seems to be the optimal level since it still retains considerable
flexibility and allows for continuous treatment of significant
quantities of river water. It is recommended that after the
screening of several alternative GAC configurations by pilot
testing, a 3.8 to 38 ML/day (1 to 10 mgd) system of the selected
design be tested to determine if scaleup effects must be incor-
porated in the design of the full scale facility. The final
element that needs to be evaluated is the economic implication of
the carbon system. Here, both initial capital investment and
long-term operating and maintenance expenses must be weighed (2).
The engineering evaluation using chemical and biological
considerations requires detailed discussion. Its implications
are demonstrated through the research findings that evolved
during the testing of Delaware River water.
CASE STUDY: DELAWARE RIVER
The Delaware River which flows past Philadelphia shares
the common plight of most developed surface water systems in the
U.S. In a 61-km (38-mi) stretch between the start of the
Delaware's estuarial portion at Trenton and the point of conflu-
ence with the Schuylkill River, more than 11 municipalities and
360 industries discharge treated wastewater. A massive effort is
currently underway to upgrade the level of wastewater treatment,
and the best available technology economically achievable for
treatment of toxic wastes will be imposed on 21 industrial cate-
gories by 1983 (3). However, trace levels of organic compounds
will continue to be present in the Delaware River as a result of
their discharge from advanced treatment plants as well as from
accidental spills and nonpoint sources.
418
-------
For more than 7 years the Philadelphia Water Department,
with the assistance of Drexel University, has.been engaged in a
major organics research program to assess the nature of the
organics problem and to evaluate treatment processes that couia
.be installed at each of its three water treatment plants. Most
of the work to date has been performed at the Torresdale Plant
[1,070 ML/day (282 mgd)], since it is the city's largest and
draws its water from the Delaware estuary. The initial years
were spent developing analytical techniques that could be used
to concentrate, extract, and identify specific organic compounds
and led to the creation of a special laboratory with personnel
and equipment devoted to this task. Treatment research first-
was performed on laboratory scale studies and in 1976, was ex-
panded to a pilot scale level with the construction of 7.5 to
30-cm (3- to 12- in.) glass columns and a 114-m3/day (30,000-gpd)
advanced water treatment pilot plant made of stainless steel,
glass, and PTFE parts.^ The activities of the past 7 years are
summarized in figure 2. The work has entailed evaluation of
oxidation, coagulation, and adsorption by granular activated
carbon and synthetic resins. In 1978, a 3-year laboratory and
pilot scale study of ozonation and carbon adsorption in the
so-called biologically activated carbon (BAG) mode was begun
under a research grant from EPA (4).
HEALTH EFFECT PRIORITIZATION
The first step in designing the carbon adsorption process
is to develop a prioritization scheme to guide the evaluation
of data. The chemical and biological analytical procedures
outlined in Figure 1 and detailed below must be evaluated by
a priority rank based on health effects and GAC operating
characteristics.
Chemical Considerations
The chemical compounds entering an adsorption water treat-
ment process consist of high molecular weight and lower molecular
weight organic compounds covering a wide range in polarity. The
National Academy of Science estimated that 90 percent of tire
total organic matter is of high molecular weight (5). This would
correspond to the natural organic compounds of little direct
toxicological importance [class 5 as defined by the EPA (6)], but
these compounds may affect the process by competitive adsorption
or support of microbial growth. A fraction of these high molecu-
lar weight organic compounds are precursors that react with
disinfectants to produce disinfectant byproducts [class 3 (6)].
The remaining 10 percent of the total organic matter includes,
but is not limited to, synthetic organic chemicals that are
present in source water and organic chemicals that are disinfec-
tion byproducts [classes 2 and 4 (6)]. A subgroup of chemicals
in the last 10 percent group are the presently known chemicals of
419
-------
N)
O
I. OEVtLOP«MT OF
ANALYTICAL »CTHOOOLOCIE$
». OUTSIDE COnnACTS
- DREXEL UNIVERSITY
B. IN-MOUSE RESEARCH
II. TREAUfNT STUDIES
A. OUTSIDE CONTRACTS
- DREXEL UNIVERSITY
B. IN-HOUSE RESEARCH:
1. PILOT I
2. PILOT II
3. PILOT III
4. FREDESIGN STUDIES
(CAC FILTERS/FILTER-
ADSORBERS)
C. PREOZCNATICN/CAC
STUDY (PILOT PLANT)
0. OTHER EVENTS
YEAR 1
YEAR 2
-H
COOffRATIVE STUOIES
-f-
YEAR 3
.LEGIONNAIRE'S .
PREDESIGN
ORGANICS. LAB CAPILLARY
GC/
OPENED
CC
DELIVERED
VEAB 1
. YEAR 2 . PILOT DM*
EVALUATION
PLAN
PLAN
NEW BCEE
ORLEANS PROBLEM
CONSTRUCTION.
PREPROPOSAL PROPOSAL
LEGIONNAIRE'S
OESEASE
C C14
+**
1972
1973
1974
1975
1976
1977
1978
1979
1980
Figure 2. Philadelphia Case History:Time
Figure From M. McGuire (15)
Schedule
-------
health concern, listed by certain organizations as suspected
animal or human health problems (5, 7-9) and by the EPA as
priority pollutants (3).
Table 1 shows the chemical groups being considered for
design of GAC systems in priority order. The primary aim for
design considerations would be to assure removal of chemicals
of known health concern - group I. Maximum contaminant levels
(MCLs) have already been proposed or established for certain
group I compounds, such as chloroform and pesticides (1, 10).
A well-designed adsorption process must be flexible enough to
accommodate the removal of other group I compounds to protect
the public health. It would be expected that when other MCLs
are developed, group I compounds would be primary targets. In
consideration of design, group IA, II, and III compounds should
be evaluated to see if, indeed, they would be removed by carbon.
Trace organic screening protocols are available for this task.
Group IV compounds fall into a different category of concern
as disinfectant type, dose, and concentration are primarily
utilized for their control. The trihalomethane formation poten-
tial (THMFP) and total organic halide tests can monitor this
category.
Group V compounds are of concern because of their effect
on Group I-III compounds. This group can be monitored by trace
organic screening protocols. The effects of group V compounds
would include those of competition and reequilibration. Compet-
itive effects occur when a strongly adsorbed compound enters the
carbon bed and displaces a weakly adsorbed compound. In this
case the concentration of the displaced compound may be higher
in the effluent than in the influent until a new equilibrium is
reached. A reequilibrium effect occurs when the concentration
of a compound in the influent decreases. Desorption of this
compound from the carbon then occurs and, again, the concentra-
tion in the effluent could be higher than in the influent until
a new equilibrium is established. Finally, a compound might
appear in the effluent that was never in the influent, due to
chemical or microbial reactions on the column involving compounds
previously adsorbed from the influent. These new compounds would
also be influenced by the breakthrough, exhaustion, and competi-
tion processes.
Table 2 lists the prioritized compounds identified in
Torresdale (Philadelphia, Pa.)/drinking water during eight
sampling surveys conducted between Feb. 6, 1975, and Jan. 3,
1977 (11). One hundred sixteen different compounds were found
including 40 isomers (e.g., xylene isomers). Compounds con-
sistently identified in concentrations greater than 1 yg/1 are
chloroform and dichlorobromomethane. Compounds periodically
identified at this level include dibromochloromethane,
1,2-dichloroethane, 1,2-dichloropropane, trichloroethylene,
toluene, and diacetone-L-sorbose. Many of the 116 compounds
421
-------
TABLE 1
Health Effect Prioritization of Chemical Parameters
Group I - CARCINOGENS AND SUSPECTED CARCINOGENS listed by
(1) National Academy of Science (1977)
(2) National Cancer Institute (1978)
Group IA - MUTAGENS, TERATOGENS AND PROMOTORS listed by
National Cancer Institute (1978)
Group II - PRIORITY POLLUTANTS listed in NRDC v. Train (1976)
Group III - OTHER COMPOUNDS OF POTENTIAL HEALTH EFFECT listed by
National Academy of Science (1979)
Group IV - PRECURSORS - Organic Compounds that react with a Disinfectant
to produce a by-product which falls into Groups I - III.
Group V - HUMIC SUBSTANCES AND NON-HEALTH EFFECT COMPOUNDS IN HIGH
CONCENTRATION which can compete with compounds of Groups I - III
and cause earlier breakthrough of'these compounds.
422
-------
TABLE 2
Categories of Organic Compounds Identified at Torresdale
HALOGENATED
a.) ALIPHATIC
b.) AROMATIC
AROMATIC
ALIPHATIC
PHTHALATE ESTERS
MISCELLANEOUS
TOTAL
TOTAL
NUMBER
24
10
50
7
7
18
116
I
CARCINOGENS
&
SUSP. CARC.
CHLOROFORM
CARBON TETRACHLORIDE
BIS (2-CHLOROETHYL) ETHER
TRICHLOROETHYLENE
0
BENZENE
0
0
0
5
la
MUTAGENS
TERATOGENS &
PROMOTERS
BROMOCHLOROMETHANE
BROMOFORM
DIBROMOCHLOROMETHANE
DICHLOROACETONITRILE
DICHLOROETHANE
DICHLOROBROMOMETHANE
1,2 DICHLOROPROPANE
1,1,1 TRICHLOROETHANE
0
0
DECANE
0
0
9
II
PRIORITY
POLLUTANTS
4
6
7
0
7
ISOPHORONE
25
423
-------
were found to be highly variable during weekly composite analy-
ses (11, 12). Prioritizing these compounds according to the
categories listed in Table 1 yields 39 compounds in groups I
and II. Of these, 32 were usually in the 1-yg/l or lower range.
Biological Considerations
The granular activated carbon used in treatment of drinking
water has been found to be an ideal habitat for bacterial growth
because of the large surface area and pore structure of GAC and
its affinity for organic compounds. Total coliforms were not
found to proliferate in GAC adsorbers and were not recovered in
their effluents. However, total bacterial levels in GAC contac-
tors and their effluents are significantly higher than those
found in sand filter effluents (13). This established microbial
flora in the GAC adsorber represents a potential for an unknown
health risk which should be carefully examined. A priority
scheme for this evaluation is proposed in Table 3.
TABLE 3
HEALTH EFFECT PRIORITIZATION OF BIOLOGICAL PARAMETERS
Group I - PATHOGENS AND OPPORTUNISTIC PATHOGENS
Group IA - TOTAL COLIFORM SUPPRESSORS
Group II - ENDOTOXINS
Group III - CHLORINE-RESISTANT BACTERIA
Group IV - HARMLESS INDICATORS OF UNSTERILE CONDITIONS
Group V - BENEFICIAL ORGANISMS TO THE BREAKDOWN OF ORGANIC
COMPOUNDS
Group I includes pathogenic organisms and opportunistic
pathogens capable of infesting an altered or already debilitated
host. Group IA are those organisms that can inhibit or suppress
the growth of existing coliform bacteria and therefore mask the
survival of waterborne pathogens. Group II are endotoxins which
are lipopolysaccharide-protein complexes produced in the cell
walls of gram-negative aerobic bacteria. Group III bacteria are
of concern because these could develop into a flora of chlorine
resistant microbes that would present possible regrowth, taste
and odor, or chlorine demand problems in the distribution system.
Group IV includes harmless microbial indicators of the treatment
process, and group V includes organisms that could be beneficial
to the breakdown of organic compounds.
424
-------
The bacterial flora in the GAG adsorption system therefore
may have either a beneficial or a detrimental effect. The role
of these organisms in the GAC column and subsequent distribution
system must be thoroughly investigated before full scale carbon
systems are implemented in the water treatment process. Table 4
lists the biological parameters that should be considered in the
design of a GAC system. These include specific analyses to
determine the taxonomy of predominant bacterial groups/ non-
specific analyses using scanning electron microscopy to investi-
gate the morphology of bacteria and their location on the carbon
structure, and surrogate procedures (e.g./ MF total coliform
test, SPC test) to measure general classes of organisms. The
primary aim of the microbiological design considerations would
be to assure the removal of all organisms falling in Group I.
It is expected that a properly designed postdisinfection system
also would remove groups II and III.
TABLE 4
MICROBIAL PARAMETERS
PARAMETER OBJECTIVE
TOTAL COLIFORM - INDICATOR OF POLLUTION
MODIF. STANDARD - DEGREE OF DISINFECTION
PLATE COUNT
BIOCHEMICAL - BACTERIAL SPECIATION
ENDOTOXINS - POTENTIAL HEALTH EFFECT
SCANNING ELECTRON - PHYSICAL CHARACTERISTICS
MICROSCOPY OF MICROBIAL GROWTH
At the present time no conclusive information is available
on endotoxins (group II). Before this parameter can be regarded
as a serious design consideration, additional research is
necessary. This experimentation must evaluate the levels of
endotoxin in GAC effluents, the effect of postdisinfection on
the endotoxin complex/ and the human health risk of ingested
endotoxins.
ENGINEERING EVALUATION
The framework of chemical and biological considerations
presented in Figure 1 must be optimized during the field testing
of activated carbon adsorbers. An engineering evaluation en-
compasses the choice of design and operational characteristics
of the GAC system. The planned carbon process must economically
meet acceptable treatment goals at startup and be restored to
its full efficiency after the carbon has been exhausted. Field
425
-------
testing in Philadelphia has included a number of hydraulic and
vessel configurations using both virgin and regenerated carbon.
These experiments are summarized in Table 5. As shown in
Figure I, both numerical and qualitative evaluations of GAC
process efficiency have been employed to ensure maximum flex-
ibility in meeting future water treatment objectives.
Chemical Considerations
Figure 1 shows that three groups of chemicals can be defined
after the health prioritization step - > 1 yg/1, transients, and
£ 1 yg/1.
Two different pilot plant monitoring approaches can be used
to determine specific organic compounds. The first approach
includes single compounds selected because of their effect on
human health. The second is a general screening procedure for
any prioritized organic compounds.
Table 2 lists the specific pollutants of primary health
concern that have been identified in Torresdale drinking water.
Specific quantitative analytical methods for each compound found
in concentrations greater than 1 yg/1 were completed during pilot
plant studies by purge-and-trap analysis (23).
The transient compounds and those compounds present in the
influent with concentrations of £ 1 yg/1 were monitored by
another analytical approach, a screening procedure utilizing a
GC computer plot program (24). A screening procedure consists
of a qualitative analysis with subsequent quantitative evaluation
of the constituents found. A general sequence which is followed
for screening of trace organics is shown in Figure 3. The flow
diagram is a general, flexible scheme describing the successive
laboratory operations necessary. The isolation method is a
critical choice of the analyses, as it defines the type of
compounds studied and the maximum recovery of a compound (25).
The screening methods have been employed to develop chro-
ma tographic profiles. Each chromatographic profile contains
the complex group of trace organics present in the extracted
water and is a fingerprint of the sample.. This constitutes an
information pattern. When many profiles are plotted in the same
manner, all the chromatographic profiles can be compared. GC-MS
identifications help evaluate the consistencies and differences
observed among profiles.
In the Philadelphia work the set of screening protocols has
included analyzing the most volatile organics by the purge-and-
trap method (23) and analyzing the less volatile higher molecular
426
-------
Table 5
Philadelphia Water Department:Organic Treatment Studies
NJ
STuor
Bench 1
Bench 2
Pilot 1
Pilot tf
Pilot III
Pilot III*
OBJECTIVE
GAC Theo-
retical
Mechanism
Oxidizing
Agents
Adsorbent
Types
Adsorbent
Types
GAC
Hydraulics
(Contact
tine ft sur-
face loading)
GAC Pretreat-
ment
(Chlorlnatlon
vs. Dechlortna
tlon)
ADSORBENTS
Ftltrasorb-
400 Carbon
Filtrasorb
400 carbon
and XAD-2
resin
Calgon BPl
t fittrasorb
400 carbons;
XAD-2. XAO-4.
* XE-340
resins; West-
vaco UV-G 1240
I ICI Hydro-
darco 1030
carbons
Flltrasorb
400 carbons
Flltrasorb
400 carbons
COLUMN *
HYDRAULICS
Isotherms;
Column Studies
Batch and
Continuous
Reactors
Carbon and
resin col urns
operated in
series and
parallel. EBCT-
3.0-9.7 min.;
CDM«31n. and
3 t 6 In.; BD-
1.75-5.71 ft.
EBCT'7.3-7.5
•In.; CD'S 9
3 in., t 4 »
6 in.; BD=
3.0-3.2 ft.
EBCT- 7. 5 to
60.0 min.;
SI-0.4-J.O
gpm/ft.2; CD-
4 * 6 In. a
1 » 3 In.;
BD-3.0 t 4.0ft.
£BCT*7.5 nln. :
SLO.O gpm/ft?
CD=2? 3 In.;
BOO ft.
CHLORINE
Chlorlnatlon
and non-
chlorination
Chlorine,
Pe manga nate.
Oione
Free Cl?
residual,
1. 5-2.0 «3/J
Combined C1Z
residual.
2.0 mg/1
Free Cl,
residual,
1.0-2.0 ng/1
Free CI,
residual
1.0-2.0 mg/1;
Dechtorination
with sod tun
sulflte
ANALYTICAL6
METHODS
4 Polarity
Probes -
Nltronethane;
Methyl Ethyl
Ketone;
n-ButanoI ;
1 ,4-Ofoxane
16 Organic
Compounds
nm. VOA.
TOC. selected
Organic Pro-
files (HRR).
GC/MS
TTW, RFH. TOC
selected Organic
Profiles (HRR).
GC/HS
TTffl. VOA. TOC.
Organic Profiles
(HRR ft CILE)
GC/HS. SPC. Micro.
TTHM. VOA. TOC
Organic Profiles
(CLLE) GC/HS. SPC,
Micro.
START
1 DURATION
1976 to
1977
1977
3/76
(18 weeks)
12/76
(15 weeks)
9/77
(IB weeks
to 1.5 yrs.)
9/77
(18 weeks)
REFERENCES
HcGutre. 1977
HcGuire, et •!.,
1978
Suffet. et al..
1978
Suffet, et al..
1978
Boettger, et al.,
1977
Cairo, et al.,
1979
HcElhaney. et al.,
1978
Yone, et al.,
1979
HcElhaney. et al.,
1978
-------
Table 5
Philadelphia Water Department:Organic Treatment Studies
(Continued)
fsJ
CO
STUOV
Pilot IV
Plant
Scale
Pilot V
Bench 2
Pilot VI
Pilot VII
BAG 1
OBJECTIVE
GAC Nodes
(filter/ Ads
vs. Contact-
or)
THM reduction
GAC He-
generation
GAC Re-
generation
(Theoretical
Capacity)
C.AC Scale-
Up (Full
Scale vs.
Pilot Scale
Fill/Ads)
CAC Scale-
Up (Full
Scale vs.
Pilot Scale
Fi It/Ads-
Regenerated)
0/one 1
Carbon
ADSORBENTS
Filtrasorb
400 » 300
carbons
Flltrasorb
400 1 300
carbons. Re-
generated
at IOOn°F 1
1750»F
Filtrasorb
400 1 300
carbons. Re-
generated at
IOOQOF 1
I7SO°F
Filtrasorb
400 carbon
Filtrasorb
400 carbon
Filtrasorb
400 carbon
COlUMN *
MVDRAUI ICS
EBCT.15.0 mln:
Sl=?.0 gpm/ft?
CO--? ft 1? in. 1
? * ?'«?.5';
fln-4.n ft.
Belmont Plant
7R MT,0
EnCTOS.O nln;
SI 1.5-1.75
guru/ ft?; C0=
5 C 6 in «
1 P 1? In.
Ill)- 37-4? In.
Isotherms t
Column Studies
lBf.l-» 9 min;
SI ? gpn/ft?
CD 1044 ft' 1
5 ft?; BD=
»?8 in.
IBf.l 9 nin;
SI ? g|)ri/fi?;
CO-1044 ft? >
5 ft?; BD=
?8 In.
BAC; carbon
alone; sand
CHIORINE
Free Clj
residual
1.0-2.0 mg/l
Chlorine ad-
dition point
and amount
altered
Free Cl?
residual
1.0-?. 0 ng/1
None
Free Residual
1.0-?. 0 ng/1
Free Residual
1.0-?. 0 n|/l
Chlorinated
» non-chlor-
inated col-
umns
ANAL»TICAlb
MIT HODS
TTHM.VOA.TOC
Organic Pro-
files (MRR)
nc/MS.SPC.
Micro.
THM. TWWP.
SPC
Tin. VOA
Organic Pro-
files (MRR)
C.C/MS. SPC
4 Polarity
Probes
T1HH. VOA
Organic Pro-
files (MRR).
P.C/MS, SPC
TTMM. VOA
Organic Pro-
files (IIRR).
fiC/MS, SPC
TTIIM. VOA
Organic Pro-
files (MRR.
CUE), T.C/MS.
SPC, Micro.
START
1 DURATION
10/77
(?0 weeks)
6/77
(1? not.)
11/78
(5 mot.)
1978
3/7B
(6 mos.)
11/78
(6 nos.)
1979
(In progress)
RCFCREMCES
Cairo, et al.,
1978
Aptoxlcz, et al. .
1978
EPA Grant
i EBCT-Empty bed contact time; SL-Surface loading, CO-Colurm dianeter, BO-Bod depth
b TTIfl-Total trihalomethanes; TOC-Total organic carhon; SPC-Standard plate count;
MRR-Macroretlcular resin; Cm-Continuous liquid-liquid entraction; GC/HS-r,as
chromatooraphy/mass spectroraetry; VOA-Vnlatile organic analysis; Bade.-Bacterial
speciation; RFM-Rapid fluorometrlc nettiiid.
-------
ISOLATION
AND
CONCENTRATION
* .
|GC PROFILE}-*JGC/MS ANALYSIS)
t
I PROFILE-COMPUTER DATA HANDLING SYSTEM |
A. Digitizer
1
B. Data
Transmission
C. Data Storage a
and
Management
D.
Data a
Retrieval
I Keznevai .
DATA MANIPULATION a
AND
PRESENTATION
j PROFILE EVALUATIONS
Chromatographer controls these steps
Figure 3. Screening Procedure for Trace Organics
429
-------
weight volatile organics by a resin* accumulator and a continu-
ous liquid-liquid extraction method (25).
Engineering Evaluation for Compounds > 1 yg/1-—
For compounds in concentrations greater than 1 vg/1, either
formed through the chlorination process (such as trihalomethanes)
or consistently present in the Delaware River, the adsorption
mechanism can be evaluated using breakthrough and loading curves
based upon quantification of trace organic compounds in the
influent and effluent to the GAC columns. Thus, in comparing the
performance of GAC in a filter-adsorber mode and in a contactor
mode for pilot test 4 (21), the breakthrough curve for chloroform
shown in Figure 4 was first developed. Leakage of chloroform in
the filter-adsorber effluent is evident in the initial week of
operations. This is in sharp contrast to the contactor effluent
which does not exhibit any significant leakage until the sixth
week. The difference in the concentration of chloroform between
the two modes of operation is approximately 15 ug/1 from week 2
until week 9, when 80 percent breakthrough for chloroform takes
place. Complete saturation of chloroform occurs in all GAC
systems by week 12.
The second step in analyzing the performance of the carbon
was to make the comparison on a mass loading basis. Figure 5
was developed by plotting the cumulative loading of chloroform
adsorbed on the carbon against the total mass of chloroform
applied to the carbon. One hundred percent removal is a straight
line whose tangent is 45 deg, and increased divergence from the
straight line indicates progressively poorer removal character-
istics. It is evident from the plot that the carbon used in the
contactor mode performed better, removing more chloroform than
the carbon used in the filter-adsorber mode. At the point of
saturation, carbon used in the contactor mode adsorbed more
than 20 percent more chloroform than the GAC used in the filter-
adsorber mode. If these data are analyzed in a different
fashion—"namely, at the instantaneous 50 percent removal point as
determined rom the breakthrough curve (indicated in Figure 5 by
asterisks)--a difference of approximately 30 percent in perfor-
mance between the two modes is found. Also shown in Figure 5 is
the fact that carbon 1** (12 by 40 mesh size) was more effective
at removing the mass loading of chloroform than carbon 5***
(8 by 30 mesh size).
A statistical modeling approach was next applied to these
data to test the significance of/the difference in performance
found between the two modes of operation. When percent break-
through is plotted as a function of time, the S-shaped curve
*XAD-2, Rohm and Haas, Philadelphia, Pa.
**Filtrasorb F400, Calgon Corp., Philadelphia, Pa,
***Filtrasorb F300, Calgon Corp., Philadelphia, Pa,
430
-------
120 i—
100 —
10 20 30 40
•0 1OO
Figure 4. CHC13 Removal by FILT/ADS versus Contactor
8
o
f
I
JJ 1* *J» 34)
•«CHCI, AWUD /^otf
Figure 5. Cumulative Mass Loading of CHC13
431
-------
shown in Figure 6 is obtained. The relationship between percent
breakthrough and time can be given by
a+bt
" - (1)
1 + ea+bt
where P is percent breakthrough; t is time at which the sample
was taken expressed as sequential day number; and a and b are the
parameters that define the curve whose values are estimated from
the data.
Equation (1) is nonlinear; however, a transformation can be
used to linearize the relationship. Defining a new variable P1
by
reduces Eq (1) to
loge p^r (2)
a + bt (3)
The correlation coefficient r ranged from 0.85 to 0.93 for
the chloroform data. The curve parameters, a and b, for each
carbon vessel were then statistically analyzed using a Student's
t test. The results indicated that the filter-adsorber and
the contactor performed statistically differently in adsorbing
chloroform. It is important to note that a direct application
of the t test to the field data would not have been appropriate,
since the character of the process is changing throughout time.
Therefore, the entire set of collected data represent a set of
single observations from many different populations, rather than
many observations from a single population. This analysis has
demonstrated that monitoring and designing a GAC plant for
compounds of > 1 yg/1 can be accomplished by mass loading and
statistical analysis.
Engineering Evaluation of Transient Compounds and Compounds
< 1 yg/l~
An engineering evaluation is much more difficult for trace
organics in concentrations less than or approximately equal to
1 yg/1, or for those transient organics present in the influent
to the carbon columns (Figure 1). For this category of organics,
competitive adsorption among specific compounds may produce
displacement and reequilibrium effects. Adjunct to this category
of organics are those that may be the product of chemical and
microbial reaction in the carbon column. These characteristics
create a seemingly nonconservative reaction on the carbon column
and its performance cannot be effectively analyzed using mass
loading curves. For these classes of compounds, a qualitative
screening approach using a time-series representation with a GC
432
-------
"*•"
A - ACTUAL DATA
100
90-
\
80-
ui
u>
*
ui
50
40-
30-
20-
10-
10 20
30
40 50 60
TIME IDAVSI
70
9O 100
Figure 6. simulated versus Actual CHC13 Breakthrough Curve for Contactor
[F-400]
-------
computer plot has been employed (12, 19), as well as analysis of
organics on the carbon particle itself (20).
The organic profile screening procedures employed on each
sample of the Delaware River typically identify 40 different
compounds of concentrations £ 1 yg/1 using a packed GC column and
more than 100 different compounds using a capillary column.
Because of this large number of compounds, a computer program was
developed to present a GC profile in a comparative mode (24).
In this program each peak of the GC profile can be plotted as a
spike proportional to concentration on a relative retention time
scale. Figures 7 and 8 show typical computer-reconstructed GC
profiles of the influent and the GAC column effluent for weeks
1 to 18. These data are from pilot test 1 (12). The data are
semiquantitative and represented on the GC profile plot by the
relative scale shown in Table 6.
TABLE 6
Relative Scale for Pilot Plant 1 Data*
GC Estimated
Integrator Profile Concentration
Area Response Units Peak He i girt Range ug/1
10,000 or higher 1.0 >1
5,000-10,000 0.5 1
0-5,000 0.25 <1
Interpolated response 0.1 0.1-<1
* Internal standard: 2-ethyl-l-hexanol •' RRT 1.00
As indicated in Figures 7 to 9, it is possible to categorize
these compounds on the basis of their appearance or disappearance
from week to week both on the influent and on the effluent from
the carbon column. Another approach would be to categorize these
compounds on the basis of the health effects priority classifica-
tion (Table 1). For compounds with enforceable MCLs or recom-
mended MCLs (nonenforceable health-based goals), a consistent
level beyond the acceptable concentration would indicate exhaus-
tion of the carbon for those compounds. Compounds in lower
priority classifications do not present long-term health effect
problems at an approximately 1-yg/l concentration, but need to be
considered as well since these may displace the high toxicity or
carcinogenic compounds.
A serious problem with this qualitative* approach occurs for
compounds that are present intermittently in the Delaware River.
In this case pilot studies for the design of activated carbon
434
-------
INFLUENT
Ul
1
— 1
_p v
O (
4
j
11.
j.
u
x>
J
r -
j c
ii
3<-
*
*
*
.
i
,i
j
"«
1 c
i
i
?!
) c
(
^
r
V
J
1 "
s^
1
•
|t
1
i
b
'9
><<
•c
U
-_ i
:i
r":
oc
* *
i
i
j
1'
iLi
~\
j
o
IT
1
4
i
L
1
'
?
3
V
4
3
t
i
1,
c
t-
(/
»-
2
!
i
*
0^
*
^ 1
1
t_
1
n w
J —
•
4*
*
Q>
" 0*
r"
1
4
4
*
i
JL
1
_ J
i
t
lit
J
ID
e e
... .»
I
{•
*
1
1
L
j
1 1
ill
50
A
* *
1 J_...
. I
J
I
j
1
a
i _
|
i.
1
A _
1.
1
*
. i
.._j
1.-
t
r
<
i
1
e
j
L .
2
> 0
s?
1^
H £
» *
b
j
ft
i
L
^
LL (^
£> m
cff 0
I ? , ^8? *
e * [ »
, *. , 1 l3
1 " L
1 ..'» , ._
iriTlr.i
e b
5
1
4
iizr
i
add a aca a a
» 4
.SO
bd dd
4 1-4 - 4
.75
1.00
1.25
b d d d a d
4 4 4 4 - t-
1.50
1.75
Z.OO
?.so
MELflTIVE HETENTION TIME
Figure 7. Computer Reconstructed GC Profile of Influent
D Tentatively identified compounds (Cl2Bu-dichlorobutane, MeEtMal-methyl maldemide)
*Compounds confirmed by GC/MS. Compounds noted as specific retention times
based upon GC/MS analysis. a) Major peaks present in all samples; b) peaks
which appear as a major peak after not being present or were previously very
small peaks; c) peaks which disappear or are very small after being major peaks;
d) peaks which change from major to minor; e) peaks present for only a few weeks.
-------
F400
.£•>
osoo
d?
nif'
L LI..
L
I .
10
a
i
_,L
. .1 1.
J
11
4
iiii
dd
4 1
.25
t
A—L
1
1 I
1
L L
.'
H
3
III
»
.50
• • .
_ i
i i
i
._ J
. 1
1
t
A
t
If
-••»
., ..i
. ..JL-
-. A . .
. -A _.
[.._.*. .
i
, L ., .
i
,IIi
b
4
b 1
! • • w
1 *- • JL IflT
- *- — — - — -• 10
13
i .... . i ... i . . t
i. _ . . . i
It
. J 1 ... I , I
.. J - 1 1 __. A • .. L_i _
9
11.1 1 . «.. . 1 1. L
.. i 1 . -.1 1. . .
» . 1 .... 1 . ........ 1 _l 1 .
1 . 1 . i . . 1. . k. .
1 . J 1 1 ... •.._ 1 ._
*
11 1 l.ll J 1
li'fiiiii [JLiir.i!— •
d c d
1 4 1 f -» 4 --» 4 4-- 1-
00 I.2S 1.50 1.75 2.00 2.?S
RtlHTIVE RtrtNTION TIME
Figure 8. Computer Reconstructed GC Profile of Effluent from Carbon Column (F400)
Letters b-e and j| same as for legend of Figure 7. Letter a compounds
not marked. Compounds noted at specific retention times based upon
GC/MS analysis.
-------
A. Halogenated
• 1. Chloroform [CHC13]
• 2. dibromochloromethane [Br-ClCH]
• 3. dichlorobromomethane [Cl-BrCH]
• 4. dichlorobenzene isomer [C12-0]
5. 1,1,1-trichloroacetone [TCA]
• 6. trichloroethylene [Cl3ethe]
• 7. 1,1,2,2-tetrachloroethane [Cl4Et]
• 8. trichlorobenzene isomer [C13O]
9. chloromethylbutene isomer [C_HgCll
B. Aromatic
1. acetophenone [Aceto]
2. benzaldehyde [CCHO]
3. C3-benzene isomer [C.,-0]
• 4. ethylbenzene or xylene isomer [C2-0]
5. phenylacetic acid [OHA 1
6. m/p-tolunitrile [Tolu]
• 7. toluene [OCH-j]
8. benzoic acid [OCOOH]
• 9. naphthalene [Naph]
C. Miscellaneous
1. 2,3:4,6-di-0-isopropylidene-L-sorbofuranose*
[L-Sorbo]
• 2. ,di-n-butyl phthalate [DBF]
3. tributyl phosphate [Bu.,F04]
4. dihydroact inid iolide [Di-act]
5. ethyl acetate [EA]
6. 1,2,3,5-d i-0-isopropylidene-D-xylofuranoset
[D-Xylo]
• 7. isophorone [Isophor]
• 8. On EPA list of priority compounds
*Alternate name: diacetone-L-sorbose
tAlternate name: diacetone-D-xylose
Figure 9. Compounds Found by GC-MS
437
-------
systems are entirely dependent on the particular river conditions
present during the test. Since many organic compounds are in
this category, several research approaches are being considered
at this time. One such approach would involve the use of theo-
retical adsorption models such as the net adsorption energy
concept (26), to classify health effect compounds and select
polarity probes representative of distinct phases. A mixture of
these probes/ each in a concentration range of approximately
1 ug/1, then could be spiked to the influent river water to the
carbon column. Although this would represent much more severe
conditions than would normally be expected in the Delaware River,
the resulting breakthrough curves would enable a conservative
estimation to be made of the carbon vessel's bed life.
Biological Considerations
Historically, the microbial quality of water has meant the
assessment of th£ vater in terms of possible transmission of
waterborne infectious disease. The common approach to the detec-
tion of pathogens in water has been to use an indicator organism
in place of the pathogen; that is, an organism whose presence
suggests that pathogens may be present. The total coliform group
was chosen for this purpose because it is part of the normal
fecal flora; the presence of a member of the total coliform group
in a water sample indicates possible fecal contamination and
therefore possible pathogens.
In GAC adsorber effluents the presence of total coliforms
has been notably lacking. However, recent research indicates
that while no coliforms have been found in the GAC effluents,
large numbers of bacteria have been recovered using a revised
standard plate count test (13, 27, 28). Thus, reliance on the
total coliform test alone as an effective indicator of microbial
disinfection is inadequate for the GAC system. The standard
plate count test which measures the total bacterial population,
may be the best indicator available today for measuring the GAC
effluent for microbial contamination and also the disinfection
process prior to the release of GAC-treated drinking water into
the distribution system. The SPC test does not, however, differ-
entiate between harmless and potentially dangerous bacteria and
cannot be used as an indicator of fecal pollution. It is there-
fore suggested that the SPC test be used in conjunction with a
total coliform test for a more complete bacterial analysis of GAC
effluents. Moreover, the identification of the predominant
bacteria in the GAC effluents also should be examined for each
GAC system to determine the nature of the organism in terms of
human pathogenicity and chlorine tolerance.
Although no significant coliform population has been ob-
served in the GAC adsorbers or their effluents, other gram-
negative bacteria are present in significant numbers. Therefore,
the possible existence of endotoxins in GAC effluents must also
438
-------
be examined. A recent review of the research completed on this
topic by Love and Symons (29) indicates that no increase in
endotoxin concentrations occurs in GAC effluents. The National
Academy of Sciences (7) has concluded that the potential health
risk of ingested endotoxins is low. Nevertheless, until further
research has been done to determine the health risk involved
with ingested endotoxins, their levels in GAC effluents should be
analyzed on a periodic basis as a precaution.
Microbial Findings—
A significant aspect of research work during the GAC treat-
ment studies of the Delaware River involved the investigation
of microbiological effects. Four parameters were used to deter-
mine the microbiological quality of the GAC system. These
included the membrane filter total coliform test, standard plate
count test, biochemical identification of bacteria, and scanning
electron microscopy.
Six primary areas of microbial concern were examined: the
physical characteristics of biological growth on granular acti-
vated carbon; total coliform levels in GAC effluents; total
bacterial levels in GAC effluents; total bacteria on carbon
particles in GAC adsorber; biochemical identification of bacteria
isolated from GAC effluents; and physical parameters affecting
bacterial levels in GAC effluents, including empty bed contact
time and chlorine vs. dechlorinated influents.
The physical surface of a GAC particle (e.g., carbon 1) from
an operating GAC column (60-min. EBCT) and accompanying biological
growth first were examined using the scanning electron mirro-
scope. The activated carbon particle surface (Figure 10) was
found to have a rough, porous texture capable of providing a
favorable and protected enviornmental surface for bacterial
growth. Figure 11 shows a typical bacterial population on a GAC
particle. Bacteria, on the GAC particles examined, were found to
be scattered over the carbon surface with heavier concentrations
in the macropore areas. Bacteria were not observed to form a
uniform biofilm over the entire surface area of the carbon
particle as has been hypothesized. Few higher forms of life,
such as protozoa, were observed on the carbon particle surface.
Although such an advanced ecosystem was observed in wastewater
GAC systems by Weber (30), lower concentrations of nutrients
in the potable water GAC system may be responsible for the
limited biological colonization on the GAC particles.
The major purpose of the microbiological GAC study was to
determine the levels of total coliform and total bacteria in the
GAC adsorber effluents and changes in these levels over time.
Evaluation of the virgin carbon using the SPC test indicated that
no bacteria were present on the carbon prior to its use. No
total coliforms were present in the influent waters and the total
bacterial levels ranged from 0 to 300/ml. In the GAC adsorber
439
-------
Figure 10. Electron Microscope Magnification of Carbon Particle (60X)
-------
Figure 11.
Electron Microscope Magnification of Carbon
Particle (4000X) Showing Bacterial Growth
441
-------
effluents no total coliforms were found throughout the test
program; however, as shown in Figure 12, total bacterial levels
ro|e sharply in the first month to maximum densities of 10 to
10 bacteria per milliliter, after which they decreased and
remained at slightly lower levels for the rest of the study. In
comparison, the effluent of a pilot plant sand filter, also shown
in Figure 12, had a total bacterial density that ranged from 0 to
91/ml throughout the study. These levels are significantly lower
than the bacterial levels found in the GAC column effluents. It
must be pointed out that for both GAC and sand filter systems the
influent water was prechlorinated to a free residual concentra-
tion of approximately 2 mg/1. While the sand filter maintained
this residual throughout the bed, the carbon effectively removed
chlorine in the top layer of the bed and thus did not inhibit
bacterial growth on the GAC.
A study was also undertaken on the bacterial densities found
on carbon particles in the GAC contactor. Figure 13 represents
the change over time in the number of bacteria on the carbon par-
ticles in a GAC column and the corresponding number of bacteria
in the GAC effluent. The upper curve represents the bacterial
population on the carbon particles. This curve shows that the
bacterial population on the carbon particles follows a sigmoidal
growth curve; that is, the bacteria after an initial short lag
phase will increase exponentially until the environment begins
to limit their growth and a negative growth acceleration phase
occurs. At this point the population will stabilize itself with
respect to its environmental limits and population changes will
occur only with shifts in the environment.
Bacterial densities in the effluent closely follow the
bacterial population dynamics on the carbon particles. The lower
curve indicates that bacterial levels in the GAC effluent are
highest during fche exponential phase of bacterial growth on the
carbon particle when the di-viding bacterial cells are most vul-
nerable. Bacterial levels decrease and level off in the effluent
only after the bacterial population in the GAC adsorber has be-
come established and has reachfd a pseudosteady-state situation.
One important aspect of this study was to identify the bac-
teria that were isolated from the GAC effluents. Unfortunately,
many of the bacteria isolated were nonreactive to the standard
biochemical tests and many were not identified. Table 7 lists
the bacteria that were isolated and identified from the GAC
columns or their effluents using th« API system and other bio-
chemical tests for enterobacteriaceae and other gram-negative
bacteria. The most predominant bacteria that were isolated and
identified belonged to the genus Pseudomonas. Members of the
genus Bacillus also were found in large numbers. The bacteria
that were identified in this study were predominantly soil or
water saprophytic chemo-organotrophs and have no known pathogenic
significance to man. However, some species of Flavobacterium,
442
-------
5YM30L
103.
II
'I
l\
I I
/',
I
I
I
4
/
10"
COL3MK
2
3
iana
\
15
15
15
9.5
72
108
TIME (DAYS)
130
Figure 12. Bacteria in the Column Effluents
443
-------
10"
ID?
i 10
^ s
•e c:
<
A CABBOK PARHCLZ
dAC COLUMN EFFLUS;?
4
(WEEKS)
Figure 13,
Bacteria On the Carbon and in the
Effluents of a GAC Column
444
-------
TABLE 7
BACTERIA ISOLATED FROM THE GAC SYSTEM,
THEIR API CODE AND COMMENTS
IDENTIFICATION
API CODE
COMMENT
Ps. putIda
Ps. fluorescens
Ps. maltophilia
Ps. pseudoalcaligenes
Pseudomonas spp. Other
N
II
n
n
n
n
n
n
it
n
H
it
H
n
Achromobacter sp.
Alcaliganes odorans
Alcaligenes sp.
Flavobacterium odoratum
CDC IV F
CDC II F
2 200 004 41
2 201 004 41
2 005 006
1 000 000 41
1 202 000 51
0 000 004 50
2 000 004 50
0
0
0
0
0
2
1
0
0
0
0
0
0
000
000
002
002
001
200
010
001
000
002
010
202
242
004
004
004
004
004
004
004
004
004
004
004
004
004
50
51
41
40
50
51
41
50
40
01
01
00
00
Acceptable
Acceptable
Excellent
Acceptable
Very Good
Best Fit
Good Liklihood but
Low Selectivity
n
n
n
Best Fit
Best Fit
Best Fit
Best Fit
Best Fit
Best Fit
Excellent
445
-------
Pseudomonas, and Alcaligenes are known to be opportunistic
pathogens, chlorine resistant, or suppressors of total coliform.
Two engineering considerations were examined in relationship
to bacterial levels on the GAC effluents: the effects of empty
bed contact time and of pretreatment (chlorination) conditions.
Figure 14 graphs the bacteria in the effluents of three
GAC contactors with EBCTs of 7.5, 15, and 30 min, respectively.
Even after the volumetric flow rate effect was taken into con-
sideration, it was found that the greater the EBCT, the higher
the bacterial levels found in the adsorber effluents. However,
experiments using an extremely long EBCT of 60 min did not indi-
cate a similar pattern. Figures 15 and 16 demonstrate the effect
of chlorinated influent water on the bacterial levels in the
carbon particle and in the adsorber effluent. As can be seen
from these figures, chlorinated influent water has a detrimental
effect on the microbial ecosystem in the GAC column and also
decreases the levels of bacteria in the GAC effluents.
Finally, it was found that the recommended standard plate
count method for analysis of water and wastewater (2-day incu-
bation at 35°C) (31) is not adequate for the enumeration of
bacteria from GAC columns. In Table 8, a statistical analysis
of the standard plate count test at 2-day and at 7-day incubation
periods indicates that bacterial recovery rates were very poor
at 2 days but increased dramatically at 7 days. Preliminary
studies also have indicated that the standard 35°C temperature
for incubation further limits the bacterial recovery rate
compared to resuJts at a 22°C or 28°C incubation temperature.
CONCLUSION
A framework has been developed for designing, and subse-
quently evaluating, the operating performance of a granular
activated carbon process. It begins with the need to classify
chemical and biological data in a health effect prioritization
scheme. This scheme includes both direct impacts to human
health and indirect impacts affecting the operating performance
of the GAC. Esthetic problems arising in the treated water
during its transmission in the distribution system should
also be considered. These chemical and biological considerations
then are used in performing an engineering evaluation and
economic analysis of the alternative GAC process configurations.
Seven years of research studies by the Philadelphia Water
Department have shown that a complex mixture of organic compounds
exists in the Delaware River at Philadelphia. A total of 116
different compounds (including 40 isomers) were found, generally
in a concentration range of 1 ug/1 or less. Only chloroform and
dichlorobromomethane were consistently found in concentrations
greater than 1 ug/1. Dibromochloromethane, 1,2-dichloroethane,
446
-------
103
SYMBOL SOLL'MI
S3CT (MLn.}
7.5
15
30
72
13C
TIME (DAYS)
Figure 14. Effect of Contact Time on Bacteria
Levels in GAC Effluent
447
-------
106
i "4
103.
• XUJIfi * JH
A COUim 7 DECHLQC7WTED
_L
0
TOP
12 IB
DOTH (IHCHIS)
30 36
BOWOT
Figure 15. Bacteria on the Carbon Particles in a GAC Column
(Chlorinated vs Dechlorinated Influent Waters)
. .i .
103
w2
I—L
loe
Figure 16.
tn
Bacteria in GAC Column Effluents (Chlorinated vs
Dechlorinated Influent Waters)
448
-------
TABLE 8
Statistical Analyses of the 2 Day vs 7 Day S.P.C. Study
2 Day
Samples No. Obs.
1 11
2 15
3 15
4 50
5 50
Mean
30.9
74.8
27.3
184
236
Variance
334.89
141.61
88.36
40555
64800
No . Obs .
11
11
14
50
50
7
Mean
409.4
459.4
312.1
3248
2628
Day
Variance
1640
4542
1900
1933983
685730
Equal
Means
• s •
NO
NO
NO
NO
No
Equal
Variance
No
No
No
No
No
-------
1,2-dichloropropane, trichloroethylene, toluene, and diacetone-
L-sorbose were periodically found at this level. The presence
and concentration of organic compounds in the river were highly
variable depending upon the amount discharged, the point of
origin, and the hydrodynamics of the Delaware estuary.
The engineering evaluation performed, thus far, has encom-
passed a number of hydraulic and vessel configurations using
virgin and regenerated carbon. Because of the variable chemical
nature of the Delaware River, both numerical and qualitative
evaluations of GAC process efficiency have been utilized. For
organics that are consistently found in concentrations greater
than 1 yg/1, a rigorous quantitative analysis has been developed
using breakthrough curves, mass loading curves, and statistical
testing of the compounds found in the influent and in the efflu-
ent. For transient compounds and those in concentrations less
than or equal to 1 yg/1, a qualitative screening approach using
a time-series representation with a GC computer plot program has
been applied. The transient compounds, depending upon their
human health impact, may be the critical elements in the design
of the GAC facility for Philadelphia. Presently, several re-
search approaches are being considered to address this problem.
The biological considerations are based upon the ability
to properly kill those organisms passed from the effluent of
the GAC process to ensure that no subsequent problems are
created in the distribution system. This does not preclude the
existence of a biological population on the carbon to achieve
the beneficial objective of assisting in the removal of organic
compounds.
Research studies performed in Philadelphia have demonstrated
that a relationship exists between the level of bacteria found on
the carbon particle and the level of bacteria in the effluent
from GAC adsorbers. Examining the physical characteristics of
the biological growth on the carbon using scanning electron
microscopy revealed that the bacteria were scattered over the
carbon surface with heavier concentrations in the macropore
areas. The most predominant bacteria that were, isolated and
identified belonged to the genus Pseudomonas. Members of the
genus Bacillus also were found in large numbers. The bacteria
that were identified in this study were predominantly soil or
water sapro^hytic chemo-organotrophs and have no known pathogenic
significance to man. However, some species of Flavobacterium,
Pseudomonas, and Alcalicfenes are known to be opportunistic
pathogens, chlorine resistant, or suppressors of total coliform.
Further microbiological studies are needed to establish the
significance of these bacteria in relation to health effect and
distribution system problems. The research also showed that a
revised standard plate count test should be incorporated into
routine bacterial analysis of GAC effluents to ensure bacterial
removal.
450
-------
The interaction of chemistry and biology in a GAC system
with predisinfection must be more carefully investigated before
any conclusive engineering evaluation and economic analysis can
be made. Research studies must determine if microorganisms can
decompose specific organic compounds of human health concern and
high molecular weight organics that compete with these lower
weight compounds for adsorption sites on the carbon. In addi-
tion, research is needed to establish if certain species of
microorganisms are more efficient in this process than others.
An understanding of these key issues then may lead to the
development of optimal operating conditions that will enhance
the growth of the most beneficial microbial population in the
carbon vessel.
451
-------
REFERENCES
1. U.S. EPA. Control of Organic Chemical Contaminants in
Drinking Walter. 43 Fed. Reg. 5756 et seq. (Feb. 9, 1978).
2. Guarino, C.R, et al. Design and Economic Considerations of
Granular Activated Carbon Systems for Removal of Trihalo-
methanes and Synthetic Organic Contaminants. Activated
Carbon Adsorption of Organics from Aqueous Phase (I.H.
Suffet and M.J. McGuire, editors).Ann Arbor Science,
Ann Arbor, Mich. (1979).
3. Natural Resources Defense Council^ et al. v. Train. Consent
Decree on Priority Pollutants, 86RC2120 (D.D.C. 1976).
4. Philadelphia Water Department. Research Investigation on
the Removal of Trace Organic Compounds by Combined Ozona-
tion and Adsorption in a Biologically Activated Carbon
Process. Research proposal funded by EPA, Philadelphia,
Pa. (Sep. 1978).
5. Drinking Water and Health. Safe Drinking Water Comm.,
Natl. Res. Council, Natl. Acad. Sci., Washington, D.C.
(1977).
6. Interim Treatment Guide for Controlling Organic Contami-
nants in Drinking Water Using GAC. U.S. EPA, Cincinnati,
Ohio (1978).
7. Report of Safe Drinking Water Committee to the Environ-
mental Protection Agency. Natl. Acad. Sci., Washington,
D.C. In press (1979).
8. Human Health Considerations of Carcinogenic Organic
Chemical Contaminants in Drinking Water. Natl. Cancer
Inst. Position Paper. 43 Fed. Reg. 29148 et seq. (Jul. 6,
1978).
9. Interagency Research Liaison Committee. Regulators Release
Chemicals Hit List. Chem. Engrg. News, _56:50 (1978).
10. National Interim Primary Drinking Water Regulations.
EPA 570/9-76-003. Ofce. Water Supply, Washington, D.C.
(1976).
452
-------
11. Suffet, I.H., Brenner, L., Cairo, P.R. Identification of
Trace Organics in Philadelphia Drinking Water During a
Two-Year Period. Water Res., in press.
12. Suffet, I.H., et al. Evaluation of the Capability of
Granular Activated Carbon and XAD-2 Resin to Remove
Trace Organics from Treated Drinking Water. Envir. Sci.
Technol., 12:1315 (1978).
13. McElhaney, J.B., & McKeon, W.R. Enumeration and Identifi-
cation of Bacterial Populations on GAC. Proc. AWWA 6th
WQTC, Louisville, Ky. (1978).
14. McGuire, M.J. The Optimization of Water Treatment Unit
Processes for the Removal of Trace Organic Compounds With
an Emphasis on the Adsorption Mechanism. Ph.D. thesis,
Drexel University (1977).
15. McGuire, M.J., et al. The Effect of Oxidizing Agents on
the Removal of Trace Organics from Drinking Water. Proc.
Intern. Ozone Inst. , Ozone Technol. Symp., Los Angeles,
Calif. (1978).
16. Suffet, I.H., et al. Evaluation of the Capability of
Granular Activated Carbon and Resins to Remove Chlorinated
and Other Trace Organics From Treated Drinking Water.
Water Chlorination; Environmental Impact and Health
Ef fects~Vol.2(R.L., Jolley, H.Gorchev, and D.H.
Hamilton, Jr., editors). Ann Arbor Science, Ann Arbor,
Mich. (1978).
17. Boettger, J., et al. Trace Organic Removal by Activated
Carbon and Polymeric Adsorbent for Potable Water. Proc.
AWWA 97th Ann. Conf., Anaheim, Calif. (1977).
18. Cairo, P.R., et al.. The Application of Bench Scale and
Pilot Scale Studies for Control of Organic Chemical
Contaminants in Drinking Water. Activated Carbon Adsorp-
tion of Organics from Aqueous Phase (I.H. Suffet and M.J.
McGuire, editors). Ann Arbor Science, Ann Arbor, Mich.
(1979).
19. Yohe, T.L., Suffet, I.H., and Coyle, J.T. Monitoring
and Analysis of Aqueous Chlorine Effects on GAC Pilot
Contactors. Activated Carbon Adsorption of Organics from
Aqueous Phase (I.H. Suffet and M.J. McGuire, editors).
Ann Arbor Science, Ann Arbor, Mich. (1979).
20. Yohe, T.L. and Suffet, I.H. Specific Organic Removals by
Granular Activated Carbon Pilot Contactors. Proc. AWWA
99th Ann. Conf., San Francisco, Calif. (1979).
453
-------
21. Cairo, P.R., et al. Development of Criteria for the Design
of Full Scale Carbon Adsorption Systems, Proc. AWWA 6th
WQTC, Louisville, Ky. (1978).
22. Aptowicz, B.S., et al. A Comprehensive Study of Trihalo-
methanes in Philadelphia's Drinking Water and Treatment
Techniques to Reduce Their Concentrations. Presented at
AWWA 98th Ann. Conf., Atlantic City, N.J. (1978).
23. Bellar, T.A. and Lichtenberg, J.J. Determining Volatile
Organics at Microgram-per-Liter Levels by Gas Chroma-
tography, Jour. AWWA, £6:12:739 (Dec. 1974).
24. Glaser, E.R., Silver, B., and Suffet, I.H. Computer Plots
for the Comparison of Chromatographic Profiles. Jour.
Chroma tog. Sci. , 1J>.:22 (Jan. 1977); 16:12 (Jan. 1978).
25. Suffet, I.H. and Radziul, J.V. Guidelines for the
Quantitative and Qualitative Screening of Organic
Pollutants in Water Supplies. Jour. AWWA, 68:10:520
(Oct. 1976); Addendum, Jour. AWWA, £9:3:174 (Mar. 1977).
26. McGuire, M.J. and Suffet, I.H. The Calculated Net Adsorp-
tion Energy Concept. Activated Carbon Adsorption of
Organics from Aqueous Phase (I.H. Suffet and M.J. McGuire,
editors). Ann Arbor Science, Ann Arbor, Mich. (1979).
27. Klotz, M., Werner, P., and Schweisfurth, R. Investigations
Concerning the Microbiology of Activated Carbon Filters.
Translations of Reports on Special problems of Water
Technology. Vol. 9~EPA 600/9-76-030 (1976).
28. Van der Kooij, D. Some Investigations Into the Presence
and Behavior of Bacteria in Activated Carbon Filters.
Translations of Reports on Special Problems of Water
Technology. Vol. 9.. EPA 600/9-76-030 (1976).
29. Love, O.T. and Symons, J.M. Operational Aspects of GAC
Adsorption Treatment. U.S. EPA, MERL, Cincinnati, Ohio
(June 1978).
30. Weber, W.J., Pirbazari, M., and Nelson, G.L. Biological
Growth in Activated Carbon by Scanning Electron Microscopy.
Envir. Sci. Technol., .12:817 (1978).
31. Standard Methods for the Examination of Water and Waste-
water.APHA, AWWA and WPCF, Washington, D.C. (14th ed.,
1976).
454
-------
32. McGuire, M.J. Feasibility Analysis and Implementation of
Synthetic Organic Chemical Control Strategies. Presented
at Seminars on Control of Organic Chemical Contaminants in
Drinking Water conducted by Public Technology, Inc., Los
Angeles, Calif. (Nov. 1978).
455
-------
OPERATIONAL EXPERIENCES WITH ACTIVATED
CARBON ADSORBERS AT WATER FACTORY 21
Perry L. McCarty
David Argo
Martin Reinhard
Water Factory 21 is a 0.66-m /s (15-mgd) advanced waste-
water treatment plant operated by the Orange County (Calif.)
Water District. It is designed to improve the quality of bio-
logically treated municipal wastewater for injection into ground-
water, thereby preventing sea-water intrusion. The processes and
concerns of this facility are similar to those of facilities
treating highly contaminated water supplies for potable purposes.
The processes implemented at the water factory are lime treat-
ment, air stripping, recarbonation, chlorination for algae
control, filtration, granular activated carbon adsorption (GAC),
reverse osmosis, and final chlorination for disinfection and
ammonia removal.
This paper describes Water Factory 21's 2-1/2-year opera-
tional experience with GAC. While GAC is generally associated
with organics removal, lime treatment, air stripping, and reverse
osmosis also are important in increasing the overall reliability
and efficiency of water treatment.
OPERATIONAL PERIODS AT WATER FACTORY 21
Figure 1 designates the sampling locations at the water
factory by a number preceded by the letter Q.
The three operational periods at Water Factory 21 are
described by Table 1. During the first period Water Factory 21
was started and the processes were tested. The influent was
trickling filter effluent from treated municipal wastewaters with
a high ammonia concentration. No water was injected into the
ground, so breakpoint chlorination was fiot used. Analysis of
organic removals by various treatment processes was begun during
period 2, while the trickling filter effluent still was being
treated. The effluent was blended with water from a deep aquifer
to reduce electroconductivity to below 900 u/cm before injec-
tion. Air stripping and breakpoint chlorination were used to
meet the ammonia limitation. During period 3, municipal waste-
water began to be treated by an activated sludge process and
flows were segregated. Thus, the concentration of trace organic
456
-------
1
LIME C02
SLUDGE STRIPPING
REVERSE
OSMOSIS
CHEMICAL
CLARIFICATION
FILTRATION
RECARBONATION
ACTIVATED
CARBON
ADSORPTION
Figure 1. Schematic of the 0.66-m3/s Advanced
Water Reclamation Plant at Water Factory 21
TABLE 1
Different Operational Periods at Water Factory 21
Period Dates Operational Characteristics
Jan. 1976 to
Oct. 1976
Oct. 1976 to
Mar. 1978
Mar. 1978 to
Jan. 1979
Trickling filter influent, no break-
point chlorination, no reverse
osmosis, no injection
Trickling filter influent, breakpoint
chlorination, no reverse osmosis,
injection
Activated sludge influent, no forced
air circulation in stripping, partial
ammonia removal by chlorination,
reverse osmosis, injection
457
-------
Material and ammonia decreased significantly. However, water
continued to be passed through the stripping towers because
trace organics removal was found to be quite effective, even
without forced air circulation. Also in period 3, a 0.22-
m^/s reverse osmosis (RO) system was added to reduce the
mineral content of the blended water.
Figure 2 illustrates one of the 17 carbon adsorption columns
in service at the Water Factory. Each is a coal tar, epoxy-lined
steel pressure vessel with an overall height of 12.5 m, a side
wall height of 7.3 m, and a diameter of 3.7 m. The column inlets
and outlets use stainless steel well screens over a 0.38-mm slot
opening to retain the carbon within the contactor. Each column
contains approximately 35 metric tons of 8 by 30 mesh GAC* with
a bulk density of 420 kg/m3. The columns are designed to
provide a 34-min empty bed contact time at design flow which
is equivalent to a surface loading rate of 3.8 mm/s.
INFLUENT TO
WASTE VALVE
DRAIN VALVE
MANIFOLD
MANIFOLD
^-INFLUENT
LINE
FLOW TUBE
EFFLUENT FLOW
CONTROL VALVE
CROSS OVER
CONNECTION
VALVE
'ASTE DRAIN LINE
Figure 2. Granular Activated Carbon Contactor Vessels
*Filtrasorb 300, Calgon Corp., Pittsburgh/ Pa.
458
-------
Originally, the columns acted as upflow countercurrent
packed bed contactors. Water was introduced through the bottom
inlet screens and flowed upward through the carbon, exiting
through the top screens. A space was left void at the top of
each column to allow backwashing by expanding the normal flow
rate in the upflow direction. The columns were piped so that
water could be introduced at the top of the contactor to clean
the stainless steel outlet screens of carbon fines and to reduce
head loss when required.
In mid-1977, while pilot studies were being conducted to
determine the feasibility of reverse osmosis demineralization
of the GAC effluent, it was discovered that operating the carbon
columns in the upflow configuration caused continual bleedoff of
carbon fines, in the size range of 25 ym or less, to the RO
system. Even at hydraulic loading rates of 3.2 mm/s there was
sufficient fluidization of the GAC bed and interparticle col-
lision to cause a slow attrition of the carbon and a carryover
of the fines. Operation of the columns by downflow gravity
tended to arrest this process. For confirmation, a series of
samples from upflow and downflow operations were analyzed with
a particle counter.* The results (Table 2) indicated that the
total particle count for an upflow GAC column was approximately
four times that of a downflow column. Most of the particles
were 25 ym or less in size. In several further tests the accu-
mulation of material on 1-ym cartridge filters during upflow and
downflow operations was compared. The downflow mode produced an
accumulation of a brown residue associated only with biological
materials, while the filter receiving upflow water was rapidly
blackened with GAC fines. Thus, all GAC columns were modified
to receive the downflow water so that the RO membranes were
protected from fouling by carbon fines.
TABLE 2
Comparison of Particle Counts Found in WF-21
Upflow and Downflow GAC Contractors
Sample
Upflow
GAC Effl
Downflow
GAC Effl
2.5-5
ym
9500
2796
Number of Particles in 10 ml
0-25 25-50 50-100 100-150
ym ym ym ym
1837
297
192
70
29
14
1
1-
Total
Particles
per ml
1560
440
'Coulter Electronics, Inc., Hialeah, Fla,
459
-------
The change from the upflow to the downflow mode made regen-
eration of GAC more difficult. In the upflow mode the bottom
carbon becomes more contaminated and is easily removed for regen-
eration, after which it is replaced at the top of the column. In
the downflow mode the upper carbon is the most contaminated, but
it is difficult to remove from the system. The lower carbon
first must be removed and stored; the top carbon then is removed,
regenerated, and placed on the bottom. This is a workable scheme>
but is more likely to result in additional carbon loss because of
the increased handling. In spite of this problem the change to
downflow was considered desirable because it improved the efflu-
ent quality.
When organic removal through a GAC column decreases signif-
icantly, the column is removed from service and the exhausted
carbon is regenerated in the system, as shown schematically in
Figure 3. Exhausted carbon is extracted in slurry form and
transferred to a dewatering bin with screen drains. About 8 kg
waterAg carbon is needed to form a slurry suitable for transfer
through a flanged pipe with large-radius bends. The dewatering
bins are coated with epoxy to protect the steel from corrosion.
Draining the carbon takes about 10 min. After draining, the
carbon retains 40 to 45 percent of its weight in moisture which
is optimal for thermal regeneration.
Subsequently, the carbon is transferred by a stainless steel
screw conveyor (40 to 200 kg/hr dry basis) for discharge into
the top of the carbon regeneration furnace. This is a gasfired,
six-hearth unit rated at 5.5 metric tons/day of dry carbon. Six
burners are provided: two each on hearths 4, 5, and 6. The
temperatures of the hearths are regulated independently by an
automatic temperature control and the furnace operates at
temperatures up to 930°C on hearths 5 and 6. Each burner is
adjusted, so that the atmosphere holds no excess oxygen that
otherwise might burn the carbon. Steam is added to hearths 4
and 6 to give a more uniform temperature distribution throughout
the furnace. The carbon is moved across each hearth by four
rotating stainless steel rabble arms. The high temperature
causes volatilization of the adsorbed organics which are com-
busted in an afterburner on top of the furnace. A venturi wet
scrubber removes carbon fines from the furnace exhaust.
After regeneration the carbon is cooled in a quench tank
and pumped by two carbon slurry pumps to a defining chamber.
Once defined, the carbon is transferred in slurry form back
to the GAC adsorbers for reuse.
The overall operation of the GAC system has been relatively
trouble free. While in the upflow mode the columns were back-
washed when head loss buildup caused a drop in flow. Usually
this happened once a week. The need to backwash was associated
with buildup of carbon fines on the top effluent screens and not
460
-------
4
EXHAUSTED CARBON SLURRY ^
CARB
FILL
CHAP
k
ON ,
IBERT
i
CARBON
COLUMN
i
t
, .YJ
^L
PRESSURE
WATER 1
\ REGENERATED
^AND MAKEUP
CARBON SLURRY
MAKEUP
CARBON r~
'A
MAKEUP
CARBON
WASH *— K
TANK
i Y
•, « ,T. M -,?, a PRESSU
\MAKEUP CARBON "*'tn
SLURRY
SCREEN
• •' •* *• • OVERFL
2 DRAIN
' j . -•—
' 5 M *
REfitNE-
^!|§N
WASH .
TANK *-h.
ll
pi « 1*1 era «i"t. ^ PRE
•
CARBON
^DEWATERING
TANK
/ SCREW FEED
1 CONVEYOR
L^n
ft
CARBON
REGENERATION
FURNACE
RE
T
"W-28f-
— ^ T
.4-. (-3, CARBON
HH &^3 SLURRY
^•^ PUMP
SSURE
1 V ™ P0"»* »• ^ WATCR
\REGENERATED CARBON
SLURRY
Figure 3. Carbon Regeneration System
461
-------
with biological growth or accumulation of suspended solids in
the GAC beds themselves. The change to downflow operation
coupled with improved quality of applied water further reduced
the need for backwashing; the GAC columns are backwashed
about every two weeks. Apparently, little buildup of biological
solids or other suspended material occurs in the carbon bed.
The water applied to the GAC columns generally is free of
suspended solids, and influent and effluent turbidities
typically are lower than 0.5 NTU. At no time during the
operation of the carbon adsorbers have anaerobic conditions
been detected. The air stripping process prior to GAC
adsorption supplies water high in dissolved oxygen.
GAC ROLE IN ORGANICS REMOVAL
GAC is just one of several unit processes used at Water
Factory 21 for organics removal. Figure 4 illustrates the
frequency distribution of chemical oxygen demand (COD) at
various sampling locations during period 3. The distribution
of COD and trace organics at various locations was found to
follow a log normal distribution (1), characterized by the
geometric mean concentration M located at the 50 percent
point, and a spread factor S represented by the slope of the
line. The spread factor is similar to a standard deviation;
68 percent of the time the concentration lies between M/S and
MS, and 95 percent of the time between M/S2 and MS2.
Figure 4 indicates that lime treatment and reverse
osmosis, as well as GAC, are effective in organics removal.
Lime treatment not only removes suspended organics but also a
portion of the dissolved organics, some of which are not
removed effectively by GAC (2). Reverse osmosis is very
efficient in removing the higher molecular weight organics
which are very poorly characterized to date.
Figures 5, 6, and 7 illustrate removals of three
representative trace organics by various processes during
period 3. The first, 1,4-dichlorobenzene, is removed
effectively by stripping (04) and GAC (Q9), but not by lime
treatment (02) or filtration (06). The second, tetrachloro-
ethylene, is removed efficiently by stripping (04), but not
by GAC or any other process. The third diisobutylphthalate,
is removed best by GAC (09). These results indicate the
importance of the total system in organics removal. During
period 3, forced air circulation was not used in the stripping
towers.
PERFORMANCE OF GAC
Table 3 and 4 summarize the performance of GAC for the
removal of COD, TOC, trace organic contaminants, and various
heavy metals during periods 2 and 3. Included in these tables
are the geometric mean concentrations, the number of samples
462
-------
100
50
20
10
_ 5
o>
E 3
Q
O
O
I
0.5
0.2
INFLUENT
—
LIME TREATEC
=====
FILTERED
EFFLUENT
I 2 5 10 20 50 80 90 95 9699
PERCENT OF TIME LESS THAN
Figure 4. Frequency Distribution of COD Concentrations at
Various Sampling Locations During Period 3
z
g
I
z
tu
o
O 0.01
0.001
I—I 1 1 1 1—I
1,4,-DICHLOROBENZENE
5 10 20 50 80 90 95
PERCENT OF TIME LESS THAN
99
Figure 5. Frequency Distribution of I,4-Dichlorobenzene
at Various Sampling Locations During Period 3
463
-------
100
10
5 o
o:
UJ
o
8
QOI -
0001
T i i i i i r
TETRACHLOROETHYLENE
0:4
5 10 20 50 80 90 95 99
PERCENT OF TIME LESS THAN
Figure 6. Frequency Distribution of Tetrachloroethylene
at Various Sampling Locations During Period 3
100
1 I0
5^
i •
i-
tr
z O.I
UJ
o
o
o 0.01
0.001
i 1 T T f I I
DIISOBUTYLPHTHALATE
01,
5 10 20 50 80 9095
PERCENT OF TIME LESS THAN
99
Figure 7. Frequency Distribution of Diisobutylphthalate
at Various Sampling Locations During Period 3
464
-------
Table 3. Removal of Organic Materials by GAC During Periods Two and Three
Period Two
Period Three
Contaminant
Inf. Cone.*
(No. Samp.)
Q6
COD 42(272)
TOC 14(156)
1,4-Dichlorobenzene 0.02(6)
1,2-Dichlorobenzene 0.01(6)
Di isobutylphthalate
Tribromomethane
Dimethylphthalate
Chlorobenzene/o- 0.09(7)
xylene
Bromodichloromethane 4.6(33)
Dibromochloromethane 1.4(29)
tt-xylene
Naphthalene
Di-n-butylphthalate
Styrene
Carbon
Tetrachloride
Ethylbenzene 0.06 (7)
Bis-[2-ethylhexyll-
phthalate
1-Methylnaphthalene
Trichloroethylene
Tetracnloroethylene 0.04J28)
Methylene chloride 1.5 (38)
1,1,1-Trichloroethane
2-Methy1naphthalene
Chloroform 8.2 (32)
% Removal
(95%CI)
60(58 to 63)
51(48 to 54)
17(-750 to 90)
82(0 to 97)
46(-5 to 72)
72(46 to 86)
84(58 to 94)
Ir.f . Cone.*
(No. Samp.)
06
45(3 to 69)
72(-260 to 98)
-7(-98 to 43)
2K-70 to 63)
% Rernovaj
(95% CI)_
24(155) 49(46 to 52)
0.07
0.02
2.0
0.41
1.3
O.li
1.8
0.65
0.05
0.05
0.59
0.03
(16)
(16)
(16)
(17)
(16)
(16)
(16)
(17)
(16)
(16)
(14)
(16)
0.07 (17)
0.02 (16)
3.4 (5)
0.02 (16)
1.0 (16)
0.16 (17)
0.16 (17)
0.02 (16)
5.3 (17)
98(43 to 100)
9K-150 to 100)
87(61 to 95)
81(40 to 94)
64(24 to 83)
63(30 to 80)
54(5 to 77)
52(2 to 77)
50(7 to 73)
50(-8 to 76)
44{-90 to 34j
40(-190 to S3!
20(-110 to 70)
17(-77 to 61)
9(-69 to 51)
0(-660 to 87)
-6(-220 to 65)
-6(-190 to 61)
-12(-430 to 75)
-18(-150 to 45)
-4K-146 to 19)
•Geometric mean cone, in ug/L except COD and TOC which are in mc/1.
Table 4. Removal of Heavy Metals by GAC During Periods Two and Three
Period Two
Period Three
Contaminant
Ba
Cd
Cr
Cu
Fe
»3
Mn
ft.
Inf. Cone.*
(No. Samp. )
06
% Removal
(95% CI)
2.6
26
1.4
29
56
(27)
(26)
(32)
(33)
(27)
105 (33)
1.9 (26)
(33)
(26)
3.2
3.0
4(-27 to 27)
0(-37 to 27)
7(-32 to 35)
38(9 to 58)
64(51 to 74)
66(41 to 80)
1K-47 to 46)
-16(-83 to 27)
27(-56 to 66)
Inf. Cone.*
(No. Samp.)
06
0.77 (6)
7.8 (6)
(6)
(6)
(6)
7.2
5.6
36
28
1.3
3.1
(6)
(6)
(6)
% Removal
(95% CI)
10(-250 to 77)
5(-150 to 64)
-32(-210 to 43)
45(-3 to 70)
56(-85 to 89)
-50(-530 to 64)
-35(-270 to 51)
68(-780 to 99)
'Geometric mean cone. ^n ug7^
465
-------
analyzed, the average percent removal, and the 95 percent con-
fidence interval for the percent removal (1). During period
2 about 14 percent of the GAC in each column was regenerated
every 40 to 70 days, while during period 3 about 50 percent was
regenerated every 6 months. The trace organic contaminant data
in Table 3 are arranged in order of removal efficiency by GAC
(except for COD) during period 3. The analytical methods used
are described elsewhere (1). In general, chlorinated benzenes,
some of the phthalates and aromatic hydrocarbons, and the
brominated trihalomethanes were removed at least as efficiently
as COD. The one- and two-carbon chlorinated compounds, on the
other hand, were not removed by GAC under the existing operating
conditions. These compounds were removed by stripping. The
confidence intervals on the percent removal are quite broad, so
care must be taken in interpreting these data.
The heavy metal data in Table 4 show that chromium, copper,
and lead are partially removed by GAC treatment. Iron was fairly
efficiently removed during period 2, but during period 3, GAC
treatment increased the effluent iron concentration. Tfeis change
is probably caused by wear on the linings and increased corrosion
of the GAC vessels. The number of samples collected for heavy
metal analyses during period 3 is low, and the resulting confi-
dence interval is broad. Thus, some Caution is warranted in
drawing conclusions from these data.
Instead of COD, TOC frequently has been proposed as a gen-
eral monitoring tool for GAC performance. At the Water Factory,
however, COD is more often used as a control parameter since the
TOC instrument was not always operable and the accuracy of TOC
analyses has been questionable. Table 5 compares COD with TOC
results. The COD/TOC ratio normally lies between 2.0 and 3.0.
FRESH VS. OLD GAC
Figure 8 summarizes the effectiveness of a single GAC column
in the removal of COD toward the end of period 2 and through
period 3. During the upflow operation in period 2 the column
influent concentration of COD generally was greater than 35 mg/1,
and an average of about 30 mg/1 COD was removed. Whenever efflu-
ent COD from a column reached 20 mg/1, about 5.5 metric tons were
regenerated and about 0.8 kg COD were removed for each kilogram
of regenerated carbon.
At the beginning of period 3 the influent COD decreased con-
siderably; accordingly, the carbon needed to remove less organic
matter from the water to maintain the COD limitation of 20 mg/1
in the effluent and did not have to be regenerated as frequently.
Assuming that 3800 m^ (10*> gal) of throughput is equivalent to
one day of operation, almost 6 months of operation were possible
before regeneration was necessary. During this time, COD
removal by activated carbon treatment averaged about 7 mg/1, for a
466
-------
Table 5. COD to TOC Ratios at Various Sampling locations for
Different Periods of (Deration at WF-21
Period
Two
Three
Parameter
COD*, mg/1
(no. sattples)
TOC*, mg/1
(no. sanples)
COD/TOC ratio
95% CI for ratio
COD* ,mg/l
(no. samples)
TOC*,rog/l
(no. samples)
COD/TOC ratio
95% CI for ratio
Chen. Eff. Filt. Eff.
Q2 Q6
52 (274) 42 (272)
21.5 (19) 14.4 (156)
2.41 2.92
2.3-2.6 2.8-3.0
27 (156)
10 (44)
2.70
2.6-2.8
GAC Eff.
08
16.6 (264)
7.0 (163)
2.37
2.2-2.5
12.3 (125)
6.2 (91)
1.98
1.8-2.2
*Gecnetric Mean Concentration
WATER FACTORY 21
TYPICAL GRANULAR ACTIVATED CARBON COLUMN W/REGEN£RATON
60
50
40
E30
»
Q
10
TRICKLING FILTER .
EFFLUENT '
ACTIVATED SLUDGE EFFLUENT
0 Z0406080COeOI40l60ie0200220a40a02803003K)340360
THROUGHPUT VOLUME, MG
13 14
— • ••••* -•• - -• -* ' - •* * - -* ' - - J ...--' J
THROUGHPUT VOLUME, 10V
Figure 8. Influent and Effluent COD for a Typical GAC Column
During the Latter Part of Period 2 and Into Period 3
467
-------
cumulative total over the period of about 1.7 kg COD/kg regen-
erated carbon.
To evaluate the effect of biological processes within the
activated carbon system, one GAC column, started during phase 2,
was operated continuously with no regeneration. The COD removal
by this system is shown in Figure 9. For the first 475,000 m3
(125 mil gal) of throughput, COD removal gradually decreased.
For the next 475,000 m3 (125 mil gal), COD removal remained
nearly constant at about 32 percent. This removal is believed to
be the result of the activity of microorganisms growing on the
GAC. The trickling filter effluent received by the Water Factory
at that time was inefficiently treated and no doubt contained a
significant amount of biodegradable organics.
With the change to the more efficiently treated activated
sludge wastewater, the fractional removal of COD by this GAC
column decreased after a short acclimation period to about
20 percent. The operation of this column indicates that bio-
logical processes are significant in GAC performance, which may
explain the high organic removals per kilogram of GAC obtained.
TRACE ORGANICS REMOVAL BY FRESH AND OLD GAC
A comparative evaluation was made of trace organics removal
during period 3 by a regenerated fresh GAC column and an old GAC
column (Figure 10). The effluent from the fresh GAC is desig-
nated as Q7-12, and that from the old GAC as Q7-5. At the start
of this evaluation (Jul. 5, 1978) about 75,000 m3 water had been
passed through the fresh GAC since the previous regeneration,
and about 380,000 m3 (100 mil gal) water had been passed through
the old GAC since it was put in service. The GAC had just been
transferred to a new vessel and about 10 percent new GAC was
added. Thus, its initial performance was similar to that of
freshly regenerated GAC. Water was passed through the column at
the design flow rate until September 1, when 100 percent of the
GAC was regenerated. The unit was put back into operation on
October 1. Data on trace organics were gathered from July 5
through December 31.
Figure 10 shows the COD of daily composite samples for the
fresh and old GAC after the complete regeneration on October 1.
The effluent COD from the fresh column was initially about 5 mg/1
which is typical for regenerated carbon, and rose at a rate of
about 0.17 mg/l/day for the first 60 days. This corresponds to
an increase in effluent TOC of about 0.08 mg/l/day.
The usual practice is to regenerate only one half of the
GAC in a bed. Accordingly, the rate of effluent COD increase is
twice that depicted in Figure 10. This figure also indicates an
apparent variation in effluent COD which parallels the variation
in influent COD. This variation is an important consideration
468
-------
60
SO
0.40
o
O 30
O
20
10
•TRICKUN6 FILTER EFFLUENT-
—ACTIVATED SLUDGE
EFFLUENT
A CARBON COLUMN INFLUENT
o CARBON COLUMN EFFLUENT
0 a 50 75 CO 125 150 175 200 225 250 275 300 3» 350 375 400 425 450
THROUGHPUT VOLUME, MG
01 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18
THROUGHPUT VOLUME, I05m3
Figure 9. Influent and Effluent COD for a GAC Column Over
an Extended Tine Period Without Regeneration
40
30
o»
E 20
o
O
o
10
FRESH GAC (07-12)
20 40 60
TIME (DAYS)
80
100
Figure 10. Comparison of Influent and Effluent COD Concentration
Over Time for Fresh and Old GAC
469
-------
when criteria for GAC performance and regeneration are being
developed.
The geometric mean concentrations and spread factors
for the trace contaminants in the influent and effluents from
the two GAC columns are summarized in Table 6.
Table 7 presents the geometric mean influent concentration
and the percentage removal of each trace constituent based upon
the differences between the influent and effluent geometric
means, together with the 95 percent confidence interval for the
percent removal (1). The constituents are arranged in order from
that with the highest calculated removal (by fresh GAC) to that
with the lowest. Many of the constituents were present in low
concentrations, very near the detection limit. Nevertheless, the
data show a sufficient indication that many but not all trace
constituents are removed very efficiently by GAC. Chloroform and
the two-carbon chlorinated solvents are not removed efficiently.
A surprising result of this analysis is that the efficiency
of trace constituents removal by old GAC is comparable to that
by freshly regenerated GAC. To evaluate this result further a
test was conducted to determine whether a statistically signifi-
cant difference existed between the GAC effluent concentrations
for any of the constituents. The data from this analysis are
given in the last column of Table 7. Numbers lower than 0.1
indicate that the differences are statistically significant.
Only for the two THMs, dibromochlorome thane and bromodichloro-
methane, and for diwethylphthalate did the data show significant
differences in performance between fresh and old GAC.
An analysis was attempted of the data obtained following
regeneration of the GAC for the Q7-12 samples. Because of the
general day-to-day variability of influent trace organic concen-
trations and the effect this has on effluent variability, no
significant tfend in most constituent concentrations could be
detected. The significant problem in detecting differences or
trends with such input variability should be recognized when
performance standards are set for GAC and for the frequency of
regeneration. Treatment objectives that are both desirable and
feasible also must be established. The current requirements of
the Orange County Water District for the combined effluent from
all GAC columns is a COD not to exceed 30 mg/1. The requirement
was met about 90 percent of the time during period 2 and has not
been exceeded during period 3. This level is practical for
day-to-day operation but probably is too high for potable
supplies used directly.
EFFECTIVENESS OF GAC FOR TRIHALOMETHANE REMOVAL
difficulty of removing •THMs and other halogenated one-
and two-carbon compounds is well recognized (Tables 3 and 7).
470
-------
Table 6. Summary o£ Trace Contaminant Geometric Mean
Concentration, M, Spread Factor, S, Number of Samples
analyzed, N, and Number of Analyses Above Detection
Limit (Nu).
Influent
M
Contaminant i>9/l
Heptaldehyde
Naphthalene
Tribromomethane
1,4-Dichlorobenzene
Oibromoch lor ome thane
Diisobutylphthalate
Styrene
Br omod i ch 1 or ome thane
Diethylphthalate
Dime thy Iph thai ate
1,1 , 1-Tr ichl or oe thane
Chlorobenzene/o-xylene
m-Xylene
p-xylene
Di-n-butylphthalate
Bis- 12-ethylhexyl) -
phthalate
Chloroform
Trichloroethylene
Ethylbenzene
Tetrachloroethylene
Pentachloroanisole
0.28
0.046
0.41
0.066
0.65
2.0
0.030
1.76
0.62
1.32
0.16
0.11
0.046
0.012
0.5*
3.4
5.3
0.98
0.023
0.16
0.078
S
1.
1.
2.
1.
3.
1.
2.
3.
2.
2.
11
1.
2.
2.
3.
1.
3.
2.
2.
2.
2.
GAC (06)
M
and
Effluent
Fresh GAC (Q7-12)
M
Nu yg/l
56
67
19
78
4
66
61
4
56
13
.3
93
24
63
04
72
0
75
42
47
00
15(11)
16(9)
17<9)
16(3)
17(16)
16(16)
14(7)
16(15)
16(10)
16(16)
17(12)
16(15)
16(15)
16(6)
14(6)
16(5)
17(17)
16(7)
16(10)
17(13)
13(8)
0.009
0.002
0.02
0.006
0.09
0.35
0.006
0.55
0.21
0.47
0.23
0.043
0.037
0.01
0.50
3.3
5.3
1.0
0.024
0.23
0.004
S
5.12
7.77
30
3.6
6.66
2.49
3.53
3.35
2.65
2.43
2.90
3.39
3.63
4.10
3.29
2.40
2.50
2.30
2.84
7.81
2.79
N
and
Nu
11(2)
11(2)
11(3)
11(3)
11(6)
13(2)
10(2)
11(10)
13(2)
13(8)
11(6)
11(9)
10(8)
10(4)
13(5)
13(6)
10(10)
11(5)
10(7)
11(9)
13(4)
Effluent
Old GAC (05)
M
P9/1
0.034
0.004
0.17
<0.02
0.62
0.26
0.014
1.88
0.27
1.72
0.26
0.07
0.016
<0.02
0.35
4.8
7.06
1.1
0.015
0.19
<0.01
5
4
3
1
2
1
1
2
1
9
1
2
2
1
2
2
1
4
S
.93
.87
.69
-
.98
.32
.93
.98
.62
.83
.06
.85
.16
-
.49
.21
.21
.32
.85
.32
--
N
and
Nu
14(5)
15(3)
14(5)
15(1)
14(14)
15(6)
13(5)
13(13)
15(5)
12(9)
14(13)
15(14)
14(8)
14(3)
13(2)
15(5)
13(13)
13(6)
14(5)
14(9)
--
Table 7. Average Percentage Removal of Trace Contaminants
by GAC and 95% Confidence Interval for
Average Percentage Removal
Contaminant
Heptaldehyde
Naphthalene
Pentachloroanisole
Tr i broroome thane
1,4-Dichlorobenrene
Dibromochloromethane
Di isobutyIphthalate
Styrene
Br omod ichlorome thane
Diethylphthalate
DimethyIphthalate
1,1,1-Trichloroethane
Chlorobenzene/o-xylene
m-Xylene
p-Xylene
Di-n-butyIphthaiate
Bis- [2-ethylhexyll-
phthalate
Chloroform
Trichloroethylene
Ethylbenzene
Tetracbloroethylene
Fresh GAC
(07-12)
Removal
%
97
96
95
95
91
86
83
80
69
66
64
63
61
20
17
15
3
0
-3
-4
-44
59
-13
82
-280
54
22
30
-85
13
-96
24
-110
3
130
-400
-300
-150
-130
-230
-190
-560
95%
CI
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
to
Old GAC
(07-5)
Removal 95%
% CI
100
100
99
100
98
98
96
98
89
66
83
93
84
71
87
82
63
55
68
63
68
88
91
>87
59
>70
5
87
S3
-7
56
-30
69
36
61
-
17
-41
-33
-9
35
-19
32
33
-65
-97
72
-30
-130
-33
-130
-87
-3
20
-200
-150
-no
-230
-51
-270
to
to
-
to
_
to
to
to
to
to
to
to
to
to
_
to
to
to
to
to
to
Level of
Signifi-
cance for
Difference
Between
Effluent
Means*
98
99
90
54
94
50
50
86
26
95
61
81
77
21
34
64
72
62
0.
>0.
-
0.
_
0.
>0.
0.
0.
>0.
0.
>0.
0.
0.
-
>0.
0.
0.
>0.
0.
>0.
5
5
5
05
5
5
01
5
01
5
5
5
5
5
5
5
5
5
Values below 0.1 indicate differences are statistically significant
471
-------
Generally, THM's are removed much better by fresh GAC than by
old GAC. The more highly brominated THM's are removed more
efficiently than chloroform. Further comparisons between the
fresh and old GAC are shown in figures 11 to 13. The water
applied to GAC had been chlorinated in the recarbonation
basin for control of algae. The higher effluent compared
with influent concentrations of chloroform frequently found
perhaps resulted from additional formation when the chlorinated
water passed through the GAC column.
After 100 days the concentration of THM in the influent
to the GAC increased as a result of a decrease in the influent
ammonia concentration and a resulting free chlorine residual
in the recarbonation basin. However, after 180 days, when the
influent concentrations of THMs decreased, the effluent
concentrations remained high for a short period during which
they exceeded the influent concentrations. These results are
similar for the fresh and old GAC, although the fresh GAC was
more efficient in THM removal.
A comparison of the relative effectiveness of new versus
old GAC is given in Figure 14. With chloroform, freshly
regenerated GAC performs a little better than old GAC. Over
time the difference is diminished and after 180 days the
chloroform concentration in the fresh GAC effluent was higher
than that in the old GAC effluent. The removal of brominated
trihalomethanes by freshly regenerated GAC was much better,
but the ability of fresh GAC to remove these materials
decreased rapidly; 50 days after regeneration little difference
in removal between fresh and old GAC was evident.
CHARACTERISTICS OF TRACE ORGANICS REMOVAL
Even though the calculated removals of the specifically
measured trace organics demonstrate wide confidence intervals
and no definite relationship between the chemical structure
of the contaminant and its removal can be deduced, some trends
in Tables 3 and 7 appear to be consistent. Within the group
of the THMs the removal increases with an increasing number
of bromines. Dichlorobenzene is better removed than
chlorobenzene and within the group of aromatic hydrocarbons,
naphthalene is more efficiently removed than the xylenes and
ethylbenzene. These increases in removal may be ascribed to
an increase in the hydrophobicity of these compounds as
measured by the n-octanol water partition coefficient,
Poct increases in the sequence
CHC13 <
and also
chlorobenzene < di chlorobenzene
472
-------
50 100 I!
TIME (DAYS)
200 250
Figure 11. Comparison of Chloroform Removal by
Fresh (Q7-12) and Old (Q7-5) GAC
O
CHCI2Br
50 " 100 150 200 250
TIME (DAYS)
Figure 12. Comparison of Bromodichloromethane Removal
by Fresh (Q7-12) and Old (Q7-5) GAC
473
-------
~ 6
4
i
OC
8 J
C H Cl Br,
REGENERATION
I
0 50 100 150 200 250
TIME (DAYS)
Figure 13. Comparison of Dibromochloromethane Removal
by Fresh (07-12) and Old (Q7-5) GAC
1.5
± 1.0
o
o
OJ5
O
CHCI,
REGENERATED
0 50 100 150 200 250
TIME (DAYS)
Figure 14.
Ratio of Effluent Trihalomethane Concentrations
for Fresh (Q7-12) and Old (Q7-5) GAC
474
-------
and
xylenes, ethylbenzene < naphthalene
This finding indicates that the hydrophobia properties of the
trace organics are directly related to their removal by GAC,
although other factors may play important roles.
The phthalate esters show an erratic behavior in such a
scheme with bis-[2-ethyIhexyl]-phthalate being the least removed
compound of this group, possibly because of a lack of specificity
in the analytical methods and as yet undetected interferences.
The low removal or increase in tetrachloroethylerte and alkylated
benzene concentrations is difficult to explain. Elevated concen-
trations may result from leaching from the tank coating which is
known to contain tetrachloroethylene. Alkylated benzenes may
originate from GAC which contains aromatic hydrocarbons in its
structure.
If individual THM removals by the old GAC column are ranked
from most efficient to least efficient, the order is the same as
for the fresh GAC, even though the absolute percent removals are
different (i.e., tribromomethane is still removed to the greatest
extent and chloroform to the least in both columns). The same
comparison holds for the chlorobenzenes and the aromatic hydro-
carbons, an indication that biological processes also may be
important in the removal of many trace organics.
Using old GAC, the phthalate esters again show an inconsis-
tent behavior, which differs when fresh GAC is used in that
dimethyIphthalate appears to be removed. The effluent concen-
trations of alkylated benzenes seem to be lower when old GAC is
used, suggesting that leakage was reduced during prolonged use
or that biological factors are important in the removal of
trihalomethanes.
SUMMARY AND DISCUSSION
For the removal of organic materials in the Water Factory 21
treatment system, GAC is one of the important processes. It is
not the only effective process for this purpose, however, and
lime treatment, stripping, and reverse osmosis also are relied
upon. The efficiency and reliability of GAC treatment must be
assessed with respect to its effectiveness as part of an overall
treatment system. It constitutes the main process for removal
of some organics, a collaborative process for others, and an
ineffective process for still others. For many of the materials
not effectively removed by GAC, stripping has proved to be an
efficient and relatively inexpensive complementary process.
Full-scale GAC systems can be operated continuously with a
minimum of operational problems. Higher efficiencies of organic
475
-------
carbon removal can be obtained by more frequent regeneration than
is generally practiced at the Water Factory 21. The requirements
of the Orange County Water District system necessitated regeneration
of about 14 percent of the carbon at 40- to 70-day intervals
during period 2, when an inefficiently treated trickling filter
effluent was received. After a well-treated activated sludge
effluent was received, carbon regeneration once every 6 months
proved sufficient.
Biological activity in GAC has proved quite effective for
the removal of a significant portion of the influent organic
materials measured as COO; it was responsible for removal of
roughly 30 percent of the influent organics during phase 2 and
20 percent during phase 3. Biological activity also may be asso-
ciated with the removal of a significant fraction of mar»y trace
organic materials. Whether these are decomposed biologically or
simply sorbed by the microorganisms attached on GAC itself was
not investigated. This phenomenon deserves further study.
The present study also has indicated the great difficulty
of detecting analytically significant differences in the removal
of trace organic materials. This is attributed to the great
variability of influent concentrations in the full-scale system
and the general lack of sufficient analytical precision when
concentrations are near the analytical detection limit. For this
reason caution should be exercised when regulations based upon
removal efficiencies of such materials are established. COD
removal is a good indicator of GAC performance and of the need
for carbon regeneration, and it can be readily evaluated by the
Water Factory personnel. However, difficulties could be expected
at water treatment plants where COD concentrations are near the
detection limit of 1-2 mg/1. Suitable GAC performance indicators
need to be evaluated further.
ACKNOWLEDGEMENTS
This study was supported by the Wastewater Research
Division, U.S. EPA, through grant EPA-S-803873; the Office of
Water Research and Technology, U.S. Department of the Interior
through grant 14-34-0001-7503; and the California Department of
Water Resources through grant B52353. Appreciation is extended
to James Graydon, Joan Schreiner, Kenneth Sutherland, Lewis Yee,
and Huong Nguyen for their assistance in sample analyses.
476
-------
REFERENCES
1. McCarty, P.L. and Reinhard, M. Statistical Evaluation of
Trace Organics Removal by Advanced Wastewater Treatment.
Ann. Conf. WPCF, Anaheim, Calif. (Oct. 3, 1978).
2. Randtke, S.J. and McCarty, P.L. 1979. Removal of Soluble
Secondary Effluent Organics. Jour. Envir. Engrg.
Div.-ASCE., 105;EE4;727.
3. Leo, A., Hansch, C., and Elkins, D. 1971. Partition Co-
efficients and their Uses. Chem. Rev., 71:6:525.
477
-------
EFFECT OF PRECHLORINATION ON ACTIVATED CARBON
ADSORPTION
Richard Sander
INTRODUCTION
Recently there has been an increasing interest in the
problem of organic contaminants in water, especially with the
chlorinated organic compounds which may be carcinogenic (1). We
know now that the concentration of chlorinated organic compounds
in drinking water should, in general, be minimized.
One of the methods recommended for the improvement of drink-
ing water is the removal of organic compounds by granular acti-
vated carbon (GAC) filters. This is because GAC has excellent
removal efficiency for many undesirable substances and is a
simple and convenient unit operation.
First Column Test
In a research program sponsored by the Ministry of Research
and Technology of the Federal Republic of Germany, a continuous-
flow water-treatment pilot plant was installed at Stuttgart.
Efforts were made to find a process scheme for combined
removal of haloforms, other chlorinated organics, and ammoni/a/
using controlled chlorine dosage in'combination with activated
carbon filters. A simplified flow diagram of the pilot plant is
shown in Figure 1.
After dosing sodium hypochlorite solution and alum directly
into the raw-water pipe, 3 m per hour of Neckar River water
were pumped into the sludge-blanket clarifier, which was operated
with an average retention time of 2.5 hours. The clarifier
effluent was then passed through a dual-media gravity filter
filled with sand and anthracite coal, followed by two GAC fil-
ters which were operated in parallel. Each adsorber consisted
of acrylic tubes, 10 cm in diameter, with a total depth of
2.4 id of granular activated carbon. Two carbon types F300 and
LSS were used for the tests. The filtration rate was 7.5 m/h
afod the total empty bed contact time was 20 minutes. Samples
were taken about once a week from raw water, clarifier effluent,
filter effluent, and carbon adsorber effluents.
478
-------
-J
vo
ALUM.
0
CHLORINE ^i
rf
X?
5mg/l
— — ^
RAW WATER
/
k_^
X^ x^1
3m3/ h
\^i
^
—
, X<]
Y
Sludge Removal ^
0 Rotameter
^t^j Sample Port
s-\
V .,
8m3 / s'
— r- — yj
"//
* 1 rtiiAi
Iyyyyy DUAL
Anthracite F'LTRA
'/////.
MEDIA
TtON
^ / Owftow NNNNN
\ ^
$
1 0*7,5 m/h
Sj_
^T
-T —» 100*
I LSS
I _^
T
GRANULAR
— ACTIVATED
CARBON
Figure 1. Schematic of Stuttgart Pilot Plant for the Treatment of
Neckar River Water
-------
In the first run, from the beginning of August 1977 to
October 1977, conventional treatment with breakpoint chlorina-
tion, using chlorine dosages of 20 mg/1 was applied.
In Table 1 the roost important analytical data for raw water
and clarifier effluent are summarized, covering the first period
of the test.
Table 1
Average values of DOC, UV Absorbance, Ammonia, Haloforms
and TOC1 during the First Test Period (Breakpoint
Chlorination)
Parameter
DOC [mg/1 )
^254 nm 'm~1]
NH+ -N [mg/1]
2CHCl3+CHBrCl2
[yg/r]
TOCl [yg/1]
Raw Water
4.9
11.8
1.2
0.5
100
Clarifier Effluent
4.4
8.8
0
48
660
Breakingpoint chlorination and subsequent clarification result
in a 10 percent removal of DOC and in 25 percent reduction of
UV absorbance. As expected, ammonia is completely removed by
breakpoint chlorination, but the concentrations of haloforms and
Total Organic Chlorine (TOCl) increase significantly from an
average of 0.5 to 48 ug/1 and from 100 to 660 ug/1, respectively.
The relative performance of the two different carbon filters
in terms of DOC and UV absorbance removal as a function of
throughput, are shown in Figure 2.
It can be clearly seen that there are considerable differ-
ences in the performance of the two carbon systems. While GAC
type F300 still removes about 50 percent of DOC at the end of
the run, the DOC reduction of LSS only amounts to 30 percent,
which corresponds to a breakthrough of 70 percent. Based on the
measurement of UV absorbance, the removal of organic matter was
quite similar to that of DOC reduction, as shown in the lower
part of Figure 2.
480
-------
*• GAC LSS
1234
Throughput in m3/! AC
•• GAC LSS
•* GAC F300
5
234
Throughput in m3/! AC
Figure 2. DOC and UV Absorbance Breakthrough Curves for Carbon After
Breakpoint Chlorination and Clarification of Neckar Water
481
-------
Table 2 contains the data for the concentration of the
total organic chlorine compounds after breakpoint chlorination,
determined by the pyrohydrolysis technique (2). It can be seen
that the fairly high initial concentrations are reduced to the
same extent as the DOC> and we can observe the same differences
between the two carbons* i.e., a better efficiency for the F300
carbon.
Table 2
reduction of TOC1 Concentration in Breakpoint Chlorinated
and Clarified Neckar River Water using Granular
Activated Carbon
Throughput
Ratio*
ar/in
140
360
860
1440
2660
3670
4180
5040
Clarifier Effl.
lug/i]
890
740
600
660
1140
400
330
510
GAC 1
lug/1]
-
-
-
-
-
330
290
270
L.SS
% of
Reduc-
tion
-
-
-
-
-
82
88
53
GAC
lug/i)
-
-
-
-
-
200
240
190
F300
% of
Reduc-
tion
-
-
-
-
-
50
73
37
*m water treated per m GAC bed volume
Figure 3 illustrates a striking contrast to these findings;
carbon F300 is less effective in removing haloforms.
One can see that the chloroform concentration in the
effluent of F300 reaches that of the influent at the end of the
run, while at the same time the effluent concentration of LS5
increases only to about 70 percent of that of the influent.
Apparently the bromine-containing bromodichloromethane is more
effectively reduced by both carbon types than is chloroform.
482
-------
o
o
GAC LSS
1234
Throughput in m3/! AC
GAC F300
CHCI3
1234
Throughput in m3/1 AC
Figure 3. Chloroform and Bromodichloromethane Break-
through Curves for Carbon, after Breakpoint
Chlorination and Clarification of Neckar Water
483
-------
Batch Tests
To confirm the results obtained with the column studies and
to get further information on the influence of chlorine concen-
tration on the removal of organics by activated carbon, a series
of laboratory batch tests were conducted using simulated natural
waters. Different dosages of sodium hypochlorite were added
to a solution of humic acid extract from the Neckar River in
Karlsruhe tap water, spiked with 2 mg/1 ammonia. After 24 hours
reaction time the solutions were dechlorinated with sodium sul-
fite and analyzed for DOC, UV absorbance, TOC1, and haloforms.
The humic acid used was extracted from Neckar River water
by a method developed by Rudek and Jekel (3). All experiments
were conducted at room temperature and at a pH of 7.5.
The effect of increasing initial chlorine concentration on
DOC, UV absorbance, and organic chlorine compounds is shown in
Figure 4. As expected, chloroform and TOC1 increase markedly
with chlorine addition while the DOC of the humic acid solutions
remains almost constant and the UV absorbance decreases 50
percent.
Batch adsorption tests were performed by adding different
doses of ground carbon F300 or LSS to a series of 2-liter
bottles containing the same solution. The sealed bottles
were placed on a magnetic stirrer for 48 hours, after which time
equilibrium concentrations of DOC, TOC1, UV absorbance, and halo-
forms were determined. The pH of the chlorinated humic acid
solutions was not affected significantly by the adsorption
process.
Adsorption Isotherms
In Figure 5, logarithmic plots of the isotherm for the
removal of DOC from the reaction mixtures by F300 and LSS are
compared.
Obviously, the data show very little influence of the ini-
tial chlorine concentration since all points can be described
by the same curve. This might be due to the fact that, in
chlorine treatment of humic acid solutions, neither the splitting
of high-molecular-weight molecules nor the production of more
soluble matter by oxidation processes takes place. Consistent
with the results of the first column test, carbon type F300
exhibits a greater capacity for the humic acid than type LSS.
The slopes of the isotherms indicate multi-component adsorption,
with a small fraction of fairly well adsorbable substances
producing the steep slopes at high equilibrium concentrations.
These problems are discussed in more detail in these proceedings
by Frick (4), using the same experimental data.
484
-------
o>
(26rog/l Humic Acid, 2ng/l AnnTonia, pH 7.5, 20"C)
' PAC LSS
100
O 50
I -
x
10
Humic Acid
- 7 mg I Chlorine
• 31 mgil
• 56mg
-------
Isotherms for the adsorption of chlorinated organic com-
pounds based on the measurement of TOC1 are plotted in Figure 6.
Again, carbon F300, which has proved more effective for
the removal of DOC, showed a higher capacity for chlorinated
humics than LSS. The isotherms show increasing slopes at high
final concentrations, i.e., low carbon dosages, thus indicating
also a strongly adsorbable fraction. At low equilibrium concen-
trations the curves for different chlorine dosages converge.
Apparently, chlorination of humic acid solutions only results
in the formation of chlorinated humics yet does not change the
adsorptive properties of these compounds significantly.
In Figure 7, UV absorbance has been used as an indicator of
organic matter in the solutions. As a result of prechlorination,
the isotherms show steeper slopes at higher initial chlorine
dosages indicating less adsorption of UV-absorbing organics
onto carbon. Because there had been almost no influence of the
chlorine concentration on the DOC isotherms, this effect might
be due to the fact that chlorine will react preferably with a
small fraction of humics which are highly effective in UV
absorbance and strongly adsorbable. Apparently the increasing
slopes at high equilibrium concentrations are attributable to
less reactive but well adsorbable matter.
The isotherms for the adsorption of chloroform are shown
in Figure 8. As expected from the first column test, carbon LSS
exhibited higher loadings than F300. These somewhat surprising
results can be explained by assuming that competitive adsorption
takes place at the carbon. While strongly adsorbable compounds
in the humic acid solution account for the very effective DOC
removal by carbon F300, this fact inversely lowers the extent of
adsorption of less adsorbable substances such as chloroform.
Second Column Test
Isotherms obtained under identical conditions using the
same test solutions for the two carbons can be quickly and
conveniently compared with each other to reveal their relative
advantages. However, granular carbon columns are dynamic sys-
tems, thus, not only are the equilibrium adsorption properties of
the carbon important but also the adsorption rates. Furthermore,
it is not possible to use batch tests for measuring the potential
effect of biological activity which may occur in a column.
Therefore, in most cases carbon performance has to be determined
by pilot tests.
In a second run, the pilot plant was operated from January
to October 1978 under almost the same conditions as in the first
run. The chlorine to ammonia ratio was in this case held con-
stant at 5:1 by weight, thus producing chloramine only. This
486
-------
I
ai
20
100
TOO
10OO
Figure 6.
Batch Adsorption Isotherms of TOC1 from
a Chlorinated Solution of Humic Acid
0.5
- PAC LSS
« • Humic Actd
• • • • 7 mg, I Chlorine
* « • «31 mg/l •
• • -56mg-l
TE
.5
0.1
0.05
Q02
0.51- PAC F300
0.1
0.05
0.02
10
20
UV Absortaanct mm
-1
50
Figure 7.
Batch Adsorption Isotherms of UV Absorbance
from a Chlorinated Solution of Humic Acid
487
-------
0.1
0.01
• 7 mo I Chtarm*
A 31 mg I
« 56 ma I
10
WO
Figure 8. Batch Adsorption Isotherms of Chloroform
from a Chlorinated Solution of Humic Acid
1.0 •
I
I
Combined Chlorine
10
Throughput in m'/l AC
w
S
&
OS
Free Chlorine
Chlorine
GAC LSS
GAC F300
0 5 X>
Throughput in m'/l AC
Figure 9. DOC and UV Absorbance Breakthrough Curves for Carbon
after Breakpoint Chlorination Compared with Breakthrough
after Chlorination below the Breakpoint
488
-------
was in contrast to the first pilot test run where breakpoint
chlorination had been used.
Average values of plant operation data are shown in Table 3.
While DOC and UV absorbance removal by chlorination and clarifi-
cation vary only slightly from the results of breakpoint chlori-
nation (shown in Figure 1), ammonia concentration only decreases
from 1.4 to 0.9 mg/1 in the clarifier effluent. As expected from
the batch tests, the concentration of chlorinated organic com-
pounds is markedly lower than the corresponding values for the
first test period.
Table 3
Average values of DOC, UV Absorbance, Ammonia, Haloforms
and TOC1 during the First Test Period (Chlorination
below the Breakpoint)
Parameter
DOC [mg/1]
254 nm
NH+ [mg/1]
-------
during the second test was indicated by the complete disappear-
ance of ammonia and oxygen from the effluent within a few weeks.
The corresponding TOC1 values are presented in Figure 10.
Obviously more reduction of chlorinated organic compounds is
achieved by GAC F300 than by GAC LSS. Depending on factors
like concentration of combined chlorine and concentration of
organics in the raw water, the TOC1 of the clarifier effluent
exhibits considerable variations throughout the period of study.
This is also reflected in the effluent of the carbon columns.
In Figure 11, chloroform concentrations and relative perfor-
mances of the applied carbons are plotted against the throughput.
In contrast to the findings mentioned above, but in accordance
with the results of the first run and with the isotherm tests,
the efficiency of GAC F300 is lower than of LSS. A somewhat
surprising observation is the existence of periods where the
effluent concentration of the carbon beds exceeds the influent
concentration. These phenomena are attributable to competitive
adsorption processes between chloroform and organic matter in the
water. This effect is especially strong in the case of carbon
type F300 which exhibits such effects at low influent concentra-
tions of chloroform. Presumably chloroform is displaced by humic
substances; carbon type F300 was shown to be very effective for
the removal of total organic matter.
CONCLUSIONS
The conclusions which can be drawn from these results are
as follows:
The formation of chlorinated organic compounds can be
minimized by chlorine dosages below the breakpoint,
resulting in chloramines only.
- Based on the measurement of DOC and TOC1 isotherms,
the effect of initial chlorine concentration on the
adsorption of humic material onto activated carbon is
almost non-existent. The better efficiency observed
in the second pilot test using chlorine concentrations
in small amounts below the breakpoint is due to
biological activity within the carbon filters.
- The relative efficiencies of the carbons, as indicated
by column tests, correspond to the results of the
isotherm test. The data show that it is very impor-
tant to Jcnow the objective for which the carbon is to
be used in order to choose the best quality carbon.
490
-------
100
80
2 60
£
| 40
20
0
Clarifter Effluent
QAC LSS
GAG F300
10 Throughput in ma/1 AC
Figure 10. TOC1 Values at Various Treatment Stages after Chlorination
of River Neckar Water below the Breakpoint
o
u
Ctorifi«r Effluent
GAC LSS Effl.
GAC F300 Effl.
10
Throughput in m*/l AC
(10,25)
10
Throughput in m'/l AC
Figure 11.
Chloroform Concentrations at Various Treatment Stages
and Chloroform Breakthrough Curves for Carbon after
Chlorination below the Breakpoint
491
-------
The results clearly indicate that a combination of low
dosage chlorination and biological action within granu-
lar activated carbon filters will induce the removal of
ammonia and organics and will lead to a much better
quality of the water than when using breakpoint chlori-
nation.
REFERENCES
1. Harris, R.H. 1976. Carcinogenic Organic Chemicals in
Drinking Water. In: Tourbier, J., et. al. Biological
Control of Water Pollution, University of Pennsylvania
Press.
2. Kuhn, W. and H. Sontheimer. 1973. Several Investigations
on Activated Charcoal for the Determination of Organic
Chloro-Compounds. Vom Wasser, 15.
3. Rudek, R. and M. Jekel. 1976. Extraction of Humic Acids
from River Water. Unpublished, Engler-Bunte-Institute of
the University of Karlsruhe.
4*
pfhfk- B' J,97?- Prediction of Multicomponent Adsorption
Behavior; Equilibrium Apects. pp. 256-269 in this Symposium,
492
-------
Removal of Purgeable Organic Chlorine Compounds
by Activated Carbon Adsorption
B. Fokken
and
R. Kurz
The increasing demand for drinking water in the city of Cologne
could no longer be met solely by ground water resources. A group
of wells was placed in a meander of the Rhine River south of
Cologne in order to draw bank-infiltrated ground water from the
river. Geohydrological conditions at that location assure that
the pumped raw water is mainly bank-infiltrated water from the
Rhine River. The relatively long underground passage, of about
600 m, provides significant pretreatment of the river water.
Biological Growth
After commissioning the wells, it was noticed in 1969, that the
number of microorganisms which form a gelaULnoas substance,
strongly increased in both wells and pipes.
In order to design treatment facilities, tests were performed
with a pilot filter. The experimental results showed that there
was a substantial increase in filter head loss after a short
time of filter run due to the slimy biological growth. Further
investigations became necessary to identify and safely control
the problem-causing bacteria.
The bacteria were identified to be of the following species:
Sphaerotilus natans or Sphaerotilus discophorus, as well as a
variation of Pseudomonas fluorscens.The first two organisms
appear to grow rapidly in activated carbon filters under typical
water treatment conditions. They cause a typical slimy coating
of strands as shown in Figure 1.
In the experimental stage, chlorine was applied to kill the
bacteria. While a concentration of 3 g/m chlorine could not
stop the growth,.complete kill was achieved with concentrations
above 5 to 6 g/m chlorine.
The existing fibre structure was almost completely destroyed so
that the remaining tiny particles which entered the filter did
not cause any measurable increase in filter head*loss. After
493
-------
Figure 1. Bacteria in Wells for Reclamation
of Bank-infiltrated Ground Water
UV-Extinction
254 nm in nv1
DOC(g/m3)
CHCI,
CHCUBr
CHCI Brz
CHBr,
CCI«
CfCltH
C,-CI4
River Rhine
8,5 ±1,0
4,4 ±0,6
6,5 ±3,4
0,47 ±0,6
O,3O±0,2
0,31 ±0,3
Bankinfiltratec
before Chlorination
1,4 ±0,2
1,1 ±0,2
0,50 ± O,1
O.16 ± O,1
O,6O ± O,1
1,6O ± O,1
1 Ground Water
after Chlorination
1,3 ±0,1
1,1 ±0,2
0,64 ±0,1
1,6 ±O,3
2,8 ± O,6
2,9 ± O,8
O,16 ± 0,1
0,61 ± O,1
1,6 ± 0,1
Halocarbon Concentrations in mg/m3
Table 1.
Influence of Bank Infiltration and Chlorination
on Concentrations of Different Substances
494
-------
treatment facilities with activated carbon filters had been
installed in the water works of Hockirchen, the system was
chlorinated with the same high concentrations that were
administered during the experimental tests. When operating
the plant it turned out that the growth could also be prevented
by smaller chlorine concentrations in the raw water.
Under today's delivery and treatment conditions, it is sufficient
to have a chlorine concentration of 1 g/m3 at the beginning of
the supply system. This chlorinate reacts with the precursor
organic substances and causes an increased formation of chlori-
nated organic-compounds. The organic carbon concentration is
about 1.1 g/m . Tests showed that bank infiltration removes
part of the chlorinated organic compounds that occur in the
river water. Chlorination, however, produces these compounds
again in even much higher concentrations.
Bank Filtration
Data characterizing the Rhine River water and the bank filtrate
before and after chlorination are shown in Table 1. One can
clearly see the removal capacity of the filtration path with
regard to the UV-extinction and dissolved organic carbon (DOC).
Furthermore it can be stated that chloroform and carbon tetra-
chloride are extensively retained by the subsoil, while tri- and
tetrachlorethylene increase. It is remarkable that due to
chlorination the brominated organics monobromodichloromethane,
dibromochloromethane, and bromoform are formed. This can be
explained by the bromide concentration of about 0.2 g/m3 in the
bank filtrate: the added chlorine reacts with bromide ions to
form bromine, which then reacts rapidly with the existing
organic compounds.
In a series of bench tests, 2 g/m of chlorine were added to
bank infiltrated ground water; a second series of tests were con-
ducted under identical conditions, except that bromide was dosed
to a concentration of 1 g/m . Figure 2 shows the corresponding
curves of the haloform development. The curves with solid sym-
bols represent the haloform development provided by a natural
bromide content of 0.2 g/m . The addition of bromide promotes
the formation of bromoform, but retards the development of bromo-
chloromethane compounds. Bromochloromethane compounds develop
better in water with a low concentration of bromide, but the
chloroform concentration (not shown in Fig. 2) varies little in
both series of tests. After a 24 hour reaction period, the halo-
form concentration is 11 mg/m with natural bromide content and
20.5 mg/m if bromide is added. This result indicates that an
intensified formation of haloform is to be expected if bank
infiltrated ground water has an increased concentration of
bromide.
495
-------
Concentration of
Haloforms in mg/ms
o
CHBrCI^ i,2g/m*Br
2 4 6 8 10 12 14 16 18 20 22 24 26 28
Reaction Time in Hours
Figure 2. Influence of Different Bromide Concentrations
on the Formation of Haloforms
496
-------
Adsorption of Organic Substance and Chlorinated Organic Compounds
Because of the high chlorine loading and the content of odour-
and taste-intensive substancesf treatment of the bank-infiltrated
ground water with activated carbon is required. In a research
program sponsored by the Ministry for Research and Technology of
the Federal Republic of Germany, the adsorption of haloforms by
activated carbon in the presence of other organic substances was
studied. Three kinds of activated carbons used in filtration
plants were tested: ROW-08-S, LSS, and LEV 693. As the iso-
therms of the UV-absorbing organic sustances in Fig. 3 show, the
activated carbon ROW-08-S is best qualified to remove these
substances. A maximum load of 0.3 g organic substances in water
(OSW) per g activated carbon may be achieved for ROW-08-S with
extrapolation of the isotherms to an inlet concentration of
1.3 g/m . In LSS the maximum load is 0,16 g OSW per g activated
carbon, under the assumption that an extinction unit of one m~
at 254 nm is equated with a concentration of 1 g organic
substances per m .
The breakthrough curves of the filters in Fig. 4 demonstrate a
behaviour which is similar to that of the isotherms. That means
it should be possible to describe the breakthrough curves .of the
filter for UV-active organic substances by the adsorption iso-
therms. Martin (2) provided a method based on the assumption that
the filter is a totally mixed vessel in equilibrium. However,
the breakthrough curves for a filter in equilibrium which were
achieved by this approach have to be adjusted to the actual
observations by a system calibration factor, since kinetics also
play a part in an ordinary carbon filter. Preliminary experi-
ments with a pilot filter showed that the breakthrough curves at
several layer depths could indeed be calculated by an adjustment
at only one point. As had been expected, the factor was the
smallest with the thinnest filter layer.
Fig. 4 also shows that the equilibrium data allow an approximate
calculation of the breakthrough curves. So far, however, it is
still unknown which parameters could theoretically define the
appointed factor; i.e., to what extent does kinetic data of both
the carbon and the substances enter this factor. This adjustment
is valid only for collective parameters like DOC, UV, and dis-
solved Organic Chlorine (DOC1) with a sudden abrupt breakthrough.
Haloforms differ widely in their solubility in water. The more
bromine they contain, the less soluble these compounds will be
in water. This characteristic is also reflected in the equilib-
rium isotherms of the activated carbon ROW-08-S, shown in Fig. 5.
At the same equilibrium concentration, the load of this activated
carbon is up to 8 times higher with bromoform than with chloro-
form. If the adsorption in the filter is mainly determined by
the position of the isotherms, it has to be expected that
497
-------
0,1
0,01
LMdingingOSW/g
Activated
MOW MS
LBV I
1,1 *7
0,12 0,97 2»
0,38 1,M) 3»
0,1
0,5 1,0
Equilibrium Concentration
ln«OSW/m'«IMar
Figure 3. UV-equilibrium Isotherms of Organic Substances
in Water for Different Activated Carbons
1,0
tt 40 M *0 TO
Throughput in lO^m3 W«t«r/m3
Activated Carbon
Figure 4. UV-breakthrough Curves for Different
Activated Carbons
498
-------
chloroform breaks through prior to bromochloromethanes. This is
confirmed by the measured filter breakthrough curves, shown in
Fig. 6. Note the areas (lobes) of the breakthrough curve above
C/C = 1. The chloroform lobe can be accounted for with the
exp?anation that chloroform is displaced by more strongly
adsorbed substances, so that higher concentrations appear in the
filter effluent than in the filter influent. These lobes are of
much smaller magnitude with monobromodichloromethane and dibromo-
chloromethane and might also be traced back to the reaction of
excess bromine and chlorine with the activated carbon. With
tetrachlorethylene, the filter still removes 75% of-the applied
concentration after a load of 60,000 m water per m activated
carbon. This is mainly due to the low aqueous solubility and
the nonpolar nature of this compound. It also turns out that
the slope of the filter breakthrough curves can be determined by
the position of the isotherms.
The lobe of the chloroform concentration is believed to be caused
mainly by displacement. Fig. 7 shows breakthrough curves of
chloroform for various filter layers. It can be seen that, as
the filter depth is increased, the breakthrough of chloroform
indeed occurs later, but the lobes are much more distinct. Based
on these results it can be concluded that relatively high chloro-
form concentrations (i.e., almost twice as much as observed in
the filter inlet), may be encountered in the filter discharge.
To avoid this problem, the filter run length of those filters
with large bed depths, as typically used in major filtration
plants, should be controlled by the breakthrough of chloroforms.
Merk (3) has proposed that the maximum lobe can be calculated
for binary mixtures by simulations based on the isotherms of
both the displaced and the displacing substances, since kinetic
effects are negligible. However, the displacing substances are
generally unknown in filter breakthrough curves for a natural
raw water, so that at best' an approximate calculation might be
possible under certain assumptions. In binary mixtures the
displacing substance is characterized by a better adsorption
behavior so that higher loads compared to the displaced sub-
stance can be reached at a given equilibrium concentration.
Different qualities of activated carbon show different break-
through characteristics of chlorofora^as can be seen in Fig. 8.
Type LSS, for example, which provide*'£ar better removal rate of
organic chlorine compounds, shows only a small lobe, whereas
higher outlet than inlet concentrations occurred with types LEV
693 and ROW-08-S. If reactivation is required when 50 percent of
the chloroform breaks through* the maximum allowable throughput
ratio is^7,000 m water per m activated carbon for LEV 693*
14,600 mVm for ROW-08-S, and 20,000 nT/m for LSS. These rela-
tive capacities reverse with simultaneous removal of UV-active
substances? i.e., LSS has a much smaller capacity for UV-active
organic substances than ROW-08-S.
499
-------
1,0
0,1
O.O3
o,oJ
LuadloB in mg Hatoforma/g
Actfv»tad Carbon
CMBr.
0,1 04 0,» 1,0 2,0 0.0 10,0
Cquilibrim Conoantratii
Figure 5. Equilibrium Isotherms of Haloforms for the
Activated Carbon, Norit ROW-08-S
Throughput in 1(>3m3
Activated Carbon
Figure 6. Breakthrough Curves of Different Haloforms
and Tetrachlorethylene for the Activated
Carbon Norit ROW-08-S
500
-------
2.0
1.5
1.0
0,5
250 »OO 7*O K>OO 12SO 1*OO 175O
Throughput in m* Water
Figure 7. Breakthrough Curves of Chloroform in Different
Layers of an Activated Carbon Filter Filled with
NORIT ROW-08-S
2,0
1.5
CMC).
V_.tOm/h
LBVM3
1,0 /
io ao M 40 M «o TO
Throughput in lO^m' W»ter/m3
Activated Carbon
Figure 8. Breakthrough Curves of Chloroform for Several
Types of Activated Carbon
501
-------
Presently we do not know of a carbon which effectively removes
haloforms as well as total organic substance (TOC).
Judging by the position of the isotherms for dibromochloromethane,
shown in Fig. 9, it can be said that LSS adsorbs this compound
much better than the activated carbons ROW-8-S and LEV-693.
Furthermore, it can be inferred from the isotherms that certain
maximum loads can be reached with a filter in equilibrium. Under
the existing conditions an effect of displacement occurs in
ROW-08-S, indicated by the negative slope of^he isotherm. The
corresponding breakthrough curves in Fig. 1 demonstrate that
a lobe also occurs in the filter effluent of ROW-08-S, which is
not observed with type LSS. Using previously made kinetic
measurements, filter breakthrough curves of haloforms probably
can be described by film diffusion if the position of the
equilibrium isotherms is taken into consideration. This assump-
tion is strengthened by the fact that the available surface area
(4) of the individual activated carbon correlates with the
existent isotherms.
An investigation of the maximum possible filter load, which can
be calculated from the breakthrough curves presented, revealed
that ROW-08-S and LEV-693 nearly attained the equilibrium load;
in the case of LSS, the equilibrium load deviated extremely from
the actual load of LSS. If, on the other hand one compares the
load of dibromochloromethane calculated from the filter 4ata with
the equilibrium loading determined in laboratory tests, then it
appears that the smaller the final load, as calculated from the
equilibrium data, the earlier this final load is reached. For
example, the load in LEV 693 filter is 0.18 g/kg activated carbon;
the equilibrium load, however, could reach 0.36 g/kg. In an LSS
filter, the possible equilibrium load is 1.42 mg/kg, but the load
reached in the filter is only 0.37 mg/kg.
The only explanation for these results is that there is a mixed-
solute system in equilibrium as far as the isotherms are con-
cerned, whereas fronts of loaded and unloaded zones prevail in
the filter. Hence, an intensified adsorption of various compounds
can partially take place due to their better kinetics, so that in
the various filter layers a higher loading can be achieved than
could have been expected based on equilibrium assumptions.
As is shown in Fig. 11 for UV-absorbance, the processes in the
filter have to be described by both the equilibrium and the
kinetics. Hence, it appears that at a high filtration rate of 20
m/h, the breakthrough occurs earlier than at 10 m/h. The worse
breakthrough behaviour at 5 m/h may be explained by kinetic
effects; i.e., substances that are slowly but strongly adsorbed
may have displaced UV-active substances.
502
-------
0.2
O.I
0,0*
O.O1
LMdln« m me CMBr.CI /g
ActivMatf Carbon
O LM
0,1 O.2
1,0 «.0 10
Equilibrium Cenewttnrtlon
InmgCHKr.CI/m'WatOT
Figure 9. Equilibrium Isotherms of Dibromochloromethane
for Different Activated Carbons
1.0-
0,5
LCV
*-ROWO«»
LS»
1O 9O
ftO «0 TO
Throughput in 1o3m3 W«ter/m3
Activated Carbon
Figure 10. Breakthrough Curves of Dibromochloromethane
for Different Activated Carbons
503
-------
3. 2Ow/h
Wm/ti
1O
2O SO
4O 90 «0 TO
Throughput in 1o3m3 Witer/m?
Activated Carbon
Figure 11. UV-breakthrough Curves for Different
Filtration Rates
2,0
1.S
1.0
0,3
C'c.
5m/h lOm/h lOm/h
10 SO 30 40 SO «0 70
Troughput in 1O'm' WM*r/m*
Activated Cw*on
Figure 12.
Breakthrough Curves of Chloroform for
Different Filtration Rates
504
-------
This theoretical assumption is confirmed by the breakthrough
curves of chloroforms in Fig. 12. Here, the chloroform is also
the first which breaks through at a filtration rate of 5 m/h,
whereas the various slopes of the breakthrough curves at 10 and
20 m/h can be related to the increased filtration rates. The
lobe of the discharge concentration of chloroform which is
reached by an increased throughput can again only be explained by
kinetics. The results with the various filtration rates are
still to be analysed in detail, as existing explanations of their
behavior are not completely satisfying.
Summary
Tests were performed on the adsorption of organic substances in
activated carbon filters. It was found that the removal of
purgeable organic compounds is possible by using ordinary
activated carbons as used in major filtration plants. However,
due to displacement effects, higher concentrations ot halotorms
than had been found in the raw water, especially chloroform, may
show up in the drinking water. It was also observed that
activated carbons, which guarantee an optimum removal of the
UV-absorbing organic substances, are not necessarily most
effective in removing haloforms. Due to the short filter run
because of chloroform breakthrough, it is necessary to change
the treatment scheme so that the formation of chlororganic
substances (notably haloforms) can be diminished or avoided
before the water enters the activated carbon filters. To "solve
these problems further research will be required.
REFERENCES
1. Mitteilungen der Landesanstalt fur Wasser und Abfall
Nordrhein-Westfalen; Ergebnisse der Gewasseriiberwachung
durch Wasserkontrollstationen und das Laborschiff "Max
Truss"; April 1977 - March 1978
2. Strack, B., and H. Martin. Experience with fluidized bed
furnace at Benrath treatment plant. pp. 658-667 in this Symposium.
3. Merk, W. Konkurrierende Adsorption verschiedener Wasser-
inhaltsstoffe in Aktivkohlefiltern. Dissertation,
Universitat Karlsruhe; 1978
4. Holzel, G. Laboratory activated carbon test methods for
water utilities. pp. 270-284 in this Symposium.
505
-------
REMOVAL OF HIGHER MOLECULAR WEIGHT ORGANIC COMPOUNDS
BY THE GRANULAR ACTIVATED CARBON ADSORPTION UNIT PROCESS
A. A. Stevens, D. R. Seeger
J. DeMarco and L. Moore
ABSTRACT
The granular activated carbon adsorption unit process in
drinking water treatment typically removes purgeable organic
compounds for time periods on the order of a few weeks. Experi-
mental evidence indicates that less volatile compounds of
generally higher molecular weight than the purgeable fraction,
but still detectable by gas chromatography, are efficiently
removed for longer periods. Field data substantiate this.
Explanatory mechanisms may include stronger adsorption affini-
ties or biodegradation. Non-gas chromatographable, higher
molecular weight materials such as humic acids, as measured
by Total Organic Carbon (TOO or trihalomethane formation
potential revert to lower removal efficiencies, Biodegradation may be
responsible for a continued long term removal of a fraction of these
materials.
INTRODUCTION
A summary of the available literature regarding the useful-
ness of the granular activated carbon (GAC) unit process for
removing specific organic compounds from drinking water supplies
has been compiled by Symons (1). Although some acknowledged
analytical weaknesses existed in the overall data base at that
time, the conclusion was confidently drawn that GAC is correctly
described as a broad spectrum adsorbent for removal of organic
compounds from water. Symons1 review acknowledged a spectrum of
removal efficiencies for the compounds, as measured by gas
chromatography, in the studies. Further, and as a general rule,
in the studies described, compounds that eluted late during the
gas chromatographic analysis were generally better removed than
the more volatile, early eluters.
The elution order of compounds in a gas chromatograph is
closely related to volatility, and that factor is related in-
versely to molecular weight in that, within a homologous series,
506
-------
as molecular weight increases, volatility decreases. Thus, from
the gas chromatographic evidence available from broad spectrum
adsorption studies, a favorable relationship between adsorb-
ability and increasing molecular weight seems to exist. Of
course, other factors are much more important in influencing ad-
sorption, such as solubility and polarity of the compound, than
volatility or molecular weight, per se. Increasing molecular
weight within a homologous series simply influences these
parameters in a favorable way as to improve adsorption, thus
explaining in a general way the apparent relationship between
gas chromatographic retention times, molecular weight, and
adsorbability. Of course/ very high molecular weight substances
that will not pass through a gas chromatograph do not fall into
this category and, therefore, this generalization cannot be made
about that class.
On-going work by the U.S. EPA Laboratories, their grantees,
contractors, and many others worldwide has been focusing on
further investigation of the effectivenes of GAC for removal of
organic contaminants, both disinfection by-products and synthetic
organic contaminants from drinking water supplies. Early
results from some of these studies plus information available
from the literature (1), have given the U.S. EPA cause to pro-
pose the required use of the GAC unit process as the best broad
spectrum approach to control synthetic organic contaminants in
certain vulnerable supplies (2). Recent data from pilot and
full-scale adsorbers has indicated, however, that certain
volatile halogenated organic compounds, generally common sol-
vents conveniently measured at the same time as trihalomethanes,
are often present in relatively high individual concentrations
and are somewhat poorly removed by GAC or are efficiently
removed for only short perids of time.
Because of the analytical convenience of measuring these
volatile organohalides, they are monitored with greater fre-
quency and quantitative accuracy than the multitude of other
synthetic organic contaminants that can be found in certain
drinking waters (approximately 460 compounds in one tap water
sample alone) (3). Because of this, the noted early break-
through of these few individual compounds tends to cause under-
estimates of the capacity of the GAC unit process to protect
the consumer from synthetic organic contaminants in general.
Much can be determined on a laboratory scale, about the
equilibrium capacities of GAC for certain individual compounds
with or without the influence of competing species. Equilibrium
and kinetic studies, both static and dynamic, can be used to
model the adsorption behavior of compounds in continuous flow
dynamic systems. Beyond the subjects of adsorption capacities,
507
-------
kinetics, and solubility parameters, however, interpretations of
field results are always severely complicated by variations of
loadings with time, relative loads of different compounds (also
varying with time), relative concentrations of substances dif-
fering in concentration by several orders of magnitude, and
finally, biological activity within the adsorber. Thus, theo-
retical aspects of adsorption will not be addressed and this
discussion will be limited to in-practice observations inter-
preted under the described inherent limitations.
IN-PLANT EXPERIENCES
Experiences with the removal of specific "higher'molecular
weight: compounds at a number of U.S. water treatment plants
were discussed elsewhere in these Proceedings. Two utilities
will be discussed in this paper; both operate full-scale GAC
adsorbers that demonstrate particularly well, the usefulness ot
GAC as a barrier against source water contaminants of "higher
molecular weight" contrasted to the GAC's performance for the
.removal of organohalide solvent co-contaminants.
Additionally, examples of biologically derived removals
of trihalomethane precursors and two synthetic organic compounds
are presented as examples of protection provided by the GAC unit
process beyond that which might be predicted by results of
studies of the adsorption phenomenon itself.
MT. CLEMENS, MICHIGAN
The water treatment plant at Mt5 Clemens, Michigan is an
8 million gallon/day plant (32,000 m /day) employing conven-
tional alum coagulation, settling in a parallel pair of basins,
followed by GAC, with a 6.2 min empty bed contact time, in-
stalled in a sand replacement mode for taste and odor control.
Historically, the GAC has been replaced every 3 years, corres-
ponding to taste and odor breakthrough. During th* period of
the most recent GAC replacement (beginning April 1978), this
Laboratory agreed, on a technical assistance basis, to evaluate
the performance of the new GAC for trihalomethane and precursor
removal. At the same time, the opportunity was presented to do
a more extensive organic analysis to further evaluate the GAC
performance.
The analyses employed in the study were: (1) the conven-
tional purge and trap GC with electrolytic conductivity detec-
tion in the specific halogen mode for the trihalomethanes and
other volatile halogenated compounds; and (2) solvent extraction
with methylene chloride of acidified, 1 gallon (3.7 liter)
samples, followed by capillary column gas chromatography with
flame ionization detection (GC/FID) or gas chromatography with
mass spectrometric detection (GC/MS) analysis of the solvent
concentrates for measurement or detection of less volatile
organic species.
508
-------
Methyl chloroform (1,1,1-trichloroethane) is a chlorinated
solvent frequently found in this utility's source water and is
an example of the group of compounds that is poorly removed by
GAG. Figure 1 is a plot of the settling basin effluent.,(GAC
influent) and GAC effluent from a single 1 mgd (4,000 m /d)
filter unit at the Mt. Clemens plant. Note the range of concen-
trations of 1,1,1-trichloroethane found within the basin all on
the same day. This is not an analytical artifact, but reflects
the highly variable load of this compound challenging the GAC at
this plant, severely complicating any attempt at a concise
quantitative assessment of the adsorber's performance. Even
given this difficulty, the data in Figure 1 show that the
adsorber was ineffective for removal of this ".ompound on the
first day of sampling, only 8 days into the run. In fact,
the adsorber effluent exceeded the influent on three of the six
sampling occasions during the first 120 days of the run, without
doubt, reflecting high variations in loadings not detected
because of infrequent sampling.
30
O)
a.
0)
c
CO
SI
+•*
fl>
O
w
o
c
O
c
V
O
c
O
O
25 -
20 -
15 -
10 -
5 -
-50 -20 0 20 50
GAC installed
100 200
Days in service
300
337
Figure 1. 1,1,1-Trichloroethane Removal at Mt. Clemens,
4/78-2/79
509
-------
The more extensive analysis performed using capillary
column GC/FID, however, demonstrated a contrasting good perform-
ance for a group of less volatile compounds. Investigations by
capillary GC/MS provide some information as to the nature of
these compounds. Some of the compounds that were identified in
this analysis have their corresponding peaks labelled in
Figure 2. Those with an asterisk are in the solvent or are
identified as an artifact of the analytical method and should
be ignored. Most important to this discussion is the large
unresolved group of compounds whose peaks are simply labelled as
hydrocarbons; the retention time of n~c18H^8 is marked to
show that the molecular weight range of these hydrocarbons is
mostly above that of clgH3p (234). Figure 3 shows a
reproduction of the same cnromatogram above a corresponding
chromatogram generated in an identical manner from a sample of
7 week old GAC modular filter/adsorber unit effluent. The
data have been normalized on the internal standard, so, within
limits, the chromatograms can be compared with one another
quantitatively. For quantitative reference, the internal
standard theoretically corresponds to 0.25 ng/1 of anthracene.
Clearly, the GAC was effectively removing this group of high
molecular weight hydrocarbons. At this time, total organic
carbon (TOC) reduction through the filter was 43 percent.
After the May 1978 episode, the hydrocarbon contamination
did not reappear until February 1979. Recently acquired GC/FID
results indicate that at this time, the GAC was still as effec-
tive as earlier in acting as a barrier to penetration of the
filters by the hydrocarbon materials, even though the GAC was
nearly one year old and TOC reductions were found to be down to
16 percent. The recurrence of the contamination has provided an
opportunity to follow up and determine the source of contamina-
tion. This investigation is now underway.
JEFFERSON PARISH, LOUISIANA
Gas Chromatographable Compounds
The second site is at the Jefferson Parish Water Department,
Jefferson Parish, La.; this site was selected as being a partic-
ularly good example of a full-scale operating GAC filter module
demonstrating contrasting performance for removal of a light
halogenated solvent as compared to higher molecular weight
compounds. At this location, the U.S. EPA has an ongoing
research project to assess the effectiveness of GAC to remove
synthetic organic contaminants from the lower Mississippi River
water source.
The treatment scheme employs a cationic polymer as a
primary coagulant, followed by potassium permanganate addition
for taste and odor control, then lime and ferrous sulfate
510
-------
100
so-
20-
V
c
(0
W g)
11
I !
« O
V
0)
c
0)
3
c
o
45
O
a
100
300
500
700
900
1100
lOO-i
60-
C18H38
T
I200
1400
T"
1(00
1100
2000
Figure 2. Reconstructed Gas Chromatogram, Setting Basin
(Effluent) Mt. Clemens, 5/16/78
511
-------
100
•-"J
SETTLING BASIN
(EFFLUENT)
TOC=2.8mg/L
100
U1
100
400
700
1000
1300
ItOO
1900
2200
1 1 1 1 I
100 400
SEVEN WEEK OLD GAC
(EFFLUENT)
TOC=1.6mg/L
- i i i i i • i i; i !••••
100 1000 1100
TJ
«0
•o
(0
ID +*
c w
s«
10 C
a :r.
1
liOO 1900 2200
Figure 3. GAC Performance, Mt. Clemens, 5/16/78
-------
addition, upflow clarification, disinfection with combined
chlorine, and rapid sand filtration. The experimental scheme at
Jefferson Parish included the testing of a single 1 mgd (4,000
m /d) sand replacement filter-adsorber and a second, identical
module as a post filter adsorber where the adsorber unit was
plumbed in series with a remaining operating rapid sand filter.
The performance of the post filter adsorber (20 min EBCT) is the
subject of this discussion.
Table 1 lists compounds that were of particular interest
for treatment studies at Jefferson Parish at the start of the
project. For this discussion, the performance of the adsorber
for removal of 1,2-dichloroethane will be contrasted with the
removal of a group of pesticides, essentially those compounds
numbered 14 to 23 in Table 1.
Table 1
Compounds of Interest, Jefferson Parish GAC Study
1. trichloroethylene
2. p_-dichlorobenzene
3. carbon tetrachloride
4. chloroform
5. bromoform
6. bromodichloromethane
7. dibromochloromethane
8. 1,2-dichloroethane
9. vinyl chloride
10. tetrachloroethylene
11. benzene
12. toluene
13. xylenes
14. BHC isomers
15. lindane
16. heptachlor
17. heptachlor epoxide
18. chlordane
19. dieldrin
20. endrin
21. DDT isomers
22. DDD isomers
23. DDE isomers
24. atrazine
25. diethylatrazine
26. hexachlorobutadiene
27. HCB
28. toxaphene
29. methoxychlor
30. strobane
31. PCB's
32. phthalates
33. PAH's
34. bis (2-chloroethyl) ether
513
-------
1,2-dichloroethane is measured during the routine purge-
and-trap trihalomethane analysis. The chlorinated hydrocarbon
insecticides (CHI) are isolated from 6-liter water samples by
use of a solvent extraction procedure followed by glass capil-
lary column chromatography with electron capture detection.
Because the concentrations of individual CHI are very low, and
occurrences are quite variable, GAC treatment effectiveness was
most conveniently evaluated using a total CHI parameter (re-
ported in picomoles/L as the units of concentration)
Figure 4 shows the performance of a GAC post-filter ad-
sorber for the removal of 1,2-dichloroethane (DCE). Although
the influent concentration of DCE was quite variable, 1 to 24
Vg/lr a fairly smooth breakthrough curve was observed for this
compound during the 180 day run. Ten percent breakthrough was
observed at approximately 44 days, 50 percent at approximately
90 days, and equilibrium (Ceff = Cinf) at about 110 days. By
contrast, however, Figure 5 shows the effluent and influent
curves for total CHI during the same run. The curves show that
only when the influent concentration of CHI decreased to near
the detection limits did GAC effluent concentrations reach those
of the influent. This occurred during the period 45 to 64 days
into the run, again between 56 and 62 days, and once again at
100 to 106 days. Following 106 days, when influent CHI returned
to relatively high concentrations, the GAC was quite effective
through the 180th day. For the duration of the run, a removal
efficiency for total CHI of 83 + 5 percent was calculated,
again demonstrating the effectiveness of GAC as a barrier to
this group of synthetic organic contaminants. This long term
removal of CHI confirms the conclusions of Robeck, et al. (4),
that GAC beds are efficient for the removal of certain pesti-
cides even after exhaustion by other criteria.
Non-Gas-Chromatographable Compounds
The parameter "Term TTHM" is a measure of non-gas-chroma-
tographable trihalomethane precursor materials obtained by
chlorination and storage of a sample under controlled conditions
and is defined as the concentration of trihalomethanes at the
end of the storage period, simulating the effects of disinfectant
contact time. Precursors are generally considered to be aquatic
fulvic and humic acids of a molecular weight range of 500 to
50,000 or more (5).
Continuing the discussion of the same Jefferson Parish
experimental filter run described above, Figure 6 is a plot of
the influent and effluent terminal concentrations of total
trihalomethanes (term TTHM). From this pair of curves, two
important observations can be made. Firstly, 10 percent break-
through is reached quite early (18 days) in the run in contrast
again to CHI. Although, with humic acid or fulvic acid from the
same source, lower molecular weight fractions generally are more
514
-------
30
o_o Adsorber Effluent
Adsorber Influent (Sand Filter Effluent)
180
Figure 4. Performance of Post Filter Adsorber for Removal of
1,2DCE (Feb.-Oct. 1977) - Jefferson Parish, La.
240
o-cAdsorber Effluent
D-OAdsorber Influent
60 80
Days of run
140 160
180
Figure 5. Performance of Post Filter Adsorber for Removal of
Total CHI (Feb.-Oct. 1977) - Jefferson Parish, La.
515
-------
adsorbable than higher molecular weight fractions (500, 10K,
50K) (6). This 10 percent fraction of material could consist of
either very low molecular weight, soluble biodegradation pro-
ducts or very high molecular weight material. Highly soluble
molecules may not be adsorbed strongly even though they are
small enough to penetrate into the carbon grains; substances of
extremely high molecular weight may be excluded by size effects
from diffusing through the internal pores of the carbon grains
to reach vacant adsorption sites.
o—o Adsorber Effluent
o-o Adsorber Influent
20
40
60
SO
Days of run
100
120
140
160
180
Figure 6. Performance of Post Filter Adsorber for Removal of
Term THM (Feb.-Oct. 1977) - Jefferson Parish, La.
Secondly, 50 percent breakthrough is not reached until 102
days. Further, equilibrium (C. f = C ff) is not reached
during the 180 day run, but a stiady state around 70 to 75 per-
cent remaining in the effluent seems to be achieved. This kind
of result is also observed when TOC is used as a measurement
parameter and has been observed at numerous other locations to
varying degrees. This means that either a 25 to 30 percent
fraction of the humic material is very strongly adsorbed, or
that some other mechanism of removal is acting. Current think-
ing is that biological activity is an important mechanism for
this continuing removal of natural organic materials. Evidence
is available that indicates a preoxidation step, which will
render the compounds more polar/ and therefore less adsorbable,
actually enhances this long-term removal phenomenon, presumably
because the oxidation products are more biodegradable (see other
papers in these Proceedings).
Recent experiments at the EPA Laboratory investigating the
treatability of certain chlorinated volatile compounds, have
indicated that the apparent biological removal process is not
516
-------
restricted to natural humic materials, that is, to trihalo-
methane precursors. The synthetic organic contaminants chloro-
benzene and o-dichlorobenzene were found to exhibit distinct
signs of biodegradation during the treatment, presumably
enhancing the effectiveness of the overall unit process.
CONCLUSIONS
For the removal of synthetic organic contaminants, GAC is a
good, broad-spectrum adsorbent. Although the effectiveness of
GAC is marginal for some easily measured low molecular weight
synthetic organic contaminants, excellent protection is still
provided for other co-contaminants, especially in a higher
molecular weight range.
Biological activity has a likely role in the continuing
long-term removal of a 25 to 30 percent fraction of natural,
very high molecular weight humic acids, and a less well recog-
nized role in removal of some synthetic organic contaminants as
well.
REFERENCES
1. Symons, J.M., "Interim Treatment Guide for Controlling
Organic Contaminants in Drinking Water Using Granular
Activated Carbon," U.S. Environmental Protection Agency,
Cincinnati, Ohio, 55 pp. plus 3 Appendices (January 1978).
2. U.S. EPA Interim Primary Drinking Water Regulations,
"Control of Organic Chemical Contaminants in Drinking
Water," Federal Register, 4_3, No. 28, 5756-5780 (Feb. 9,
1978).
3. Coleman, W.E., R.G. Melton, F.C. Kopfler, K.A. Barone,
T.A. Aurand, and M.G. Jellison, "The Identification of
Organic Compounds in a Mutagenic Extract of a Surface
Drinking Water by GC/MS/COM." Submitted to Environmental
Science and Technology (April 1979).
4. Robeck, G.G., K.A. Dostal, J.M. Cohen, and J.F. Kreissl,
1965, "Effectiveness of Water Treatment Processes in
Pesticide Removal." Jour. Am. Water Works Assoc., 57 (2)
181.
5. Stevens, A.A., C.J. Slocum, D.R. Seeger, and G.G. Robeck.
1976. "Chlorination of Organics in Drinking Water," Jour.
Am. Water Works Assoc., 68, 615-620.
6. Snoeyink, V.L., J.J. McCreary, and C.J. Murin. "Activated
Carbon Adsorption of Trace Organic Compounds." EPA-600/2-
77-223, Env. Prot. Tech. Series, Dec. 1977.
517
-------
v>EPA
NATO CDSM
A
•OTAN CCMS
NATO-CCMS
518
-------
MONITORING TECHNIQUES FOR THE
CONTROL OF THE ADSORPTION PROCESS
519
-------
&EPA
COM
•OTAN CCMS
NATO-CCMS
520
-------
A REVIEW OF ANALYSIS TECHNIQUES FOR
ORGANIC CARBON AND ORGANIC HALIDE
IN DRINKING WATER
Y. Takahashi
INTRODUCTION
Most drinking water, especially that which comes from
surface sources, is contaminated with a modest amount of or-
ganics. Typically these organics come from natural sources and
are reasonably innocuous. In recent years, however, it has been
discovered that many drinking waters contain organics which are
potentially hazardous to human health. These can come from
industrial sources as well as from the chlorination process
itself. Accordingly, extensive work is now underway in many
countries throughout the world to develop treatment processes
that will effectively remove these contaminants.
To support this work and to better understand the health
effects of trace organics in drinking water, sophisticated and
expensive analytical techniques have been developed to detect
and identify these contaminants. To date, more than 700 trace
organics have been identified in various raw and treated drink-
ing waters. Despite this, a large portion of the organics
present are still unidentified.
Thus, while the complete identification and quantification
of trace organics in drinking water is desirable, it appears
that such comprehensive information is neither technically nor
economically feasible as a routine process evaluation and control
tool. Accordingly, several group parameters have been studied as
surrogates for undesirable contaminants in water. Two of these
which are particularly useful are total organic carbon (TOC) and
total organic halides (TOX).
ORGANIC CARBON DETERMINATION
Total organic carbon (TOC) in water exists in a number of
forms and phases. Thesve can be classified as purgeable organic
carbon (POC), dissolved nenptirgeable organic carbon (NPOC), and
suspended organic carbon (SOC). The NPOC is defined as the
organic carbon fraction which passes through a 0.45 micron filter
and is then nonpurgeable.
521
-------
The determination of TOC in water containing SOC is always a
problem. The reproducibility of the analysis is poor even if the
SOC is broken down to smaller particles or homogenized prior to
the TOC determination. Some TOC techniques may not quantita-
tively respond to the SOC (1).
When the measurement of TOC is used to evaluate the effi-
ciency of organic removal in drinking water treatment systems,
the NPOC determination throughout the treatment process should
provide the most useful information since most of the SOC is
rather easily removed by simple coagulation and/or filtration.
The remaining carbon (NPOC) can be a measure of the amount of the
additional treatment required.
On the other hand, if trihalomethane formation is compared
with the TOC in raw water where prechlorination is performed,
the SOC should be taken into account. Some trihalomethane
precursors exist as SOC or as adsorbed on the SOC (2).
The instrumental method for determining TOC was first
reported by Van Hall, et al., in 1963 (3). Since then, numerous
TOC techniques have been developed and many different TOC
analyzers are commercially available. It is beyond the purpose
of this article to review all available techniques and minor
variations used by these analyzers. However, TOC methods which
represent the fundamental trends in this analysis field are
summarized in Table 1.
Initially the TOC parameter was introduced to correlate
with biochemical oxygen demand (BOD) and chemical oxygen demand
(COD) for industrial and domestic waste waters. The TOC of
these waste samples is typically 5 mg/1 and higher. Van Hall's
method is well suited for this purpose and has been widely used.
The method is based upon the high temperature oxidation
of carbonaceous matter in water to carbon dioxide (CO ) followed
by CO- detection with a nondispersive infrared (NDIR) analyzer.
A small amount of water sample is injected into a combustion
chamber maintained at 900°C. All carbonaceous matter is oxidized
to C02 in the presence of oxygen and a catalyst. Oxygen
carries the gaseous products through the NDIR which measures the
CO- content of the gas stream. The detector output is graphi-
cally displayed on a strip chart recorder; the peak height is
proportional to the carbon content of the sample.
Since all carbonaceous materials in the sample, including
inorganic carbon (1C), liberate CO., 1C has to be either
removed prior to the analysis or determined separately.
Another common high temperature method is based on the
reductive pyrolysis of organics to methane- and flame-ionization-
detection (FID) (4). In this method an acidified sample is
522
-------
Table 1
TOC Method Comparison
HIGH TEMPERATURE METHODS (~800°C)
METHOD DETECTOR
Catalytic Oxidation NDIR
Catalytic Reduction FID
Oxidation then FID
Reduction
DETECTION LIMITS(rog/1)
NPOC POC
2
1
0.1
Lost
0.5
Lost
REFERENCES
Van Hall, et al., (3)
Takahashi, et al., (4)
Croll (5)
01
to
to
LOW TEMPERATURE METHOD (~60 C)
METHOD
Persulfate Oxidation
Silver-catalyzed
Persulfate
Ultraviolet Oxidation
UV-Persuifate Oxidation
UV-Persulfate Oxidation
UV-Persulfate Oxidation
then Reduction
DETECTOR
NDIR
NDIR
NDIR
NDIR
Conductivity
FID
DETECTION LIMITS(mg/1)
NPOC POC
0.2
0.05
0.001
0,01
0.1
0.01
Lost
Lost
Lost
0.05
Lost
0.001
REFERENCES
Menzel and Vaccaro (6)
Baldwin and McAtee (7)
Wolfel and Sontheimer (8)
Goulden and Brooksbank (9)
Ehrhardt (10)
Takahashi (11)
-------
injected into a sample boat containing an oxidizer and is heated
at 115°C. CO. (from 1C) and volatile organics (VOC) are
transferred into the gas phase; the gases are passed through a
column packed with porous polymer resin. VOC is retained on the
column, but CO^ is eluted and discarded. Then, the column is
backflushed ana the VOC is carried into the high-temperature
reduction zone where the VOC is converted to methane in the
presence of hydrogen and a catalyst. Methane is detected by the
FID. The residual organic carbon (ROC) remaining in the boat is
then heated to 800°C, pyrolyzed, and oxidized. The pyrolysis
products are reduced in the reduction zone mentioned above and
the resulting methane is detected by the FID.
Both of the methods described above have been selected
as standard methods by the American Society for Testing and
Materials (ASTM) (12).
The majority of high temperature TOC analyzers lack
precision and accuracy below 2 mg/1. Certainly the FID is
sensitive enough to detect less than 0.1 mg C/l with a 50
yl water sample. The principal problems are that there are
organic contaminants in the carrier and reactant gases and trace
carbon contaminants in the catalyst. When a discrete water
sample is injected, steam at high temperature reacts with
carbonaceous contaminants in the pyrolysis chamber to cause an
apparent carbon reading. This is generally defined as the
method blank or the system zero bias which often varies between
1 and 5 mg/1.
In the method reported by Croll (5), the erratic system
blank problem is minimized when a sample is continuously intro-
duced at a constant rate. An acidified and sparged sample is
introduced at the rate of 0.6 ml/min into a metal chamber
(800°C) packed with solid oxydizer (CuO). The CO- produced
is carried with nitrogen into a high temperature reduction
chamber containing nickel catalyst. The methane produced is
detected with an FID. The reading is taken when a signal has
reached a plateau. The relative standard deviation of +_ 38, _+
7, + 1.5 and +_ 1.7 percent were reported at the organic carbon
levels of 0.15, 0.65, 5.15 and 50.15 mg/1, respectively (5).
An advantage of the high-temperature TOC methods is the
short analysis time, 2 to 5 minutes per analysis. Also, because
of exhaustive oxidation and reduction conditions at above 800°C,
complete destruction of organics in water is expected.
A disadvantage of the high temperature methods is that
the TOC values in river water, well water, and treated drinking
water, including granular activated carbon (GAC) filter efflu-
ent, are below its detection limit.
524
-------
To achieve low-level TOC analysis for such applications,
the following criteria must be satisfied:
1. Efficient inorganic carbon removal.
2. High sensitivity, from 10 to 50 yg/1.
3. Repeatable and low system blank.
In order to satisfy these requirements, low temperature
oxidation techniques are much more attractive. Selected low
temperature TOC methods are also listed in Table 1.
In all the low temperature techniques mentioned, the
oxidation reactions are carried out in the aqueous phase under
severe oxidizing conditions. Such conditions are created by
addition of strong chemical oxidizing agents such as peroxides
or persulfates, or by irradiating the water with high energy
sources, such as ultraviolet light.
Initially trace level TOC was of concern only to oceano-
graphers for sea water studies. Their main interest in seawater
TOC was not primarily for pollution-oriented ^reasons but -rather
to understand the interactions of dissolved organic matter and
the living and nonliving organic matter of the ocean environment.
The range of TOC in uncontaminated seawater is 0.3 to
3.0 mg/1. The low level of TOC and high salt contents in
seawater make TOC analysis by high temperature techniques almost
infeasible on a routine basis.
In 1964,.Menzell and Vaccarro (6) reported the wet chemical
oxidation with potassium persulfate (K2^o0s^' tnis subse-
quently became the standard method in limnology and oceanography,
In this method, 1 to 5 ml of seawater is acidified and
sparged in a glass ampule. Persulfate is then added to the
ampule. The solution is then autoclaved at 130°C for 30 minutes
to oxidize organics to CO . The CO- is purged with an
inert gas and the concentration is aetected with an NDIR.
Recently, the kinetics of persulfate oxidation of organic
materials in fresh water was thoroughly studied by Goulden and
Anthony (13). This work showed that organics in fresh water can
be oxidized at 90 to 100°C in a few minutes.
The differences in the oxidation of organics in fresh
water and seawater seemed to be twofold. The first difference
is their slow oxidation reaction in the presence of approxi-
mately 3 percent salt. Skopintsev, et al. (14), reported that
the organic matter in fresh water, including humic substances,
525
-------
is completely oxidized to CO2 by persulfate. However, sodium
chloride addition to the same sample interfered with the oxida-
tion so much that complete oxidation was not achieved within 15
minutes at 100°C.
The second difference is the nature of organics in sea-
water. These can be highly refractory against persulfate
oxidation. Takahashi and Mar (22) compared the oxidation rates
of organics in seawater and in fresh water spiked with salt.
This work confirmed the presence of very refractory organics in
natural seawater.
The efficiency of persulfate oxidation of organics in
natural water samples is still in question. Some TOC values
obtained by various oxidation techniques are summarized and
compared in Table 2. It indicates that the average TOC obtained
by persulfate oxidation is 80 to 90 percent of the values
obtained with high temperature or photochemical oxidation.
It has long been known that when water is exposed to high
intensity ionizing radiation, such as X-ray or fast proton
recoils formed by the passage of fast neutrons, hydroxyl radi-
cals (OH) are produced in high concentration. The hydroxyl
radical is also produced by ultraviolet (UV) irradiation." Such
radiation can be easily and conveniently produced with a mercury
discharge lamp. The benefit of the hydroxyl radical is that
it is an extremely strong oxidizing agent. Accordingly, many
organics are oxidized to CO when exposed to UV radiation.
The oxidation is probably aided by a number of mechanisms in
addition to that caused by the hydroxyl radical. Some may
undergo photodecomposition directly. Organics that absorb UV
energy are raised to an excited electronic state that leaves
them more susceptible to OH oxidation.
Even though photolytic oxidation of trace organics to
CO- in fresh water was investigated by Beattie, et al. (15),
much literature about this technique has been investigated and
reported for the oxidation of organics in seawater. [Armstrong,
et al. (16), and Armstrong and Tibbits (17)'].
Recently, Woelfel and Sontheimer (8) obtained 0.001 mg/1
sensitivity using this photooxidative technique for fresh
water. A low pressure mercury lamp was immersed into a sealed
reaction vessel with a sparge gas inlet and outlet.
In this technique the sample is first acidified and
sparged, then the mercury lamp is energized. Organic matter is
oxidized to CO which is then detected by an NDIR. The
reaction time may vary depending upon the type of organics in
the sample being analyzed; the typical reaction time is less
than 5 minutes. The apparatus for the oxidation of organics and
the typical CO generation is illustrated in Figure 1.
526
-------
Table 2
TOC Data Comparison
ro
Number
of
Determinations
24
68
38
7
TOC Range
(mg/1)
0.35-1.22
0.57-1.74
1.49-3.08
0.13-1.63
0.65-1.42
RELATIVE MEAN VALUES
High Temp. Persulfate UV UV-Persulfate REFERENCES
(100)
(100)
(100)
(100)
90
78
97
86
(100)
95
101
107
Willaims (20)
Sharp (21)
Goulden and Brooksbank
Gershey, et al. (1)
Takahashi and Mar (22)
(9)
(100) Data normalized to this value.
-------
APPARATUS
SAMPLE
INLET
UVLAMP
SAMPLE
UV REACTOR
STOPCOCK
NDIR RESPONSE
EASY TO OXIDIZE
DIFFICULT TO OXIDIZE
UVon
Figure 1. TOC By UV Oxidation
528
-------
The addition of persulfate to the photooxidation process
increases the oxidation rates of many organics. This is due to
the excitation of organics with UV radiation and formation of
the active species SO. as well as OH at higher concentration.
Ehrhardt (10), and GoUlden and Brooksbank (9) used this techni-
que for seawater, and fresh water, respectively.
During the last 5 years or so, the formation of trihalo-
methanes during disinfection, mainly chlorination of drinking
water, has been intensively studied.
With respect to trihalomethane detection, some researchers
are interested in measuring the POC to find out whether or not
POC and total trihalomethanes are related.
The POC produced by chlorination is 0.005 to 0.020 mg/1;
typically the NPOC is 2 to 4 mg/1. Accordingly, a method to
determine the POC at yg/1 levels is needed along with a separate
NPOC measurement.
Takahashi (11) developed a method to detect POC and NPOC
sequentially with the detection limits of 1 and 10 yg/1, res-
pectively .
The operation principle of this method is as follows.
A 10 ml sample of water is mixed with an acidified potassium
persulfate reagent. The sample is transferred to a glass sparge
chamber where a flow of helium sweeps out carbon dioxide from
inorganics and those organics that are purgeable (POC) under the
temperature and flow conditions used. These materials are
transferred by the carrier gas to a very efficient scrubber that
removes the carbon dioxide but allows the POC to pass through.
The POC is converted to methane and measured using the same
catalytic reduction and flame ionization detector measurement
procedure as described in reference (4). The water sample is
then pushed through a quartz coil positioned around a very in-
tense mercury discharge lamp. The combination of the persul-
fate oxidant and the intense ultraviolet light oxidizes the
remaining organics to carbon dioxide. After oxidation, the
sample passes to another glass sparge chamber where the resul-
ting carbon dioxide is purged, converted to methane and measur-
ed. The final readout is the TOC content of the sample. The
advantages of this procedure for low levels relate to use of a
large water sample and a method background blank that is
typically less than 0.03 mgC/liter (Figure 2).
The currently available techniques, UV or UV-promoted
persulfate oxidation, satisfy most of the requirements for
measuring trace TOC in drinking-water-related samples routinely
and without highly skilled operators. The techniques are
sensitive enough, for example, to distinguish the TOC in high
529
-------
T
COjSCHUMEM
CO,
MJMEAM.I OKGANIC CAMON
Mi 12)
»A(M»tH(1l MIACTION UV »AMOEII 121
POI1. «MMCI
UV - PERSULFATE OXIDATION
Figure 2. UV - Persulfate Oxidation
530
-------
quality deionized water from that in regular quality deionized
water. TOC concentrations in selected water samples are sum-
marized in Table 3.
Now the technology is available to utilise these photo-
oxidation techniques to monitor the NPOC continuously.
Table 3
Typical POC and NPOC in Various Types of Water
SAMPLE POC CMgC/f) NPOC (mgC/1 )
River Water <2 2-6
Well Water <2 0.1-2
Lake Water <2 0.3-3
Drinking Water 1-15 ' 0.3-3
GAC Effluent <2 0.05-0.3
Deionized Water <2 0.03-0.2
Distilled Water <2 0.05-0.3
Reverse Osmosis <2 0.05-0.3
It has been demonstrated that the NPOC in GAC column
influent and effluent can be measured at constant intervals to
indicate performance of a GAC column (Figure 3).
The TOC indicates the maximum organic concentration level
in water but does not distinguish whether it is due to naturally
occurring organics or to pollutants.
Studies have been made to increase the selectivity of TOC
measurements to indicate the degree of pollution-oriented
organics. This specificity was obtained.by making carbon
isotope measurements. The value of the C/ C ratio
varies depending upon the nature of organics in water. Calder
and Parker (18), and Games and Hayes (19) studied the variation
of the C/ C ratios in the various water samples.
Whether or not this type of additional information is indeed
worthwhile remains to be seen.
ORGANIC HALIDE DETERMINATION
It has been well established that many halogenated organics
are toxic to human health and some are carcinogenic. Many of
these toxic substances have been found in drinking water
supplies. The majority of halogenated organics in raw water are
either chemicals discharged by various industries or are agri-
culturally related compounds, such as pesticides and herbicides
leached from the soil. Naturally occurring halogenated organics
in water are very limited. Thus, the total organic halides
531
-------
OJ
to
3.0 n
2.0 -
r 1.0.
GAC COLUMN INFUENT"
PUMP
*
TAP WATER
LJ
GAG
COLUMN
GAC COLUMN EFFLUENT
•
3
i
4
HOURS
Figure 3. Continuous NPOC Monitor
-------
(TOX) in raw water are closely related to manmade contaminants
that are toxic.
Interest in measuring TOX was stimulated by finding
chloroform in the drinking water of New Orleans, LA (33). Rook
(34) traced the origin of chloroform in finished drinking water
to the reaction between chlorine and naturally occurring fulvic
and humic acids in water. Rook (35) also attempted to detect
any significant amounts of -intermediate chlorinated organics
resulting from chlorination and found no other gas chromato-
graphable compounds besides trihalomethanes. However, the
question still remains whether or not any non-gas chromatograph-
able chlorinated organics are produced and, if so, are they
produced in any significant amounts.
TOX is still a relatively new water quality parameter, and
the measurement is a little more difficult and time consuming
than the TOC measurement. However, the technology of measurement
has been improved lately to the point where routine measurement
of TOX is feasible.
Difficulties with TOX measurements are mainly due to
the following reasons:
1. The range of TOX in surface water is 1 to 100 yg/1.
This requires a high sensitivity detector or the
solution has to be concentrated.
2. None of the detection methods except neutron activa-
tion analysis, which requires highly sophisticated
instrumentation, responds equally to all types of
halogenated compounds. Thus, all halogenated organics
have to be converted to a common species.
3. Inorganic halide (IX) concentration in surface water
is 1,000 to 100,000 times higher than the TOX. Thus,
the TOX method has to have an IX rejection ratio of at
least 500,000 to one.
The following is a review of four TOX techniques presenting
the advantages and disadvantages of each.
Liquid-Liquid Extraction Method
Liquid-liquid extraction is the simplest and most con-
ventional technique to concentrate the TOX in an appropriate
organic solvent and leave the IX in the aqueous phase. Wegmen
and Greve (23) investigated extractable organic halides (EOX) in
the Rhine River over the period of 1973 to 1975 (23), and later
extended this work until 1977 (37), using the following extrac-
tion technique.
533
-------
Organics from surface waters were extracted into petro-
leum ether more than three times. The combined extracts were
dried and concentrated. A small aliquot of combined extracts
was injected into a furnace and burned in an excess of oxygen.
The hydrogen halides (HX) formed were passed into a titration
cell where they were microcoulometrically titrated with silver
ions in the cell electrolyte. The detailed discussion of the
microcoulometric titration of HX is described elsewhere (39).
In this method the EOX is concentrated by a factor of 1000 and
the minimum detection limit is 1 yg/1.
Extraction efficiency of various organics from water and
formation of microcoulometrically titratable species by this
conventional combustion technique are discussed in this work.
It was reported that a significant fraction of total organics
were either not extracted or were evaporated along with the
ether.
A variation of the liquid-liquid extraction technique
is to use a small amount of organic solvent and avoid the
evaporation step. This technique is commonly used to determine
the trihalomethanes by GC (32). This kind of extraction is
possible for the TOX determinations; however, the extraction
may take a long time because of the slow kinetics to reach the
partition equilibria. This is especially true for high mole-^
cular weight organics. Also, a solvent has to be selected by
considering the most favorable partition coefficients for as
many organics as possible between the solvent and water to
achieve the highest extraction efficiency.
XAD Resin Method
Glaze, et al. (24) used a macroreticular resin (XAD)
to adsorb organics from water and then eluted these organics
with a small amount of ether.
In this method 200 mg of 100/120 mesh XAD is packed in
a small glass column. About 120 ml of water is forced through
the column at the rate of 25 ml/min. Then, the XAD column is
washed with ether until 1 ml of ether is eluted. The organic
halides in the ether are then determined by combustion and
microcoulometric titration as described earlier.
XAD resin is expected to adsorb a greater fraction of
organics than the fraction extracted by an organic solvent from
water. In this technique the ether was not evaporated to
concentrate the eluant, thus minimizing the loss of volatile
organics. Furthermore, during the ether elution step, a gas-
tight vial placed in ice water was used to collect the ether
eluent. Therefore, overall recovery from water is expected to
be better than the EOX.
534
-------
By this method the organics in water are concentrated
by a factor of 100, and the detection limit is 2 yg/1.
This method is free from IX interferences, but adsorption
of organics into the XAD resin and the elution of adsorped
organics from the resin by ether must be carefully evaluated.
Thurman, et al. (26), studied the relationship between
solubility and adsorption capacity for selected organic com-
pounds on the porous acrylic resin XAD-8. The adsorption
capacity of the resin is important for predicting the optimum
ratio of sample to column size in the preconcentration step. If
the same relationship is applicable to XAD-2 resin/ only a small
fraction of organics will be retained on the column using the
method described by Glaze, et al. (24). The adsorptivity of
trace organics from natural water by the XAD-2 resin was also
investigated by Suffet, et al. (25). The report indicated
that granular activated carbon adsorbed nonpolar trace organics
much more efficiently than XAD-2 on an equivalent bed volume
basis.
Powdered Activated Carbon/Pyrohydrolysis Method
In 1973, Kiihn and Sontheimer (27) reported on a method
for organic chlorides based on powder carbon adsorption and
pyrohydrolysis of carbon along with adsorbed organics. The
pyrohydrolysis technique was first developed to study the
efficiency of organic removal by pilot-scale or full-scale
granular activated carbon (GfcC) columns by measuring the TOX
on the carbon at various depths in the column bed.
In this method two grams of powdered carbon is mixed
with 20 1 of sample, the pH is adjusted to 5, and the mixture
stirred for 1 hour to adsorb the organics onto the carbon
surface. After 1 hour of mixing, the pH is raised to 7 and
some coagulants are added.
The carbon is then removed by filtration and saved. The
solution is then mixed with another 2 g of powdered carbon to
repeat the adsorption and filtration steps. The two portions
of carbon are combined and transferred to a quartz boat for
pyrohydrolysis.
In the pyrohydrolysis step a quartz boat is placed in
the first section of a quartz pyrolysis tube. The second
section of the tube is maintained at 1000°C. Oxygen and water
vapor are passed through the pyrolysis tube and the pyrohydro-
lyzate is collected through the condenser. The temperature of
the first section is programmed to 700°C.
535
-------
The organic halide portion of the adsorbed organics is
pyrolyzed to HX and X2 and these are adsorbed in the pyrohy-
drolyzate. Halide concentration in the pyrohydrolyzate is then
determined by an ion selective electrode (Figure 4).
To make the inorganic halide correction, the entire adsorp-
tion process is repeated and chlorides in the carbon are washed
out using a nitrate solution. The chloride concentration in the
nitrate solution is determined by an ion selective electrode.
Dressman, et al. (28), repeated the Kiihn and Sontheimer
method and reported some improvements. These improvements are:
1. Remove adsorbed Cl"" as well as the Cl" in the pore
water of the GAC by washing with nitrate solution prior
to pyrohydrolysis. This is defined as "direct" carbon
adsorbable organic halides (CAOX) method.
2. Use a microcoulometric detector to measure halides in
pyrohydrolyzate. They found that the chloride selective
electrode gave erroneously high Cl~ readings and was
affected by addition of CuSO and Kj0?'
3. CAOX is applied to the nonpurgeable fraction only.
Only 70 percent of the CHC1 was recovered.
Also, the chloride interferences were thoroughly investi-
gated and it was found that nitrate ion effectively removed the
sorbed chloride ion from the PAC.
The Kiihn and Sontheimer method as modified by Dressman is
an attractive method, but it still presents some difficulties.
These are:
1. Lengthy carbon adsorption step,
2. Loss of purgeable organic halides such as CHC1_,
3. Response to bromides has not been investigated.
Activated Carbon Column/Combustion Method
Takahashi and Moore (30) utilized a carbon column adsorp-
tion technique, but with shorter analysis time.
The analytical procedure consists of three steps. These
are: 1) adsorption of organics from water into granular activated
cartoon (GAC) packed in a microcolumn; 2) desorption of inorganic
halides by washing the GAC with nitrate solution; 3) combustion
of sorbed organics along with the GAC, and coulometric titration
of halides thus produced with silver ion. Water samples of 25
to 100 ml are used. Adsorption time is typically 15 to 30 min,
536
-------
Ol
u>
100°C-»700°C 1000°C
02
200ml/min
Steam
Generator
~0.8 mi/min
Water
Condenser
ION SELECTIVE ELECTROOl
- PYROHYOROLYZATE
Kuhn andSontheimer (27)
Micro-
coulometer
Dressman et al (28)
r "2
CO
1 °2
H2O
CO2
HX
PYROHYDROLYZATE
Figure 4. TOX Determination by Pyrohydrolysis
-------
and combustion and determination time is typically 10 min. The
detection limit of this method is 6 X 10"^ mole/1 (or 2 jug/1
of chlorine) and the apparent TOX from 100 mg/1 of chloride
ion was lower than the detection limit.
A schematic diagram of the adsorption apparatus is shown
in Figure 5. One hundred to two hundred mesh GAC is packed
in glass columns (2 mm I.D. x 5 cm long), and two columns are
connected in series. A water sample is poured into the sample
reservoir and is then forced through the GAC columns by pres-
surizing the reservoir. A typical sample flow rate is 2 to
3 ml/ min. Four samples are easily processed simultaneously.
The adsorption efficiency on the GAC in the first column is
checked by measuring the TOX on the GAC in the second column.
The carbon is then oxidized along with sorbed organics from
water/ and the produced halides are coulometrically titrated.
It has been reported that, when using the conventional
oxidation technique with 0^, the combustion products from
brominated organics are not all titratable with silver ion. In
this work/ combustion is first carried out in an atmosphere of
steam from water on the GAC and carbon dioxide. The GAC is then
oxidized with oxygen.
This combustion technique provides not only the quantita-
tive recovery from brominated organics, but also some information
on the characteristics of the organic halides adsorbed on the
carbon. Figure 6 shows the typical response from various
samples. Peak X is obtained when the carbon is heated at 200°C
with a CO. carrier gas. Here, low molecular weight organic
halides wnich are weakly adsorbed on the carbon are vaporized.
Peak Y is obtained when the carbon is heated at 800°C with C02
and the high molecular organics are vaporized. Finally/ the GAC
is completely oxidized along with residual organics with O_
(Peak Z). Part of this combustion process is similar to tne
regeneration of GAC. Therefore/ peaks X/ Y, and Z may provide
some information about regeneration of GAC after treating water.
The relative contribution to peaks X, Y, and Z for selected
organics is also shown in the Figure 6.
Similar results are expected with the two carbon adsorption
techniques except for purgeable organic halides. However/ no
direct comparison has yet been done.
The organic halide fraction in water measured by these
four techniques are graphically illustrated in Figure 7. There
is uncertainty about the response to high molecular weight
organics for each method. The solid adsorbents become less
effective to adsorb large molecules due to unfavorable kinetics
538
-------
N2
SAMPLE
RESERVOIR
L
|| GAC COLUMN 1 |) GAG CLEANUP
¥ COLUMN
£} GAC COLUMN 2
GAC ADSORPTION
NITRATE
RESERVOIR
Micro-
coulometei)
FURNACE
T1TRAT10N
CELL
PYROLYSIS
Figure 5. TOX by GAC Column Adsorption
539
-------
Chloroform
Bromoform
200°C! 800°C
Raactant gas
Temperature
COMPOUNDS X Y 2 COMPOUNDS X Y 2
I I
Bromobanzene
2,4,6—Trichlorophenol
m-dichlorobenzene
I I
3—bromobenzoic acid
Chlorinated
humic acids
drinking water
Figure 6. Pyrogram of Compounds Adsorbed on GAC
540
-------
Ul
*>.
2
CARBON ADSORPTtGft
XAD/ETHER
ETHER EXTRACTIC
(T) Chloroethanol
(2) Chloroacetone
(5) Chlorofonn
(4) Chlorocyclohexane
(§) Chlorobenzene
(?) Dodecyl Chloride
(?) Hexachlorobenzene
(?) Pesticides
CO) Humic Acids
MOLECULAR WEIGHT
Figure 7. OX Determined by Various Methods
-------
of adsorption. In liquid-liquid extraction methods/ the parti-
tion coefficients and equilibrium kinetics of such large mole-
cules are expected to be unfavorable.
Activated carbon adsorbs th*j largest fraction of total
organics from water. The organic halides in natural waters
determined by various methods are summarized in Table 4.
Unfortunately, none of the samples were analyzed by all methods.
Thus, the direct comparison of TOX values on these samples
cannot be made. However, it is clear that TOX values by the
carbon adsorption method are highest. Also, the high values are
not due to inorganic halide interference.
There is no doubt that in the near future better TOX
methods will be developed. Meanwhile, the TOX should be meas-
ured by the method which responds to the largest fraction of
total organic halides. As one can see in Figure 7, the acti-
vated carbon will adsorb organics most effectively, and the
carbon column adsorption technique can be performed routinely
and conveniently.
The efficiency of the carbon adsorption has been evaluated
by measuring the TOC in the sample before and after the adsorp-
tion, using various water samples. However, this area needs to
be investigated more thoroughly.
In Table 5, the organic level in various waters is summar-
ized. Four interesting results are found in this table. The
TOX in many U.S. surface waters is less than 15 ygCl/1. After
chlorination, the TOX increases as one might expect because of
trihalomethane formation. However, the concentration of non-
purgeable organic halides (NPOX) also increases.
Another example of organic levels in water is demonstrated
in Figure 8. POX, TOX, and NPOC were measured at various stages
of the drinking water treatment process including the GAC
filter. The samples were obtained from the Philadelphia Water
Department, Torresdale, PA.
There is potential for determining the TOC1 and TOBr
separately. Once the chloride ion is removed from the carbon,
the TOC1 and TOBr can be determined by neutron activation or
perhaps by X-ray fluorescence without pyrolysis. Alternatively,
after the pyrolysis of the carbon, the HC1 and HBr, from the
TOC1 and TOBr, may be separately determined by ion chroma-
tography (36).
It is generally agreed that TOX is a very important group
parameter for drinking water. The carbon adsorption and micro-
coulometric detection technique offers a convenient method for
closely approximating the true concentration of TOX in drinking
water.
542
-------
Table 4
Organic Halides by Various Methods
(pg/1 as Cl)
SAMPLE
Raw
City 1
Finished
Raw
City 2
Finished
S Raw
w City 3
Finished
Lake of Constance
Rhine above Basle
Rhine of Cologne
Rhine at Duisburg
TTHM BOX
2.92
49.2
0.88
152.7
1.0
123.2
3
27
37
55
XAD/ether CAOX
1.8
73.2(2.o|a)
5.5
113. 4(11. 0)(a)
4.7
98.7
6
93
192
228
REFERENCES
Glaze, et al. ,(24)
Kuhn, et al.,(29)
Average vali1
for *•?;/!
Rhine at Loblth
18
(range 7.7-50)
Wegtnan & Greve (23)
Average value
for 1974
(a) Nonpurgeable
-------
Table 5
Organic Levels in Various Waters
POX
TOX
NPOC
REFERENCES
SAMPLE (u«/l««« Cl)
Raw 1
City A
Finished 59.5
Raw 1
City B
Finished 96
City C Raw 1
Raw HO
City D
Finifhed 9.8
Raw 0.5
City E
Finished 64
Raw 17.8
City F
Finished 182
Raw ND
City G
Finished 26.6
Untreated 1
Chlorinated 76
CIO. treated 1
Oj Created 1
Ozonation followed
by chlorination
Rhine Water
Sandbank filtrate
Chlorinated Sandbank
Filtrate
Activated Carbon
Filtrate
(U»/l «• Cl) («g/l)
6.0 2.99
178 2.14
11.5 3.29 Takahaahi and
Moore (30)
312 2.25
21.0 6.50
11
27
11
116
29
347 EPA Report (31)
11
125
6 2.99
356
23.5 Takahashi and
4 1.94 Moore (30)
154
78 4.60
17 °-»° Kuhn & Sontheimer
85 1.00 (29)
2
544
-------
POX NPOX NPOX NPOC
ug Cl/l ug Cl/l POX mg C/l
11.5
RAW WATER
BASIN
FeCJ3
19
126
SEDIMENTATION
BASIN
.- 104
214
- - 98 214
-- 26
3.15
6.6
3.29
2.1 2.30
2.2 2.25
52 2.0 0.98
..... 15 59 3.9 1.21
Figure 8. Change in Organic Carbon and Organic
Halides During Treatment Process
545
-------
ACKNOWLEDGMENT
The author thanks Jim Coyle of the Philadelphia Water
Department for supplying a few samples for TOX measurements,
546
-------
REFERENCES
1. Gershey, R. M., M. D. MacKinnon and P. J. leB. Williams,
Marine Chem. In Press.
2. Stevens, A. A., C. J. Slocum, D. R. Seeger and G. G.
Robeck, Proceedings of Conference on the Environmental
Impact of Water Chlorination. Oak Ridge/ Tennessee,
October 1975.
3. Van Hall, C. E., J. Safranko and V. A. Stenger. 1963.
Anal. Chem., 35, 315.
4. Takahashi, Y., R. T. Moore and R. J. Joyce. 1972. Amer.
Lab., £, 31.
5. Croll, B.T., 1972. Chem. Ind. 386.
6. Menzel, D. W, and R. F. Vaccaro, 1964. Limnol. Oceanogr.,
9_, 138.
7. Baldwin, J. M. and R. E. McAtee. 1974. Michrochem. J.,
_19, 179.
8. Wolfel, P. and H. Sontheimer. 1974. Vom Wasser, 43, 315.
9. Goulden, P. D. and P. Brookbank. 1975. Anal. Chem., 47,
1943.
10. Ehrhardt, M.. 1969. Deep Sea Res.,16, 393.
11. Takahashi, Y. Proceeding of Water Quality Technology
Conference, San Diego, California, Dec. (1976).
12. 1978 Annual Book of ASTM Standards, Part 31, D-2579,
p. 541, American Society for Testing and Materials,
Philadelphia,
PA. (1978).
13. Goulden, P. D. and D. H. J. Anthony. 1978. Anal. Chem.,
50 953.
14. Skopintsev, B.A., E. S. Bikbulatov, and N. Y. Melnikova.
1977. Oceanography, 16, 630.
547
-------
15. Seattle, J., C. Bricker and D. Garvin. 1961. Anal. Chem.,
33, 1890.
16. Armstrong, F. A. J., P. M. Williams and J. D. H. Strickland,
1966. Nature, 211, 481.
17. Armstrong, F. A. J., S. Tibbitts. 1968. J. Mar. Biolo.
Assoc., 48, 143.
18. Calder, J. A. and P. L. Parker. 1968. Environ. Sci.
Tech., 2, 535.
19. Games, L. M. and J. M. Hayes. 1976. Anal. Chem., 48,
130.
20. Williams, P. M. 1969. Limnol. Oceanogr., 14, 297.
21. Sharp, J. H. 1973. Marin. Chem., _!, 211.
22. Takahashi, Y. and D. Mar, Presented at Pittsburgh Con-
ference, Cleveland, Ohio, March (1979).
23. Wegman, R. C. C. and P. A. Greve. 1977. Sci. Total
Environ., 7, 235.
24. Glaze, W. H., G. R. Peyton and R. Rawley. 1977. Environ.
Sci. Tech., 11, 685.
25. Suffet, I. H., L. Brenner, J. T. Coyle and P. R. Cairo.
1978. Environ. Sci. Tech., 12, 1315.
26. Thurman, E. M., R. L. Malcolm and G. R. Aiken. 1979.
Anal. Chem., 50, 775.
27. Kiihn W. and H. Sontheimer. 1973. Vom Wasser, 41, 65.
28. Dressman, R.C., E. F. McFarren and J. M. Symons, Paper
presented at Water Technology Conference, Kansas City,
Missouri, Dec. (1977).
29. Kiihn, W., H. Sontheimer, L. Steiglitz, D. Maier and R.
Kunz. 1978. J. of Am. Water Works Assoc., 70, 326.
30. Takahashi, Y. and R. T. Moore, Paper presented at Am.
Chem. Soc. National Meeting, Hawaii, April (1979).
31. Study of Exposure to Organics via Drinking Water in 7
U.S. Cities, EPA Report in preparation by Health Effect
Research Lab.
32. Mieure, J.P. 1977. J. of Am. Water Works Assoc., _69, 60.
548
-------
33. Bellar, T.A., J. J. Lichtenberg and R. C. Kroner. 1974.
J. Am. Water Works Assoc., 66, 703.
34. Rook, J. J., 1976. J. of Am. Water Works Assoc., 68, 168.
35. Rook, J. J. 1977. Environ. Sci. Tech., 11, 478.
36. Wegman, R. C. C. and P. A. Greve. 1977. Sci. of the Total
Environ. , 7_, 235.
37. Small, H., T. S. Stevens and W. C. Bauman. 1975. Anal.
Chem., 47, 1801.
38. Wegman, R. C. C. and P. A. Greve, Paper presented at the
Int. Symposium on the Analysis of Hydrocarbons and
Halogenated Hydrocarbons, Burlington, Ontario, May (1978).
39. Drushel, H. V. 1970. Analytical Letters, 3, 353.
549
-------
USE OF ULTRAVIOLET (UV) ABSORPTION FOR CONTROL OF
ADSORPTION PROCESSES
Friedrich Fuchs and Heinrich Sontheimer
One of the most important and difficult problems encoun-
tered when using activated carbon filters for drinking water
treatment is the choice of the control method. The analytical
criteria chosen for this purpose are crucial for carbon grade
and quality, filter design, and for the operation conditions;
e.q., regeneration frequency* As can be seen from other papers
in this symposium, there are many different analytical methods
used to control activated carbon filters. The most important
and commonly used technique in Germany is ultraviolet (UV)
absorbance.
UV-absorbance measurements are convenient and allow continu-
ous, automatic control. In many waters, UV-absorbance correlates
fairly well witb the concentration of adsorbable organic* and
for this reason, has proven in many cases to be more practical
for filter calculations than measurements of dissolved organic
carbon (DOC) or chemical oxygen demand (COD).
However, UV-absorbance also has some important disadvan-
tages. It does not give any information on toxic organic sub-
stances. It covers primarily aromatic organics such as humic and
fulvic acids, which present no direct health problem apart from
their role as precursors for haloforms and other chlorinated
organic compounds.
Since humic substances constitute the bulk of total organics
in most of our natural waters, a good correlation between UV and
the total organics measurement, such as DOC, usually is observed.
<
Only a few seconds are needed to measure UV-absorbance and
the measurements are very accurate. DOC measurements are more
time-consuming and require more eomplex instruments. Thus, it
seems reasonable to replace DOC- by UV-absorbance measurements.
For transforming UV-values to DCte-concentration$^ a function
derived from Rhine River data "covering one year" has been used
(Figure 1).
550
-------
12
11
10
I 9
Z
e 8
g
vA
<
uj 6
u
z
2 5
O
£ «
uv
DOC
= 2.0
D Alpenrhein at Lustenau
• Schaffhausen
O Basel
• Karlsruhe
A Wiesbaden
A Koln
x Wesel
km
km
km
km
km
km
359
508
684
814
2315
DISSOLVED ORGANIC CARBON (DOC) IN mg I
Figure 1. Correlation of UV-Absorbance and
DOC Values for River Rhine
Figure 1 shows that there is a fairly good correlation
between DOC concentration and UV-absorbance along the Rhine
River. Some scattering appears which is to be expected since
the Rhine water contains many different organics. Furthermore,
there is no constant ratio between the two analytical measure-
ments. This is shown by deviations from the constant UV/DOC
ratio line of 2 in the diagram. The ratio is lower^han 2 for
less polluted waters, whereas with higher pollution, UV-absorbance
increases to a larger extent than DOC does so that the ratio
exceeds 2. The same observation is shown by Table 1, in which
average values trom ditterent sources on the Rhine and other
rivers are listed. These illustrations clearly point out that
the higher the DOC, the higher the specific UV-absorbance.
551
-------
Table 1
Typical DOC and UV-Absorbance Values
for River Waters in Germany
River
Rhine
Rhine
Rhine
Rhine
Rhine
Danube
Kocher
Schussen
Place
Au/Lustenau
Schaffhausen
Basel
Mainz
DUsseldorf
Leipheim
Schwab.Hall
Lake of Constance
DOC mg/1
1.5
1.7
2.3
4.2
4.5
0
0
3.
5.
6.4
UV m
1.8
2.8
4.4
8.3
10.9
5.2
10.3
17.1
-1
UV/DOC
1.2
1.6
1.9
2.0
2.4
1.7
2.1
2.7
In many cases, we ~an observe a linear relationship between
UV-absorbance and DOC with a positive intercept on the DOC axis.
This indicates that a few organic substances exist that have
some DOC but no UV-absorbance. This can be seen in Figure 2,
based on data taken from control measurements in WuppertaL.
DOC = O.U8 + 0.28E
Pure Water
• Sandfiltration
A Bankfiltration
1.0 2.0 3.0 U.O
UV-EXTINCTION (E) AT 25U nm, m-1
5.0
Figure 2. Relation Between UV-Absorbance
and DOC at Wuppertal
552
-------
In this special casef the UV and DOC values for the bank-filtered
raw water as well as for the treated waters are on the same line,
thus allowing a linear correlation between the two measurements.
But this is not always the case, especially after other treatment
steps. In order to understand this better, Figure 3 shows data
from the plant in Dusseldorf which treats water from the Rhine
River.
01
E
O
O
u
E
U
01
O
O
O
13
12
10
9
8
7
6
5
4
3
2
1
COD
COD
DOC
UV
D"SC
2.94
2.48
2.56
2.38
2.36
1.19
2.39
1.01
Figure 3. Changes in Collective Organic Parameters
J>uting Rhine-River-Water Treatment at
Puteseldorf
553
-------
The diagram includes data for DOC, COD, and UV-absorbance
at 254 nm as well as the ratios COD/DOC and UV/DOC. The data are
average yearly-lvalues for the different treatment steps. The
results show that the UV/DOC and COD/DOC ratios decrease after
ozonation. However, the COD/DOC ratio remains nearly constant
after activated carbon filtration, whereas the UV/DOC ratio
decreases further. This shows that UV-absorbance, combined with
occasional DOC measurements, will help to shed light on the
changes taking place in the organics during different treatment
steps. Another application is in comparing data obtained during
different years, as in Figure 4.
O
ST.
CO
3
•a
U
01
u
O
a
17
16
15
11
13
12
11
10
9
2
7
6
5
4
3
2
1
UV
197«
1978
UV
*- DOC
DOC
Figure 4. UV-Absorbance and DOC Values for Rhine-River-Water
Treatment at Duisburg For Two Different Years
In 1974, the total organic pollution of the Rhine River had
been much higher than in 1978. This is shown by the DOC and
UV-absorbance measurements. The DOC and UV-absorbance values
for drinking water were the same in 1978. This was not the case
in 1974, due to the greater pollution of the untreated water.
This example shows that UV measurements give information on the
amount of the organics and the treatability of 'the polluted
water similar to that obtained by DOC measurements.
This is especially true if we look at adsorption data,
where DOC as well as UV measurements can be used to produce
adsorption isotherms. As a typical example, such data are shown
554
-------
in Figure 5 for two different carbons without pretreatment, and
for one of the carbon grades before and after flocculation.
UV-absorbance, measured at 254 nra wave length and calculated for
aim path, has been converted to an equivalent concentration of
UV-active substances, calculated as mg/1.
O
3- 100 —
£
K
O
Carbon-LSS
After Flocculation
Before Floccuiatton
Carbon F300
Befo-r Floctulalion
'
1.0
c IN mc I OR mo DOC I
Figure 5. Adsorption Isotherms of Rhine River Water
Comparison between the two carbons reveals that the iso-
therms are similar when using UV values, but are different when
using DOC values. Since UV data represent the better adsorbable
organics, we find that the two carbons have similar capacities
for these substances. However, this is not the case for all
organic substances measured by DOC. Here the carbons and the
isotherms are more different. At the same time we find steeper
DOC than UV isotherms.
Preflocculation preferably removes polar organics otherwise
competing with the less polar ones for adsorption sites on the
carbon. Removing the more polar substances by flocculation prior
to activated carbon adsorption will therefore reduce competition
and lead to less steep and higher adsorption isotherms.
The slopes of the isotherms in Figure 5 explain why lower
concentrations in the influent of the carbon filter will lead to
lower loadings on the carbon too. This becomes obvious if we
compare carbon loadings obtained in 1974 to values obtained in
1976-78, when the influent concentrations were somewhat lower.
Some of these comparison data can be seen in Figure 6.
Figure 6 shows non-linear regression curves between loading
and throughput for the high pollution level observed in 1974,
compared to measured values for the period 1976-78. Here the
555
-------
quantity of IMF extract has been used as a measure of the total
organic concentration (Figure 6). It is quite evident that we
now have lower loadings for the same volumetric throughput than
we had before 1975. This allows a higher safety margin for the
adsorption of undefined and possibly toxic organics for a larger
throughput. It must not be forgotten that a decrease in the
influent concentration of pollutant also implies a decrease in
equilibrium adsorption capacity. The data in the lower half of
Figure 6, showing the same comparison, represent the chlorinated
organic loading. These data from the treatment of Rhine River
water lead to the conclusion that the chlorinated organic com-
pounds (Figure 6) are adsorbed with very similar results to those
observed (Figure 6) for the overall organics for the changes in
UV-absorbance.
WATER WORK DUISBURC '.VITTLAER
CAC FILTER WITH F 100
Figure 6. Comparison of Carbon Loadings with Higher (1974)
and Lower (1976/78) Pollution in the River Rhine
While we can generally compare breakthrough curves in
carbon filters by UV-absorbance as well as by using DOC, there
are some differences, as illustrated in Figure 7. This diagram
shows breakthrough curves measured at a carbon filter plant.
The data suggest similar conclusions if we compare DOC or UV
breakthrough curves, but there is a substantial difference
556
-------
between the curves for the two parameters. The removal effi-
iency is always higher for the same throughput when measured by
UV-absorbance rather than DOC values.
The curves also show that a temporary improvement of the
water quality in the influent of the filter can cause desorption
of organic substances on the carbon, whereupon the efficiency of
the filter is reduced considerably. This is exemplified in
Figure 7 by the incident that occurred at a throughput of
approximately 23 m /kg. The amount of polar compounds covered
by DOC is larger and because polar substances desorb more easily,
the DOC values measured in the effluent are temporarily greater
than the inlfluent values, resulting in a negative removal
efficiency.
O.c
o.r 't
, u o. •. —
o. i
rUKOUC.tlt'UT IN ill k
-------
measurements explain the results fairly well, as the P630 carbon
used in the first filter has a very flat isotherm and a high
adsorption rate. With this we have to expect a small working
zone in the carbon filter. One can conclude from the data that
UV measurements are useful for easy, rapid control of activated
carbon filters in operation. This also has been proven by
experience in practice.
0.8
0.6
0.4
0.2
CAC . Hydraffin P 630
Carbon Loading At 57 m /kg
CAC Throughput
Upper Layer. 96 g>'kg
Lower Layer. 7 g/kg
0.8 h-
0.6—
Bedheight 1.9m
0.2r-
Bedheight 0.6m
CAC: LSS
Carbon Loading At 59 m /kg
CAC Throughput
Upper Layer: 50 g/kg
Lower Layer: 23 g/kg
Bedheight 1.9m
70
50 60 70 80 SO 60
3/kg GAC THROUGHPUT m37kg CAC
80
m
Figure 9. Filter Efficiency at Different Bed Heights
in GAC Filters with Two Different Carbons
(UV-Measurement)
SUMMARY
Summarizing all the available experience from different
types of investigations, the following conclusions can be made:
1. UV-absorbance has proved useful as a control parameter for
granular activated carbon filter performance. It is very
easy to measure and very often runs parallel with other
more significant but more difficult-to-measure criteria,
such as dissolved organic chlorine.
560
-------
Generally speaking, UV-absorbance can be used in those
cases where parallel tests using other parameters and
analytical methods have shown that the UV data correlate
sufficiently well with other more important criteria. This
is often, although not always, the case. There may be
exceptions, for instance if the removal of pure substances
like haloforms in small concentration is crucial for the
carbon filter operation.
In most cases, carbon filter control can best be done by
using a convenient method, such as UV-absorbance, for
routine control of the filters and by comparing these data
from time to time with the results of other more sophis-
ticated methods. This will allow conclusions from easily
obtainable data and will give information in such a short
time that special tests could be made in case of unusual
values and relationsips. In this respect, UV-absorbance
measurements can be worthwhile and helpful for carbon filter
control.
561
-------
BIOASSAY TECHNIQUES FOR EVALUATING
THE POSSIBLE CARCINOGENICITY OF ADSORBER EFFLUENTS
R.J. Bull, M.A. Pereira
and K.L. Blackburn
INTRODUCTION
Considerable controversy exists concerning the efficacy of
granular activated carbon (GAC) as a generally applied measure
to protect the public health from the trace organic material
that contaminates many surface and some groundwater sources of
drinking water. Critics of proposed regulations that would
mandate such treatment of drinking water argue that significant
reduction of hazards, by GAC treatment, has not been sufficiently
demonstrated to justify the capital costs of implementing such
treatment.
A limited feel for benefits associated with GAC treatment
of drinking water can be obtained by demonstrating the reduction
of total organic carbon and removal of specific drinking water
contaminants by such treatment. This approach is limited by
the ignorance of the toxicological significance of the organic
compounds removed by GAC and by the realization that most docu-
mented individual organic contaminants of drinking water occur
at negligible concentrations in terms of acute toxicity to
humans. Moreover, known contaminants of drinking water comprise
a very small' portion of the organic consitutents present. These
difficulties have given rise to the suggestion that a more
accurate indication might be obtained with biological testing of
hazards associated with organic compounds in drinking water and
a more comprehensive means of assessing the degree to which GAC
might reduce these hazards.
CLASSICAL METHODS OF ASSESSING CARCINOGENIC RISKS
Carcinogenic and mutagenic activities of chemicals in
drinking water are of particular concern because the risk from
chemical carcinogens is generally believed to be linear with
dose (Hoel, 1977). The traditional use of long-term carcino-
genesis bioassays in experimental animals is limited in its
application to the organics problem in drinking water for
several reasons. First, to be of manageable size, whole animal
experiments must deal with relatively high response rates
562
-------
(Hartley and Sielken, 1977). To rule out the possibility of
relatively low levels of risk, the concentration of very large
volumes of water is required. Second, the experiments require
n long time, either 2 years or for the lifetime of the animal.
Tnis limits the ability to take into account the periodic vari-
ations that occur in source waters. Third, the expense of this
type of testing would considerably limit the ability to study
varying source waters and thus leave questions about the gener-
ality of the results.
In recent years, considerable progress has been made in
understanding certain phenomena which underlie carcinogenic and
mutagenic responses. The somatic nutation theory of chemical
rarcinogenesis, and the demonstration that a large number of
known chemical carcinogens could directly or indirectly (by
activation with liver microsomal enzymes) induce point mutations
in bacteria (McCann, et al., 1975), has led many to suggest that
bacterial or in vitro mutagenesis testing could serve to assess
the carcinogenic activity in drinking water. However, the
relationship between carcinogenic activity and mutagenic activi-
ty does not appear to be quantitative (Bartsch, 1976; Meselson
and Russell, 1977). To judge the efficacy of a treatment pro-
cess one must be able to deal with reduction in risk. It is not
clear that 50 percent reduction in mutagenic activity in a_
drinking water produced by GAC treatment can be interpreted as
a significant reduction in carcinogenic risk.
The purpose of this presentation is to discuss this problem
and outline the approach being taken by the Drinking Water Pro-
gram of the U.S. EPA.
BACTERIAL AND IN VITRO SYSTEMS AS INDICATORS OF CARCINOGENIC
RISK
The simplest and most broadly applied system for muta-
genesis testing is the Salmonella tester strains developed by
Ames (Ames, McCann and Yamasaki, 1975). Numerous other systems
have also been evaluated utilizing other bacterial species,
yeast, mammalian cell culture, and the classical Dosophila
model. The endpoints employed in these alternative systems also
differ from the reversion assay utilized in the Ames test.
The major shortcoming of bacterial and in vitro testing
systems as risk assessment models is that they do not adequately
take into account the critical role that absorption, distribu-
tion, metabolism, and excretion play in the development of
toxicological responses in general (Gillette, 1977; Watanabe,
Young and Gehring, 1977). The potency of a compound in vivo
would be expected to be dependent on the kinetics of steps
leading to the formation and destruction of the "active" met-
abolite as well as alternate pathways of metabolism (Gillette,
1974; Oesch, Raphael, Schwind and Glatt, 1977; Bentley, Oesch
563
-------
and Glatt, 1977; Billings, McMahon, Ashmore and Wagle, 1977)
which are not adequately taken into account with metabolic
activatina systems applied in vitro (Flesher, Harvey and Synder,
1976; Biggerf Tomaszewski and Dipple, 1978; Oesch, Bentley and
Glatt, 1976; Baird, Dipple, Grover, Sims and Brookes, 1973).
Similar reservations have been expressed even with pure com-
pounds with respect to predicting risk of heritable mutation
using bacterial and in vitro systems by the DHEW Working Group
on Mutagenesis Testing (Flamm, et al. 1977). The absence of
pharmacokinetic influences in bacterial and in vitro systems is
self-evident and requires no further elaboration. It may be
possible in the future to take such factors into account by
fairly simplistic and short-term animal experiments when dealing
with pure compounds. However, there is not even a clue as to
how such variables might be taken into account when dealing with
complex mixtures such as those found in drinking water.
It is apparent from work supported by our laboratory that
drinking water devoid of mutagenic activity is not likely to be
found. In Table 1 the mutagenic potency of organics isolated
from six drinking waters are compared. Details of the sample
preparation have been previously reported by Kopfler, et al.
(1977) and the Ames test results by Loper and Lang (1978). The
activities represent the total mutagenic activity that could be
associated with the indicated drinking waters, excluding cyto-
toxic fractions, and normalized to recovered organic material.
It is notable that the most potent mixture of chemicals was
recovered from Seattle drinking water. Seattle obtains its
drinking water from a protected watershed. New Orleans was con-
sidered to be one of the worst possible situations before this
work was initiated, since it derives its water from the indus-
trially impacted Mississippi River, yet it also ranks behind
Tucson, where the concentration of total organic carbon is so low
as to be unmeasurable. These data are placed in a somewhat
different perspective if mutagenic activity is normalized by the
volume of water from which they are obtained, however, Seattle
still ranks ahead of New Orleans. These results have alerted us
to possible carcinogenic hazards occurring in a number of drink-
ing waters. But, can we relate these results to the relative
carcinogenic risk posed by drinking water in these cities? We
do not think so.
Several investigators have called attention to the fact
that quantitative relationships between carcinogenic activity
and mutagenic activity in the Ames test are difficult to achieve.
These difficulties are evident within the major carcinogen
classes such as the polycyclic aromatic hydrocarbons (Andrews,
Thibault and Lijinsky, 1978a) and the nitrosamines (Meselson and
Russell, 1977; Andrews, Thibault and Lijinsky, 1978b) as well as
across chemical classes (Ashby and Styles, 1978a; Ashby and
Styles, 1978b).
564
-------
Table 1. Mutagenic Activity of Organic Compounds
Found in Municipal prinking Water
(Adapted from Lope- e\4 al., 1978)
in
TA 100
New Orleans
Miami
Philadelphia
Seattle
Ottumwa
Tucson
Type of
Water Supply
Surface
Ground
Surface
Surface
Surface
Ground
Presumed Source
of Contamination Net Colonies/mg
Industrial
Natural and
Industrial
Natural and
I ndustrial
Natural
Agricultural
Uncontaminated
41.4
46.4
193.0
282.0
6.9
91.4
Net Colonies/1
49.1
240.2
181.5
66.0
• 10.0
3.2
-------
Meselson and Russell (1977) were able to obtain 14 chemi-
cals for which the carcinogenic potency was sufficiently well
established to compare with mutagenic potency in the Ames test.
Without consideration of route of administration, a reasonably
good correlation2between carcinogenic and mutagenic potency
could be made (r = 0.88 for log potencies by linear regression)
as long as 4 nitrosamines were removed from consideration.
Inclusion of the nitrosamines reduced the correlation consider-
ably (r = 0.12 on linear regression). As an extension of this
same data we have selected 13 of these 14 compounds which had
been tested by the oral route. Where test results included
multiple species the average carcinogenic potency was utilized
and the results depicted in Table 2.
Table 2. Mutagenic and Carcinogenic Potency for
13 Compounds Tested by the Oral Route
(Adapted from Meselson & Russell, 1977)
Mutagenic Carcinogenic
potency M potency K
Aflatoxin B^
4-Aminobiphenyl
Benzo(a)pyrene
Dibenz (a, h) anthracene
4-4 ' -Methylene-bis- 2-
chloroanaline (MOCA)
Methyl methanesulfonate (MMS)
2 -Naphthalamine
1,3-propane sultone
Sterigmatocystin
Diethylnitrosamine
Dimethylnitrosamine
Di-N-butylnitrosamine
N-nitrosomethylurea
230
1.8
4.8
0.40
0.10
0.057
0.59
0.54
28
0.00093
0.0025
0.0097
0.43
405
0.31
5.5
0.47
0.41
0.032
0.083
1.4
21
9.3
6.7
0,32
0.10
r =0.14 based upon log potency
M = 1/pg required to produce 100 revertant colonies.
K = [-ln(l-I/Dtn+1)]
I = single risk incidence of tumor
D = dose rate
t = time in 2 year intervals
n = 3
As can be seen the correlation between Ames test potency and
carcinogenic potency is very low (r =0.14). It is apparent
that the Ames test has not yet progressed to the point that
changes in its activity can be taken to reflect comparable
changes in carcinogenic risk. On the other hand, the qualitative
566
-------
relationship between carcinogenic activity and mutagenicity in
the Ames test indicates cause for concern.
SHORT-TERM IN VIVO BIOASSAYS FOR CARCINOGENIC ACTIVITY
The somatic mutation theory opens the way for shorter term
in vivo assays for carcinogenic effects of chemicals as well as
supplying a rationale for bacterial and in vitro systems. A
number of experimental systems have been and are being developed
which are capable in the short-term of addressing the relative
carcinogenic potential of chemicals in vivo. These methods are
of essentially two types, involving either measurements of DNA
damage or cellular response (transformation). The former methods
may or may not quantitatively relate to a carcinogenic response;
the latter actually involve preneoplastic changes that should be
proportional to carcinogenic potency. Consequently, there are
certain advantages in interpretation in the latter case. The
relationship between levels of DNA damage and carcinogenic
potency is at present unresolvable because of questions relating
to tissue specificity, repair processes, the large variety of
eletrophilic intermediates generated with certain carcinogens,
and the multiplicity of potential nucleophilic binding sites for
such intermediates of DNA.
In Vivo Transformation Assays
Development of this group of bioassay methods have derived
much of their impetus from evidence indicating there are two
distinct stages involved in chemical carcinogenesis, first
demonstrated in the mouse skin by Berenblum (Berenblum and
Shubik, 1947). The stages are referred to as initiation and
promotion. Initiation, the first stage, is the irreversible
alteration of a cell to a potential cancer cell and can be
accomplished by a single application of a chemical. This stage
is hypothesized to result from a somatic mutation. Development
of tumors could then be brought about or accelerated by contin-
uous application of a second agent, referred to as a promoter.
Some chemicals possess exclusively one property or the other.
However, others possess both activities and are referred to as
complete carcinogens. Higher total doses of an individual
compound invariably are required for complete carcinogenesis
than for initiation. These initial observations of Berenblum
eventually resulted in the now classical bioassay system for
tumor initiation and promotion in the mouse skin.
The mouse-skin system is a very convenient model. It is
quite sensitive to certain classes of carcinogen and tumor
development is quickly evident in the form of papillomas without
sacrifice of animals. Papilloma formation appears as early as
5-6 weeks after beginning promotion, peaking at about 12-14
weeks. Development of carcinomas takes longer, approximately
one year. However, there is a very high degree of correlation
567
-------
between the initial development of papillomas and subsequent
development of carcinomas (Burns, et al., 1978). The disadvan-
tages of the system are that the skin is not the target organ
for many classes of chemical carcinogens; and that it lacks the
systemic pharmacokinetic influences. Thus, the data are of
limited usefulness for risk assessment of chemicals involving
the oral route of exposure.
Development of useful short-term systems for other target
organs entailed finding the counterpart of the papilloma in
these organs. These were first demonstrated by Farber (1973) as
regenerating or hyperplastic nodules in the livers of animals
fed different hepatocarcinogens. These nodules were shown to
develop into carcinomas if treatment was continued, but many
disappeared if treatment was terminated, a situation very similar
to that observed with skin papillomas. It was subsequently shown
that, although the nodule disappeared, transformed foci remained
which could be appropriately stimulated to yield carcinomas
(Farber, 1976). Prior to this finding (Friedrick-Freksa, Gossner
and Borner, 1969) and subsequently (Solt and Farber, 1976; Farber,
1978), it has been shown that within a short time following
initiating doses of hepatocarcinogens, small foci of altered
cells can be demonstrated histologically. These cells possess
alterations in morphology which appear identical to those
served in fully developed carcinomas (Sherer and Emmelot, 1975).
Subsequently, it has been demonstrated that the appearance of
enzyme-altered islands and their development into hepatocellular
carcinomas can be increased and accelerated by a variety of pro-
moting agents (Peraino, Fry and Grube, 1978; Pitot, Barsness and
Kitagawa, 1978; Farber and Solt, 1978). Phenotypic comparisons
of foci and hepatocellular carcinomas are compatible with a
progression of foci, to hyperplastic nodule, to tumor, and
substantiate operation of a two-stage process of chemical car-
cinogeiiesis in'the liver (Farber, 1973; Solt and Farber, 1976;
Farberjl976; Sirica, Barsness, Goldsworthy and Pitot, 1978).
The ability to promote lesions is extremely important to
the development of short-term in vivo assays. The number of
initiated cells produced does not vary between promoted and non-
promoted systems, but the ability to detect initiation depends
upon replication of the transformed cells. In the case of the
liver, it has now been well demonstrated that promotion substan-
tially increases the number of histochemically observable foci,
the development of hyperplastic nodules, and ultimately the
number of tumors developed in long-term experiments (Farber and
Solt, 1978; Peraino, Fry and Grube, 1978; Pitot, Barsness and
Kitagawa, 1978; Sirica, Barsness, Goldworthy and Pitot, 1978).
The enzyme-altered foci induced by carcinogens and subjected
to promotion are easily detectable at low power magnifications
within 3-4 weeks of carcinogen exposure. The sensitivity of a
bioassay based upon such systems can be estimated by examining
568
-------
the ratio of foci and tumors developed by such procedures vs. the
responses obtained without such procedures. Farber and Solt
(1978), using a promotion regimen involving subcarcinogenic doses
of 2-acetylaminofluorene (2-AAF) in the diet for 1 week followed
by partial hepatectomy, found approximately 103 foci/liver
in fischer rats resulting from a dose of 200 mg/kg diethylnitros-
amine (DEN). Seventy percent of animals so treated developed
hepatocellular carcinomas in nine months. With the same dose of
DEN and observation period no tumors were observed in Fischer
rats which had not been subjected_to 2-AAF and partial hepatec-
tomy. In another study, 4.9 x 10 foci/liver were observed
after 3 months in animals given 150 mg/kg DEN on a choline-
deficient diet, whereas animals subjected to the same dose of
DEN with a choline-sufficient diet displayed no foci. The effect
of choline deficiency could be increased by another 25-fold by
including partial hepatectomy in the experimental procedure
(Lombard! and Shinozuka, personal comm.). These results indi-
cate: measurement of foci is about 1000X as sensitive as long-
term carcinogenesis studies conducted with the same treatments,
even though results can be obtained within a month; and develop-
ment of foci is predictive of subsequent tumor development.
Change in the target organ specificity of a chemical carcin-
ogen is another feature of promotion. For example, dibenzanthra-
cene and urethane are initiators, but not complete carcino5ens
in the mouse skin. However, the skin does become a target' organ
for these compounds; that is, tumors are developed if initiation
by these compounds is followed by phorbol ester promotion.
Recently, Shinozuka and Lombard! (1978) demonstrated that the
choline-deficient diet greatly enhanced development of hepato-
cellular carcinomas and decreased evidence of pancreatic adeno-
carcinomas in rats exposed to azaserine. Without such promotion,
pancreatic tumors predominate in azaserine carcinogenesis
(Longnecker and Curphey, 1975). Consequently, it may be possible
to base a short-term in vivo bioassay for carcinogenic potential
on the basis of a single,'or a few, target tissues. Despite this
possibility the approach would be considerably enhanced if it
could be applied to other target organs. Although not as yet
extensively studied as the liver, identification of promoters
active in other target tissues [e.g., colon (Reddy, Weisburger
and Wynder, 1978), lung (Witschi and Lock, 1978), bladder (Hicks,
Chouaniec, and Wakefield, 1978), and pancreas (Shinozuka, Popp
and Konishi, 1976)] promises rapid progress in developing paral-
lel assays in other target organs.
At the Health Effects Research Laboratory of EPA in
Cincinnati, we are presently examining the utility of the two
established initiation/promotion systems in assessing the
hazards associated with organic chemicals in drinking water.
In preliminary experiments we have found that treatment of
coagulated, settled, and filtered Ohio River water with various
disinfectants used in drinking water treatment increased the
569
-------
carcinogenic potency of chemicals present in the mouse skin.
These experiments were conducted by injecting a total of 1.5 ml
of 100-180 fold reverse osmosis concentrates (celluose acetate)
subcutaneously in SENCAR mice followed by promotion with phorbol
myristate acetate (PMA). Preliminary results showing papillomas
data obtained through 18 weeks of exposure are shown in Table
3. Ranking of the potency of the reaction products was ozone >
chlorine • chloramines > chlorine dioxide. The specific activ-
ity of the compounds found in such water was approximately 1-2
percent of 7,12-dimethylbenzanthracene/ based on the total
organic carbon content of the original water. It is not possible
to determine what this means in terms of potential human health
hazard, although potencies of this magnitude are certainly not
negligible. Ames tests of these concentrates were negative
(Loper, personal comm.), most probably because of insufficient
concentration of the material present. Similar concentrates are
presently being used to confirm these results in mouse skin and
to extend the analyses to enzyme-altered foci in the liver of
rats in cooperation with Drs. Lombardi and Shinozuka of the
University of Pittsburgh. An additional aspect in the present
studies is to examine the influence of GAC removal of precursor
substances on the carcinogenic responses induced by disinfectant
treatment.
Table 3. Papillomas Resulting from Treatment of SENCAR-
Mice with Reverse Osmosis Concentrates3 of
Waters Treated with Alternate Disinfectants
and Promotion with PMAC.
Number of
Concentration Animals with Fraction Total Number
Treatment Factor*1 Tumors/Number of Tumors
Treated.
Non-disinfected
0.9% saline
7, 12-Dimethylbenz-
anthracene
C102
C12
ClxNH(3-x)
03
102
-
- .
168
106
142
186
0/25
0/25
16/25
0/25
4/25
5/2S
7/25
0
0
0.64
0.00
0.16
0.20
0.28
0
0
35
0
b
8
9
aDose of concentrate 6 x 0.25 ml s.c. injections over 2 weeks.
Coagulated, settled and dual-media filtered Ohio River Water.
CPMA = phorbol myristate acetate 2.5 vg in acetone to the skin
3 x weekly for 18 weeks.
Initial volume/concentrate volume.
570
-------
A second study has been initiated using SENCAR mice which
is attempting to determine the relative carcinogenicity of
organics extracted from the drinking water of five cities.
Concentration techniques were the same as indicated previously
for Ames testing. Early results of this study are shown in
Table 4. Although it is too early in the experiment to form
any firm conclusions, it is interesting to note that to date
an entirely different order of carcinogenic potency is emerging
from this work than was obtained with Ames testing. Whereas
Salmonella mutagenesis ranked the potency of chemicals derived
from the five cities (in descending order) as Seattle >
Philadelphia > Miami > New Orleans >Ottumwa, the initiating
potency appears to be New Orleans > Miami > Ottumwa >
Philadelphia > Seattle. Comparisons of the ROE and XAD frac-
tions of the five cities also reveal marked inconsistencies with
the mouse skin and Ames test in scoring relative potency. In
fairness it must be pointed out that these assays were not run on
identical samples, so that exact agreement is perhaps too much to
expect. On the other hand, blind spots created in Ames testing
of complex mixtures as a result of cytotoxicity may prevent
direct comparisons from being made. We have yet to encounter
toxicity problems in any of our SENCAR mouse studies.
DNA Damage
The somatic mutation theory of chemical carcinogenesis.
suggests that the initial (but not necessarily sufficient) event
leading to cancer involves an alteration in genetic material,
now known to be DNA. Consequently, if methods were available to
assess the damage done to DNA as a result of carcinogen exposure,
it should be possible to correlate such damage with the develop-
ment of tumors. In point of fact, this assumption underlies all
in vitro assays which utilize mutation as a measure of carcino-
genic activity. If one assumes that the differences in potency
of chemicals in vivo and in vitro is solely dependent upon
pharmacokinetic and metabolic considerations, measurement of DNA
damage in target tissues in vivo should rectify the differences.
A number of factors prevent immediate adoption of available
techniques for measuring nucleic acid damage in vivo as a sim-
plistic answer to the short-term assessment of carcinogenic risk.
Chemical carcinogens or their metabolites interact with
DNA at a variety of sites forming covalent bonds with purine and
pyrimidine bases or the phosphodiester backbone of the DNA chain
(see e.g., Singer, 1977, on nitrosamine adducts; Miller and Miller,
1977). Immediately after the damage is done, repair processes
are initiated which may be of either high fidelity or error-
prone. A question, the discussion of which is beyond the scope
of this presentation, exists as to whether initial DNA damage,
residual damage, or repaired damage would be the best measure of
carcinogenic potency.
571
-------
Table 4. Papillomas Resulting from Treatment of SENCAR
Mice with ROE and XAD Fractions of Municipal
Drinking Waters and Promotion with PMA for
16 Weeks
Sample
Number of Animals
with Tumors/Number
Treated
Total Tumors
ROE + XAD
Miami
ROE
XAD
Ottumwa
ROE
XAD
Philadelphia
ROE
XAD
Seattle
ROE
XAD
New Orleans
ROE
XAD
Solvent
DMBA 25 yg
a = Reverse osmosis
4/40
4/40
4/40
2/40
2/40
2/40
0/40
1/38
1/40
3/40
2/40
35/40
concentrate
4
5 9
5
2 7
2
2 4
1
1 2
3
7 10
2 2
261 261
extracted sequentially with pentane,
b =
methylene chloride and acidifyed methylene chloride taken to
dryness and combined (Kopflfcf 3t al. 1977).
Residual reverses-osmosis concentrate placed on XAD-2 and eluted with
ethanol (Kopflefr «t al. 1977).
c = Doses 6 X 25 rag/Kg over a 2 week period for all ROE and XAD samples.
572
-------
Only very limited attempts have been made to relate DNA-
damage induced in vivo with subsequent tumor development. The
alkaline elution method has been used to detect DNA damage
induced by single doses of N-nitrosodimethylamine, N-methyl-N-
nitrosourea, 1, 2-dimethylhydrazine, and cycasin (Parodjt
-------
in vitro mutagenic activity represents a proportionate reduction
in carcinogenic^hazard.
This fact has led us to the conclusion that evaluation
of treatment processes for drinking water must involve in vivo
methods. Consideration of the quantities of sample materials
required and the high cost of long-term carcinogenesis bioassays
precludes the use of these classical methods on a wide basis.
Consequently, our Laboratory is concentrating its efforts on
detecting the initiating activity of chemical carcinogens in
short-term in vivo animal experiments. Many of these bioassays
require less than 1 month to perform and approach in vitro
techniques both in terms of quantity of sample required and in
overall cost.
It is important not to be misled by the apparent simplic-
ity and lower costs of bacterial and in vitro test systems for
general screening of complex mixtures. In many instances,
inexpensive in vivo systems, such as the mouse skin initiation/
promotion assay, are likely to be more cost effective. This is
particularly true where cytotoxicity prevents determination of
activity by in vitro methods. The greater adaptive capabilities
of the whole animal allows carcinogenic potency to be assessed
where extensive fractionation, with the attendant problems of
recovery and artifact, would be required for adequate Ames
testing. The studies with SENCAR mice reported here indicate
that sensitivity of such systems is not necessarily inferior
to in vitro and bacterial systems, as is commonly supposed.
Consequently, demands on the amount of sample required per mouse
is certainly no more than that required per plate in the Ames
test.
The in vivo methods available are not limited to those
specifically addressed in this presentation. We have specifi-
cally chosen techniques which have high sensitivity to minimize
the concentration of water samples. Extensive concentra-
tion magnifies problems of recovery and artifact. If required
on a large scale, as would be required for long-term animal
experiments, concentration of organic constituents from drinking
water becomes the major cost involved. Although methods such as
in vivo cytogenetic analysis and chromosome aberrations will
undoubtedly be utilized, the sensitivity of such systems to low
levels of risk in vivo are clearly limited (see e.g., Trzos,
Petzold, Brunden and Swenberg, 1978).
Even using in vivo methods for determining carcinogenic
potency, the extrapolation of risk from experimental animals
to man is not without its difficulties (Schneiderman and Brown,
1978; Hartley and Sielken, 1977; Clayson, 1977). In some
respects we see the utilization of short-term in vivo assays
coupled with more adequate dosimetry as one means of improving
the ability of cross-species extrapolation of carcinogenesis
574
-------
data. If animal experiments and human epidemiology studies can
be normalized by better dosimetry, much better comparison of
carcinogenic potency may be realized in the context of the
following scheme:
Human
Animal
Target tissue
DNA-binding
I
Cancer
incidence
Alkylated
hemoglobin
Target tissue
DXA-binding
I
Short-term in_
vivo bioassays
I
Long-tern
Carcinogenesis
bioassav
The proposed dosimeter, alkylated hemoglobin, is based upon
the principle that the majority of chemical carcinogens appear
to be, or are metabolized to, electrophilic agents which bind
more or less indiscriminately with nucleophiles present within
the body (Miller and Miller, 1977). Systemicaiiy administered
carcinogens have been shown to bind to histidine and cysteine
residues in hemoglobin (Osterman-Golkar, et al., 1976; Osterman-
Golkar, et al.,~1977; Segerback, et al., 1978, Calleman, et al.,
1978; Pereirar unpublished results) in both man and experimental
animals. Since the bound residues are potentially traceable to
the exposure and represent an integration of both dose and metab-
olic activation, levels of alkylated hemoglobin can be adjusted
both in man and experimental animals to take into account the
influence of differing pharmacokinetic and metabolic factors.
Thus, controlling the level of alkylated hemoglobin has the
potential of being a powerful means of cross-species normaliza-
tion of dose-response information. Given this tool, the ability
to normalize initiating activity with eventual tumor incidence in
experimental animals can be extended to human populations with
documented and quantitated exposures to particular chemicals.
This will allow direct comparison of results from animal experi-
ments and human epidemiology.
The within-species validity of short-term in vivo assays
can be easily established if there is a consistent relationship
with tumor development in long-term carcinogenesis bioassays.
575
-------
To a very large extent this can be done by comparing results in
the short-term systems with already existing long-term data.
Development of epidemiological data on the basis of comparisons
of tumor incidence and levels of alkylated hemoglobin can then
serve as the means of comparing man's sensitivity to particular
carcinogens with that of animals. If this approach proves valid,
we will be in a position to assess with confidence the hazards
associated with chemically undefined mixtures found in drinking
water.
In conclusion, it must be recognized that there are a wide
variety of toxicological hazards potentially associated with
drinking water. A combination of factors, ranging from public
perception of the problems to the emergence of a critical level
of fundamental knowledge, which allows development of appropriate
short-term methods, has focused attention on the carcinogenic
and mutagenic hazards found in drinking water. As imperfect as
they are, these are the basis for the approach outlined here.
There are limitations in the basic research information available
to address other toxicological hazards such as potential develop-
mental toxicities, cardiovascular disease risk, neurotoxic
hazards, etc. resulting from organic chemicals in drinking water.
As the ability to address such problems effectively is developed,
these other toxicological endpoints will also have to become part
of the assessment.
576
-------
REFERENCES
1. Ames/ B.N., Ourston/ W.E., Yamasaki, E. and Lee, F.D. 1973,
Carcinogens and Mutagens: a simple test system combining
liver homogenates for activation and bacteria for detection.
Proc. Natl. Acad. Sci. (USA) 7J), 2281-2285.
2. Ames, B.N., McCann, J. and Yamasaki, E. 1975. Methods
for detecting carcinogens and mutagens with the Salmonella/
mammalian-microsome mutagenicity test. Mutation Research
31, 347-364.
3. Andrews, A.W., Thiebault, L.H. and Lijinsky, W. 1978a.
The relationship between carcinogenicity and mutageni-
city of some polynuclear hydrocarbons. Mutation Research
51, 311-318.
4. Andrews, A.W., Thiebault, L.H. and Lijinsky, W. 1978b.
The relationship between mutagenicity and carcinogenicity
of some nitrosamines. Mutation Research 51, 319-326.
5. Ashby, J. and Styles, J.A. 1978a. Does carcinogenic
potency correlate with mutagenic potency in the Ames
Assay? Nature 271, 452-455.
6. Ashby, J. and Styles, J.A. 19785. Factors influencing
mutagenic potency in vitro. Nature 274, 20-22.
7. Baird, W.M., Dipple, A., Grover, P.L., Sims, P. and
Brookes, P. 1973. Studies on the formation of hydrocarbon-
deoxyribonucleaside products by the binding of derivatives
of 7-methylbenz(a)anthracene to DNA in aqueous solution
and in mouse embryo cells in culture. Cancer Research 33,
2386-2392.
8. Bartsch, H. 1976. Predictive value of mutagenicity tests
in chemical carcinogenesis. Mutation Research 38, 177-190.
9. Bentley, P., Oesch, F. and Blatt, H. 1977. Dual role of
epoxide hydratase in both activation and inactivation of
benzo(a)pyrene. Arch. Toxicol. 39, 65-75.
10. Berenblum, I. and Shubik, P. 1947. A new, quantitative
approach to the study of the stages of chemical carcino-
genesis in the mouse's skin. Br. J. Cancer !_, 373-391.
577
-------
11. Bigger, C.A.H., Tomaszewski, J.E. and Dipple, A. 1978.
Differences between products of binding of 7,12 dimethyl(a)
anthracene to DNA in mouse skin and in a rat liver micro-
somal system. Biochem. Biophys. Res. Comm. 80, 229-235.
12. Billings, R.E., McMahon, R.E., Ashmore, J. and Wagle, S.R.
1977. The metabolism of drugs in isolated rat hepatocytes.
A comparison with in vivo drug metabolism and drug metabo-
lism in subcellular liver fractions. Drug Metabolism and
Disposition J>, 518-526.
13. Brusick, D.J. 1978. The role of short-term testing in
carcinogen detection. Chemosphere £, 403-417.
14. Buecheler, J. and Kleihues, P. 1977. Excision of O -
methylguanine from DNA of various mouse tissues following
a single injection of N-methyl-N-nitrosourea. Chem.-Biol.
Interactions 16, 325-333.
15. Burns, F.J., Vanderlaan, M., Snyder, E. and Albert, R.E.
1978. Induction and progression kinetics of mouse skin
papillomas. In: Carcinogenesis A Comprehensive Survey
Vol. 2 (Eds. Slaga, T.J., Sivak, A. and Boutwell, R.K.)
pp. 91-86, Raven Press, N.Y.
16. Clayson, D.B. 1977. Relationships between laboratory and
human studies. J. Environ. Pathol. Toxicol. 1, 31-40.
17. Farber, E. 1973. Hyperplastic live nodules. In: Methods
in Cancer Research Vol. 7 (Busch, H., ed) pp. 345-375,
Academic Press, N.Y.
18. Farber, E. 1976. On the pathogenesis of experimental
hepatocellular carcinoma. In: Hepatocellular Carcinoma,
(Okueda, K. and Peters, eds.) pp. 1-24, Wiley, New York.
19. Farber, E. and Solt, D. 1978. A new liver model for the
study of promotion. In: Carcinogenesis Vol. 2. Mechanisms
of Tumor Promotion and Carcinogenesis (Slaga, T.J., Swak, A,
and Boutwell, P.K., eds). pp. 443-448. Raven Press, N.Y.
20. Flamm, W.G., Chairman DHEW Working Group on Mutagenicity
Testing 1977. Approaches to determining the mutagenic
properties of chemicals: Risk to future generations.
J. Environ. Pathol. & Toxicol. 1^, 301-352.
21. Flesher, J.W., Harvey, R.G. and Synder, K.L. 1976. Onco-
genicity of K-region epoxides of benzo(a)pyrene and 7,12-
dimethylbenz(a)anthracene. Int. J. Cancer 18, 351-353.
578
-------
22. Friedrich-Freksa, H., Gossner, W., and Bonner, P. 1969.
Histochemische Untersuchungen der Cancerogenese in der
Rattenleber nach Dauergaben von Diathyl-nitrosamin.
Z. Krebsforsch. 72, 226-241.
23. Gillette, J.R. 1974. A perspective on the role of
chemically reactive metabolites of foreign compounds in
toxicity - II. Alterations in the kinetics of covalent
binding. Biochemical Pharmacology 23, 2927-2938.
24. Gillette, J.R. 1977. Kinetics of reactive metabolites
and covalent binding in vivo and in vitro. Biol. React.
Intermediates (Proc. IntT C~ohf. ) pp. 25-41.
25. Hartley, H.O. and Sielken, R.L. Jr. 1977. Estimation of
"safe doses" in carcinogenic experiments. J. Environ.
Pathol. and Toxicol. .1, 241-278.
26. Hoel, D.G. 1977. Some problems in low-dose extrapolation.
In: Origins in Human Cancer. (Hiatt, H.H., Watson, J.D.
and Winsten, J.A., eds). pp. 1391-1396. Cold Spring Harbor
Laboratory Conferences on Cell Proliferation. ,
27. Kopfler, F.C., Coleman, W.E., Melton, R.G. and Tardiff, R.G.
1977. Extraction and identification of organic micropollu-
tants: Reverse Osmosis Method. Ann. N.Y. Acad. Sci. 2»8,
20-30.
28. Koropatnick, D.J. and Stich, H.F. 1976. DNA fragmentation
in mouse gastric epithelial cells by precarcinogens, ulti-
mate carcinogens and nitrosation products: An indicator for
the determination of organotrophy and metabolic activation.
Int. J. Cancer 7_» 765-772.
29. Longnecker, D.A. and Curphey, T.J. Adenocarcinoma of the
pancreas in azaserine-treated rats. Cancer Research 35,
2249-2258.
30. Loper, J.C., Lang, D.R., Schoeny, R.S., Richmonoc, B.B.,
Gallagher, P.M. and Smith, C.C. 1978. Residue organic
mixtures from drinking water show in vitro mutagenic and
transforming activity. J. Toxicol. Environ. Health 4,
919-938.
31. McCann, J., Choi, E., Yamasaki, E., Ames, B.N. 1975.
Detection of carcinogens as mutagens in the Salmonella/
microsome test: Assay of 300 compounds. Proc. Nat. Acad.
Sci. (USA) _72, 5135-5139.
579
-------
32. Meselson, M. and Russell, K. 1977. Comparisons of carcino-
genic and mutagenic potency. In: Origins of Human Cancer
Book C. (Hiatt, H.H., Watson, J.D. and Winsten, J.A., eds).
pp. 1473-1481. Cold Spring Harbor Labortory Conferences on
Cell Proliferation.
33. Miller, J.A. and Miller, E.G. 1977. Ultimate chemical
carcinogens as reactive mutagenic electrophiles. In:
Origins of Human Cancer (Hiatt, H.H., Watson, J.D. and
Winsten, J.A., eds.) pp. 605-628. Cold Spring Harbor
Laboratory Conferences on Cell Proliferation.
34. Oesch, F., Bentley, P. and Glatt, H.R. 1976. Prevention of
benzo(a)pyrene-induced mutagenicity by homogeneous epoxide
hydratase. Int. J. Cancer 18, 448-452.
35. Oesch, F., Raphael, D., Schwind, H. and Glatt, H.R. 1977.
Species differences in activating and inactivating enzymes
related to the control of mutagenic metabolites. Arch.
Toxicol. 3_9, 97-108.
36. Parodi, S., Taningher, M., Santi, L., Cananna, M.,
Sciaba, L., Maura. A. and Brambella, G. 1978. A practical
procedure for testing DNA damage in vivo proposed for a
pre-screening of chemical carcinogens. Mutation Research
54, 39-36.
37. Peraino, C., Fry, R.J.M . and Staffeldt, E. 1971. Reduction
and enhancement by phenobarbital hepatocarcinogenesis in-
duced in the rat by 2-acetylaminofluorene. Cancer Research
31, 1506-1512.
38. Peraino, C., Fry., R.J.M. and Grube, D.D. 1978. Drug-
induced enhancement of hepatis tumorigenesis. In:
Carcinogenesis Vol. 2. Mechanisms of Tumor Promotion and
Carcinogenesis (Slaga, T.J., Sivak, A. and Boutwell, R.K.,
eds). pp. 421-432. Raven Press, N.Y.
39. Petzold, G.L. and Swenberg, J.A. 1978. Detection of
DNA damage induced in vivo following exposure of rats to
carcinogens. Cancer Research 38, 1589-1594.
40. Pitot, H.C., Barsness, L. and Kitagawa, T. 1978. Stages
in the process of hepatocarcinogenesis in rat liver. In:
Carcinogenesis Vol. 2. Mechanisms of Tumor Promotion
and Carcinogenesis. (Slaga, T.J., Swak, A. and Boutwell,
R.K., ed.). pp. 433-442. Raven Press, N.Y.
41. Scherer, E. 'and Emmelot, P. 1975. Kinetics of induction
and growth of precancerous liver-cell foci, liver tumor
formation by diethylnitrosamine in the rat. Eur. J.
Cancer 11, 689-701.
580
-------
42. Sheridan, R.B. Ill and Huang, P.C. 1977. Single strand
breakage and repair in eukaryotic DNA as assayed by S
nuclease. Nucleic Acids Research £, 399.
43. Shinozuka, H. and Lombardi, B. 1978. Change in organ
susceptibility of azaserine carcinogenesis in rat fed a
diet devoid of choline. In J. Cancer 22, 36-39.
44. Shinozuka, H., Lombardi, B., Sell, S. and Immarino, R.M.
1978. Early histological and functional alterations of
ethionine liver carcinogenesis in rats fed a choline
deficient diet. Cancer Research 38, 1092-1098.
45. Shinozuka, H., Lombardi, B., Sell, S. and Immarino, R.M.
1979. Enhancement of ethionine liver carcinogenesis in rats
fed a choline devoid diet. J. Natl. Cancer Inst. In press.
46. Shinozuka, H., Popp, J.A. and Konishi, Y. 1976. Ultra-
structures of atypical acinar cell nodules in rat pancrease
induced by 4-hydroxyaminoquinoline-l-oxide. Laboratory
Investigation 34, 501-509.
47. Singer, B. 1977. Sites in nucleic acids reacting with
alkylating agents of differing carcinogenicity or mutagen-
icity. J. Toxicol. and Environmental Health 2^, 1279-1295.
48. Sirica, A.E., Barsness, L., Goldsworthy, T. and Pitot, H.C.
1978. Definition of stages during hepatocarcinogenesis
in the rat: potential application to the evaluation of
initiating and promoting agents in the environment. J.
Environ. Pathol. Toxicol. 2_, 21-28.
49. Solt, P. and Farber, E. 1976. New Principle for the analy-
sis of chemical carcinogenesis. Nature 263, 701-703.
50. Trosko, J.E. and Chang, C.-C. 1978. Environmental carcino-
genesis: An integrative model. Quarterly Review of Biology
15,3, 115-141.
51. Trzos, R.J., Petzold, G.L., Brunden, M.N. and Swenberg, J.A.
1978. The evaluation of sixteen carcinogens in the rat
using the micronucleus test. Mutation Research 58, 79-86.
52. Watanabe, P.G., Young, J.D. and Gehring, P.J. 1977. The
importance of non-linear (dose-dependent) pharmacokinetics
in hazard assessment. J. Environ. Pathol. & Toxicol. 1,
147-159.
581
-------
SPECIFIC ANALYSIS OF TRACE ORGANICS IN WATER
USING GAS CHROMATOGRAPHY AND MASS SPECTROSCOPY
Martin Reinhard, Joan E. Schreiner,
Tom Everhart, and James Graydon
INTRODUCTION
Organic contaminants in water must be identified specific-
ally and measured quantitatively if the mechanisms and efficien-
cies of their removal and formation in treatment processes are
to be investigated. General parameters such as Total Organic
Carbon, Chemical Oxygen Demand, and Total Organic Chlorine give
a good indication of the overall performance of a process or a
plant. However, they do not reflect the behavior of all the
classes of organic compounds normally present in waste and
polluted surface waters (1).
The need for specific compound analysis stems also from
governmental, legal, and regulatory action. The "EPA Consent
Decree" requires limitations or guidelines for as many as 129
pollutants, of which 114 are organic. The analytical implica-
tions and problems encountered in backing up these regulations
have been discussed recently by L.H. Keith and W.A. Telliard (2).
It is apparent that no straightforward approach is available
today to measure all 114 pollutants.
This paper presents an approach for the analysis of a broad
range of trace organic compounds, as it has evolved during a
study on trace organics removal at the advanced wastewater treat-
ment plant Water Factory 21 in Orange County (1). Four different
extraction and chromatographic procedures have been combined in
an integrated analytical scheme using gas chromatography for
quantitation and gas chromatography/mass spectroscopy for
qualitative analysis.
ENRICHMENT OF ORGANIC POLLUTANTS
The procedures for the extraction and enrichment of organic
compounds from water samples are based on favorable thermodynamic
equilibria. To predict their behavior in an analytical scheme,
it is necessary to know the thermodynamic constants which control
these equilibria. Constants for predicting the partitioning in
a water/gas or water/solvent system include the Henry's Law con-
stant and the n-octanol/water partition co-efficient (Koct). in
582
-------
Table 1 these constants are given for a set of representative
compounds that are frequently found in water (3).
The Henry's Law constant relates the partial pressure of a
volatile solute to its concentration in a pseudo-ideal dilute
solution. The value of the constant can be approximated with
reasonable accuracy from the ratio of the vapor pressure of the
pure compound to the solubility limit of the solute in water [4].
The higher this constant, the better a compound volatilizes from
water.
The n-octanol/water partition coefficient has been tabulated
by Hansch and coworkers (5). The logarithm of the partition co-
efficient is the Hansch-parameter, P0ct- Tne higher the value
of Poet tne more strongly a compound partitions into a hydropho-
bic organic solvent and the less solvent will be required for an
efficient extraction.
Enrichment by Purge-and-Trap
A popular approach to concentrate volatiles from water
samples is the purge-and-trap technique, where a continuous gas
stream is bubbled through the water sample and the purged
organics are trapped with a suitable adsorbent. The kind and
amount of adsorbing material has to be chosen and optimized care-
fully. Bellar and Lichtenberg use TENAX adsorbent to measure
the highly volatile compounds ranging from methylene chloride to
chlorobenzene (6). Similarly, Grob and coworkers use milligram
amounts of activated carbon (AC) to enrich a broad variety of
compounds (7).
_, All compounds in Table 1 with a Henry's Law constant above
10 atm • mol m are purged by gaseous stripping; we can assume
that compounds with somewhat lower constants are at least pa£z
tially stripped. However, compounds with constants below 10~
are expected not to be recovered efficiently, as is the case for
phenol (7). The overall recovery is affected not only by the
stripping process, but also by the efficiency of the adsorption
onto the AC filter and by the desorption from it. "Highly vola-
tile compounds are not well adsorbed (e.g., methylene chloride
and chloroform) and therefore need to be trapped with an adsor-
bent having a relatively large surface. Desorption from a high-
surface adsorbent with a solvent or by heating in a gas stream
leads to a relatively high dilution and to a lesser concentration
effect.
The Grob procedure uses the headspace of the apparatus
(Figure 1) to purge the volatiles and a pump to recycle the
air. Thus, no gas need be introduced from outside and contamina-
tion from gas supplies is eliminated. This is essential since
583
-------
Table 1. Physical Constants of Representative Organic Pollutants3
Substance
Dichlorome thane
Chloroform
Benzene
Toluene
Tetrachloroethylene
Phenol
in Naphthalene
00
** Benzoic Acid
f
DDT
Molec-
ular
Weight
84.9
119.4
78.1
92.1
165.8
94.1
128.2
122.1
354.5
Boiling
Point
Solubil-
ity in
Water Vapor
(°C) (mg/1) Pressure
1 atm
40
62
80
111
121
182
218
249
— .
ai
K. Verscheueren, Handbook of Environmental
20°C (mm)20°C
20,000
8,000
1,780
515
140
82,000d
33
2,900
-3
1.2x10
Data on Organic
349
160
76
22
76
0.2e
0.23
0.26
_7
1x10
Chemicals,
Henry ' a Law
Constant
1 "i
atm»mol • m
1.9x10~3
3.1x10~3
4.4x10~3
5.2x10~3
-2
2.2x10
-3x10~7
-3
1.2x10
1.1x1o"6
-"5
3.9x10
Von Nostrand
>
Poct
1.59
1.97
2.13
2.69
2.88
1.46
3.20
1.87
rt
6.19g
Reinhold Co. ,
Analh
(A)
A
(B)
B
A
E
B
G
C
If.
1977.
b
Approximated from the ratio of the vapor pressure of the pure compound to the solubility limit of cue
solute in water (4).
c
P = logarithm of n-octanoI/water partition coefficient tabulated in Reference 5. (Underlined values
are calculated based on substituent constants.)
'Value for 15°C.
p
Kirk-Othmer, Encyclopedia of Chemical Technology, 2nd ed., Wiley & Sons Interscience, New York, 1967.
D. Mackay and P.J. Leinonen, Environ. Sci. Tech., 9(13), 1178 (1975).
R.D. O'Brien, Environmental Dynamics of Pesticides, Hague and Freed, eds., Plenum Press, New York, NY,
1975.
See Table 2 for code of designations for analytical methods.
-------
the purging time can then be extended so that slowly purged sub-
stances are trapped quantitatively at the activated-carbon
filter.
CLOSED-LOOP GASEOUS AIR STRIPPING
GAS
HEATER
HEATER
ACTIVATED CARBON
FILTER HOLDER
L,
•^
MH
fc.
>>
1
)
WAT
SAM
L
^
I
ER
RLE
J
VVVV 1 |
PUMP
THERMOSTATIC WATERBATH
Grob and Ziircher
J. Chromatogr. 1976
Figure 1. Schematic of Closed-Loop Stripping Apparatus
The flux J of a solute i across the phase boundary A, when
liquid film resistance dominates, is given by Eq. 1 (4):
P.
T
Ji
dt
OL
(i)
585
-------
where Q is the mass of the solute removed, KQL is the overall
mass transfer coefficient, C. the concentration of compound i
in the liquid phase, P. is tne partial pressure in the bulk gas
phase, and H. is the Henry's Law constant of compound i. It is
evident from this equation that if H. becomes too small, the
flux is reduced to such a degree thai extremely long purge times
would have to be applied to achieve a significant transfer. On
the other hand, if the activated-carbon filter does not adsorb
the volatiles from the headspace, they are recycled with the air
into the water sample, and no significant removal occurs. K
is not known; however, it was found that smaller compounds of low
molecular weight are removed within minutes when the purge gas is
flowing.
To achieve optimum reproducibility, a number of measures
have to be taken that are discussed by Grob and Zurcher [8].
These include temperature stabilization of the sample, preheat-
ing of the air stream prior to adsorption to reduce relative
humidity, and preconditioning of the adsorbent filter. For
quantitative determinations, recovery factors for each compound
must be determined. This is particularly true for compounds
which belong to a different compound class or differ consider-
ably in their physicochemical constants from those for which
the method has been calibrated previously.
For heavily loaded waters, smaller samples must be used
since overloading of the AC filter may occur. Smaller samples
are stripped in smaller flasks with equal headspace. Variations
in the relative volume of the gas and liquid phases severely dis-
tort the recovery of the compounds. This is evident from Figure
2, which depicts the ratio of the peak areas of 1-chlorooctane
(Cl-C ) to 1-chloro-dodecane (Cl-C..-) as a function of increased
headspace and decreased water volumes. It is evident that the
recovery of the more volatile Cl-Cg decreases as headspace
increases. This effect is expectea to be similar for all vola-
tile compounds in the system and was, in fact, observed by Grob
and coworkers. ' For benzene, these authors found a recovery of
only 20 percent (7), presumably due to its insufficient recovery
on the AC filter.
Elution of the organics from the AC filter has thus far been
quantitative and has not been found to limit the range of the
analysis if 20 yl carbon disulfide are used.
Enrichment by Solvent Extraction
Organic compounds partition from water into an immiscible
organic phase in constant ratios. A comprehensive collection of
such partition coefficients has been assembled by Leo, Hansch,
and Elkins (5). They chose the partition coefficient in
n-octanol/water, K . as a reference and calculated incremental
substituent constants'. With these substituent constants it is
586
-------
possible to estimate the partition coefficient for a compound
when experimentally determined value for a parent compound is
available. Log Koct (or Poct) is a measure of the lipophilic
nature of a compound. The higher its Poct, the more a compound
will partition into a hydrophobic phase. DDT has a Poct greater
than 6 and is easily extracted from one liter of water with
2 x 15 ml hexane. In contrast, tor tne extraction of the more
hydrophilic phenol, (P = 1.46), repetitive extraction with
approximately several BSndred ml of methylene chloride is re-
quired. For acidic and basic compounds, the pH must be adjusted
prior to extraction. This effect is used for a preseparation of
the organics in acidic and basic fractions.
I.I
P**oreoCI-C. ^ ^^ jn ^^
Peek oreo Ct— C,2
2.0.9
u
r o
0.7
u
I
o 0.5
0.3
0 IOO 200
D.|. 0 18 36
300 INCREASE IN HEADSPACE
55 % (550ml «IOO%)
Figure 2. The Influence of the Headspace Volume (in ml) on the
Relative Recovery of the Internal Standards Cl-C and
Cl-C-2. The dotted line represents the value of
direct injection (D.I.) of standard solution.
THE ANALYTICAL SCHEME
These considerations and experiences have led to an inte-
grated analytical scheme for the analysis of a broad range of
compounds, in which four different extraction and enrichment pro-
cedures are combined (Figure 3). Highly volatile compounds are
analyzed as 50-ml samples by pentane extraction. The analysis is
termed volatile organic analysis (VGA) and is specific for the
haloforms and chlorinated compounds with one and two carbons
(Group A).
587
-------
add IS A
SOmlsamote
— >| pent. ex.
GC/ECD
add IS B-G
1000 ml sample
.
-
CLS 2hrs
\
OY ?r
f
i ..i r.c
i
t
pHU7
hex. ex.
i
1
cone.
and or
ctean-up
pH*
, 1
MeCI2 ex.
,
r
evaporate t
dissolve et
I GC/FID or MS
methylation with
diazomethane
GC/RD,MS,ECD
Figure 3. Analytical Procedures Applied for Compound Groups
A through G (see Table 2)
Compounds of medium volatility and Itfw water solubility
(Group B) are measured by the Grob Closed-loop Stripping Analy-
sis (CLSA) utilizing 100-1000 ml samples. Heavier non-polar
compounds (Groups C and D) are extracted with hexane at pH 7.
Finally, the more polar and acidic compounds are extracted with
methylene chloride after lowering the pH to 1. With this latter
procedure, aliphatic and aromatic acids are extracted along with
phenolic and other compounds having greater water solubility.
For quantitation, the internal standards for Groups B through G
are added to the water sample prior to closed-loop stripping if
the entire analytical scheme is to be applied. Internal stan-
dards, groups of compounds analyzed, and procedures are summa-
rized in Table 2. The applicability of this scheme depends on
the type of water to be analyzed, other types of compounds
present, and other factors. It is still preliminary and can by
no means be considered final or universal.
EXPERIMENTAL
Group A; Volatile Organic Analysis (VOA)
The procedure to analyze volatile organics was adopted from
Henderson, Peyton, and Glaze (10) without modifications. The
588
-------
50-ml samples were shipped in sealed sample bottles and ex-
tracted in the same bottles after replacing 1 ml of the water by
1 ml of pentane. The pentane had been cleaned on an aluminum
column (Woelm basic), activity I (ICN Pharmaceuticals, Inc.,
Life Sciences Group, Cleveland, Ohio 44128). 5 yl of the ex-
tract were then injected onto a 2 m x 4mm i.d. column packed
with 10% squalane on Chromosorb W/aw (80-100 mesh) at 67°Cg3
Argon/methane was used for a carrier gas and a linearized Ni
detector (Tracer, Austin, Texas 78721) with a Spectra Physics
System I integrator was used for detection and quantitation.
Table 2. Summary of Procedures, Compounds, and
Internal Standards
Group
A
3
V.
Q
£
r
G
Procedure
VOA
CLSA
HEA
HEA
MEA
ME A
MEA
Compound
haloforms
1- and 2-carbon
halogenated
solvents
aromatic hydro-
carbons
chlorinated
hydrocarbon
pesticides
poly chlorinated
oiphenyls
polynuclear
aromatic
hydrocarbons
aliphatic acids
aromatic acids
phenolic com-
pounds
Internal Standard
1, 2-d ibromoe thane
1-chlorooctane (Cl-Cg)
1-chlorododecane (Cl-C,.)
1-chlorohexadecane (Ci-C^ )
a, i, a1, =" -2,3,5, 6-octa-
chloro-p-xylene
3,6-dimethylphenanthrene
6-bromohexanoic acid
m-bromosalicylic acid
p-bromophenol
Supplier
M
E
E
E
A
AN
A
A
A
Reference
110)
18]
•
Suppliers: M = MC/B, Norwood, Ohio 45212; E = Eastman, Rochester, New York 14650;
A = Aldrich, Milwaukee, Wisconsin 53201; AN = Analabs, Inc., No. Haven,
Connecticut 06437.
The compounds for which calibration factors, detection
limits, and standard deviations have been determined are listed
in Table 2.
The detection limits are near 0.1 yg/1. This has been
shown to be sufficient to measure their removal in advanced
waste treatment systems (1). The number of very volatile com-
pounds that can be detected with this procedure is limited
because of the interference of pentane. If, for instance,
methylene chloride is to be determined by the VOA procedure,
pentane is replaced by octane which elutes later, and thus makes
the detection possible.
589
-------
Group B; Closed-Loop Stripping Analysis (CLSA)
The closed-loop stripping analysis was adapted from Grob
and coworkers (7,8). Gas-washing bottles, 100, 250, 500, and
1000 ml volumes were modified so that they could be connected to
an all-steel or teflon diaphragm pump (Metal Bellows, Sharon,
Mass. 020677 -a~s shown in Figure 1. After placing the sample in
the bottle and adding the appropriate amount of internal stand-
ards, an AC filter was placed into the filter holder and the
headspace purged by running the pump for 30 seconds to remove
contaminants from jthe headspace air. Following this, a clean AC
filter was placed in the filter holder and proper air flow,
approximately 1.5-1 min, was established. After two hours the
AC filter was extracted sequentially with_ 3 portions of 7 ul
carbon disulfide, approximately 14 yl of which could be recov-
ered. Of this extract, 1.5 ul was then injected, splitless (9),
for 42 seconds onto a 50 m UCON HB column (purchased from Jaeggf
GC Laboratory, Trogen, Switzerland) at 40°C. After 4 minutes,
the temperature was increased by 3°C per minute to 180°C and held
there for 10 minutes. ^l-C,- was used as a standard for
quantitation since this standard has shown the most reproducible
recovery. The signal from the flame-ionization detector was
recorded simultaneously by a strip-chart recorder (Linear
Instruments 300 series) and a recording integrator (Perkin-Elmer,
Sigma 10) for computer areas and concentrations.
Groups C and D; Hexane Extraction Analysis (HEA)
One liter samples were extracted in a 2 liter separatory
funnel twice with 15 ml of redistilled hexane. The two phases
were separated after allowing the extracted sample to stand for
a minimum of 6 hours. The two extracts were then combined and
dried over sodium sulfate. After concentration to one ml with a
gentle nitrogen stream, the extracts were chromatographed on a
small florisil column (10). Chlorinated hydrocarbon pesticides,
PCBs, and similar compounds were eluted with 10 ml hexane/ether
185/15 v/v) and reconcentrated to 1 ml or less. The concentrated
extract was then analyzed on a 50-m SE 54 glass-capillary column
(same supplier as UCON HB) which was mounted in a GC (Finnigan
9610) equipped with a Grob injector (E.M. Becker Co., Bala
Cynwyd, Pennsylvania) and a wide-range electron-capture detector
(Valco Instruments, Houston, Texas). Samples from sufficiently
clean water supplies did not require pre-separation on florisil
and were analyzed directly after drying and concentration of the
combined hexane extracts.
Aliquots of 1.5-2.0jul were injected splitless for 42
seconds onto the' 170°C column. After 13 min isothermal phase,
the temperature was raised to 230°C and held ^t this temperature
for four minutes. The injector temperature was held at 220°C
and" * be •detector temperature was set at 250°C. Helium was used
as carrier gas with the column inlet pressure adjusted to 1 atm.
590
-------
Detector scavenger gas was introduced by means of a heat ex-
changer, fabricated with 1/8 inch stainless steel tubing and a
SwagelockR fitting. The straightened capillary end was inserted
all the way through the heat exchanger and into the detector
cell;, The BCD signal was recorded by an integrator (Perkin-
Elmer, Sigma 10) for quantitation. Group D compounds (poly-
nuclear aromatic hydrocarbons) were analyzed only qua litatively
by GC/MS.
Groups E, F, and G; Methylene Chloride Extraction at pH 1 (MEA)
The samples were acidified to pH 1 or less by the addition
of hydrochloric acid and then extracted four times with 20 ml
redistilled methylene chloride. The amount of methylene chloride
used for extraction has been increased in subsequent analyses in
order to quantitatively extract the hydrophilic phenols. Emul-
sions which formed in heavily polluted water samples were sepa-
rated by centrifugation. The extracts were combined and dried
over anhydrous sodium sulfate, filtered through sintered glass,
and evaporated to a volume of 10 ml in a rotary evaporator at a
water-bath temperature of 30°C and a pressure of 1/2 atm. After
transferring the sample to a 10-ml evaporative concentrator tube,
it was taken to dryness by the airstream-water-bath method.
(R)
Diazomethane for methylation was produced from Diazaldv '
(N-methyl-N-nitroso-p-toluenesulfonanide, Aldrich Chemical Co.,
Milwaukee, Wise. 53233) by the following procedure:
An apparatus for downward distillation was constructed from
a Diazald^ ' Kit (Aldrich Chemical Co.), featuring Clean-
Seal * joints, according to the configuration suggested by
Aldrich. Approximately 2.15 g (10 mmoles) of Diazald were
transferred to the 250-ml distilling pot along with a magnetic
stirring bar, followed by the addition of 30 ml of ethyl ether.
As the pot was cooled in an icewater bath, the Diazald was
stirred into solution, and 13.0 ml of a 0.81N solution of KOH in
95% ethanol was slowly dropped into the yellow solution. Upon
completion of the addition, the flask was allowed to warm to
room temperature and the ethereal-alcoholic diazomethane solution
was slowly distilled into the dry-ice-cooled receiving flask.
The yield from 10 mmoles of Diazald is 320 to 350 mg or 7.6 to
8.3 mmoles.
The sample residue in the concentration tube was redissolved
in one ml of a 50/50 (v/v) ether-methanol solution, and 0.5 ml of
methanol was added to the reconstituted sample, followed by at
least two Pasteur pipettes full of the reagent solution. After
mixing, the sample was allowed to sit in the hood overnight, and
the next day, the solution was concentrated to a volume of 100 ml
by a steam-water-bath method. The volume was made up to 200 ml
by the addition of 100 ml of methanol and after mixing, the
sample was ready for analysis by GC and GC/MS. Approximately
2 ml were injected splitless for 42 seconds onto a 50 m SE 54
591
-------
glass capillary column at 80°C. After 3 min, the temperature
was programmed to 250°C at a rate of 3°C/min. The isothermal
period was 20.0 min. The pressure of the carrier gas (helium)
was adjusted to 1 atm.
RESULTS
Chromatographic Analysis
In general, the extracts obtained from polluted water
samples are complex mixtures which require high resolution chro-
matography and/or specific detectors for the analysis of speci-
fic compounds. In many cases, even the combined application of
high-resolution gas chromatography and specific detectors does
not provide enough specificity, and combined gas chromatography-
mass spectroscopy must be used for specific detection of indi-
vidual compounds. The applicability of each separation and the
detection method used depends on the type of water to be analyzed
and must be adapted and tested in each case.
The VOA procedure uses packed columns with an electron-
capture detector. The specificity of the method seems adequate
for advanced treated wastewater as encountered in Water Factory
21. However, interferences from unknown peaks occasionally have
been encountered.
For the CLSA extracts, high-resolution chromatography was
deemed essential: Figure 4 depicts the chromatograms obtained
from extracts of Water Factory 21 influent Ql, air-stripping
tower effluent 04, and the final chlorinated effluent 09. Three
internal standards were added (1-chloro-octane, 1-chlorododecane,
and 1-chloro-hexadecane) prior to closed-loop stripping; the
corresponding peaks are labeled in the chromatogram as IS-9,
-15, and -20. The compounds listed in Table 3 have been quanti-
fied with respect to Cl-C^ as an internal standard.
Figure 5 shows a chromatogram of the standard mixture of
compounds analyzed by the SEA method. The temperature was opti-
mized such that all the compounds to be quantified were sepa-
rated. The peaks labeled with letters belong to the profile of
aroclor 1242. The internal standard used was«<,oe,
-------
i
!
i
20 !
IS [
1
, •!
1 d 1 i
i
S
20
I IS
1 :
|
1
20
v— '^A,* \,. 'i IS
~^J ' \
' — • —
ISO '?°"
50,67
19
{
1
-'j
16
17
in
o
1
I
'! ! ii r
|: ' l! . »
46
17
18
1
(HI.
/L^^^^
16
— ^u_^_
;/^
— *^-A™i
21
—— -_
Q 1
IS > 5OOng/{
15
IS
14 13
i i
J : |!' '
i 1< i! i
. ^^U^l^'WM^^
11 Q4
IS 'SOOng/*
^i-t-^UV1
Bill 9 1
I'
f
ij
I:
i: .
i
,'L
Ji
14 12
' '? } ''
^^-U-M,J J
15
is Q9
IS > 200ng/t
14
1
TEMP CO
^— ' H.
«
f II
is
4 ;
6^
7l
8
^vWln
9
IS
10
^-'V*^^'
8
i
l
j
9
IS
10
-*'i AAnAJ..^-" ^A
T
/I'
RATE 3Vmin
TIME (min)
1 1
II
-
.
-
-
-
1 ,
-
i. 1
M 1
'i,i
^ II 'f
V'V1^' , -
76
iis
jjL
44:
H'h :
'
••i/ i
i ^
I;
J -
6 ;
v
iy ~
1 —
80
70
60
50
40
30
20
10
-O
50
40
30
20
10
-0 "
6O
50
40
30
20
10
-Q
ISO4O*
4 min
CO
u_
&e
PLITUDE
^
-<
Figure 4. Chromatograms of CLS-extracts of Water Factory 21
waters (samples taken 11/14/78) (1):
Ql, plant influent (secondary treated);
Q4, stripping-tower effluent;
Q9, chlorinated effluent. Quantified peaks are
indicated in Table 3. Attenuation: 16 (top);
8 (middle and bottom).
593
-------
3. Detection Limits and Standard Deviations
for Trace Organic Analysis, Groups A, B, C (12)
Compound
A: \A3A Conponents
Trihalome thanes
Chloroform
Bronodi ch lorore thane
Dibranochloratne thane
Bronoform
Other Volatile Organics
Carbon tetrachloride
1,1, 1-Trichloro-
ethane
Trichloroethylene
Tetrachlorcethylene
B: CLSA Conponents
Chlorinated Benzenes
Chlorobenzene
1 , 2-Dichlorobenzene
1, 3-Dichlorobenzene
1 , 4-Dichlorobenzene
1,2,4-Trichloro-
benzene
Aromatic Hydrocarbons
Etnylbenzene
m-Xylene
p-Xylene
Naphthalene
1-Methylnaphthalene
2-Methylnaphthalene
K
HEA Conponents
Dime thy Iphthala te
Diethylphthalate
Di-n-toutylphthalate
Di isobuty lph,thalate
Bis- [ 2-ethy Ihezyl ]
phthalate
ft>ly chlorinated bi-
phenyls (Aroclor
1242)
Linda ne
Detec-
tion
Limit
ug/1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.02
0.02
0.02
0.02
0.02
0.01
0.02
0.02
0.02
0.02
0.02
0.3
0.3
0.5
0.3
4:0
0.3
0.05
Standard
Deviation
ug/la
0.085C+0.25
0.048C+0.09
0.064C+0.06
0.18C +0.2
0.039C+0.05
0.083C+0.09
0.085C+0.09
0.12C -t-0.05
0.22C +0.002
0.15C t-0.029
0.44C t-0.002
0.19C +0.002
0.28C +0.014
0.94C +0.004
0.87C +0.004
0.87C +0.004
-
-
-
0.42C
1.1C
0.83C
0.66C
0.61C
0.14C
0.09C
Applicable
Concen-
tration
Range, \.g/l
0.1-10
0.1-8
0.1-5
0.1-19
0.1-1.5
0.1-20
0.1-13
0.1-8
0.02-3
0.02-14
0.02-5
0.02-14
0.02-3
0.01-0.1
0.02-0.15
0.02-0.05
-
-
-
0.3-6
0.3-3
0.5-5
0.3-8
4.0-17
0.3-0.6
0.05-0.15
Number of
Paired
Sanples
Analyzed
95
114
115
98
58
95
83
89
21
12
16
12
9
18
18
9
-
-
-
9
6
7
10
5
4
6
Peak
No.
Fig. 4
4
12
10
11
13
2
3
14
Figs.
6 & 7
1
4
14
10
24
6
8
Values of the constants at and b. for estimating the standard
deviation according to Equation T2); C is the concentration of
the compound in yg/1.
5HEA refers to Solvent Extraction Analysis.
594
-------
en
VO
80
70
60
CD
LU 50
Q
/^Xi
0 40
LU
^^N
!=J 30
H
^j 20
Q_
^ 10
0
-
-
-
-
-
_
=-- •
Z]
_j
i
I
o.
u
o
Q.
O
J
/•'
<
1
11
U AjlU
I
o^
-------
80
70
GO
50
4O
30
20
10
0-
70
6$ x
UJ «°
§ »
1:
0-
60
50
40
30
20
10
0-
c
•
IO
i
IS • 200 ng/|
3
15
' 5 I W
1 «
I
' (i
, 6 6 i 12 '4 : f 176 B 20 ; ;
— ^
_7 o a 02
IS>200ng/l
3
• 1] 1
1
t
II 20
5 9
!' 14 '
! c i l7 19 '
8 6 ''I 16 1 * . 24
1 1 J> !W~ " | .__...
—Li '
i
I 1 '
7 04
ISMSOng/l
'°l. * 'S
.
t •'
* ' 1
J . L
a ( ! o- .e,,85
| |( ?'^< ' ,| : M
.. ._ TEMPCC)
'SO 170' , RATE 3V (tin .180230*
15 35 40
TIME (mm)
Figure 6. BCD chroma tograms of hexane extracts, Water
Factory 21 waters (l) (samples taken 11/3/78).
Ql, AWT-plant influent; Q2, lime-treated effluent;
Q4, stripping -tower effluent. Quantified peaks
are listed in Table 3. Baseline setting: 350,
Attenuation: 2 .
596
-------
06
IS '100 ng/l
TIME(rnin)
35 40
Figure 7. BCD chromatograms of hexane extracts, Water
Factory 21 (samples taken 11/3/78). Q6, Q7-5,
carbon column influent and effluent; Q9, chlori-
nated effluent. Quantified peaks are listed in
Table 3. Baseline setting: 375, Attenuation:
597
-------
on removal of the compounds which are detected by this procedure:
(1) PCB removal in lime treatment and ammonia stripping, (2) the
effectiveness of activated carbon, and (3) formation of chlori-
nation products following water chionnatiorr as indicated by
numerous peaks appearing early in the chromatograms of Q6 and
Q9. More in-depth discussions on removal of these materials are
given elsewhere (1,12,13,14,15).
Precision
The procedures described in this paper have been used to
study the removal of trace organic material in an advanced waste
treatment plant (1,11). The most prominent compounds of groups
A, B, and C have been quantified; the remaining compounds were
only qualitatively identified by GC/MS. To determine the preci-
sion of the analytical procedures, duplicate samples were
analyzed for compounds of groups A, B, and C. Standard devia-
tions were then calculated as a function of the contaminant
concentration, the detection limits, and errors resulting from
sampling and analysis (Table 3). The relative standard devia-
tions (S ) appeared to increase as compound concentrations
approached their detection limits. The standard deviation could
be described by a linear function of the form:
standard deviation(i) = a-C. + b- (2)
where a. and b. are constants and C. is the concentration of
compound i. Values of a. and b^^ are given in Table 3 for sel-
ected compounds. The concentration range for which the relative
standard deviation has been determined is given in column 4 of
Table 3. Relative standard deviations may be lower at higher
concentrations if no saturation effects are encountered. The
upper limit of a concentration range was the highest value found
in Water Factory 21 water samples.
Good results were obtained with the VGA procedure; S was
less than 10 percent. Only the compounds eluting late in the
chromatogram (e.g., brontoform and tetrachloroethylene) were
measured with Sr above 10 percent.
For the compounds determined by the CLSA procedure, gener-
ally higher relative standard deviations were determined. The
chlorinated benzene isomers were among the major contaminants:
S ranged from 15 percent (1,2-dichlorobenzene) to 44 percent
(1,3-dichlorobenzene). The nonchlorinated aromatic hydrocarbons
were measured with much higher relative standard deviations.
This is mainly due to the fact that the concentrations were
closer to their detection limits and therefore quantification
was affected by the failure of the integrator to calculate the
correct area. For naphthalene and the methylnaphthalene isomers^
no standard deviation could be calculated since not enough
values were measured above the detection limit.
598
-------
For the compounds extracted with hexane (HEA) and detected
with the BCD, a simplified equation (b^ « 0) was found to
describe the standard deviation satisfactorily. Only two
chlorinated materials were quantified; Aroclor 1242 and lindane.
Aroclor 1242 was identified based on its typical pattern and
quantification was based on the more reliable peaks in this
pattern. Since Aroclor 1242 is quantified as the whole mixture^,
but detected by individual components in this mixture, the total
detection limit is approximately 20 times higher than the
detection limit of one isomer.
Phthalate esters were detectable by the ECD due to their
two ester functions. The response, however, was not very
strong; hence, high detection limits result for this class of
compounds. The relative standard deviations were large since
all the values measured are relatively close to the detection
limit.
Analysis by GC/MS
With the specificity of the mass spectrometer it is possi-
ble to overcome partially the limitations of gas chromatography
as a separation technique. Figures 8 and 9 show a multimass-
chromatogram (MMC) which is specific for certain groups of. com-
pounds. The chromatogram at the bottom represents the total ion
or reconstructed gas chromatogram (RGC) in which all the major
compounds are detected. The use of mass chromatograms enables
the analyst to determine quickly various classes of compounds
present in the sample. For example, in the MC 91, C -benzene
isomers (alkylated benzene isomers with n aliphatic carbons) are
detected along with the internal standards. MC 43 and 57
detects compounds of aliphatic nature and MC 35 is specific for
some compounds, containing chlorine atoms. To enhance the
sensitivity for chlorine-.containing compounds, m/e 35 was
integrated for 64 milliseconds. In Figure 9, MC 104, 138, 83,
and 127 have been selected to detect numerous compounds formed
in chlorination. The chlorinated styrene isomers and styrene
dichloride were formed upon chlorination of styrene, since these
products are not present prior to chlorination.
The GC/MS analysis of the methylated methylene-chloride
extracts has led to the detection of numerous substances. Among
the most prominent are the higher fatty acids and many as yet
unidentified compounds. Figure 10 shows the MMC of methylene
chloride extracj: pf activated sludge effluent (sample taken on
23 February 1979, at Palo Alto Water Pollution Control Plant).
The MMC depicts the mass chromatograms of characteristic ions of
long-chain, saturated, fatty-acid methyl esters (m/e 74, 83,
147) along with the total ion chromatogram (RGG). This analysis
has been used for qualitative determinations only ^nd the
response factors of the compounds to be quantified remain to be
established.
599
-------
Q| 2-27-79 CLS EXTRACT IS*500ng/l
o
o
INI.xiao
as
19
43
fiSS
57
345
91
ISO
FCC
Tit
*.*_ -u. ^.i dichtorobenzene isomers
tetrochtoroefhytene .
i^-frichloroethane1',3 1 |
O_/L . \ A. . \ .
A «-
V.
ethylbenzer
N-xytent
1?
A MJL
V
ce
i
ies
s
Cp — benzene
19 20 22
A .
""tSJTTs IS°merS pentachtoroanisole
Cl-
A ^
1
pentadecane
isomers
.
25 3
-cB a-
V ...
hexadecane
JttJtjt^
O 33 38 42 44
III! 1
^ _ — . — *V_-i_ -a* i_ _r n r *i -••• •-- _.._-- _ r.i* in
-c-
^. — ,„. ...... ^..t
SO ICC ISO 3TO »3 300 390 400 HSO SOO S50 6OO 6SO 70O 7^ 800 BSD 9OO 3SO 1QOO
SCRN »
Figure 8. Multi-Mass Chromatogram (MMC) with Reconstructed-Gas-Chromatogram
(RGC) (bottom) for detection of alkylated benzenes (m/e 91),
aliphatic compounds (m/e 57, 43)f and chlorinated compounds
(m/e 35). MS conditions: EE, 70 eV; Mass Range, 29-31, 35-35,
36-135, 136-280; Integration Time, 4, 64, 4, 10. Substances
identified are given in (1).
-------
Q9
CLS EXTRACT IS»2CX)ng/l
T./too
roc
A _
^styrene
t, t.1 1 •*
A A
chtorostyrene^ j-styrene didhtoride
A R. * V . . « . n . 1, « I „ , . ,. fl n. A » . n - - - - -
brornodicNoromethone
1,1,2,2-tetrochloroethone
1
T «|
VL
/dibromochkrornethone dibromokxiomefhane
bromochlorolodomethane |
Cl-C0 I"'
! BO | M4 «5
^— *^^- JV- _> f
CI-CB a-c,
124 ' 190 132 134 '
100 ISO
800
90O
SCflN *
Figure 9. Multi-Mass Chromatogram (MMC) with Reconstructed-Gas-Chromatogram
(RGC) (bottom) for detection of some chlorination products and
styrene. MS conditions: EE, 70 $V; Mass Range, 29-31, 35-35,
36-135, 136-280; Integration Time, 4, 64, 4, 10. Substances
identified are given in (1).
-------
O
10
Figure 10.
MMC for Specific Detection of Fatty Acids in MEA with RGC (Bottom)
of Secondary Effluent (Palo Alto Wastewater Treatment Plant taken
2/23/79). For GC conditions, se& text. MS conditions: EE 70 eV;
Mass Range 30-31, 35-35, 36-235, 236-535; Integration Time 4, 64,
4, 6. Fatty acids are labeled with F (x = chain length)
-------
SUMMARY AND CONCLUSIONS
For the characterization of a broad range of compounds in
water, various methods have been combined into a procedure. The
reliability and limitations of this approach have been discussed
in this paper. This combined procedure'represents only a pre-
liminary solution to the problem of organic materials in water.
Future improvements will be possible as new columns, instru-
ments, and software become available. Physico-chemical con-
stants of trace organic compounds have been used to predict and
interpret their extraction and pre-separation in an analytical
scheme where purge/trap and solvent extraction methods are
combined to measure a broad range of organic materials. This
approach is limited by the scarcity of thermodynamic data;
such data must be generated if predictive models of pollutant
behavior in analytical as well as environmental systems are
to be developed.
In summary, the basic requirements for analyzing trace
organic compounds in tertiary effluents and drinking waters
include:
o The analysis must be based on a small sample size (up to
1 liter).Samples of one gallon or more are costly to
ship and difficult to process and store. Samples ef one
liter or less can be shipped at moderate cost and stored
in medium-sized refrigerators.
o The detection limit for most trace contaminants must be
in the lower nanogram per liter range.At detection
limits of 1 yg/1 and above,the removal (or formation)
of many contaminants cannot be determined precisely.
o For conclusive results, variabilities of the procedure,
the compound, and concentration at a sampling point have
to be considered and investigated.Generally, approxi-
mately 30 samples are needed to establish a concentra-
tion distribution (1).
ACKNOWLEDGEMENTS
This research was supported by: the Orange County Water
District; the U.S. Environmental Protection Agency through
Research Grants EPA-S-83873-011 and EPA-R-804431-01-0; the OWRT,
U.S. Department of the Interior through Grant 14-34-0001-7003;
and the California Department of Water Resources through Grant
No. B52353.
603
-------
REFERENCES
1. McCarty, P.L., M. Reinhard, J. Graydon, J. Schreiner, K.
Sutherland, T. Everhart, and D.G. Argo. The Performance and
Reliability of Advanced Treatment for Wastewater Reclamation
at Water Factory 21, Technical Report No.236, Civil Engi-
neering Department, Stanford University, Stanford, Calif.,
in press, 1979.
2. Keith, L.H. , and W.A. Telliard. 1979. Environ. Sci. Tech.,
13_(4), 416.
3. Shackelford, W.M., and L.H. Keith. Frequency of Organic
Compounds Identified in Water, EPA-600/4-76-062, December
1976.
4. Mackay, D., and P.J. Leinonen. 1975. Environ. Sci. Tech.,
9(13), 1178. ~
5. Leo, A., C. Hansch, and D. Elkins. 1971. Chem. Reviews,
71(6), 525.
6. Bellar, T.A., and J.J. Lichtenberg. 1974. J. Amer. Water
Works Assoc., 66, 739.
7. Grob, K., K. Grob, Jr., and G. Grob. 1975. J. Chromatog. ,
106, 299. '
8. Grob, K., and F. Zurcher. 1976. J. Chromatog., 117, 285.
9. Grob, K., and K. Grob, Jr. 1978. J. High Resolution
Chromatography & Chromatography Communications, !_, 57.
10. Glaze, H.W., J.E. Henderson, and G. Smith. Chapter 16 in
Identification and Analysis of Organic Pollutants in Water,
L.H.Keith,ed., Ann Arbor Science, Ann Arbor, Michigan
(1976).
11. Law, M.L.R., and D.F. Goerlitz. 1970. J. Assoc. Official
Anal. Chemists, 53_(6), 1286.
12. McCarty, P.L., and M. Reinhard. Statistical Evaluation of
Trace Organics Removal by Advanced Waste Treatment, paper
presented at the Annual Conference of the Water Pollution
Control Federation, Anaheim, Calif., October 3, 1979.
604
-------
13. Reinhard, M., C.J. Dolce, P.L. McCarty, and D.G. Argo.
1979. J. Env. Eng. Div., ASCE, 05(EE4), 675.
14
• * McCarty, P.L., D.G. Argo, and M. Reinhard. Operational
Experiences with Activated Carbon Adsorbers at Water
Factory 21, pp. 456-477 in this Symposium.
15. McCarty, P.L., D.G. Argo, and M. Reinhard. Reliability
of Wastewater Treatment, paper presented at the Water Reuse
Symposium, Washington, D.C., March 28, 1979.
605
-------
DISCUSSION
Tuesday Evening Session
DR. KUHN: This evening's papers gave an overview about
analytical control methods. We heard about some very practical
parameters for routine monitoring. We heard about bioassay
tests as a characterization method. Finally, we heard about a
very complicated analysis scheme for specific compounds that is
necessary for research work, legal cases, and cases of spills.
I think the water chemists and water analysts use many instru-
ments to control not only the water, but also the efficiency of
activated carbon or other adsorption systems.
Q DR. COTRUVO: I have a couple of questions. When comparing
the different group parameters, e.g., UV versus TOG, for measur-
ing performance of the system or water quality, there is always a
problem in trying to relate the parameter to an effect. I wonder
if Dr. Bull sees a potential for using any of those techniques or
indicators to justify the ultimate level of water quality that we
are seeking, other than generally improving the quality of water.
A DR. BULL: I do not see any easy way of using group parame-
ters for that purpose, at present, except to develop correlations
on an empirical basis. Maybe by coupling samplings of water in
a bioassay sense you could do a correlation, such as has been
done with the relationship between the TOG and the UV, but I
don't believe it would be easy.
Q DR. COTRUVO: Dr. Bull, I have another question related to
the data that you showed on the slide that looked at different
disinfectants, the laboratory mouse, and the number of papillomas
produced. Let's assume that it's valid—I know that you haven't
had a chance to reproduce that experiment yet. As I recall, the
undisinfected water produced no tumors, and various disinfectants
did produce some tumors: ozone, chlorine, and chloramines—and
chlorine dioxide did not produce tumors. Do you think that
provides sufficient evidence to conclude that the disinfection
process is toxicologically more significant than raw water
quality in terms of cancer risk?
606
-------
A Dr. Bull: Well, within the realm of that one experiment the
answer I suspect is yes, but the problem is that it was one
instance on the Ohio River. I really don't think you can gener-
alize from the one experience. I think that would have to be
repeated a number of times before we reached the conclusion that
the disinfectant byproduct problem is the most important problem
in drinking water. I suggest that we ought to look at that
aspect more carefully. One thing I didn't make clear in my pre-
sentation was that by reverse-osmosis concentration, using the
aqueous concentrate, we get good recovery of the total organic
carbon. But, it specifically excludes reaction products like the
trihalomethanes. I want to emphasize that because low molecular
weight compounds would not be concentrated by reverse osmosis.
The oause is more complex than just trihalomethane formation,
that's the point.
607
-------
&EPA
NATO COM
• OTAN CCMS
NATO-CCMS
608
-------
ACTIVATED CARBON REACTIVATION AND COSTS
609
-------
xvEPA
NATO COCM
•OTAN CCMS
NATO-CCMS
610
-------
THEORY AND PRACTICE OF REGENERATION
Dr. JCirgen Klein
and
Dr. Harald Juntgen
INTRODUCTION
Liquid-phase and gaseous-phase adsorptive purification
processes have been known for years and may now be regarded as
classic. This technology is also used for dealing with the
increasing problems of drinking water preparation and wastewater
treatment. The practical use of adsorptive processes depends on
the availability of suitable adsorbents and regeneration proc-
esses. Since activated carbon is a suitable adsorbent for water
treatment, successful regeneration of loaded adsorbents is the
key problem which ultimately determines the feasibility and
cost-effectiveness of any adsorption process. Loaded activated
carbon may undergo regeneration by distillation, extraction, or
thermal treatment, depending on the material adsorbed (1-5). In
Germany, thermal treatment in particularly advanced, fluidized-
bed regeneration processes has become predominant.
DEFINITION
Regeneration may be defined as a process by which adsorbates
are removed from the loaded adsorbent so that the latter can be
reused (6).
In the case of physical adsorption the adsorbate is rever-
sibly bound to the adsorbent's surface. As shown schematically
in Figure 1, the adsorbate can be freed and removed completely
(and recovered if necessary) by application of a desorption
energy E_, which is normally equal to the heat of adsorption,
QA. 'For this process, called "desorptlon," not only thermo-
dynamics, but above all kinetics are the key features. Normally^
desorption is considerably slower than adsorption so that undet
industrial conditions complete removal of adsorbate is not
achieved and a residual load remains.
611
-------
REGENERATION
DESORPT10N REACTIVATION
ADSORPT1VE
ADSOR8AIE
ADSORBENT
Figure l> Terms of Regeneration
Residual load problems may be aggravated if a chemical
reaction within the adsorbate or between the adsorbate and
the solid adsorbent takes place. In such cases, the desired
reversibility of the adsorption/desorption processes is lost.
The desorption energy E_(see Figure 1) enables only partial
regeneration/ so that the residual load consisting of a chemi-
cally changed adsorbate can only be removed by application of
additional chemical energy Q_. This chemical conversion
process of irreversibly bouna residual load is called "re-
activation."
FUNDAMENTALS OF DESORPTION
In contrast to adsorption, desorption is a process which
always requires activation energy. Desorption is a first-
order reaction; its rate is proportional to the load q, at the
time t (7).
The rate is given by:
-k
(1)
612
-------
where k is the velocity constant, which is a function
of temperature according to the Arrhenius-equation:
k = kQ exp (-Ej/RT) (2)
Where k = frequency factor and £_ = activation
energy of desorption.
These processes may be investigated successfully by means
of non-isothermal measuring methods (8). Thus a single test
enables the calculation of the reaction-kinetic parameters E
and k (see Figure 2). By means of this method, desorption
was investigated for a number of systems (9). The measured
desorption curves differed from the theoretically expected
curves (Figure 2), as shown in Figure 3 for the desorption of
phenol from activated carbon. In contrast to simple theory, it
was found that desorption velocity is a function of the initial
load. With increasing load, the temperature at which maximum
desorption occurs is reduced to lower values. This permits the
conclusion that desorption comprises overlapping processes with
different activation energies. As adsorption heat depends on
load, the activation energy of desorption increases with decreas-
ing residual load. Calculation techniques were developed for the
non-isothermal method presented here. This enables preliminary
calculation of desorption curves for a number of systems, pro-
vided the adsorption isotherms are known.
For industrial practice it is important to know the influ-
ence of heating rate and adsorption temperature on desorption.
The influence of the heating rate is shown in Figure 4 for the
example of phenol-loaded activated carbon. According to the
theory of non-isothermal reaction kinetics, the curves are
shifted to higher temperature values with increasing heating
rates. As shown in Figure 5 for phenol load, the adsorption
temperature can be of utmost importance. The course of the
desorption curve changes considerably with the adsorption
temperature, due to the chemical conversion of the adsorbate
during adsorption, which also implies a partial increase of
desorption energy. Such conversion is in many cases the cause
for the existence of a non-desorbable residual load. The
presence of residual load means that, in subsequent adsorption
cycles, the full adsorption area on the internal surface of
the activated carbon is no longer available, which in turn means
reduced adsorption capacity. In this case thermal desorption
is no longer sufficient. As shown in Figure 6, activated carbons
of different degrees of activation had been loaded with phenol
and had subsequently undergone thermal desorption. After four
adsorption/desorption cycles the adsorption capacity of A-38
and A-50 carbon was reduced by 50 percent and, after six cycles
no further economical use of these carbons was possible. In
such cases the original adsorptive capacity of activated
613
-------
- 15 kcal/moi
grd"1
s J mmol/g
50
Figure 2.
200 r rc;
Desorption Curve Calculated
Under Simplified Conditions
dn
dT
x10"2
8,
fi-
o
L •
2,
Q.
Fmrnol]
[g.Co'J
*
I,
• *.
* •
* •
• •
• •
* *
/A
* J *
r// ,
*//
Vs = 0.63 &£•
m - o grd
m = f. -*—. —
mm
. ««^ m
•. — no
V: M0
\| nO
'^ n0
"N\ n0
^
\ HOI
'- — \
\
i -
-------
AC A-50
* 9.5°C/min
m2= 37 °C/min
°C/min
'0 100 200 300 400 500 600 700
T [°C]
Figure 4. Influence of Heating Rate [m]on
Desorption of Phenol
400 500 600 700T[?C]
Figure 5. Influence of Adsorption Temperature
on Desorption of Phenol
615
-------
carbon can only be regained by chemical treatment; i.e., reacti-
vation. Carbon type A-79 retained its original capacity over a
greater number of regenerations than did types A-38 and A-50.
T =800°C
t = 30min
8
10 12 14
N2 OF CYCLES
Figure 6. Change of Adsorption Capacity
after Desorption
REACTIVATION MECHANISM
Thermal reactivation occurs by chemical reaction of the
carbonaceous compounds constituting the residual load, by
means of an oxidation process; e.g., partial steam gasification,
If suitable conditions are achieved (i.e., temperature range of
700 - 900°C where only the chemical reactions control the
velocity of the total reaction rate), a partial gasification
reaction according to the overall equation can be achieved
intentionally.
C + H2O
•> CO +
+ 31.4 kcal/mol
(3)
616
-------
Detailed concepts of the mechanism of this chemical reac-
tion have been developed (Wicke and Rossberg ) (12). The basic
proposed mechanism is shown in Figure 7. Steam is first ad-
sorbed and then reacts with the carbon, resulting in disinte-
gration of the H2o molecules. The oxygen is bound to the
carbon and forms a surface oxide which can react with an addi-
tional H-O molecule, subsequently producing adsorbed H~
and C0_. The goal of regeneration is to gasify only tne
carbon contained in the residual load and not the carbonaceous
adsorbent. Thus the reactivity of the several forms of carbon
plays an important role.
c c c c c ""TTLTTT
Figure 7* Mechanism of Steam Gasification
Carbon conversion can be described by the equation
r » k * 9,, ^
(4)
where
- exp (-E/RT)
K • P
H20
l+K*Pt
617
-------
Our investigations (10,13) have shown that the dependence
on the steam-partial pressure can be approached sufficiently by
the following equation:
k1
(5)
where n ranges between 0 and 1. For partial pressures from 10
to 100 kPa (0.1 to 1 atm) the value of n is 0.5, so that under
these conditions equation (5) is reduced to
r = k'
-------
The conversion rates of virgin and phenol-loaded activated
carbon are shown in Figure 9. It is evident that the loaded
carbon has a higher reactivity than the virgin carbon. This
permits the conclusion that steam reacts more easily with the
compounds of the residual load. After a certain "period of time,
which depends on the reaction temperature (e.g., 20 minutes at
800°C), the conversion rate decreases to the corresponding value
for the virgin carbon, suggesting that the total residual load
is converted. This hypothesis is confirmed by investigation of
the change in adsorption capacity and weight when varying this
period. After 20 minutes at 800eC the change of weight is
nearly zero and the adsorption capacity of virgin carbon is
restored as shown in Figure 10. However, the results obtained
for phenol-loaded carbon do not necessarily apply in general.
The reaction conditions to achieve regeneration depend on the
properties of the adsorbate and the quality of the carbon used.
However, the above described technique enables investigation of
regeneration conditions for any case in industrial practice,
REGENERATION PRACTICE
For industrial use of the regeneration process two essen-
tial conditions must be met: the activated carbon's original
adsorption capacity must be restored; and the carbon losses
caused by chemical reactions and attrition during regeneration
should be minimized.
During liquid-phase adsorption the activated carbon takes
up organic matter as well as considerable quantities of water,
so that bulk density of the wet loaded carbon is double that of
virgin dry carbon. Heor.e, the regeneration process comprises
several steps: drying; desorption; decomposition; and steam-
carbon reactivation.
The water on the external and internal surfaces of the
carbon particles is evaporated at temperatures near 110°C.
Approximately 60 to 70 percent of the total regeneration heat
is spent for this step. After drying, temperature is increased
for desorption of reversibly adsorbed compounds, which subse-
quently can be either destroyed or recovered. At higher tem-
peratures, especially above 400°C, decomposition or cracking
reactions may occur, forming a residual load of nearly irre-
versibly adsorbed matter; in the final step (i.e., reactivation)
this residual load is removed by chemical reaction; e.g., with
steam.
In industrial practice the following equipment is predomi-
nantly used: rotary kilns; multiple hearth ovens; and more
recently, especially in Germany, fluidized bed systems (Figure 11.)
619
-------
0.6
'1*1
0.4
T*eso*c
« charged with ptonol
» uncharged
V
• T.MO'C
T.750»C
T.TOO'C
(U
\
tS 60 75.. ,90
tim* 1 mm I
r I
T.850«C ! i
« chargtd in ptenofc wo«t«
T. 800 »C
xamm^&g^iM.
*
SO
90 Hm |min)
Figure 9. Reactivities of Virgin and Loaded Carbons
Chang* in adsorption activity I'A]
cyelm of rvaetivation
-t.ZOmin
T . 800 «C
0 n U
cyclt* of rtactivatMn
Figure 10, Influence of Residence Time on Success of Regeneration
620
-------
FLUIOIZED BED
MULTIPLE HEARTH
(2) SPENT CARBON
(2> REGENERATED CARBCK
Q) PROCESS -0AS
SAS
Figure 11* Equipment Used for Regeneration
Compared to rotary or multiple-hearth systems, the
fluidized-bed system has the special advantages of improved
heat and mass transfer, which result in smaller reactor vol-
ume and shorter residence times. In addition the fluidized-
bed system results in less attrition and increased flexibility
with respect to control of throughput and of reaction condi-
t ions.
An inherent disadvantage of the fluidized-bed when used
in continuously operated plants is the wider range of resi-
dence time for the individual particles; however, this can
be sufficiently mitigated by process-engineering improvements
as discussed below.
621
-------
The fluid!zed-bed principle developed by Winkler in the
1920's has since been adapted to a variety of technically impor-
tant processes (14-17). Since 1955, Bergwerksverband has used a
15-stage cylindrical fluidized-bed system for production of
activated carbon. Due to the increased use of activated carbon
for drinking water preparation, Bergwerksverband began, in 1964
to regenerate spent carbon in a single - or two-stage fluidized-bed
system. This technique was also applied directly by waterworks
(18). The Du'sseldorf, Germany waterworks operate a regeneration
unit using this process (19), and our manufacturer has built
eight additional plants abroad (20).^ Other fluidized-bed systems
are used in the Wuppertal (21) and Zurich (22) waterworks.
Due to simplification and increased cost effectiveness, the
process is significantly improved. Stringent regulations for
treatment of industrial effluents gave impetus to further devel-
opment of fluidized-bed systems. The conditions prevailing in
this field, however are quite different from those in drinking
water preparation with regard to pollutant concentrations and
regeneration frequency. Pollutant concentrations in wastewater
are orders of magnitude higher and thus require other regenera-
tion conditions; e.g., higher temperatures, different hea.t-up
conditions » and larger residence times for the solid particles.
Moreover/ in comparison to the adsorption cycles of 4 to 6 months
that are common in waterworks/ in wastewater treatment the
carbon is spent after days or even hours. Thus, carbon losses
are a considerably more important economic factor.
For the above-mentioned development work, particular impor-
tance is attached to the residence time distribution for the
solid particles, which is to be kept as narrow as possible. For
heterogeneous gas/solid reactions such as carbon regeneration,
the solid particle's residence time in the reactor plays an
important role. Only in case of ideal plug flow is each particle
in contact with the reaction gas over the same period. In
fluidized-bed systems, however, stream profiles, bubble formation,
and back-mixing result in a distribution of residence times for
the individual particles. Figure 12 displays such residence time
distribution curves; n * 1 designates a one-stage fluidized-bed,
whereas n « 4 and n = 10 signify fluid-bed systems with 4 and 10
theoretical stages, respectively (9 =£• is defined as reduced
or dimensionless time).
622
-------
S(t)
Figure 12.
Residence Time Distribution Curve
Ideal Plug Flow (n=00) Compared to Fluid
Bed with n equivalent theoretical stages
(n-1,4,10)
0 * Relative Residence Time
S * Fraction of input concentration in
response to a step-change stimulus.
Laboratory-scale tests were conducted with a batchwise
operated fluidized-bed. In this mode, the contact time is ex-
actly defined for each solid particle, and the number of theo-
retical plates corresponds to n *». It was found that activated
carbon losses under the optimal process conditions investigated,
as defined by the temperature T and mean residence time T, are
negligibly low and depend only on the carbon reactivity. In the
upper diagram of Figure 13 the empirically determined correlation
between T and T for a loaded carbon used for purification of
phenolic wastewater is shown. Under conditions defined by that
characteristic curve, the carbon's original adsorption capacity
is restored and the carbon loss is nearly zero. (13)
In continuous operation, therefore, the carbon losses to
a large extent will be determined by the period of contact in
the reactor beyond the desired mean residence time T .
623
-------
33U
onn .
aUU
pnn .
701 .
IM
?nn •
1
^^^
0 2
*^
0 3
**N,
0 4
i
•»X
05
«^^
^•^
0 7
^,
5 1C
ite
10
RESIDENCE TIME [mm]
750°C
0 10 i 20 30 40 50 '60 70 80 90 100
tt t2 t3 RESIDENCE TIME [mini
Figure 13. GAG Loss as Function of
Residence Time and Temperature
As shown in the lower diagram of Figure 13, for a given
temperature, chemical losses prior to the optimal^, are due to
the reactivity of the remaining carbonaceousi adsorDate towards
the steam. After T the carbon has no residual adsorbate and
continued conversion affects only the carbon structure of the
adsorbent. We may therefore expect that in continuous systems
this carbon loss can be cut back by narrowing the residence time
distribution; i.e., by increasing n, the number of theoretical
plates.
The residence time distribution can be influenced signif-
icantly by design parameters (length/width ratio, multistage
design, special built-in equipment, etc.) and furthermore is a
function of fluidizing gas flow for a reactor of given design.
624
-------
These considerations raise the question of the reasonable
maximum investment in design features. Probably the cheapest
improvements of the classic circular shaped reactor were the
rectangular reactor (Figure 14) and the introduction of immersed
weirs. Experimental runs of fluidized-bed models have shown
that the rectangular design when compared to the circular type
results in an improvement of residence time distribution which
can be represented as n = 2 for the rectangular design versus
n = 1 in a circular type. To optimize cost effectiveness, the
length/width ratio (L/W) should be in the range :
-------
Hence, in view of the minimum carbon loss, the costs for the
design features involved need not be extremely high. The prob-
able carbon loss can be determined as a function of the number
of theoretical plates n, on the basis of laboratory and pilot
plant scale regeneration tests as well as from investigations on
models. Figure 15 shows the total carbon loss as a function of
n. If we suppose, for example, that under ideal process condi-
tions a carbon loss of approximately 1 percent results when
n « 4, it would not be necessary to increase n above a value of
6 considering the economics of such a regeneration process.
We found by model tests that an optimum value exists for
the fluid gas velocity relative to the particle's minimum
fluidization velocity. Given this value, the number of theo-
retical stages, dependent on the reactor design, can be reached.
In case of lower or higher gas velocities the residence time
distribution curve broadens again, implying lower values for n.
A rectangular reactor for carbon regeneration based on the
design principles described above is shown schematically in
Figure 16. The system may be adapted by means of suitable dis-
tribution plate design, to achieve the required temperature
profile as well as optimal fluidization despite the changing
physical states of the activated carbon during thermal treatment.
TECHNICAL APPLICATION
Rectangular reactors for GAC-regeneration developed as
described above have been used in industrial applications. The
first commercial plant was built for regeneration of about 104
kg GAC per day used in adsorptive treatment^f plant effluents
from the Friedrich Heinrich coke oven plant~at KampLintfbrt (see
Figure 17). Operation results confirmed that the above stated
conditions for regeneration were met: i.e., restoration of the
activated carbon's original adsorption capacity, and a minimi-
zation of the carbon losses during regeneration. The carbon
losses could be reduced to less than 2 percent per cycle (18,23).
This regeneration process has also been used at plants in Italy
and Japan. The information given by the operators of these
plants confirms the success of this regeneration process scheme.
The following is a brief discussion of the economics of
fluidized-bed processes (24). Table 1 shows comparative
figures for capital and operation costs of multiple-hearth ovens
and rectangular type fluid-bed reactors for different carbon
throughputs. The figures for multiple-hearth ovens refer to an
American plant (25), those for the rectangular type are taken
from plants using the Bergbau-Forschung (BF) process. According
to these estimates the rectangular type appears more economic,
especially for larger plants. It should be stated, in addition,
that the lower carbon loss in the rectangular type (3 percent
626
-------
CO
CO
CO
DETERMINED BY REGENERATION
OF GAC FROM COKERY WASTE
WATER TREATMENT
T*800°C
1= 30min
0 1 2 3' 4 5 6 7 8 9 10 11 12
n
Figure 15. GAC-Loss as Function of Number of
Theoretical Plates (n)
AFTER BURNER
AIR
FUEL
AIR
FUEL
COMBUSTION
STEAM/WATER CHAMBER
COOLING
T
WATER
Figure 16. BF-Process for GAC-Regeneration
by 2-Stage Flu idbed Reactor
627
-------
compared to 7 percent for the multiple-hearth furnace) accounts
for the greatest cost advantage.
Figure 17. Photograph of Kamp-Lintfort Plant for
Treatment of Coke Oven Effluents
THROUGHPUT [kg/h]
CAPITAL COSTS (RELATlV) [DM]
OPERATION COSTS I % ]
GAS ']) (0,11 OM/rr?)
ELECTRICITY (0,06 OM/kWh)
.STEAM (22.16 DM/t)
GAC LOSS21 (2.54DM/kg)
LAB /STAFF31
CAPITAL SERVICE (15,5% /a)
1) Hu=4100 kcal/frt3
MULTIPLE HEARTH
FURNACE"
a
190
100
15.60
3.47
2.60
25.14
19.07
34,12
100,00
b
570
211
20.02
3.78
3.40
32.87
8.31
31,62
100,00
b:a
15,31
2.89
2.60
25.14
6,25
24,18
76,47
FLUIOIZEO BEOS'
c
234
55
21.58
2.99
2.38
13,98
3S29
19.78
100,00
C:Q
16.47
2.2S
1.82
10.67
30.00
15,10
76.36
d
400
60
19,43
3.49
0.81
2799
23.03
25,25
100.00
da
7,41
1.33
0,31
10.67
8.78
9.63
38,14
a and b = 7 VGl%
c and d = 3 VGL'.'a
ab andd =1 MAN/YEAR
C =2 MAN (YEAR
I) CHEMrENG SEPT 12 1977 S 95
II) c: DETERMINED AT BF-PRQTQTYP&flANT
KAMP UNTFCRT
d CALCULATED FOR A PLANT UNDER CONSTRUCTION
Table 1. GAC Regeneration Costs Multiple
Hearth/Fluidized-Bed-Oven
628
-------
REFERENCES
1. Himmelstein, K.J., R.D. Fox and T.H. Winter. Chem. Eng.
Progr. 69, (11)65 (1973).
2. Maier, D.. Vom Wasser 38, 255 (1971).
3. Mann, T. Chem. Ing. Techn. 46(8), 345 (1974).
4. Cohen, J.M. and J.N. English. AIChE Symp. Ser. 1.44,
Vol. 70, 1974.
5. Juhola, A.J. In Proceedings of the Water Research
Assoc. Conf. on Activated Carbon in Water Treatment,
Univ. of Reading, England, 3-5 April 1973.
6. Klein, J. Staub - Reinheit - Luft 36, (7), 292 (1976).
7. Hayward, D.O., B.M. Trapnell, Chemisorption, Butter-
worths, London 1964.
8. Juntgen, H., K.-H. van Heek. Fortschr. Chem. Forsch.
_13, 60 (1970).
9. Seewald, H., H. Juntgen. Ber. Bunsenges. phys. Chem. 81_
(7), 638 (1977).
10. Juntgen, H., J. Reichenberger. Chem. Ing.-Techn. 49, 159
(1977).
11. Juntgen, H. Carbon 15, (5), 273 (1977).
12. Wicke, E. and M. Rossberg, Zeitschr. Elektrochm. 57,
641 (1955) p. 641/645.
§3. Juntgen, H., J. Klein ancj j. Reichenberger. Paper at
GVC/AIChE Joint Meeting, Munchen 17. - 20.9.1974.
14. Schytil, F. Wirbelschichttechnik, in Verfahrenstechnik
in Einzeldarstellung 9, Springer Verlag Berlin 1961.
15. Beranek, J., K. Rose ..and G. Winterstein. Grundlagen
der Wirbelschichttechnik, Krausskopf-Verlag, Mainz 1975.
16. Reh, L. Chem. Ing.-Techn. 46, (5), 180 (1974)
629
-------
17. Werther, J. 1977. Chero. Ing.-Techn. 49(3), 193.
18. Klein, J. and H. Jiintgen Uwmelt Mr. 6. 1977. p. 465/470,
19. Poggenburg, W. Haus der Technik - Vortragsveroffent-
lichungen 404 (1977) p. 50/59.
20. Unpublished Information from Lurgi Technik.
21. Strack, B. Translation of Reports on Special Problems
of Water Technology Volume 9 - Adsorption, p. 284, EPA,
Cincinnati, Ohio 45268, EPA-600/9-76-030 (1975).
22. Schalekamp, M. Translation of Reports on Special Pro-
blems of Water Technology Volume 9 - Adsorption, p. 128
EPA, Cincinnati, Ohio 45268, EPA-600/9-76-030 (1975).
23. Klein, J. Haus der Technik - Vortragsveroffentlichuna
404 (1977), p. 59/68.
24. Hutchins, R.A. 1975. Chem. Eng. Progr. 71(5), 80.
25. Remirez, R. Chem. Eng., 15, Sept. 12 (1977).
630
-------
OPERATIONAL EXPERIENCES WITH THE
CARBON REGENERATION FURNACE AT
CHURCH WILNE
D.J. Osborne and C.A. Kennett
INTRODUCTION
The decision to build a regeneration furnace at Church Wilne
was made on economic grounds since it represented the least cost
alternative among throw-away use of powdered or granular carbon/
regeneration at the manufacturer's works, and on-site regenera-
tion. The bid specification did not state that regeneration of
the carbon had to be on site; it could be transported to the
manufacturer's works if this proved economic.
COST COMPARISON
The 1973 costs* for a 91,000 m3/day flow and 20 mg/1 equiv-
alent dose rate were as follows:
Operating:
Off site regeneration (including
250 miles transportation)
On-site regeneration
Capital:
Extra capital cost of the
regeneration system
63.28/1000 m fc!64/tonne
fcO. 73/1000 m S>36.5/tonne
£60,000
A discounted cash flow calculation using a write-off period
of 15 years and an interest rate of 8 percent, with a water flow
increasing from 45,500 m /day to 91,000 m /day over the period,
gave the following results:
Present (1973) investment to
buy the regeneration plant and
operate for 15 years
£211,000
*Costs are given in British pounds (fc), 1 i = $US 2.40.
631
-------
Present (1973) investment to
provide funds for offsite
regeneration for 15 years
£724,000
These costs have increased considerably, but a similar dif-
ferential for on-site over off-site regeneration exists today
(1979) where an expected regeneration rate of between 0.5 and
0.75 tons/day is exceeded.
REGENERATION SYSTEM
On becoming spent, one full charge of carbon is transported
hydraulically by eductor from the adsorber to the spent or "foul1
carbon hopper (See Figure 1). The carbon in both clean and
exhausted (foul) carbon hoppers is kept permanently under water.
At the base of the foul hopper, a timer-operated ball valve
controls an intermittent carbon flow to the suction side of an
eductor. The carbon slurry is lifted to an Archimedean screw-
type dewatering device at the top of the furnace. The dewatered
carbon falls vertically from the screw onto hearth 1 of the
furnace.
Figure 1»
Carbon Adsorption -and Regeneration System
CLARIFIED SANO
FILTERED
WATER
SUPPLY
BACKWASH
REJECT
SOFTENER
TOWNS WATER
FUEL
FINAL
CHLORINATION
CHLORINE
MAICEUP CARBON
FROM BAGS
OR TOTE BINS
COMBUSTION f~T
AIR »
M.W.— MOTIVE WATER
The 1.37 m (54") diameter furnace is a seven-hearth,
Herreschoff type with rotating rabble arms. It is sized for
632
-------
a maximum throughput of 3 tons per day dry carbon, but normally
operates at 2 tons per day. Carbon is rabbled from the outside
of hearth 1 to a circular annular hole at the center and then
falls to hearth 2. Here it is rabbled to the outside and falls
through holes to hearth 3. This system is continued until the
carbon emerges at the outside of hearth 6 and falls down a chute
into the quench water. There are two burners on hearths 2 and
4 and two on the top hearth (hearth zero). The zero hearth
contains no carbon, but is a strongly oxidizing afterburner stage
for the flue gases. Carbon from the quench tank is educted to
the clean carbon hopper.
On hearths 1 and 2, the wet carbon is dried by the rising
hot gases; on hearth 3, regeneration begins at approximately
600°C. Here easily volatile compounds are driven off. On
hearths 4, 5, and 6, the temperature is gradually increased and
steam is injected. The excess oxygen content at the oil-fired
burners is kept to controller maximum, around 2 percent 02. Extra
air can be injected on the hearths without burners 3 and D, as
necessary. By hearth 6, all adsorbed organics have been pyro-
lised to carbon within the granule and with a controlled oxygen
input this outer layer of carbon can be preferentially burnt
before the carbon of the granule structure. The temperature of
the final hearth is 950 to 975°C.
The maximum fuel input to each burner is 6.2 kg per hour.
In operation, it was found that hearth 4 burners require about
4.3 kg per hour each,'that only one hearth 6 burner is necessary
at 4.1 kg per hour and that the zero hearth burners are both
operated at maximum rate (6.2 kg per hour).
The furnace consumes about 230 kg per hour of steam at
310 kPa (45 psia) from an oil-fired boiler installation. Steam
is used for direct injection, center shaft cooling, and non-
operating burner cooling duties.
The boiler is provided with feedwater treatment including
a base exchange softener and tannin and polyelectrolyte dosing.
All excess water in the furnace area from the eductor oper-
ations is collected in a furnace area sump and then pumped into
the main dirty backwash water sump. This water is pumped back
to the raw water reservoir, thereby utilizing any carbon fines
as a rough pretreatment.
Commissioning
The first run of the furnace in January 1976, was plagued
with minor mechanical and constructional problems. However,
these were overcome as they arose. The decreasing carbon losses
on each subsequent regeneration show the improvements; 25 percent
633
-------
on the first run, 6 percent on the second, and the current
routine losses of 4-1/2 to 5 percent.
The main problem, initially, was defective alignment of the
rabble system center shaft. This led to two problems; an air in-
leakage at the bottom seal (subsequently corrected by realignment
and a different packing material), and a cooling air inleakage
through a flange half way up the hollow center shaft. This sec-
ond problem was difficult to solve, but was eventually alleviated
by using steam for center shaft cooling. There were also many
instances of high torque cut-out caused by the rabble tines
touching the hearth bricks as these expanded upwards; many tines
were cut and ground to give sufficient clearance.
One persistent problem was encountered on the burners.
Those on hearths 4 and 6 are at 950°C in an often slightly
reducing atmosphere. They have removable caps and jets that
require fairly frequent cleaning. The cap threads are about
50 mm from the burner tips and were often found to be completely
seized. This problem was not overcome until the whole burner was
redesigned to put the cap threads well down the body and thus
removed from the hottest zone. At normal throughputs, it was
found necessary to close down one hearth 6 burner. A system was
eventually designed to allow steam to be passed into the atomiz-
ing air line to any burner to keep it cool while not in use
without using the atomizing air, which would burn carbon unneces-
sarily. Any burner left in the furnace with no cooling flow was
rapidly and permanently made inoperable by carbon, from pyrolysis
of the residual*fuel, causing changes in the high temperature
steel structure.
These experiences have convinced the design engineers (H&G)
that, where possible, future installations should be fired by
pipeline natural gas or site-stored LPG. The difficulties in
using fuel oil in this small but very hot burner have been over-
come, but the 2:1 mechanical ratio turndown system on the oil
burners causes more operating difficulties than a fully modulat-
ing 10:1 turndown system which is available for natural gas
burners.
On a new installation, it should also be possible to use
a water seal at the bottom of the center shaft and so eliminate
the frequent need to repack the bot%om gland to prevent air
inleakage.
Carbon Quality
At the time of writing, some 550 tons of carbon have been
regenerated with no deterioration in quality.
The regenerated carbon quality is controlled by three tests;
the apparent density, the methylene blue number, and the iodine
634
-------
number. The apparent density of fresh F 200 carbon is about
570 g/1; the density at regeneration averages 640 g/1 and after
regeneration 560 g/1. The pore size distribution is an important
feature in the adsorption capacity of the carbon. On repeated
regenerations, the small-sized pores tend to enlarge. The iodine
number gives a measure of the smaller pores and the methylene
blue number the somewhat larger ones. The iodine number of fresh
F200 carbon is about 950. This slowly drops on each regenera-
tion; after three or four regenerations it is around 800 to 850.
The methylene blue number of fresh F200 is approximately 220 and
the regenerated carbon has always been between 195 to 230, so it
seems that the subsequent adsorption of taste and odor compounds
can be related to the methylene blue number. The routine control
of regeneration is by apparent density and iodine number. A sud-
den increase of iodine number during a regeneration run with no
change in apparent density has been correlated with increased
oxygen in the furnace atmosphere. A recent addition has been the
inclusion of zirconia-type, continuous recording oxygen probes in
several hearths of the furnace. These are very useful when
setting up the furnace at the start of a regeneration campaign.
Many manual Orsat analyses were previously required to ensure
that the furnace box was airtight and the burners correctly
adjusted to give a 1-1/2 to 2 percent oxygen atmosphere before
introduction of the carbon feed. When operating, no readings of
oxygen are give/n? any inleak is indicated by abnormally high
temperature on tl>e leaking hearth.
Some quality parameters for regenerated carbon are:
F2QQ Apparent density Iodine No. Methylene Blue No.
g/ml
Fresh 0.556 950 220
Fouled 0.610 666 165
After 3 regenerations 0.554 827 230
Fresh 0.520 1,000 270
Fouled 0.584 821 182
After 1 regeneration 0.525 1,020 260
The 15 tons of F400 was mixed with the main plant F200
carbon after this one regeneration.
Control System and Alarms
The furnace alarm and trip system developed for Church Wilne
allows the furnace to be left unattended. An alarm in the main
plant control room calls an operator to deal with any emergency.
635
-------
A trip system shuts down the furnace for:
flame failure at any burner;
fuel oil low pressure;
high hearth temperature (any hearth above 1000°C) ;
high torque on the furnace rabble arms;
or build-up of carbon in the quench tank.
The control feature which has given the regenerator the most
problems has been the interrupted flow on/off ball valve at the
foul carbon hopper eductor. This valve has Teflon seats which
require changing every few months owing to abrasion. The com-
plete stainless steel valve requires changing every year. One
unnecessary feature was an alarm on the valve actuator to indi-
cate seizure; the downstream eductor suffers more blockage
problems and the consequent furnace carbon feed failure is seen
some few minutes later as a rapid rise in the normally very
stable hearth 3 temperature of 675°C.
The one measurement problem has been the setting of the
furnace carbon feed rate. This is done manually using a clear
plastic tube to measure the carbon volume in a 2 minute catch
sample of the intermittent slurry feed. An instrumental method
would be preferable. If a rubber pinch suction valve were used
to control a constant flow of slurry, a density meter (ultra-
sonic or gamma ray) could be used to measure the feed rate. A
normal low flow could be interspersed with a full open period to
clear bridging or blockages as required. The fouled Church
Wilne carbon does not have the same sticky nature and tendency
to form clumps as fouled sugar industry carbon.
Carbon Exhaustion Rate, and Furnace Utilization
The carbon exhaustion rate predicted for the design 20 mg/1
equivalent dose rate at 91,000 m /day was 1.82 tons/day. The
furnace is capable of 3 tons per day throughput, but was set up
at 1.92 tons/day and has always been operated at this rate.
Initially, seven or nine adsorbers were on line at flow
rates ranging from 35,000 to 45,000 m /day giving 6 to 8 rain.
contact time. (This number has to be on line to cope with the
flow surges from the gravity filters.) The equivalent dose rate
for taste and odor (T and 0) in practice was in the range 7.6 to
9.7 mg/1 giving an exhaustion rate of only about 0.32 tons/day
and furnace operation was required only some 60 days/year.
Since 1977, all 12 adsorbers have been operated to deal with
the trace organics problem. The equivalent dose rate has risen
to 15 mg/1 and the exhaustion rate for these trace organics,
rather than taste and odor control, has risen to 0.68 tons/day,
requiring 130 days per year for furnace operation. The furnace
636
-------
is normally operated in two regeneration campaigns of 130 tons
each at 6-monthly intervals.
Costs
Base costs for a plant similar to Church Wilne are given in
Table 1 using as many real data from the plant as could be
abstracted from the records. The following aspects deserve
special mention:
Labor is estimated as that necessary in excess of require-
ments to run the main works. When the furnace is in
operation, shifts are extended from 8 to 12 hours to leave
one man free to attend the furnace at any time if required.
This is not strictly the time spent at the plant, which is
nearer 6 man-hours per day.
The fuel cost is high because of the use of steam from the
boiler used to cool the center shaft. If the air inleakage
problem had not arisen this steam demand would be 150 kg/hr
lower.
The wastewater includes a generous 10 min. backwash of all
adsorbers every da^f although this is only necessary in
summer. In winter,, wastewater usage on the adsorber section
is approximately 1.1 percent while in summer, it is up to 2 2
percent. "
The carbon is, in cneory, as good at year 15 as at year 1
and so perhaps should not be 'written off1 in this way.
The maintenance materials cost is high because the cost
period taken included purchase of many spares~which have
not yet been used.
Wastewater is assigned a cost of fcO.045 (4.5p) per m .
It is in fact recycled to the raw water reservoir and is
not lost, but passes again through the main plant.
With benefit of hindsight, the regeneration cost on the
1977 basis could probably be reduced to around £150 per
ton.
COST SUMMARY
It can be seen that the GAC plant adds about 20 percent to
the cost of the water leaving the treatment plant but adds only
some 10 percent to the cost of water to metered consumers.
Costs to metered consumers in 1977 were 10 to 12p per m
comprising of 4 to 5p for water treatment and 5 to 7p for distri-
bution and administration.
637
-------
Table 1
Throughput per day
Electric Ibwer Adsorbed
Cost Experience
Adsorber Section
~ I Vday
Regeneration
fc/day
45,500 m 2 tons
3.5
2. Oh/day
3.0
12.0
0.5
1.25
18.75
131.3
66.0
1000 3m/day 45.0 770n3/day 34.7
To overcome head loss 19 KW
Furnace drives/fans & dewatering screw - 6M?
Carbon fluid transport system 3IW
Total at 1.6p per KWi 19KW 7.3
Labor
Analytical services 0.5h/day
Supervisor 2.0
Shift Operator 3.0
Instrument fitter 0.25
Mechanical fitter -
Total at average fc7 per manhour 5.75 40.25
Maintenance Materials
Waste Water
Fuel Oil for Burners,
Afterburner & Boiler at 8.84p per Kg 941Rg/day 83.0
Make-Up Carbon 6% Losses 120Kg/day 96.0
Boiler Chemicals
Total Daily Cost &92.S5
Cost per ton regenerated
Capital Cost
The capital cost on a 1977 basis for a plant similar to Church Wilne
would be approx. 6730,000 including the regeneration facility and
initial carbon change but excluding Civil Costs.
Operation Costs
Using an interest rate of 10%, a write-off period of 15 years, a flow
of 45,50dm/day at 9.7 min. contact time and 250 tons/year regeneration
rate gives:
0.33 Vday 0.7
fe415.2
Total Costs (pa = per annum)
Debt Charges (Factor 0.1315pa)
Adsorption plant
Regeneration plant
95,995pa
33,781
52,000
fc!81,776pa
fc 0.01094 per m3, or approximately $0.083 per U.S. gal.
638
-------
SUMMARY AND CONCLUSION
From the commissioning and operating experience at Church
Wilne we are confident that carbon regeneration at a waterworks
site is a completely feasible proposition for the normal person-
nel, who were very quick to assimilate techniques new to the
industry.
The costs of the use of GAC for taste and odor removal, or
for removal of specific organic compounds where contact times of
up to 10 minutes are necessary, are not as high as is generally
supposed.
If total organics removal to a 1 mg/1 TOC standard were
required, the GAC treatment costs would perhaps be multiplied
by a factor of 2 to 3. The use of such a broad parameter is no
substitute for knowledge of the specific nature of organic
micropollutants of industrial origin and their biodegradability,
toxicity, and adsorption properties.
If industrial discharges are causing a micropollutant
problem at a drinking water works, it is probably best from an
environmental as well as a cost point of view to use GAC first
at the industrial discharge, and then as a final clean-up or
emergency safeguard as the last treatment stage before supply
to consumers.
Consider a hypothetical example (Figure 2) of a complex
organic industrial effluent, or 1000 m /day, containing 10 mg/1
of a non-biodegradable micropollutant, being discharged to a
sewage works of 100,QOO-m /day throughput, which then discharges
to a river of 1 x 10b m /day total flow, of which 100,000 m /day
is withdrawn downstream for drinking water.
For total equivalent costs, alternative plants could prob-
ably be built and operated as follows:
1) At the industrial site, 40 m carbon with a contact time of
1 hr needing a 5 ton/day regeneration rate to give removal
from 10 mg/1 to 0.5 mg/1.
2) At the waterworks, 700 m of carbon with a contact time of
10 min. needing a 1 ton/day regeneration rate to give
removal from 10 *ig/l to 1 >ig/l.
The industrial plant alternative removes 9.5 kg/day and a
level of 0.5>ig/l is found in the drinking water. The water
works plant removes only 900 g/day, 9.0 kg/day passing on down the
river and 1 >ig/hr is present in the drinking water.
In both of these examples, the compounds of interest would
be in a small minority compared with the total of adsorbable
639
-------
1000 m Id
Industrial
Wastewater
Toxic Pollutant
10 kg/d
Concentration 10 ma/I
40 m Granular
Activated Carbon
1 Hour
Contact Time
9.5 kg7d Removed
0.5 mg/l
Municipal
Sewage Works
Sewage.
100,000 m Id
Municipal
Sewage Works
River —»-900,000 m Id i»50 old — >• River — »• 900,000 m /d 9 kg/d
0.5 ug/l
" 900 g/d Removed
t
Water Works
No Carbon
10 yg/1
700 m Granular
Activated Carbon
10 Minute
Contact Time
100,000 m id
O.Syg/l
i
/d
50 g/d
100 g/d
Figure 2. Basis for Hypothetical Total Cost Example
640
-------
species and complex, competing adsorption would occur. Full
pilot scale investigations arft always necessary before any
design work can be undertaken for Such systems, but the examples
are of the right order of magnitude for many real situations.
It is probably a good insurance policy to construct a GAG
plant at the waterworks, with a 5 to 10 minute contact time to act
as a safeguard against unknown taste and odor problems and pollu-
tants or spillage, but removal of non-biodegradable organics from
known source industrial discharges should be done at the highest
concentration possible.
ACKNOWLEDGEMENTS
This paper is given by kind permission of W.H. Richardson,
C. Eng, MICE, Mi. Mun. E, MRTPI, Dip. TP, Divisional Manager
Lower Trent Division of the Severn-Trent Water Authority and of
Humphreys and Glasgow Limited. The authors would like to thank
the operations and scientific staff of Severn-Trent and commis-
sioning staff of H&G, on whom the successful running and careful
monitoring of the plant depended and whose work is reported in
this paper. The comments made and conclusions drawn are those of
the authors alone and not those of Severn-Trent W.A. or H&G.
641
-------
EXPERIENCE ON THE REACTIVATION OP ACTIVATED CARBON WTTH A
DOUBLE-STAGE FLUIDIZED-BED FURNACE AT THE HOLTHAUSEN
TREATMENT PLANT, D'tfsSELDORF, FRG
Wilhelm Poggenburg
INTRODUCTION
The Stadtwerke Du'ss'eldorf AG are treating Rhine River bank
filtrates with ozone arnd activated carbon in the waterworks of
Staad, Flehe, and Holthausen (Figures 1 and 2). The daily
capacity is about 400,000 m . Granular activated carbon is
capable of eliminating from the water many organic compounds as
well as substances causing taste and odor.
In recent years, the number, concentration, and danger of
the pollutants and contaminants have increased. Some of the
resulting effects have been:
- Today activated carbons are exhausted faster than
20 years ago
- Activated carbons of much higher quality are needed in
order to achieve a drinking water of constant quality
- As a rule, the contact time has to be extended and
hence, the carbon quantity increased.
The resulting demand for larger quantities of higher quality
and therefore more expensive activated carbons does not permit
only one-time use. The activated carbons must be reactivated and
then reused. Initially, this was done in the manufacturer's
works; however, there are now several waterworks which operate
their own reactivation plants. The studies revealed that, in
addition to economic considerations, the following factors are of
major importance:
- The waterworks must be independent of the available
reactivation capacities of the manufacturing companies.
- Carbon quality can be adapted to the requirements of
each individual type of drinking water treatment.
Reactivated carbons usually differ in quality, and
642
-------
U)
horizontal
well
max. water level
mean
water level
protecting clay layer
drawdown curve
water-bearing sands and gravels
impervious base
J
pressure
pipeline
-submerged pum
—_-^-_^rrr__-_•-—: teHtjgry sands ——— ------z
Figure 1. Water Collection in a Horizontal Well Near River Rhine
-------
ozonized air
from ozonizer
I
ejector
pressure
water
adsorber
raw water influent
intermediate storage tank
^- upper layer
(carbon)
— - lower layer
(activated
carbon)
drinking water
effluent
Figure 2. Flow Diagram of Water Treatment Plant
-------
often the quality is unsatisfactory when regeneration
has been performed at the carbon manufacturing plant.
REACTIVATION OF ACTIVATED CARBON FOR WATER TREATMENT
Saturated activated carbons can be regenerated by chemical
or thermal means. Thermal reactivation is the only process used,
to a large degree, in the waterworks, "This involves the follow-
ing process steps: drying, desorption, and pyrolysis.
Drying is achieved by evaporation" of absorbed and pore water
from the carbon particles at temperatures up to 100*C. Subse-
quently, highly volatile compounds are desorbed at temperatures
up to about 300*C. Finally, irreversibly adsorbed organic com-
pounds are pyrolyzed, while the cracking products escape as
gases. This results in pyrolyzed coke as residue on the carbon
which can be eliminated only by oxidation in the presence of
H2o, CO,, or O~ at temperatures starting from about 700°C.
The following furnace types are available for the thermal
reactivation:
Shaft furnace
Rotary kiln
Multiple hearth furnace
Single-step fluid-bed reactor
Two-step fluid-bed reactor
*Fluidizing launder
THE REACTIVATION PLANT AT DUSSELDORF
A total of approximately 900 "t (t » metric ton * 2200 Ibs)
of activated carbon, which must be reactivated up to three times
per year, are used in 42 adsorbers.
In the adsorbers, granular carbons of varying quality, with
grain size of 0.5 to 2.5 mm and with apparent densities of be-
tween 360 and 480 kg/m , are predominant. Lesser quantities
of shaped carbons with a diameter of about 0.9 mm and about a
2 mm length are also used.
A two-stage, fluid-bed reactor was erected directly adjacent
to the largest of the three treatment plants in Holthausen.
More than half of the total 2,100 m of activated carbon is
used here.
*Single-stage fluid-bed reactor where the carbon is flowing
in a channel (Wirbelrinne).
645
-------
The reactor (Figure 3) is the main part of the reactivation
plant. Its daily capacity is rated at approximately 6 t of re-
activated product. The reactor has an outside diameter of 1.6 m
and is approximately 4 m high. With an inside diameter of 0.9 m,
the grate surface area is 0.636 m .
A pulse-controlled screw feeder, in an inclined position
for preliminary dewatering, feeds the wet saturated carbon with
a water content of 45 to 55 percent from the inlet tank to the
upper part of the reactor. Here the carbon is dried at tempera-
tures between 200 and 300°C and passes through a rotary valve
into the reaction chamber where it is reactivated at temperatures
between 700 and 800'C. The red-hot, reactivated carbon flows
through a discharge pipe into a water-filled quench tank. The
average residence time of the carbon in the reactor is about
20 minutes.
The process gas is generated in a combustion chamber under-
neath the reaction chamber. Natural gas serves as fuel. The
desired process gas temperature of 900°C is controlled by water
addition. There must always be sufficient steam for gasifica-
tion. A substoichiometric combustion prevents oxygen from
getting into the reaction chamber, thus guarding against combus-
tion of carbon.
The process gas flows from the combustion chamber through
the perforated bottom plate of special steel into the reaction
chamber. With sufficient flow velocity, the activated carbon
bed is raised and fluidized intensively. Although the tempera-
ture in the fluid-bed is generally maintained at about 780°C, it
always can be adjusted to the specific properties of carbon and
adsorbate.
The temperature of the reactivated carbon coming from the
reactor and its residence time can be varied for a given process
gas temperature by changing the carbon feed rate. Relatively
uniform residence time of the carbon particles is achieved by
weirs on the lower perforated bottom plate in the reaction
chamber which avoids intermixing.
The process gas passes through the upper perforated plate
with a temperature of about 700*C, enters into the drying chamber
and fluidizes the wet carbon. During this drying phase, the gas
temperature drops to the range of 200 to 300°C. The process gas
is then cleaned in a special unit.
The material flows in the plant are shown in Figure 4.
Spent activated carbon from the Holthausen treatment plant is
flushed into the storage tank through a system of pipes. The
storage tank of concrete with plastic lining has a capacity of
45 m . From this tank, the carbon to be reactivated comes
646
-------
carbon inlet •*•
-•• waste gas outlet
carbon outlet
process gas
generator
carbon
overflow
inspection glass
Figure 3. Two-Stage Reactor
647
-------
CO
spent carbon
silos
from the adsorption
plant
water
reactivated
carbon silo
to the adsorption
plant "
dewotering screw
feeder
stack
natural gas
Figure 4. Flow Diagram of Diisseldorf Reactivation Plant
-------
through the PE-pipes into a 0.85 m stainless steel holding
tank, from which the screw feeder charges the reactor. The
reactivated carbon coming out of the reactor is transported
the quenching tank through the pipe system into another 45 m
storage tank.
Carbons from the treatment plants Flehe and Staad are moved
in bulk by trucks.
For the hydraulic transportation of the carbon, by ejector
pumps, the ratio of water to carbon is adjusted to about 3:1;
the flow velocity of this water-carbon mixture is about 1.3 m/s.
The mechanical controls in the pipe system consist of stainless
steel ball valves.
The process gas is emitted from the reactor at a tempera-
ture of 200 to 300*C; it contains carbon dust and the materials
discharged during reactivation. The dust is separated in a
multicyclone and drawn off with the aid of a water-jet pump.
The odorous materials partly contained in the process gas are
burned in a post-combustion chamber at about 900*C. The hot gas
leaving the chamber passes through two heat exchangers. The
first serves to preheat the process gas from the reactor, the
second, to preheat the combustion air for the post-combustion
chamber. The cleaned waste gas, having a temperature of approxi-
mately 640°C, is released into the atmosphere through a 16.5 m
high'jpteel stack. This relatively expensive waste gas treatment
guarantees that the standards of environmental protection, as
laid down by legislation and regulatory agencies, will be met at
all times.
The operation of a continuous process without constant
attendance requires the complete automation of the reactivation
plant. A part of this automation is the automatic "shut-down"
of the plant to a safe condition in case of failure. Facilities
for manual operation are included for the purposes of mainten-
ance, tests, and troubleshooting.
In fully automatic operation, a process computer monitors
and controls the startup and shutdown, the operation of the
reactor, and the handling of the carbon in the reactivation
plant.
A mosaic panel in the control room, manned only part-time,
shows all important parts of the plant in a flow diagram.
Optical signals of the plant condition and faults, switches and
buttons for possible manual operation, and the most important
readings are integrated in this control panel, which indicates
at all times the actual state of the process.
649
-------
The following conditions were specified for the planning
of the reactivation plant:
- In several fluid-bed reactors, which are independent*
from each other, a total of up to 500 kg of reactivated
product must be obtained per hour.
- The plant is to be operated continuously and is to be
staffed by personnel only during working days.
- The storage and handling system must permit charging
the reactors from all three treatment plants.
- The waste gas cleaning must meet the anti-pollution
air requirements of the authorities concerned.
Since no experience on the operation of a two-stage fluid-
bed reactor with fully automatic process control was available
to the waterworks, the expansion was arranged in two phases.
The possibility of erecting the equipment without housing
was rejected for reasons of environmental protection and in the
interests of humane working conditions. All equipment is housed
in a building of steel frame construction finished with a light
sheeting of shaped metal.
The construction of the plant started in Spring 1975; the
first activated carbon was reactivated in August 1976. Since
then, 2,000 t, or about 5,000 m , of carbon have been reactivated
in one two-stage, fluid-bed reactor. Table 1 shows some tech-
nical data for the reactivation plant.
Table 1
Technical Data of the Reactivation Plant
Capacity
First stage of completion
Final completion
250 kg/h
500 kg/h
Reactor (first stage of completion)
Natural gas consumption ,
(fuel value: 9.7 kWh/Std. nr)
Power requirement
Water requirement
Approx. 60 m /h
490 kW
Approx. 50 1/h
Afterburner (first stage of completion)
Natural gas consumption
(fuel value; 9.7 kWh/Std. m3)
Approx. 21 m /h
Temperatures
Process gas
Reaction zone
Drying zone
Afterburner
Stack
850 - 950*C
700 - 800°C
200 - 300eC
Approx. 900*C
Approx. 640"C
650
-------
For plant safety, the monitoring, measuring and control
equipment, and the "EMERGENCY-OFF" controls all work independ-
ently of the computer.
FIELD EXPERIENCE
It was not expected that such a plant would go into opera-
tion without any difficulties. Considerable efforts and constant
readiness of the staff of contractors and plant personnel have
led to the result that, after about 2 years, the reactivation
plant is giving satisfactory service in automatic operation and
is supplying a reactivated product of quality that consistently
meets specifications.
On the basis of this past experience, only one additional
fluid-bed reactor is planned for future installation to reacti-
vate all spent activated carbon arising from the three water-
works. Furthermore, fresh activated carbon of low quality can
be up-graded by the fluid-bed reactivation, although only with
higher losses.
The performance .test in the second half of 1976 showed an
extraordinarily high failure rate although only proven standard
material had been used. Because of frequent plant failures it
was still impossible, one year later, to sustain sufficient
continuous operation to allow the determination of optimum reac-
tivation conditions. However, substantial knowledge and experi-
ence on function and service life of all equipment was gained
during this time, and large quantities of spent activated carbon
were reactivated. Surprisingly, the failures occurred most
frequently in those parts of the plant which had proved to be
particularly successful in other applications under similar
conditions. Considerable changes were also necessary in the
reactor itself and in the control program.
IMPROVEMENTS IN THE REACTOR
With the original internal components in the bottom of the
reaction chamber, it was impossible to obtain a uniform retention
time for the carbon. Widely varying and often unsatisfactory
qualities of the reactivated product were the result. The high
initial adsorption capability of the carbons declined very
rapidly in operation, probably because the micro-pore system
was insufficiently reactivated. Various reactor internal config-
urations had to be tested until a complete recovery of quality
was obtained in combination with only mild attacks on the acti-
vated carbon structure. With the current solution, there is only
a small deviation in the retention times of the various carbon
particles. The presently established thorough exchange of mass
and heat enables a precise temperature control of the process.
651
-------
No significant effect upon the hardness of the carbon can
be detected, although the screen distribution is gradually
shifting toward smaller particle sizes.
The originally installed ceramic nozzle bottoms have not
met expectations. After a short time of service, the slots in
the nozzle heads were clogged by sediment from the water, by
material from the brick lining, and by foreign matter in the
carbon resulting in an unacceptable pressure increase. Perfor-
ated bottom plates of heat-resistant, alloy steel therefore have
been installed in the drying and reactivation chambers. They
can be run for months without cleaning. Their service life
until now has been satisfactory.
An early test of water injection for control of the process
gas temperature in the burner space revealed considerable oper-
ating problems. The water with a carbonate hardness of 100 to
110 mg/1 as CaCO., and with a maximum chloride content of 250 mg/1
is now softened and introduced by way of a cascade. Current
experience indicates no need for further treatment.
At first, minor quantities of gravel inadvertently charged
into the reactor together with the carbon were found to interfere
with the reactivation process. Sintering on the trays resulted
very quickly in constriction of the slots or perforations. The
result was that it was no longer possible to achieve the gas flow
rate needed for fluidization and reactivation after a short time
of operation. Bar grates in the feed system to the reactor are
now retaining the gravel. Danger to the feeding screw is thus
eliminated and the sintering on the trays is satisfactorily
minimized.
EXPERIENCE WITH CARBON TRANSPORT
No problems are encountered in the pipe system if the
hydraulic transportation of the carbon is carried out correctly.
An insufficient water-to-carbon ratio or an excessively high
flow velocity will lead to clogging or increased wear. Pre- and
reflushing with water are important prerequisites for trouble-
free carbon handling.
Stainless steel, plastic, and shaped steel pieces with
inside enamel coating or cement mortar lining were used for the
transport lines on a test basis. All these materials gave satis-
factory service when properly made and assembled.
Originally installed, cast iron ball valves were removed
because of severe erosion and replaced by ball valves of stain-
less steel. With tanks and pipelines it was found that, in case
of improper execution or insufficient quality, stainless steel
in this unusually corrosive environment will fail within a very
short time.
652
-------
A particular problem is the reliable determination of the
filling level of activated carbon under water, especially under
corrosive conditions or high temperatures. Improved methods of
measuring and indicating filling level are presently under
trial.
THE WASTE GAS SYSTEM
The ceramic nozzle bottoms installed originally in the
reactor cause higher pressure losses than the current perforated
bottom plates. In order to avoid excessive stress of the reactor
shell by over-pressure and to prevent the leakage of gas, a waste
gas fan was installed which produced a slight suction of about
500 Pa (0.07 psi) at the reactor top. This waste gas fan was
extremely susceptible to failure and caused numerous total break-
downs .
After the perforated bottoms had been installed, the reactor
was run successfully at a slight over-pressure without utilizing
the waste gas fan. The fan was therefore removed and an over-
pressure of about 6 kPa (0.9 psi) now exists in the burner
chamber in normal operation.
During commissioning there were occurrences of undesirable
dust deposits on the heat exchanger pipes aft**1 extended opera-
tion. A cleaning device consisting of chains, which are operated
intermittently by an electric motor, was later installed and is
now removing these deposits.
VERIFICATION OF THE GUARANTEES
The manufacturers' guarantees were verified by tests in June
1978.
Guarantees;
1. Reactivation of 250 kg/h granular activated carbon with
a bulk density of virgin carbon (unsaturated) of about
430 kg/m and a grain size range of 0.5 to 2.5 mm,
or shaped carbon with about 0.9 mm diameter and about
2 mm length. This guarantee applies at a discharge
temperature of 720*C.
2. The quality of the reactivated product must at least
meet the standard of service reactivation presently
offered by carbon manufacturers.
3. The loss in the reactor must not amount to more than
5 percent by volume for the activated carbon grades
F300, LSS, and ROW 0.8. The undersize fraction (<0.5
mm) fed into the reactor is disregarded in calculating
the loss. •,
653
-------
4. Energy and process water requirement:
Natural gas consumption.
(Hu * 7,600 kcal/std. m3) 130 m3/h
Electric energy consumption 50 kW
Process water consumption 10 m3/h
Results;
1. The carbons defined here are all reactivatible. In the
case of granular carbons, the guaranteed throughput of
250 kg/h was achieved and often exceeded. Further tests
are necessary to determine whether the throughput can
also be achieved in the reactivation of shaped carbons.
2. The reactivation quality is uniform and satisfactory.
It matches or exceeds the qualities obtained in con-
tracted reactivations. The quality of the reactivated
products are, in some cases, even better than those of
virgin carbon. Further tests are necessary to determine
the optimum operating conditions of the reactor.
3. The losses in the reactor are not easily determined.
As yet, this problem has not been solved satisfactorily.
The contractor's reactivations show widely fluctuating
losses; the total losses (operating, handling, and reac-
tivation losses) are between \2 and 25 percent. In the
instance of on-site reactivation, losses are currently
11 to 12 percent and relatively constant. Between the
storage silo holding the material to be reactivated and the
silo holding the reactivated carbon, the losses amount to
about 9.5 percent; approximately 2 percent of the carbon is
lost during transport and backwashing in the filter plant.
A first test, with a loss balance in the various parts of the
reactivation plant, leads to the conclusion that the losses
in the reactor are below 5 percent by volume. A significant
change of the mean particle diameter was not detected here.
Common knowledge indicates that it should be possible
to achieve further reduction of the reactivation losses
and especially of the handling losses. Tests of these
possibilities are planned. This program will also try
to monitor the loss rate in the various parts of plant
with more accuracy than before.
654
-------
4. The consumption of gas and electric energy is much
lower than the guaranteed values. Hence, the operation
of the proprietary reactivation plant is much more
economical in terms of energy costs than calculated
originally (Figure 5). The process water consumption
is higher than expected and therefore must be reduced,
although it does not significantly influence the
economy. An essential portion of the process water is
that used for transport.
REACTIVATION COSTS
The reactivation costs consist of the carbon losses, the
energy consumption, costs for personnel, and the depreciation as
listed in Table 2.
Table 2
Reactivation Costs at Holthausen
Cost
Cost Basis DM per 103kg AC
1. Natural gas 75 m3/h 132
2. Electrical power 35 kW 14
3. Water for transport, etc. 14 m /h 17
4. Manpower and maintenance 67
5. Capital interests, etc. 264
Reactivation costs 494
6. Carbon losses
a) Reactivation furnace <5%
b) Transport in the
reactivation plant,
multiclone
discharge Approx. 4.5%
c) Transport and
backwashing in the
filter plant Approx. 2%
Total losses Approx. 11.5% 345
Total costs including carbon replacement 839
The reactivation costs may be reduced still further after the
second reactor is installed.
655
-------
heat output
heat exchanger I
heat exchanger I
afterburner outlet
combustion
preheated air from
heat exchanger H
preheated air from
heat exchanger I
water evaporation
carbon overflow
secondary combustion
carbon outlet
gasification
carbon inlet
water evaporation
cooling of combustion
chamber
heat balance in kW
input output
384
46
139
255
46
139
129
161
23
55
23
66
13
87
35
heat input
436
Figure 5. Heat Diagram
656
-------
SUMMARY
The thermal reactivation of activated carbon in the water-
works has been successful in satisfying widely varying require-
ments. Most importantly, reactivation plants can today be
reliably operated fully automatically. The on-site reactivation
plant has come to be a valuable element in guaranteeing the
supply of safe, palatable drinking water at all times, even
under adverse conditions.
657
-------
EXPERIENCE WITH A FLUIDIZED BED FURNACE AT
BENRATH TREATMENT PLANT
Dr. B. Strack and H. Martin
DESCRIPTION OF THE PLANT
Benrath waterworks have been delivering drinking water from
the river banks of the Rhine for the town of Wuppertal for
exactly 100 years. Raw water from the sand banks of the Rhine
River requires careful treatment, which has been done with ozone
and granular activated carbon filtration since 1967. The carbon
must be replaced or reactivated between two or three times per
year, depending on output and water quality. The sixteen filters
yield about 1,000 metric tons yearly of spent carbon to be
reactivated.
The activated carbon was formerly reactivated by the carbon
manufacturer; since 1977, our own activated carbon reactivation
plant with a capacity of 5 metric tons per day has been in opera-
tion. Such a plant in the waterworks"area offers the advantages
that the filter carbon can be exchanged quickly, there are short
lines of transfer, and the manager of the waterworks is indepen-
dent from forwarder and carbon manufacturer. The fluidized bed
furnace employed for reactivation allows fast starting as well
as changes of reactivation conditions during its operation.
Furthermore, it is possible to achieve the carbon qualities
required for adsorption of the impurities present in water.
A flow diagram of the Benrath reactivation plant is shown
in Figure 1. Spent carbon from the filters is hydraulically
transported into a rectangular storage tank made of concrete.
Underneath the tank is a separating vessel in which heavy
impurities such as gravel, which could cause troubles in the
furnace, are separated.
The carbon is transported via an intermediate tank to a
conveyor screw, which feeds the furnace continuously.
The superficially dewatered but still wet carbon is fed into
the fluidized bed (Figure 2) and is reactivated for about 45
minutes at a temperature of 600°C. The furnace is fired with
natural gas. The temperature is controlled by injection of
658
-------
storage tank
spent carbon
fluidized bed furnace
afterburner storage tank
reactivated carbon
Figure 1. Flow Diagram of the Benrath Reactivation Plant
carbon inlet waste gas outlet
hot gas
inlet
Figure 2. Cross-Section of the Fluidized Bed Furnace
659
-------
water. Afterwards the carbon is quenched and washed with water,
then stored in a tank, and transported back to the filters as
needed.
The exhaust gas from the furnace passes a cyclone which
removes dust and is purified in a combustor to.meet air pollution
control requirements. Thus, less than 20 mg/m of dust and 3 ppm
of carbon monoxide escape as pollutants.
During the time this plant has been operating, extensive
experience has been gained in estimating the quality of the
reactivated carbon and in identifying and solving operational
difficulties. This is discussed in more detail below.
QUALITY OF THE REACTIVATED CARBON
The activated carbon has several simultaneous tasks in the
process used at Benrath. The water treated with ozone and fil-
tered over gravel is passed through the carbon filter primarily
for removal of dissolved organic matter by adsorption. In addi-
tion, microorganisms in the carbon bed remove ammonium and some
of the organic substances. Furthermore, residual ozone is decom-
posed in the top-layer of the carbon filter and a small amount of
manganese, not completely removed in the preceding sand bed, is
retained.
The experience of many years shows that the granular acti-
vated carbon filters perform these simultaneous tasks if a good
quality of carbon is used. Hence, quality control at the reac-
tivation step is important. During recent years, detailed
investigations have been carried out with the support of the
German Federal Minister for Research and Technology.
The total concentration of the dissolved organic substances
is often determined as DOC. Measurement of the UV-adsorbance on
the other hand is faster, simpler, and gives similar information.
Table 1 illustrates mean values of DOC and UV absorbance for
Rhine River water, river bank filtrate, and drinking water.
Table 1. Mean Values of Dissolved Organics
Rhine River bank Drinking
River filtrate water
UV (nT1) 10.1 3.5 1.0
DOC (g/m3) 3.9 1.6 1.0
660
-------
Because of its advantages the UV method is used for continuous
surveillance of the filters at Benrath waterworks. In this
manner the efficiency of the filters can be monitored and deci-
sions made as to the necessity of carbon replacement. Both the
efficiency and frequency of replacement depend on the quality of
the carbon and the composition of the raw water^ The filters
should be operated so that a DOC value of 1 g/m in the drinking
water reservoir is not exceeded; this will ensure that all unde-
sirable substances are removed.
The efficiency of a filter is a practical criterion for
evaluating performance at the waterworks. It is defined by the
ratio of the amount of matter removed by the filter to the amount
being initially present. It is determined by UV measurements at
the inlet and the outlet of a filter. The efficiency decreases
continuously as more water is treated, eventually reaching a
specified lower limit at which the carbon in the filter is
exchanged with reactivated or fresh carbon.
The speed of the deterioration depends on the quality of
the carbon, since a carbon with a high adsorption capacity will
treat more water before the efficiency limit is reached than one
with low capacity. Of practical interest is how much water can
be treated with a given type of carbon in the present filters.
This measure of quality must also be valid for the reactivated
carbon.
As an example, Figure 3 shows the efficiency of different
filters recorded over the throughput. It is evident that effi-
ciency can be described as a function of throughput or time of
operation. The data, though somewhat scattered, seem to be
log-linearly related. One of the tests with our own reactivated
carbon yields a somewhat better efficiency at low throughput,
suggesting the possibility of a more suitable quality. However,
as throughput continues our two reactivated carbons behave essen-
tially the same as the virgin carbon and the carbon reactivated
by the producer.
At the waterworks it takes a relatively long period for the
adsorption behavior of the carbon to be recognized. Therefore a
rapid test of the quality of the reactivated product is needed
to allow prediction of the behavior of the filters in practice.
Conventional tests such as Iodine Number, Methylene Blue Number,
and benzene loading were assessed. These methods can supply some
indirect information about the porous structure, but no regular
relation to the behavior of the filters in the waterworks could
be found. Only the Iodine Number gave, in some cases, useful
hints for predicting the performance of the carbon.
661
-------
O virgin carbon
A react, by producer
D from own
O reactivation plant
1.5
—•—
throughput in
million m3
Figure 3. Breakthrough of UV-Absorbing Substances
in Benrath AC Filters
After these negative results, a practical test proposed by
the study group of Prof. Sontheimer was further developed and
applied in our research program. In this method the adsorption
equilibrium between organic matter in the water to be treated
and a carbon sample is investigated using a UV-measurement. One
obtains "isotherms as shown in Figure 4a. To calculate the ad-
sorption capacity in practice, or the amount of water which can
be treated before equilibrium is reached, it is necessary to
assume an acceptable residual concentration or a ratio of organic
compounds to be removed.
As a second part of the method, paranitrophenol (PNP) is
added to the raw water, since PNP is thought to adsorb in a
manner similar to some chlorinated organic compounds. The iso-
therms of raw water and PNP (Figures 4a and 4b) show that carbon
sample 1, which behaves least favorably in raw water/ is the best
with PNP. Thus selection of a carbon and operating conditions
for removal of chlorinated organics cannot be based solely on
isotherms describing overall organic compound removal from raw
water. Instead/ it is necessary to consider the adsorption
behavior of both raw water and of PNP and then select operating
conditions which yield satisfactory overall results.
662
-------
200
0.100
"70
O sample 1
V sample 2
A sample 3
a sample 4
O sample 5
residual concentration
ing/m3
Figure 4a. Isotherms of Raw Water with
Several Carbon Samples
note: The residual concentration is based on UV-extinction in
m"1 which, in this case, corresponds nearly exactly _
with the total concentration of organic matter in g/m .
The latter concentration units are used to be consistent
with the units of g/kg used for loading.
100
.£40
01
I
§20
© sample 1
V sample 2
A sample 3
O sample i
O sample 5
residual concentration
0.03
0.1
0.3
in g/m3
Figure 4b. isotherms of PNP with Several Carbon Samples
663
-------
A new method has been evaluated for the analysis of the raw
water isotherms which has proven very useful. In this method the
activated carbon filter is assumed to be a completely mixed re-
actor in which equilibrium is always attained during the passage
of the water. In this manner a theoretical prediction based on
the isotherms can be made for the filter efficiency as a function
of the treated quantity of water. Furthermore, a comparison of
the theoretical prediction with the standard sample can be made,
the behavior of which is known in the filter. In Figure 5, the
calculated and observed efficiency are plotted versus specific
throughput in m water per kg carbon. The efficiency values
shown are for total organic substance in the raw water, inferred
from UV-absorbance measurements. It is evident that the actual
data agree well with the prediction resulting from this rather
simple approximate method.
100-
90-
80-
TO-
GO-
50-
Ol
1
"oi
30'
0
a Filter 11
V Filter U
rZZ calculated
V
20
60
80 specific throughput in m3/kg
Figure 5. Measured and Calculated Breakthrough
of Filters with Reactivated Carbons
Hence the adsorption test using the actual raw water is a
fast and practical method, at least for the conditions in Benrath
waterworks. With this method it is also possible to calculate
the amount of carbon that can be saved if a better quality is
achieved through reactivation.
Further investigations are planned to determine how the
described method must be changed or extended to ensure useful
results under other presuppositions and boundary conditions.
For example/ the effect of biological processes occurring on the
carbon will also be taken into consideration.
Another important aspect of the quality of the reactivated
carbon is its mechanical strength. The carbon grain is stressed
during backwash, hydraulic conveyance, and reactivation.
664
-------
Consequently every grain becomes smaller by attrition. Figure 6
shows, for example, the screen analysis for virgin, spent, and
reactivated carbon. The fine grain fraction inrr«»A«!«>s with *.ime,
which results in carbon loss when the very fine particles are
rinsed out of the filter. Tfferefore care has to be taken to
ensure that this decrease of the grain size is minimized. Past
experience indicates the carbon loss depends mainly on hardness.
Thus, the virgin carbon must be as hard as possible. Presently
the overall loss, including conveyance to and from the filters,
separation of heavy particles, thermal reactivation, and washing,
totals about 15 percent. Efforts are still being made to reduce
this carbon loss.
wei
100 -I
80 -
60 -
40 -
20 -
particle
diameter
mm
2.0
16
1.25
1.0
0.8
0,68
virgin spent
reactivated
carbon
Figure 6. Screen Analysis of Activated Carbon
665
-------
PRACTICAL EXPERIENCE WITH PARTICULAR PARTS OF THE PLANT
The reactivation plant has been operating since 1977. During
that time, we modified the plant to deal with some of the problems
that were encountered. They are summarized in Table 2.
Table 2. Summary of the Improvements at Benrath
I~IProblems resulting from water hardness
a. Desalination of injected water
b. Quick exchange of quenching water
2. Process gas composition control
3. Distribution of wet carbon
4. Use of special injectors
5. Use of special steel quality
PRETREATMENT OF PROCESS WATER
One problem concerned the quality of the water used for
the reactivation process and for conveyance. The process gas
contains a considerable portion of steam originating from liquid
water sprayed into the burner chamber. Originally, softened
water was used. However, the remaining salts, mainly sodium
carbonate, reacted with the fine sand taken in with the carbon
and formed a slag which gradually obstructed the nozzles. Now
the water is completely demineralized by a reverse osmosis plant
and the described troubles no longer occur. Another problem
arose in quenching the carbon after the reactivation to fill the
pores with water. The hot reactivated carbon causes both an
increase of the pH of the water iato the alkaline range and a
rise in temperature. As a result, calcium carbonate precipitates,
depositing itself as a dense layer on the wall and the nozzles or
the rinsing vessel. Remedial measures included changing the
water frequently and inserting less easily fouled nozzles.
GAS PRESSURE
According to past experience, the composition of the process
gas must be held constant to allow optimum conditions for the
reactivation process. Variation of supply pressure can result in
different quantities of fuel gas and air being delivered, which
is undesirable since excess oxygen burns the carbon and causes
unnecessary losses. Therefore a control vent for the exhaust gas
was installed to hold the pressure constant and a supplementary
gas analyzer was included to allow simple control of the gas
composition.
666
-------
CONTROL OF RESIDENCE TIME
The wet heavy carbon coming from the conveyor screw can
fall through the fluidized bed and stay on the bottom. There-
fore the wet carbon has to be spread by a simple device so that
it is at once thoroughly mixed. By this means the distribution
of the temperature in the bed and the residence time of the
carbon is more steady and therefore the reactivation is better.
EJECTOR DESIGN
For hydraulical conveyance, jet ejectors are installed at
several places. The carbon flows from above into the horizontal
driving water jet. This results in attrition at the water
nozzle. However, there are better ejectors designed especially
for hard solids in which the nozzle is retracted out of the
carbon stream and the diffuser consists of an exchangeable shell
made of a hard alloy.
CONSTRUCTION MATERIALS AND FABRICATION
Considerable corrosion has occurred in the plant where
stainless steel 1.4571 (equivalent to AISI 316 Ti) was not used
or where the welding was poor. The presence of spent carbon and
water containing chloride resulted in pitting corrosion at the
welding seams and crevice corrosion at the flanges. Conse-
quently, some of the welding seams had to be rewelded as well
as etched and passivated. The welding operator must be very
familiar with handling stainless steel. The flanges should be
absolutely plane and the seal must stav elastic and tight so that
no hollow space is formed. In summary the use of high quality
materials, and very careful design ai^T/nanufacturing, are neces-
sary to prevent problems of corrosion under the conditions
experienced at Benrath.
SUMMARY
The encountered difficulties were overcome during the first
2 years of operation. Our present results show that the compact
and not very complicated reactivation plant, based on a single
stage fluidized bed type furnace fed with wet carbon, works
successfully. The reliability of the waterworks performance has
improved, thereby enabling optimization of plant operation so
that a good water quality can be achieved in an economic manner.
667
-------
EXPERIENCE WITH AN INFRARED FURNACE FOR
REACTIVATING GRANULAR ACTIVATED CARBON
O. Thomas Love, Jr.
and
Wendell R. Inhoffer
Granular activated carbon is a broad-spectrum adsorbent
that must be kept in an "active" state to be effective. One
method of achieving this is with thermal reactivation. The
status of an EPA sponsored" research project to evaluate an
infra-red tunned -furnace ror reactivating granular carbon used
in drinking water is the topic of this report.
WHY AN ELECTRIC FURNACE?
Two reasons why an electric furnace might be an attractive
option for reactivation are:
1. Some water utilities in the United States have limited
fossil fuels available, yet adequate electric power.
Some of this power may be hydroelectrically generated
at the wate-rworks.
2. An electric furnace may be started or shut down
without long warm-up or cool-down periods.
Further, a tunnel furnace conveys the adsorbent on a
traveling belt and this gentle handling should reduce mechanical
attrition losses in reactivation.
LOCATION AND DESCRIPTION OF STUDY SITE
The utility selected for this study is the Little Falls
Filtration Plant in Totawa, New Jersey. This facility is owned
and operated by the Passaic Valley Water Commission; the source
of water is the Passaic River.
This location has a history of trace organic contamination
(1) (the Passaic River has hundreds of industrial discharges
upstream of the Little Falls intake). Furthermore, in recent
668
-------
years several experiments have been conducted using granular
activated carbon adsorption (2). Thus, detailed characteriza-
tion of both the source and the treatment plant existed.
The Little Falls Filtration plant employs prechlorination,
alum coagulation, settling, filtration (anthracite coal over
sand), and_at times dechlorination. The average flow is 55 mgd
(208,0002m /day). A side stream of treated water [2.2 mgd
(8,300 m /day)] was diverted to three pressure vessels, each
containing a different type granular activated carbon, with an
empty bed contact time (EBCT) of approximately 8 minutes. The
granular activated carbon adsorbers filled with virgin adsor-
bents were operated continuously without backwash from March to
May 1978, and then reactivated.
FURNACE
The furnace selected is by Shirco, Inc.* of Dallas, Texas.
Design characteristics are given in Table 1 and the unit is
shown in Figure 1.
Table 1. Electric Furnace, Little Falls
Filtration Plant
Dimensions 4 feet (w) x 20 feet (1) (1.2m x 6m)
Activated Carbon Residence 20 to 25 minutes
Time
Temperature
drying zone 1,150°F (620°C)
activation zone 1,650°F (900°C)
Power required 100 kW
Warm-up time 30 minutes
Weight 17,000 Ibs (7727 kg)
Reactivation capacity 100 Ib/hr (45 kg/hr)
*Mention of commercial products does not constitute approval or
endorsement by the U.S. EPA.
669
-------
REMOTE CONTROL PANEL
GAC IN
DISCHARGE
"MODULE"
FEED _J DRYING, PRYROLYSIS
MODULE AND ACTIVATION MODI
TO EXHAUST SYSTEM
REACTIVATED GRANULAR
ACTIVATED CARBON
TO QUENCH TANK
Figure 1. Cross Section of Infra-Red Tunnel Furnace
670
-------
The adsorbents were then sequentially educted into a
storage bin, fed into the furnace, and returned to the adsorbers
for quenching. As of May 1979, the granular carbon has been
reactivated twice; plans call for at least two more exposure/
reactivation cycles before the project is completed.
Adsorption and reactivation performance is being judged
on the parameters shown in Table 2. Additionally, data are col-
lected on power consumption, labor, and adsorbent attrition
losses.
Table 2. Performance Parameters
Parameter
Analyzed by
Procedure Used
Total Organic Carbon (TOC) A
Total Organic Halogens (TOX)* B
Specific organics (39 total) C
UV absorbance A
Fluorescence A
Iodine Number B,D
Apparent Density B,D
Phenol Number B,D
Bacteriological A
Dohrmann DC-54 (3)
Pyrolysis, micro-
coulometric detector
(4,5,6)
GC/MS
254 nm (7)
Rapid fluorometric
method (8)
AWWA Standards (9)
Standard plate count
35°C coliform
analysis (10)
A - Passaic Valley Water Commission
B - Havens and Emerson, Consulting Engineers
C - Rutgers University
D - Shirco, Inc.
*Includes influent and effluent to adsorbers, plus top and
bottom GAC samples.
671
-------
For the initial reactivation, the contact time for the
granular carbon within the furnace was 20 minutes at a tempera-
ture of 1,650°F (900°C). These conditions were selected by the
furnace vendor on the basis of thermographic analysis and Iodine
Numbers.
The first carbon to be reactivated in this system was the
lignite-based HD-1030 (a product of ICI America, Inc.). Unfor-
tunately, because of inexperience, problems were encountered in
handling the adsorbent and in operating the furnace resulting
in atypical losses. Further, performance of the lignite acti-
vated carbon could not be equitably judged. The characteristics
of the two coal-based granular carbons (Westvaco WV-W and Calgon
F-400) before and after reactivation are shown in Table 3.
Iodine numbers were restored to within 12 percent of the virgin
granular carbon and apparent densities increased about 3 percent
through the cycle.
Table 3. Comparison of Virgin To Reactivated
Granular Carbon
Cycle 1
Virgin Exhausted Reactivated
Iodine Number
HD-1030 601 416 (See text)
WV-W 850 711 761
F400 1,023 790 890
Apparent Density
HD-1030
WV-W
F400
0.44
0.57
0.47
0.44
0.61
0.51
(See text)
0.57
0.49
The real test was how the granular activated carbon per-
formed during actual treatment of Passaic River water. The
total organic carbon (TOC) concentration in the influent to the
virgin activated carbon ranged from 1.5 mg/1 to 3.9 mg/1 with a
mean of 2.7 mg/1. When the adsorbents were reactivated and put
back into service, the TOC had a mean concentration of 4.4 mg/1.
Because of the different loadings, the comparison of virgin to
reactivated carbon is obscured, but if these differences are
neglected, the decline in TOC removal is approximately 12
percent for the 11 weeks in service after reactivation (Figure
2). Subsequent cycles should allow for more direct comparisons.
672
-------
1.0
.2
1st. Reactivation
.. V
*^"*»^^
Inf. TOC:).5-5.3mg/l
EBCT = 8min.
4 6 8 10
TIME IN SERVICE,WEEKS
12
Figure 2. Effectiveness of Reactivation Measured by
Total Organic Carbon
Organic halogen (TOX) concentrations on samples of activated
carbon collected from the top and bottom of each adsorber were
averaged, with the results shown in Figure 3. Again, the reacti-
vated carbon was about 10 percent less effective than the virgin
material. Because of a modification in the TOX analysis the
influent and effluent TOX concentrations before and after reacti-
vation cannot be compared directly. For perspective, however,
the nonpurgeable organic halogen concentrations applied
to the adsorbents after reactivation ranged from 106 yg/1 to 260
ug/1 and the breakthrough pattern was very similar to the TOC
curve.
Although both UV absorbance and fluorescence data were
collected, the latter parameter is more consistent and is
potentially a good operational monitor for this system. Figure
4, a plot of fractional removal of fluorescence, shows about
4 percent difference between virgin and reactivated granular
carbon.
DISCUSSION OF RESULTS
The study by the Passaic Valley Water Commission at the
Little Falls Filtration Plant is providing very useful information
for understanding the handling and reactivating of granular
carbon. The infrared furnace in the first reactivation restored
the adsorptive properties to within 85 to 96 percent of virgin,
depending on which parameter performance is judged on.
673
-------
E
a
^
9
E
*>
X
Q
Note: These values are
averages of top
and bottom samples
Virgin GAC
10
12
TIME IN SERVICE , WEEKS
Figure 3.
Effectiveness of Reactivation Measured
oy Total Organic Halogen (TOX)
.2 --
Figure 4
2 4 6 8 10
TIME IN SERVICE,WEEKS
Effectiveness of Reactivation Measured
by Fluorescence. RFM = Rapid Fluorometric
Method (Ref. 8}.
674
-------
Most of the published reactivation data from European
installations use UV absorbance as a measure of performance. As
an example, Weissenhorn (11) showed that virgin activated and
reactivated carbon were nearly equivalent. Interestingly, the
parameter showing the least change in performance in the Little
Falls study is fluorescence, so any surrogate measurement will
be closely examined during the remainder of this project.
The second reactivation has been completed and currently
(May, 1979) the twice reactivated carbon is being exposed to
filtered water. Furnace conditions were slightly changed for
the second reactivation. The temperature in the activation zone
was raised to 1,700°F (92/6°C) a,nd the retention time lengthened
to 25 minutes. the Iodine Numbers and apparent densities im-
proved and when compared to the first reactivation, these pre-
liminary results indicate that the removal of other parameters,
such as TOC and TOX, will also improve.
PRODLEMS
The methods used to transfer the exhausted carbon from the
adsorbers to the storage hopper are somewhat archaic and cumber-
some. The greatest problem has been the discharge of activated
carbon from the hopper because of arching within the sloped
sides of this unit causing erratic feed rates into the
furnace. The storage system is being redesigned to include an
internal vibrator. This should minimize, if not eliminate, the
feed problem. With improvements in the feed system, the capacity
of the furnace could be fully utilized; presently the throughput
is only about one-half the rated capacity.
Since the infrared furnace is still in the development
stage, assessing its mechanical durability at this time would be
unrealistic. Corrosion has been troublesome where #304 stainless
steel was in contact with either the wet activated carbon or the
scrubber water. This should be solvable with a different type
stainless steel (probably #316), or possibly a liner or coating
as a preservative.
SUMMARY AND FUTURE WORK
The performance of the electric furnace at the Little Falls
Filtration plant (Passaic Valley Water Commission) is very en-
couraging. Problems with activated carbon handling and corrosion
have occurred, but these are minor and are being corrected. In-
formation currently (May, 1979) available on the operation of this
unit is summarized in Table 4. The preliminary data indicate
that the second reactivation is improved from the first. Moni-
toring attrition losses in discrete processes is not possible;
675
-------
hence/ the losses shown are aggregate values. These losses can
perhaps be reduced as more experience is gathered on operating
the system. The power requirements are also reassuring.
Assuming a power cost of $0.03/kW-hr the cost of running the
furnace would be slightly over 2 cents/lb ($0.045/kg) of carbon
reactivated. This compares very favorably with the reported
power costs from installations using other type furnaces (12).
Table 4. Comparison Summary for Virgin and
Reactivated Carbon
Iodine Number
Apparent Density
Total Organic Carbon
Flourescence
Total Organic Halogen
Attrition Losses*
Power Required**
Cycle 1
Percent
Change
-12
+ 3
-11
- 4
-10
7%
0.7 kW/lb
Cycle 2
Percent
Change
-6
+ 1
Not yet available
Not yet available
Not yet available
4%
0.7 kW/lb (1.6 kW/kg)
*This is the entire system — sum of losses from initial
backwash through transporting, reactivation, and quenching.
**This is the electrical power for the furnace only.
Steps are now being taken to improve the activated carbon
handling. Also an effort will be made to monitor off-gasses
before and following an afterburner. This effort is part of a
much broader program within the U.S. EPA to study reactivation
of granular carbon used in drinking water applications. Other
sites in various stages of study are shown in Table 5. Details
on any of these projects are available from either the project
officer or the principal investigator.
ACKNOWLEDGEMENTS
The authors thank Dr. F.K. McGinnis of Shirco, Inc. and
Mr. Fred Menzenhaur and his staff at Havens and Emerson for
their help. Particular thanks is expressed to Mr. Dick Roby and
William Reed as well as their coworkers at the Little Falls
Filtration Plant for their assistance.
676
-------
TABLE 5
ONGOING REACTIVATION STUDIES
FURNACE AMOUNT FIRST EPA PROJECT PRINCIPAL
SITE TYPE SIZE, Ib/hr (kg/hr) REACTIVATED lb(kg) REACTIVATION OFFICER INVESTIGATC
Little Falls, Shirco 100 (45) 45,000. (20,500) June 1978 Love, T. Inhoffer, H
New Jersey Infrared
Manchester, Westvaco 500 (225) 320,000(145,500) May 1979 DeMarco, J. Kittredge, D.
New Hampshire Fluid Bed
Cincinnati Westvaco 500 (225) 350,000(159,000)Sept. 1979 DeMarco J. Miller, R.
Ohio Fluid Bed
Evansville, Shirco 100 (45) 1,500 (680) June 1979 Lykins, B. Mills, D.
Indiana* Infrared
EPA Lab Several processes on 40 (18) July 1979 Love, T. Love, T.
Cincinnati a bench scale
*Exhausted GAC will be reactivated using furnace located at
Little Falls, New Jersey
-------
REFERENCES
1. Inhoffer, Wendell R. and DeHooge, Frank J. "Free Residual
Chlorination of Passaic River Water at Little Falls, New
Jersey — A Hobson's Choice." Proceedings/ 94th Annual AWWA
Conference, Boston, Massachusetts, June 1974.
2. Inhoffer, Wendell R. "Use of Granular Activated Carbon at
Passaic Valley." Proceedings 2nd Annual AWWA Water Quality
Technology Conference, Atlanta, Georgia, Dec. 1975.
3. Takahashi, Yoshihiro. "Ultra Low Level TOC Analysis of
Potable Waters," Proceedings AWWA Water Quality Technology
Conference, San Diego, California, December 1976.
4. Kiihn, Wolfgang and Sontheimer, Heinrich. 1973. "Several
Investigations on Activated Charcoal for the Determination
of Organic Chloro-Compounds," Vom Wasser, 41, 65-79.
5. Kiihn, Wolfgang and Sontheimer, Heinrich. 1973. "Analytic
Determination of Chlorinated Organic Compounds with
Temperature-Programmed Pyrohydrolysis," Vom Wasser, 4,
1-15. ~
6. Dressman, R.C., McFarren, E.F. and Symons, J.M., "An Evalu-
ation of the Determination of Total Organic Chlorine (TOC1)
in Water by Adsorption onto Ground Granular Activated
Carbon, Pyrohydrolysis and Chloride Ion Measurement." (With
addendum for samples which contain purgeable organic halogen
compounds.) Proceedings AWWA Water Quality Technology
Conference, Kansas City, Mo., Dec. 1977.
7. Dobbs, Richard A., Wise, Robert H. and Dean, Robert B.
1972. "The Use of Ultraviolet Absorbance for Monitoring the
Total Organic Carbon Content of Water and Wastewater."
Water Research, £, 1173-80.
8. Sylvia, Albert. 1973. "Detection and Measurement of
Microorganics in Drinking Water," Jour. New England Water
Works Assn., 87, No. 2.
9. AWWA Standard for Granular Activated Carbon - B604-74,
January 1974, JAWWA, 11., 672-681, Nov. (1974).
10. Standard Methods for the Examination of Water and Waste-
water, 14th Edition, APHA, AWWA, WPCF (1975).
678
-------
11. Weissenhorn, F.J. "Testing of Activated Carbon Filters in
Waterworks/" In: Translation of Reports on Special Prob-
lems of Water Technology - Vol. 9 - Adsorption, Conference
held in Karlsruhe, Federal Republic of Germany, 1975,
pp. 238-250, EPA 600/9-76-030, U.S. Environmental Protection
Agency, Cincinnati, Ohio (Dec. 1976), 454 pp.
12. Clark, Robert M. and Dorsey, Paul. "Influence of Operating
Variables on the Cost of Treatment by GAC Adsorption."
pp. 705-742 in this Symposium.
679
-------
COST OF GRANULAR ACTIVATED CARBON
ADSORPTION TREATMENT IN THE U.S.
Robert C. Gumerman, Russell L. Gulp
and Robert M. Clark
The use of granular activated carbon (GAC) as a means of
treating drinking water was proposed by the U.S. Environmental
Protection Agency (EPA) in January 1978 (1). Since that time
arguments both for and against the use of GAC as a treatment
method have been widely published (2). Serious challenges and
questions have been raised regarding EPA's cost estimates for
GAC use.
In order to respond to some of these questions, EPA's Drink-
ing Water Research Division initiated a carefully designed study
to establish water supply unit process cost curves on a con-
sistent and understandable basis (3). The study focused on proc-
esses for the removal from potable water of contaminants included
-in the National Interim Primary Drinking Water Regulations
(NIPDWR) and on the development of construction, and operation and,
maintenance cost curves for these processes. The final report
for this project contains cost curves for 99 different unit
processes. These cost curves were divided into two categories:
(1) large water treatment systems applicable to flows between 1
and 200 mgd (3,785 and 757,000 m3/d), and (2) small water
treatment systems applicable to flows between 2,500 gpd and 1 mgd
(9.46 and 3,785 m^/d). A computer program for retrieving, up-
dating, and combining the cost data was also developed as a part
of this study. This paper reports on that portion of the study
related to cost curves associated with GAC applications.
COST CURVES FOR ACTIVATED CARBON APPLICATIONS
Twelve of the 99 cost curves included in the final report
are specifically for activated carbon applications. Eleven of
the processes discussed are applicable to large systems (1 to
200 mgd); the package GAC columns pertain to small systems
(2,500 gpd to 1 mgd).
Gravity Carbon Contactors - Concrete Construction
Concrete gravity carbon contactors are essentially the same
as concrete gravity filtration structures. Costs were developed
for carbon bed depths of 1.5 and 2.6 m (5 and 8.5 ft), which
680
-------
provide empty bed contact times (EBCT) of 7.5 and 12.5 mift
respectively, at an application rate of 3.4 mm/s (5 gpm/sq ft).
The contactors were assumed to be completely housed. The costs
incorporated are for the contactor only, and exclude surface wash
and backwash pumping facilities, the initial carbon charge, and
carbon handling equipment outside of the pipe gallery.
Gravity Carbon Contactors - Steel Construction
Steel gravity carbon contactors are large diameter, field-
erected structures. Costs were developed for 6.1- and 9.1-m
(20- and 30-ft) diameter contactors which were 10.7 m (35 ft)
deep and were suitable for a carbon bed depth of 6.1 m (20 ft).
The contactors were sized for downflow operation and an applica-
tion rate of 3.4 mm/s (5 gpm/sq ft), which provides a 30-min
EBCT. Spent carbon is removed from the contactors through
multiple carbon drawoff pipes in the underdrain support plate.
Regenerated carbon is returned to the top of the contactors,
which are completely housed. The costs exclude surface and
backwash pumping facilities, the initial carbon charge, and the
carbon handling equipment outside of the building.
Pressure Carbon Contactors
Pressure carbon contactors are shop-fabricated, cylindrical
pressure vessels having a diameter of 3.7 m (12 ft). Costs were
developed for contactors with carbon bed depths of 1.5, 3, an$3
6.1 m (5, 10, and 20 ft) that give EBCTs of 7.5, 15, and 30 min
at application rates of 3.4 mm/s (5 gpm/sq ft). The contactors
were designed for downflow operation and employed a nozzle type
underdrain system for rapid spent carbon removal. The contactors
were completely housed. The costs exclude backwash and surface
wash pumping facilities, the initial carbon charge, and carbon
handling equipment outside of the building.
Conversion of Sand Filters to Carbon Contactors
An inexpensive, but temporary method of providing carbon
contact is afforded by removing the sand media and replacing it
with activated carbon. Generally the underdrain and support
gravel design can be retained without modification. The instal-
lation of a spent carbon collector and transport system and a
return system for regenerated carbon are the only additions
required. Extensive piping replacement and alteration of exist-
ing filter rate controls and instrumentation are usually neces-
sary if the application rate is to be increased beyond the
original filtration rate.
Granular Activated Carbon - Material Cost
Virgin granular activated carbon is required for both
the initial carbon charge and to replace carbon lost during
681
-------
regeneration. For quantities weighing less than 18,145 kg
(40,000 Ib), carbon is normally purchased in 0.056-m (2-cu ft)
bags; larger quantities are purchased in bulk.
Multiple Hearth Granular Carbon Reactivation
Multiple hearth reactivation uses a multiple hearth furnace
operated under closely controlled conditions of temperature,
oxygen, and moisture content. The required furnace size is a
function of the required reactivation frequency, carbon dosage
(which is a function of the nature of the organics adsorbed),
allowable hearth loading, and anticipated downtime. The costs
that were developed include the basic furnace, cooling fans,
spent carbon storage and dewatering equipment, regenerated-
carbon handling system,'quench tank, exhaust scrubbing system,
and all necessary instrumentation. The furnace was assumed to
be housed. Operation and maintenance cost curves are based on
continuous operation, and adjustment must be made for operation
times falling below 100 percent.
Infrared Granular Carbon Reactivation
Granular carbon can be reactivated by using infrared energy
to generate heat. The principal advantage of infrared reactiva-
tion is the ability of the furnace to be rapidly put into or
taken out of operation without furnace damage or excessive oper-
ational cost. The carbon moves through the furnace on a conveyor
belt, and reactivation time is varied by changing the conveyor
speed and varying the depth of the carbon on the belt. The cost
curve includes the furnace, spent carbon storage and dewatering
facilities, quench tank, afterburner and scrubber, all required
electrical equipment and controls, and an enclosure for all
equipment. The operation and maintenance cost curves assume
operation 100 percent of the time, and adjustment must be made
for lesser operation times.
Fluid Bed Granular Carbon Reactivation
This type of reactivation uses hot gases both to fluidize
and reactivate carbon. No inert heat source such as2sand is
required. Reactivation rates as high as 342 kg/hr/m (70 Ib/
hr/sq ft) can be accomplished. The cost curve includes spent
and reactivated carbon storage, carbon dewatering equipment, the
fluid bed reactor, fluidizing air blower, quench tank, particu-
late scrubber, instrumentation, and controls. All facilities
were assumed to be housed. The operation and maintenance cost
curves are developed on the basis of 100 percent operation;
adjustment must be made for less operational time.
682
-------
Atomized Suspension Powdered Carbon Reactivation
The atomized suspension process operates by spraying
a carbon slurry through a spray nozzle, where the slurry is
atomized with steam, into the top zone of a three zone reactor.
In the top zone of the reactor all water is evaporated, in the
middle zone the organics adsorbed on the carbon are pyrolyzed to
a residual char deposit, and in the third zone the residual char
deposit is removed. The reactor diameter can vary from 0.3 to
0.9 m (1 to 3 ft) and the height can vary from 3 to 15.2 m
(10 to 50 ft), depending upon the reactivation capacity. The
cost curves developed include a spent carbon slurry storage
tank, the basic reactor, a feed pump, exhaust scrubbing, reac-
tivated carbon recovery equipment, and all required instrumen-
tation and controls. The operation and maintenance curves are
for 100 percent operation, and adjustment must be made for
lesser percentages of operation.
Fluidized Bed Powdered Carbon Reactivation
Powdered carbon can be reactivated in a fluidized bed
of graded silica sand that is maintained at 704°C (1,300°F).
The fuel source and fluidizing air are added at the bottom of
the reactor. Wet spent carbon is added from above the sand
layer and reactivated carbon is separated from the fluidizing
sand and withdrawn at the base of the reactor. The costs
developed include the fluidized bed reactor, fluidizing air
blower, carbon feed and removal equipment, and required
instrumentation and controls. The operation and maintenance
cost curves assume 100 percent operation, and adjustment for
operating time less than 100 percent is required.
Of f-Site_ Regional Carbon Reactivation
Small water treatment plants employing activated carbon
may find reactivation at a large plant (or a regional reacti-
vation facility) more economical than purchasing virgin carbon
to replace spent carbon or providing on-site reactivation. The
construction costs include only granular carbon dewatering-
storage bins. The costs do not include the cost of reactivation
at the regional site. The operation and maintenance curves
include the cost of hauling to the regional reactivation location
but do not include the cost of reactivation.
Package Granular Activated Carbon Columns
Factory-assembled package GAC carbon columns may be used in
small water treatment applications. Costs were developed for
pressurized, downflow steel contactors designed for manual oper-
ation. The columns are skid-mounted and both supply and backwash
pumps are included, as is housing. Typical carbon bed depths are
1.5 m (5 ft); application rate is 3.4 mm/s (5 gpm/ sq ft).
683
-------
DERIVATI-ON OF COST CURVES
Relating Costs to Design Parameters
The construction cost for each unit process is presented as
a function of the process design parameter which was determined
to be the most useful and flexible under varying conditions such
as loading rate, detention time, or other conditions that can
vary because of designer's preference or regulatory agency
requirements. For example, the contactor construction cost
curves are presented in terms of cubic feet of contactor volume,
an approach that allows various empty bed contact times to be
used. Contactor operation and maintenance curves are presented
in terms of square feet of surface area since operation and
maintenance requirements are more appropriately related to
surface area than to contactor volume. Reactivation facility
cost curves are presented in terms of square feet of hearth area
for the multiple hearth furnace and pounds per day of reactiva-
tion capacity for the other reactivation concepts. This allows
the loading per square foot of hearth area to be varied for the
multiple hearth furnace (and other furnaces) for carbon reactiva-
tion rates less than the maximum capacity for which the system is
designed. This approach provides greater flexibility in the use
of the cost curves than if the costs were related to water flow
through the treatment plant.
Methods Used to Develop Cost Curves
The construction cost curves were developed using equipment
cost data supplied by manufacturers, cost data from actual plant
construction, unit takeoffs from actual and conceptual designs,
and published data. The costs developed* then were checked by
a second firm**, using an approach similar to what would be
utilized by a general contractor in determining the construction
bid. Any discrepancies that existed between the initial estimate
and the second estimate were then resolved.
Construction Cost Components
The costs for eight principal construction components were
developed and then aggregated to give the construction cost for
each unit process. The eight components are (1) excavation and
sitework, (2) manufactured equipment, (3) concrete, (4) steel,
(5) labor, (6) pipe and valves, (7) electrical and instrumenta-
tion, and (8) housing. These categories also provide enough
detailed information to permit accurate cost updating.
*Culp/Wesner/Culp, Santa Ana, Calif.
**Zuheide-Herrmann, Inc., St. Louis, Mo.
684
-------
The construction cost curves are not the final capital cost
for the unit process. The construction cost curves do not
include costs for general contractor overhead and profit, admin-
istration, engineering and legal fees, fiscal determinations,
and interest during construction. These items are more directly
related to the total cost of a project than to the cost of the
individual unit processes. Therefore, they are more appropri-
ately added following summation of the cost of the individual
unit processes, if more than one unit process is required. Four
example calculations are presented in Tables 1 to 8 to illustrate
the recommended method for the addition of these costs.
Operation and Maintenance Cost Components
Operation and maintenance requirements were developed for
building-related energy, process energy, maintenance material,
and labor. The separate determination of building energy allows
for regional variations. Energy requirements are presented in
kilowatt-hours per year for electricity, standard cubic feet
per year for natural gas, and gallons per year for diesel fuel.
Labor is presented in hours per year, allowing local variations
to be incorporated into the operation and maintenance cost
calculations. Maintenance material cost is given in dollars per
year, but does not include the cost of chemicals. Chemical costs
are added separately as shown in Tables 1 to 8.
EXAMPLE COST CURVES
For the majority of the unit processes, three separate
figures were used to present construction and operation and
maintenance curves. The first graph presents the construction
cost, the second graph presents energy (electrical, natural gas,
and diesel fuel) and maintenance material requirements, and the
third graph presents labor requirements and total operation and
maintenance cost. Figures 1, 2, and 3 present the cost curves
that were developed for pressure carbon contactors.
UPDATING COST CURVES
For many engineering purposes a single cost index such as
the Engineering News Record (ENR) Construction Cost Index (CCI)
can be used to update construction costs. When this approach
provides sufficient accuracy, a single index is certainly the
simplest and easiest approach. In the case of water treatment
construction costs, especially those for carbon treatment, use
of the CCI is not recommended because the construction components
used in its development are not related to those required for
water treatment plants. For example, the CCI does not include
process equipment or electrical equipment for instrumentation and
control. For water treatment plants, the use of several indices
is recommended. For example, this could include the Bureau of
685
-------
Table 1
Cost Calculation and Operation and Maintenance Requirements
for a 2 mgd Pressure Granular Activated Carbon Plant
Design Construction Operating Energy Natural Gas Diesel
Parameter Cost Parameter (kW-hr/yr) (acf/yr) Fuel (gal/yr)
1670 ft3/
contactor
86,890 Ib
1356 gjro
3342 ft3
880 sq ft
—
—
—
—
—
—
—
—
—
—
—
$ 225,770 226 ft2 133,900 0 0
total area
53,720 — 000
43,810 — 000
60,730 434,450 Ib/yr 0 0 1,680
111,350 5 JO sq ft 38,110 3,973,960 0
0 26,067 Ib/yr 00 0
495,380 — 172,010 3,973,960 1,680
24,770 — — — —
0 ~ — — —
0 — — — —
520,150
62,420 — — — —
582,570 — — — —
58,260 — — — —
640,830 — — — —
4,000 — ~ — —
12,060 — ' — — —
24,830 — — — —
681,720 — — — —
Maintenance Labor
Material ($/yr) (hr/yr)
1,860 1,057
0 0
0 0
170 141
530 424
16. 840 0
19,400 T7552
— ~
_
~~ ~~
— —
— —
—
_ _
_ _
— —
— —
— —
System and Design Criteria
Pressure Carbon Contactors - 18 min.
E.B.C.T., 5 gpu/ft2
Initial Carbon Charge - 26 Ib/ft3
Backwash Pusping - 12 gpn/ft2
Off-Site Regional Reactivation
Transportation - 20 miles
Off-Site Regional Regeneration*
Make-up Carbon - 6%/regen.,
5 regen./yr
Subtotal
00 Site-Work, Interface Piping,
<" toads « 5*
Subsurface Considerations
Standby Power
Ibtal Construction Cost
General Contractor's Overhead and
Profit
Subtotal
Engineering g 10%
Subtotal
Land, 2 acres ?2,000/acre
legal, Fiscal, and Administrative
Interest during Construction - 8%
TOTAL CAPITAL COOT
'Assumes this plant uses 5% of the regional regeneration facilities, and that the regional facilities operate at 70 percent of full capacity
using nultiple hearth granular carbon regeneration.
-------
Table Z
Annual Cost for 2 mgd Pressure GAC Plant
Item; Total Cost/year
Amortized Capital @ 8%, 20 years $ 69,430
Labor, 1,622 hr, @ $10/hr (total labor 16,220
c»sts including fringes and benefits)
Electricity, 172,010 kW-hr @ $0.03/kW-hr 5,160
Natural Gas, 3,973,960 scf @ $0.00175/scf 6,950
Diesel Fuel, 1,680 gal @ $0.45/gal 760
Maintenance Material 19,400
TOTAL ANNIAL COST* $117, 920
*Cents per 1, 000 gal treated - $117,920 (100)
1,400 (365)
23.1 £/l,000 gal
687
-------
TABLE 3
Cost Calculation and Operation and Maintenance Requirements
for a 20 mgd Pressure Granular Activated Carbon Plant
co
System and Design Criteria
Design
Parameter
Construction
Cost
Operating
Parameter
Pressure Carbon Contactors - 18 rain. 2,089 ft3/
E.B.C.T., 5 gpm/ft3 Contactor
Initial Carbon Charge - 26 lb/ft3
Backwash Punping - 12 gpm/ft2 1,356 gpm
Infrared Carbon Reactivation 4,000 Ib/day
Make-up Carbon - 6%/regen., —
5 regen/Yr
Subtotal —
Sitework, Interface Piping,
Roads e 5%
ixibsurface Considerations
Standby Power
Total Construction Cost
General Contractor's Overhead
and Profit
Subtotal —
Engineering @ 10% —
Subtotal
Land, 5 acres e $2,000/acre
Legal, Fiscal, and Administration
Interest during Construction - 8% —
TOTAL CAPITAL COST
Energy Maintenance Labor
(kH-hr/yr) Material ($/yr) (hr/yr)
$ 1,961,850 1,810 ft2
total area
618,660
523,550 —
43,810 —
339,460 4,000 Ib/d
0 260,690 Ib/day
$ 2,868,670
143,430
0
0
§ 3,012,100
301,210
$ 3,313,310
331,330
$ 3,644,640
10,000
35,240
238,010
8,890
1,816,760 178,340
2,308
1,
0
0
198, 100
0
0
0
10,640
158,810
0
0
2,519
0
"4,128
-------
Table 4
Annual Cost for 20 mgd Pressure GAC Plant
Item: Total Cost/year
Amortized Capital @ 8%, 20 years $400,060
Labor, 4,828 hr, @ $10/hr (total labor 48,280
costs including fringes and benefits)
Electricity, 1,816,760 kW-hr @ $0.03/kW-hr 54,500
Maintenance Material 178,340
TOTAL ANNUAL COST* $681,180
*Cents per 1,000 gal treated
14,000 (365)
13.3 jzf/1,000 gal
689
-------
Table 5
Cost Calculation and Operation and Maintenance Requirements
from a 75 mgd Gravity, Steel Granular Activated Carbon Plant
vo
o
System and Design Criteria
Gravity, Steel Contactor - 18 rain.
E.B.C.T., 30' dia.. 5 gpn/ft2
Initial Carbon Charge - 26 Ib/ft3
Backwash Puiping - 12 gpn/tt2
Multiple Hearth Carbon
Reactivation Furnace
Make-up Carbon, 6% regen.,
5 times/yr.
Subtotal
Site-work, Interface Piping,
Roads e 5%
Subsurface Considerations
Standby Rawer
Total Construction Cost
General Contractor's Overhead
and Profit
Subtotal
Engineering e 10%
Subtotal
Land 15 acres 9 52,000/acre
Legal, Fiscal, and Administrative
Interest during Construction - 8%
TOTAL CAPITAL COST
Design Construction Operating Biergy Natural Gas Maintenance labor
Parameter Cost - $ Parameter (kW-hr/yr) S.C.F/yr Material ($/yr) (hr/yr)
12,533 ft3/
Contactor
3,258,690 Ib
7,520 gpra
800 ft2
$ 3,029,620 125,330 2,596,330 0 16,460
1,907,800 — 0 00
101,710 — 0 00
2,151,230 800 ft2 909,700 109,446,800 11,630
0 yr 0 0 588,130
3 7, 150, 3 66 — 3,506,030 109,446,800 bl6,i20
359,520 — — — —
0 ~™ ™~" ~™ ** *~
0 — — — —
5 7,549,880 — — — —
754,990 — — — —
$ 8,304,870 — — — —
830,490 — — — —
5 9,135,360 — — — —
30,000 — — — —
65,100 — — — —
742,550 — — — —
$ 9,973,010 — — —
5,971
0
0
10, 956
0
16,921
-------
Table 6
Annual Cost for 75 mgd Gravity, Steel GAG Plant
Item; Total Cost/year
Amortized Capital @ 8%, 20 years $1,015,750
Labor, 16,927 hr, @ $10/hr (total labor 169,270
costs including fringes and benefits)
Electricity, 3,506,030 kW-hr @ $0.03/kW-hr 105,180
Natural Gas, 109,446,800 scf @ $0.00175/scf 191,530
Maintenance Material 616,220
TOTAL ANNUAL COST* $2,097,950
*Cents per 1,000 gal treated = $2,097,950 (100)
52,500 (365)
= 10.9 jzf/1,000 gal
691
-------
Table 7
Cost Calculation and Operation and Maintenance Requirements
for a 110 mgd Gravity, Steel Granular Activated Carbon Plant
Ol
\O
N)
System and Design Criteria
Qravity, Steel Contactor - 18 min.
E.B.C.T., 30' dia., 5 qpm/tt2
Initial Carbon Charge - 26 Ib/ft3
Backwash Punping - 12 gpn/ft2
Multiple Hearth Carbon
Reactivation Furnace
Make-up Carbon, 6% regen.,
5 times/yr.
Subtotal
Site-work, Interface Piping,
Roads e 5%
Subsurface Considerations
Standby Rawer
Ibtal Construction Cost
General Contractor's Overhead
and Profit
Subtotal
Engineering 3 10%
Subtotal
Land 20 acres 0 52,000/acre
Legal, Fiscal, and Administrative
Interest during Construction - 8%
TOTAL CAPITAL COST
Design Construction Operating Ehergy Natural Gas Maintenance Labor
Parameter Cost - $ Parameter (kW-hr/yr) S.C.F/yr Material ($/yr) (hr/yr)
11,489 ft3/ $4,503,440 183,824ft3 3,720,810 0 21,590
Contactor
4,779,400 Ib 2,746,810 — 000
7,520 gpn 101,710 — 000
1,173ft2 2,467,640 1,173ft2 1,111,310 158,149,050 12,740
— 0 1,433,620 0 0 857,740
Ib/vr
— S 9,819,600 — 4,832,120 158,149,050 892, 070
— 490,980 — — — —
— _ rt „,,. wm.^ _ ^_,
_ o — — — —
— $10,310,580 — — — —
— 927,950 — — — —
— $11,238,530 — — — —
— 1,123,850 — — — —
— $12,362,380 — — —
— 40,000 — — — —
— 79,070 — — — —
— 1,10«,240 — — — —
— $13,589,690 — — — —
8,362
0
0
13,824
0
22, 186
—
—
—
—
—
—
—
—
—
—
-------
Table 8
Annual Cost for 110 mgd Gravity, Steel GAG Plant
Item; Total Cost/year
Amortized Capital @ 8%, 20 years $1,384,110
Labor, 22,186hr, @ $10/hr (total labor 221,860
costs including fringes and benefits)
Electricity, 4,832,120 kW-hr @ $0.03AW-hr 144,960
Natural Gas, 158,149,050 scf 6 $0.00175/scf 276,760
Maintenance Material 892,070
TOTAL ANNUAL COST* $2, 919, 760
*Cents per 1,000 gal treated = S2!9.19.:?60 (j:OQ)
77,000 (365)
= 10.4 jzf/1,000 gal
693
-------
9
8
7
6
3
2
9
8
e
5
4
3
2
-wl.OOO.OC
> 9
1- 8
in 7
s :
s
»- 3
o
=>
£ 2
to
z
o
o IOO,0(
9
8
7
6
5
4
3
2
)0
)0
_^*
-^^!
^^
_^
^'^
f*
«•
*
•
^-^
^^
*<»
t
100 2 3456789 IOOO 2 3 456789 iqpOO 2 3 456789
INDIVIDUAL CONTACTOR VOLUME -ft3
\ i i 1
10 100
INDIVIDUAL CONTACTOR VOLUME - m3
IOOO
Figure 1. Construction Cost for
Pressure Carbon Contactors
694
-------
Ul
o
Z
<
Z
Ul
1
7
6
5
4
3
2
100,0
1
6
s
4
3
Z
IO.OO
9
8
7
6
5
4
3
2
101
9
8
7
6
S
4
3
2
- *
7
6
5
4
3
2
DO 1,00
E I
6
5
4
3
2
0 100, (
9
;*».
- f 3
- 1 4
i
- & 3
(T
- % 2
Ul
30 10,0
9
8
7
6
5
4
3
2
1000
1
•
0,000
a*
30O
**>
00
X
.-•s:
x^
X
X*
x
X
X
X
/
^
x
/
— T
^
/
y
^
/
S
*
t
4
9
f
j
I/
s
^
S
X
,
x
s
/
s
jf
S
f
S
/
/
jr
r
/
/
V
/
/
/
/
/
f
/
/
/
/
f
/
/
\ • —\ - - •• •
1 1
BUILD
MA1N7
MATE
ING
ENA
?IAI
PROCESS
E
NC
E
i i
Nf
E
^JE
:F
G
t
R
j
"
i
Y
t
_U-
IOO
10
3 4 367891000 234 56789OOOO 2
TOTAL CONTACTOR AREA-ft 2
•4-
3 456 789
100 IOOO
TOTAL CONTACTOR AREA -m2
Figure 2. Typical Activated Carbon Column Installation
695
-------
8
o
I"
7
6
S
4
3
2
1,000
I1
6
S
4
3
2
100,0
9"
8
7
6
5
4
3
2
10,00
9
8
7
6
5
4
3
2
000
r g
7
6
5
4
•»
2
poo
: I
6
5
4
3
2
00 . 100
9
8
6
- S.5
-^4
' i*
O
CO
- 32
0 IO.OC
9
8
7
6
5
4
3
2
IOOO
'
000
- — •
)0
— • • •
.» •**"
-— -!
,--
__.
»•—
.«••
***
**
*•
*
i*
»
ff
*
,*
^*^~
^
*^\
^"
S
x^
XI
^^
r
X1
X1
x1
X
X
/
/
/
x
"^
T
i^
^ LAE
^
)TA
OR
.
COS!
f
,
IOO 2345 67891000 2 3456 789IOOOO 2 3 4567
TOTAL CO<|TACTOR AREA -ft 2
89
ib 160 lobo
TOTAL CONTACTOR AREA - m2
Figure 3. Operation and Maintenance Requirements for
Pressure Carbon Contactors - Labor and Total Cost
696
-------
Labor Statistics (BLS) (4) indices in conjunction with the ENR
Building Cost Index and Skilled Labor Wage Index.
Table 9 shows the indices recommended for application to
each of the eight categories used to derive construction costs.
This approach allows updating to be done proportionally to
changes in the cost of each component. The eight construction
cost components represent the major items of material and labor
affecting the cost of water treatment plant construction.
The principal disadvantages of this approach are the lack of
geographical specificity of the BLS indices and the additional
effort involved in the use of seven indices rather than a single
index. The approach that was used to develop the cost data pro-
vides the user of the data with a choice of updating procedures
to be used. The individual cost indices can be applied to the
components and component costs then added to obtain the unit
process cost. As an alternative, the total unit process cost can
be updated in a single step by applying the ENR Construction Cost
Index to the sum cost of the eight components.
UPDATING OPERATION AND MAINTENANCE COSTS
Operation and maintenace (O&M) costs are updated somewhat
differently. Energy requirements are presented in terms of
kilowatt-hours per year for electricity, standard cubic feet per
year for natural gas, and gallons per year for diesel fuel; labor
is presented in hours per year. Therefore, updating of energy
and labor costs is accomplished simply by using prevailing unit
costs for energy and labor. Maintenance material costs, which do
not include chemicals, are best updated using the new Producer
Price Index for Finished Goods. The maintenance material costs
presented in the final report are based on October 1978,
Producer Price Index for Finished Goods of 199.7 (4).
THE COMPUTER PROGRAM
A computer program was developed as a portion of this
project. This program and data file that contain the construc-
tion costs as well as operation and maintenance requirements are
presented in the final report. The purpose of the computer pro-
gram is to allow costs to be retrieved, updated, and analyzed.
The program allows a number of other factors to be varied
in addition to the indices used for updating. Other factors
that can be varied to examine their impact on project costs are:
(1) the operating parameter for the process; (2) sitework and
interface piping costs; (3) special subsurface requirements;
(4) standby power requirements; (5) general contractors overhead
and profit percentage; (6) engineering costs; (7) land cost;
(8) legal, fiscal, and administrative costs; and (9) interest
during construction and capital cost amortization.
697
-------
Table 9
Recommended Indices for Derivation of Construction Costs
Construction Cost Components
Indices
Excavation and sitework
Manufactured equipment
Concrete
Steel
Labor
Pipes and valves
Electrical and
i ns trume ntat ion
Housing
ENR wage index (skilled labor)
BLS general purpose
machinery and equipment
code 114
BLS concrete
ingredients code 132
BLS steel mill products
code 1013
ENR wage index (skilled labor)
BLS valves and fittings
code 114901
BLS electrical machinery and
equipment code 117
ENR building cost index
698
-------
EXAMPLE CALCULATIONS FOR GRANULAR ACTIVATED CARBON PLANTS
To demonstrate the application of the construction and
operation and maintenance cost curves that were developed for
granular activated carbon and other processes, examples were
developed for four different capacity GAC postadsorber systems.
These examples use varying concepts for the carbon contactor
design and for the method used to reactivate spent carbon. The
four example GAC plants and their capacities are: (1) 2-mgd
plant - pressure carbon contactors and off-site carbon reactiva-
tion at a regional facility; (2) 20-mgd plant - pressure carbon
contactors and on-site infrared carbon reactivation; (3) 75-mgd
plant - gravity, steel carbon contactors and on-site multiple
hearth furnace carbon reactivation; and (4) 110-mgd plant -
gravity, steel carbon contactors and on-site multiple hearth
furnace carbon reactivation.
The design criteria/ capital cost calculation, and operation
and maintenance requirements for these examples are presented in
Tables 1, 3, 5, and 1, respectively. The annual cost calcula-
tions for the examples are presented in Tables 2, 4, 6, and 8,
respectively. Construction costs are in October 1978 dollars.
Unit costs used to calculate operation and maintenance costs
were $0.03/kWh (electricity), $10.00/hr (labor), $0.00175/scf
(natural gas), and $0.45/gal (diesel fuel).
The examples are only for hypothetical conditions. The
design criteria and costs presented are general, and are not
necessarily applicable to all plants with the same capacity and
operating conditions.
As expected, economy of scale plays a major factor in the
calculation of the unit cost of water produced. As shown in
Tables 2, 4, 6, and 8, the unit cost of water produced by the
example plants ranged from $0.23/1,000 gal treated for the 2-mgd
plant to $0.10/1,000 gal for the 110-mgd plant.
FACTORS INFLUENCING CAPITAL AND OPERATION AND MAINTENANCE COSTS
In addition to deriving detailed capital and operation and
maintenance costs, the data can be used to study the sensitivity
of unit process costs to changes in specific design parameters.
For example, if it is desired to perform a sensitivity analysis
for a specific set of design variables, unit process costs can
be set at predetermined levels and then varied individually over
an appropriate range. The following analysis for postfilter
adsorption systems illustrates this technique.
Assume a set of design variables for a postfilter adsorption
system (contactor plus multiple hearth reactivation furnace) as
given in Tables 10 and 11. Table 10 contains the levels assumed
699
-------
Table 10
Design Parameters for Post-filter Adsorption
Design Parameter Level
Activated carbon cost $ 0.60/lb
Activated carbon loss per reactivation cycle 6%
Fuel cost $ 1.75/mil Btu
Electric power cost $ 0.03/kw hr
Construction cost index 265.4
Producer price index 199.7
Direct hourly wage index $ 10/hr
Amortization rate 8%
Amortization period 20 years
Loss in absorptive capacity 0%
Design capacity 70%
Empty bed contact time 18 min
Reactivation frequency Every 2.4 months
Table 11
Assumptions for Separate Post-filter Adsorption
Design Capacity (mgd)
Item
Number of contactors
Diameter of contactors (ft)
Depth of contactors (ft)
Volume of GAG per
1
3
8
13
653
5
6
12
13
1,469
10
12
12
13
1,469
100
40
20
14
4,396
150
60
20
14
4,396
contactor (ft3)
Minimum empty bed contact 18 18 18 18 18
time (min)
700
-------
as input variables and Table 11 contains assumed design configu-
rations for various sized systems. Table 12 contains the amor-
tized capital and operating and maintenance costs in /d/1,000 gal
resulting from these assumptions.
An important variable related to both cost and quality of
finished drinking water is the period between carbon reactiva-
tions. Figure 4 depicts the cost in /z
-------
Table 12
Amortized Capital and Operating and Maintenance
Cost for Post-filter Adsorption
(cents/1,000 gallons)
Design Capacity (mgd)
Item 1 5 10 100 150
Amortized capital cost:
Filter adsorber 29.4 12.7 9.8 4.9 4.6
Operating and Maintenance
Cost:
Filter adsorber 22.1 12.6 10.8 7.4 7.1
Total Cost:
Filter adsorber 51.5 25.3 20.6 12.3 11.7
702
-------
8
o
g
P
o
i
cc
Q.
23456
PERIOD BETWEEN REACTIVATION, months
Figure 4. Production Cost for Post-filter Adsorbers
versus Period Between Reactivation
(100 mgd design capacity)
703
-------
REFERENCES
Interim Primary Drinking Water Standards - Control of
Chemical Contaminants in Drinking Water, Fed. Reg. 43:28
(Feb. 9, 1978), p. 5756.
Pendygraft, G.W., F.E. Schlegel, and M.J. Huston. 197$.
The EPA-Proposed Granular Activated Carbon Treatment
Requirement: Panacea or Pandora's Box, Jour. AWWA, 71:2:52.
Gumerman, R.C., R.L. Gulp, and S.P. Hansen. Estimating
Costs for Water Treatment as a Function of Size and Treat-
ment Efficiency - Interim Report. MERL, Ofce. Res. Dev.,
USEPA, Cincinnati, Ohio, EPA 600-12-78-182 (Aug. 1978).
Producer Prices and Price Indexes, Bureau of Labor Statis-
tics, U.S. Dept. of Labor (Oct. 1978).
704
-------
INFLUENCE OF OPERATING VARIABLES ON THE COST
OF GRANULAR ACTIVATED CARBON TREATMENT
by •
Robert M. Clark and Paul Dorsey
On January 9, 1978, the Environmental Protection Agency
(EPA) proposed a regulation designed to protect the public from
organic chemical contaminants in drinking water (1). The
proposed regulation consists of two parts: Part A, a maximum
contaminant level (MCL) of 0.10 mg/1 (100 parts per billion) for
total trihalomethanes (TTHM) including chloroform, and Part B, a
treatment technique requiring the use of granular activated
carbon (GAC) to keep synthetic organic chemicals associated with
industrial pollution and urban and agricultural runoff from
contaminating drinking water supplies.
Much concern over these regulations has been expressed by
the potentially affected water supply utilities. The general
focus has been on the costs and benefits associated with the
utilization of GAC as a water treatment process (2). To clarify
GAC economics, a study was undertaken to obtain the best possi-
ble information regarding the construction and operation of GAC
systems (3). Data from this study has been utilized in this
paper to examine the influence of changes in the level of
operating variables on the cost of GAC treatment.
GAC USE IN WATER TREATMENT
Treating water with granular activated carbon involves two
major and separate process operations: filtration and reacti-
vation (4). The water comes in contact with the activated
carbon by passing through a structure filled with activated
carbon granules. Impurities are removed from the water by
adsorption when sufficient time is provided for this process.
The structure can be either a water treatment filter shell in
which the filter media has been replaced with GAC, or an inde-
pendent activated carbon adsorption system, A separate acti-
vated carbon adsorption system usually consists of a number of
columns or basins used as contactors and connected to a re-
activation system. When the activated carbon's adsorptive
capacity is exhausted, it is taken out of service and either
replaced with new activated carbon or reactivated by combustion
705
-------
of the organic adsorbate. Activated carbon is routinely added
to the system to replace that lost during hydraulic transport
and reactivation. These losses include those due to physical
deterioration and burning during the reactivation process.
This economic analysis will include both the use of GAG as
a replacement for existing filtration media (sand replacement)
and as a separate adsorption system (post-filter adsorption).
In the first part of this analysis, on-site multiple hearth
reactivation will be assumed. The economics of using truck
transport combined with regional or centralized reactivation
furnaces will also be considered. Standard levels of key de-
sign parameters will be set at predetermined levels and then
one variable at a time will be changed to determine its effect
on system cost.
The need to consider the cost of separate GAC contactors
is eliminated if GAC is assumed to replace sand in existing
filters. For the purposes of the sand replacement analysis, a
water treatment plant is assumed to consist of an integral
number of 1-mgd (4000 m3/d) filters with an effective volume
of 18.5' (5.6 m) X 18.5' (5.6 m) X 2.5' (0.8 m) or 856 ft3
(24.4 m ). Design parameters assumed tor the sand replacement
systems are shown in Table 1 and design assumptions for post-
filter adsorber systems are shown in Tables 1 and 2.
For sand replacement, a GAC loss of 10 percent per
reactivation cycle is assumed, but a GAC loss of only 6 percent
per cycle is assumed for post-filter"adsorbers. These two
assumptions are intended to reflect differences in the operation
of the two systems. Sand replacement systems are labor inten-
sive and increase the possibility of GAC loss because the
activated carbon is changed manually. In post-filtration
systems, the activated carbon is assumed to be changed hydrauli-
cally, leading to fewer possibilities for handling losses.
Table 3 shows the unit costs for both types of systems for 1, 5,
10, 100 and 150 mgd plants (4,000, 20,000, 40,000, 400,000 and
600,000 m /d) at 70 percent capacity.
The costs shown in Table 3 can be further divided into
reactivation costs for sand replacement systems and into reacti-
vation and contactor costs for the post-filter adsorbers. Table
4 contains amortized capital and operations and maintenance
costs for sand replacement and post-filter adsorption systems.
As can be seen, a large portion of the capital cost for both
processes is associated with reactivation. Post-filter adsorp-
tion is more capital intensive in general than is the sand
replacement mode.
706
-------
Table 1 -
Design and Cost Parameters for Granular Activated Carbon
Design Parameter Level
Sand replacement:
Activated carbon cost $ 0.60/lb ($1.32/kg)
Activated carbon loss per reactivation cycle .... 10%
Fuel cost , $ 1.75/million BTU ($1.66/kJ)
Electric Power Cost $ 0.03/kw hr
Construction cost index 265.A
Producers price index 199.7
Direct hourly wage rate $10/hr
Amortization rate 8%
Amortization period 20 yrs, ,
Volume per filter 856 ft (24 m)
Loss in adsorptive capacity 0%
Design capacity 70%
Empty bed contact time 9 min
Reactivation frequency Every 1.2 months
Post-filter adsorption:
Activated carbon cost $ 0.60/lb (?1.32/kg)
Activated carbon loss per reactivation cycle . . . . 67,
Fuel cost $ 1.75/million BTU ($1.66/kJ)
Electric Power Cost $ 0.03/kw hr
Construction cost index 265.4
Producers price index 199.7
Direct hourly wage rate $10/hr
Amortization rate 8%
Amortization period 20 years
Loss in adsorptive capacity 0%
Design capacity 70%
Empty bed contact time 18 min
Reactivation frequency Every 2.4 months
707
-------
Table 2
Assumptions for Separate Post-Filtration Systems
Design capacity (mga)
(m3/day)
Item
Number of contactors
Diameter of contactors (ft)
(n>)
Depth of contactors (ft)
(m)
Volume of GAC per
contactor, (ft )
(n )
Minimus empty bed contact
time (iiiin)
1
(4,000)
3
8
(2. A)
13
(4.0)
653.1
(18.5)
18
5
(20,000)
6
12
(3.7)
13
(4.0)
1469.5
(41.6)
18
10
(40,000)
12
12
(3.7)
13
(4.0)
1469.5
(41.6)
18
10
(400,000)
40
20
(6.1)
14
(4.3)
4396.0
(124.4)
18
150
(600,000)
60
20
(6.1)
14
(4.3)
4396.0
(124.4)
18
Table 3.
AMORTIZED CAPITAL AND OPERATING AND MAINTENANCE
COST FOR GAC SYSTEMS
Cost (*!/!, 000 gal)*
Design Capacity (mgd)
(roVday)
Item
Amoritized Capital
cost:
Sand replacement
Filter adsorber
Operating & Maint.
Cost:
Sand replacement
Filter adsorber
Total Cost:
Sand replacement
Filter adsorber
1
(4,000)
24.8
29.4
19.5
22.1
44.3
51.5
5
(20,000)
9.0
12.7
13.0
12.6
22.0
25.3
10
(40,000)
6.2
9.8
11.7
10.8
17.9
20.6
100 150
(400,000) (600,000)
2.0 1.7
4.9 4.6
8.7 8.4
7.4 7.1
10.7 10.1
12.3 11.7
*To convert from £/1000 gal to
-------
Table 4
Amortized Capital and O&M Costs for GAC Systems (£/l,000 gal)*
o
VO
i
(4,000)
(20,000)
Plant Capaciy (mgd)
(m3/day)
10
(40,000)
100
(400,000)
150
(600,000)
Capital O&M Captlal O&M Capital O&M Capital O&M Capital O&M
Sand Replacement Reactivation 24.0 27.6 8.3 13.0
Activated
carbon fill
0.8
0.0
0.7 .0
5.4 11.7
0.7 0.0
1.4 8.7
0.6 0.0
1.1 8.4
0.6 0.0
Filter-Adsorbers Reactivation 24.0
Contactor 3.7
Activated
carbon fill
1.8
17.1
5.1
0.0
8.4 10.0
2.9 2.9
1.5 0.0
To convert from c/1,000 gal to c/m multiply by 0.26
5.5 9.3 1.4 6.6 1.1 6.3
2.9 1.5 2.4 0.8 2.4 0.8
1.4 0.0
1.2 0.0
1.2 0.0
-------
SENSITIVITY ANALYSIS
In this section, the effect of changing variable levels
will be considered. Standard values shown in Table 1 and the
resultant unit costs shown in Tables 3 and 4 will be -assumed.
The effect of changing the design parameters around the standard
values will be examined for GAC sand replacement and post-filter
adsorbers separately (5). A comparison will be made between 100
mgd (400,000 m /day) sand replacement and filter-adsorption
systems. Mathematical equations describing these systems will
be presented.
Figure 1 shows the cost of both sand replacement and post-
filter adsorption systems versus plant capacity,* based on the
assumptions in Tables 1 and 2. Figures 2 and 3 show the O&M and
amortized capital costs for both system types respectively.
Significant economies of scale exist with respect to plant
capacity. Similar relationships hold with respect to loading
rate as well. Unit costs will rise dramatically for a plant of
given capacity as loading rate decreases. Figure 4 shows the
effect of reactivation frequencies on a total production cost
for both post-filter adsorption and sand replacement systems.
As can be seen, cost rises dramatically with increased reactiva-
tion.
O&M Cost Effects
Some operating variables influence only operating and
maintenance cost, some only amortized capital cost, and some
variables affect both O&M and capital cost. The first set of
variables to be examined are those that influence O&M cost
only. These variables are: hourly wage rate in $/hr; activated
carbon loss per reactivation cycle in percent; fuel cost in
$/therrn; producers price index; and electrical power cost in
$/kWh. This analysis is performed for 1, 5, 10, 100, and 150
mgd (4,000, 20,000, 40,000, 400,000 and 600,000 nr) systems
operating at 70 percent of capacity.
Figure 5 shows that changes in hourly wage rate have a
greater impact on O&M cost in small plants than in larger
plants. For example, it can be seen that as the hourly wage
rate increases from $6/hr to $10/hr, the O&M cost for a 1 mgd
*Note: The proposed regulation (Part B) was to apply to
utilities serving 75,000 people or more [approximately 10 mgd
(40,000 mj/d)] (1).
710
-------
35
=•30
ra
CO
o
i- 25
\
O
020
c
o
I 15
•o
o
2 10
o
5--
Sand Replacement
20 40 60 80 100
Plant Capacity in mgd
120
140
160
Figure 1. Total Production Cost Versus Plant Capacity for
Post-Filter Adsorption and Sand Replacement Systems
T Post-Filter Adsorption
Sand Replacement
40
60
80
100
120 140
160
Plant Capacity in mgd
Figure 2. Capital Cost Versus Plant Capacity for Post-Filter
Adsorber and Sand Replacement Systems
*To convert from ^f/1,000 gal to jzf/m multiply by 0.26
711
-------
35 T
30--
Figure 3.
Sand Replacement
— — _ , j
Post-Filter Adsorption
20 40 60 80 100 120 140 160
180
Plant Capacity in mgd
0 & M Cost Versus Plant Capacity for Post-Filter
and Sand Replacement Systems
35 T
. 30--
ro
a>
O
c
o
3
O
oZ
5--
Figure 4.
12345678
Reactivation Frequency in Months
Total Production Cost Versus Reactivation
Frequency in Months for Post-Filter Adsorption
and Sand Replacement Systems (100 mgd)
*To convert from £/l,000 gal to
multiply by 0.26
712
-------
-J
M
GJ
35-r
30-
03
cn
O
O
O 20
O
O
CJ
O
4 6 8 10 12
Direct Hourly Wage Rate ($/hr)
5 mgd
TO mgd
150 mgd
14
16
Figure 5.
O & M Cost Versus Wage Rate for 1,5,10,100, and
150 mgd Sand Replacement Systems
(To Convert from jzf/1000 gal to
Multiply by 0.26)
-------
plant (4*000 m ) increases from slightly over 17^/1,000 gal
(4.4 £/ur) to slightly less than 19. 5jzf/l, 000 gal (5.1 g/nT).
The same wage rate changes in a 10 mgd plant (40,000 m )
increases the O&M cost from approximately 11£/ 1,000 gal
(2.9 jzf/nT) to 11.6^/1,000 gal (3.0 jzf/m3). Figure 6 shows
the changes in O&M cost resulting from increases in activated
carbon loss per reactivation cycle. Figures 7 through 9 show
changes in O&M resulting from changes and wage rates, fuel cost,
wholesale price index, and electrical power cost.
Capital Cost Effect
The group of variables that influence amortized capital
cost are as follows: construction cost index; amortization
interest rate; amortization period in years. Figure 10 illus-
trates the impact of varying construction cost index on amor-
tized capital cost in £/l,000 gal. Figure 11 shows the effect
of changing interest rates on amortized capital cost. Figure 12
shows the effect of changing 'amortization period. As can be
seen, the impact of changing these parameters is also greater for
small than for large plants.
O&M and Capital Cost Effect
Several of the standard design parameters affect both
amortized capital and O&M cost. These parameters are activated
carbon cost and reactivation frequency. Figures 13 and 14 show
the effect of activated carbon cos.t variations on capital and
O&M cost and Figures 15 and 16 show the effect of changing
reactivation frequency.
Key Variables
Our previous analyses show that the cost of GAC systems
are more sensitive to some variables than to others. Most
variables have a greater impact on smaller than on large sys-
tems. The only exception is the impact of reactivation fre-
quency on capital cost.
O&M Effects —
The variables having the most effect on O&M costs are
system size, reactivation frequency, activated carbon loss, and
activated carbon cost. Small changes in these variables may
have large effects on system cost. Fuel cost and wage rate have
a smaller effect on system cost. System cost is relatively
insensitive to power cost and producers price index. Power cost
has a greater impact on smaller systems than do some of the
other variables.
714
-------
35 T
30--
10
05 25
20 •
o
o
15--
00
O
25
5 75 10.0 12.5 15.0
Carbon Loss Per Reactivation Cycle (%)
17.5
20.0
Figure 6. 0 & M Cost Versus Carbon Loss Per Reactivation Cycle
for 1, 5, 10, 100, and 15Q mgd Sand Replacement
35T
30-
I 254
o>
x 20+
o
W
O
o
15-
10-
5-
150 mgd
.05 1.0 1.5
Fuel Cost in $ per Therm
20
Figure 7. 0 & M Cost Versus Fuel Cost in $/Therm for 1, 5,
10, 100, and 150 mgd Sand Replacement Systems
*To convert from jd/1,000 gal to
multiply by 0.26
715
-------
35T
30
=5 25
o>
5 20-
u)
o
U
S 15
o
10 •
1 mgd
150 mgd
.5
1.0
1.5
2.0
25
3.0
3.5 4.0
4.5
Producers Price Index (x 100)
Figure 8. 0 & M Cost Versus Producers Price Index for 1, 5,
10, 100, and 150 mgd Sand Replacement Systems
35T
30
S 25--
20
o
fj
00
O
15-
10-
5-
1 mgd
150 mgd
.01
.03
.04 .05
.06 .07
.08
Power Cost in $/kw hr
Figure 9. 0 & M Cost Versus Power Cost for I, 5, 10,
100, and 150 mgd Sand Replacement Systems
*To convert from jzi/1,000 gal to £/m3 multiply by 0.26
716
-------
1.0
1.S 2.0 2.5 3.0 3.5 4.0 4.5
Construction Cost Index (x 100)
Figure 10. Capital Cost Versus Construction Cost Index for 1,
5, 10, 100, and 150 mgd Sand Replacement Systems
35T
6 8 . 10
Interest Rate (%)
12 14
16
Figure 11. Capital Cost Versus Interest Rate for 1, 5,
10, 100, and 150 mgd Sand Replacement Systems
*To convert from ^f/1,000 gal to jt/m multiply by 0.26
717
-------
35T
00
Amortization Period (yrs)
Figure 12. Capital Cost Versus Amortization for 1,5,10,100,
and 150 mgd Sand Replacement Systems
*(To Convert from £/1000 gal to
Multiply by 0.26)
-------
35
30 •
a. 25
20
15-
10
100 mgd
2 3
.4
.5
150 mgd
.8 .9
Carbon Cost (S/lb)
Figure 13. Capital Cost Versus Carbon Cost for 1, 5, 10,
100, and 150 mgd Sand Replacement Systems
o
»
35
30
25
20
10
5-
3 4 .5 .6
Carbon Cost in t /Ib
1 mgd
.8 9
Figure 14. 0 & M Cost Versus Carbon Cost for 1, 5, 10,
100, and 150 mgd Sand Replacement Systems
*To convert from £/l,000 gal to jzJ/m multiply by 0.26
719
-------
2 3 4 5 6
Regeneration Frequency (months)
1 mod
100 mgd
ijISO mgd
8
Figure 15. Capital Cost Versus Regeneration Frequency for 1,
5, 10, 100, and 150 mgd Sand Replacement Systems
1 mgd
S mgd
10 mgd
•100 mgd
150 mgd
34567
Reactivation Frequency (months)
Figure 16. 0 & M Cost Versus Reactivation Frequency for 1,
5, 10, 100, and 150 mgd Sand Replacement Systems
*To convert from jzi/1,000 gal to £/m3 multiply by 0.26
720
-------
Capital Cost Effects—
Plant capacity, amortization period, construction cost
index, and interest rate have a significant impact on system
cost. System costs are relatively insensitive to activated
carbon cost. Capital costs are moderately sensitive to reacti-
vation frequency.
Cost Equations
From the previous analysis, some variables have a greater
impact on cost than do others. For example, both O&M and
capital costs are extremely sensitive to plant capacity and
loading rate. Reactivation frequency also has a significant
impact on cost.
In the previous sections, the effect of changing individual
parameters was studied. Some of these variables may change
simultaneously, however, complicating the relative impact
of each variable on total cost. Individual regression equations
have therefore been developed relating key design variables to
annual cost. For capital cost, these calculations have been
made in terms of dollars per year at 8 percent over a 20-year
amortization period. Conversion to another base, such as
£/l,000 gal, or to a different annualized cost, is a simple
calculation.
These equations were derived by performing a regression
analysis on key design variables (3). The variables were
randomized and several hundred points chosen in order to arrive
at each equation. The O&M and Capital Cost equations for steel
downflow gravity contactors and multiple hearth reactivation
furnaces follow:
Contactor O&M Cost— .. ..
CVQ = 92.2 TVOLU-bbPRU'*UDHRU'**PPIU'±:> (r^ = 0.98) (1)
where
CV = annual operating and maintenance cost for steel
downflow gravity contactor vessels in $/yr
TVOL = total contactor volume in ft
PR * power cost in $/kW hr
DHR = direct hourly wage rate in $/hr
PPI = Producers Price Index divided by 100
721
-------
Contactor Capital
where
CV =9.6 CNTNM'"VOL1U'WJ>CCIW> " (r = 0.99) (2)
CV = annual capital cost for steel downflow gravity
contactor vessels in $/yr
CNTN = the number of individual contactors
VOL1 = the effective volume in ft of a single contactor
CCI = Engineering News Record Construction Cost Index
divided by 100
Reactivation O&M Cost—n Qfi n ,- n ,_ 0
RE = 1008.1 RRTEU*0bCRBCU*0/CRBLU*o::> (rZ = 0.99) (3)
where
RE = annual reactivation operating and maintenance
cost in $/yr for a multi-hearth furnace
RRTE = reactivation rate in Ibs/day
CRBC = carbon cost in $/lb
CRBL = carbon loss in percent per reactivation cycle
Reactivation Capital Cost--
RE * 1991.7 RRTE0'JaCCI°'99 (r2 = 0.99) (4)
c
where
RE = annual reactivation capital cost in $/yr for
c a multi-hearth furnace
RRTE = reactivation rate in Ibs/day
CCI = Engineering New Record Construction Cost Index
divided by 100
Initial Carbon Fill Cost—
The capital cost for the initial carbon fill is given by
the following:
CF = (TVOL)(UN)(AM) (5)
where
CF = annual capital cost for carbon fill in $/yr
722
-------
TVOL = total initial volume of carbon in Ibs
UN = unit cost in $/lb
AM = capital recovery factor
Design Implications
This analysis can provide some guidance for design. One
can consider the most sensitive variables when making both
design and operating decisions. Examples of how these data
might be used are provided in the following sections.
REGIONAL REACTIVATION
The previous analysis shows that unit costs associated
with smaller capacity plants are extremely high (Figure 1). A
significant portion of the amortized capital costs are for on-
site reactivation facilities, which suggests the possibility
that for small plants some alternative to on-site reactivation
such as regional reactivation should be explored. To illustrate
the economic trade-offs considered with a regional reactivation
site, the following case study is hypothesized.
Case Study
The case study consists for four water utilities of
various sizes. Three options are considered for each site as
follows:
1. establish a granular carbon reactivation system at
each utility;
2. establish a centralized or regional granular carbon
reactivation system;
3. discard the spent activated carbon and replace it
with virgin GAG.
Costs associated with the three options will be calculated
on the basis of $/yr. Activated carbon requirements for the
various water treatment plants are in tons/yr and furnace sizing
is based on Ib/day of granular carbon reactivated. Table 5
contains the utility identification number, average water flow,
and demand for activated carbon in tons/yr.
Three cost components are normally considered in a regional
system. These components are the cost of on-site storage, the
cost of transportation, and the cost of reactivation. The cost
of on-site storage is given below.
723
-------
Table 5
Design Assumptions for Case Study
Utility No.
1
2
3
4
Average Flow
mgd (m /d)
40
5
100
20
(16,000)
(20,000)
(400,000)
(80,000)
Reactivation Demand
tons/yr (t/yr)
3114
389
7785
1557
(2,825)
(353)
(7,062)
(1,412)
TABLE 6
Variable Levels for Equation 8
Design Parameter
TSVOL
CCI
TCRB
DI
DHR
PPI
RRTE
CRBC
CRBL
Z
Level
34,126 ft3
265.4
•a
207,600 ftJ
$0.65/gal
$10/hr
199.7
70,384 Ib/day
$0.60/lb
6%
0.24
(966 m3)
•3
(5,875 nT)
($0.17/L)
(31,919 kg/day)
($1.32/kg)
724
-------
OSC = 1.35 TSVOL°'98CCI°-98 (r2 = 0.99) (6)
O
where,
OS = Annual storage cost in $/yr
O
TSVOL = Total storage volume in ft
CCI = Engineering News record construction cost index
divided by 100'
Transportation cost is given by the following equation:
TRC = O.Q06 TCRB0'99MIL0-75DI0-57DHR°-33PPI0-082
(r^ = 0.99) (7)
where,
TRC = Transportation cost in $/yr
TCRB = Total Volume of carbon transported in ft /yr
DI = Cost for diesel fuel in $/gal
MIL = Distance to reactivation site in miles
DHR = Wage rate in $/hr
PPI = Producers price index divided by 100
Reactivation costs are given by equations 3 and 4.
Combining equations 3 and 4 with equations 6 and 7 yields
an equation for total cost to a utility as follows:
REGC = 1.35 TSVOL0*98CCI0*98 (8)
+ 0.006 TCRB0-99MIL0-75DI0-57DHR°-33PPI0-082
+ Z(1008.1 RRTE°'86CRBC°'67CRBL°*65
+ 1991.7 RRTE°'38CCI°'99)
where,
REGC = the cost to a given utility in $/yr for regional
reactivation
Z = the portion of the reactivation cost charged to the
utility
725
-------
Given the levels assigned to the variables in equation 8
(Table 6), the costs associated with off-site regeneration as
a function of distance are shown by equation 9.
REGC = 583,000 + 1,950 MIL0*75 (9)
Equation 9 was plotted as shown in Figure 17 and compared
against on-site reactivation and disposal of spent activated
carbon. Off-site reactivation can be a cost-effective solution
particularly for small utilities. As the utility becomes larger
the advantages of off-site reactivation diminish.
DESIGN IMPLICATIONS
The sensitivity analysis performed for operating variables
has some important design implications. To illustrate these
implications/ the following examples will be considered: trade-
off between contactor volume and number of contactors, activated
carbon quality and cost, contact time and activated carbon use
rate.
Contactors—Volume and Number
The annual capital cost for contactors is represented by
equation 2, as follows:
CV =9.6 CNTN0-99VOL10'81CCI°'99 (2)
The partial of equation 2 with respect to CNTN is:
3CV
3 CNTN
-0.01x7nT10.81rr,,0.99 (10)
9.50 CNTN "•UJ-VOL1U*°-LCCI
The partial of equation 2 with respect to VOL1 is
» 7.78 CNTN0'99VOL1~0'19CCI°'99
If a constant total volume is to be maintained a minimum
cost solution will result from fewer contactors of larger
volume. The cost for additional increments of volume is lower
than for increased number of contactors (equations 10 and 11).
To illustrate this effect, assume two design configurations for
contactors and volume as shown in Table 7. Figure 18 is a plot
of equation 2 assuming the values in Table 7. As can be seen,
configuration 2 (fewer contactors, larger volumes) is always
less expensive even though the total volume of activated carbon
is the same for both design configurations.
726
-------
15.0
On-Site Reactivation Cost
5^SUe~Reactivation Cost
To 20 30 40 50 60 70 80 90
Miles to Reactivation Site
*To convert from */1.000 gal to ^/m3 multiply by 0.26
10J
Figure 18,
10* 10s
o
Total Volume in ft.
Comparison Between Design Configurations
727
-------
Table 7
Assumed Design Configurations
Item
Number of
Contactors
Volume per ,
Contactor in ft
(m3)
Design Configuration
1 2
6 3
1500 3000
(457) (914)
1
12
1500
(457)
2 1
6 40
3000 4400
(914) (1341)
212
20 60 30
8800 4400 8800
(2682) (1341) (2682)
728
-------
Carbon Cost and Quality
The previous analysis shows that carbon cost/ carbon
loss, and reactivation frequency have a significant impact on
cost. A manager interested in minimizing total system cost
might choose to purchase the minimum cost activated carbon.
Figure 19 shows total system cost versus reactivation frequency
for a 100 mgd (400,000 m ) post-filter adsorption system for
three different carbon costs. Assuming 60^/lb carbon for the
system being reactivated at a frequency of 1.1. months the total
system cost is approximately 16^/1,000 gal (P-,) (4.2 0f/m ). If
carbon cost is increased from 60^/lb (132.3 £/kg) to 80£/lb
(176.4£/kg) , the reactivation period would have to increase
from 1.1 to 1.7 months (P2)r for costs to remain constant. If
the reactivation period increased beyond 1.7 months as a result
of the more expensive carbon, system cost would be lowered.
Equation 3 can be used to study the trade-off between reactivation
frequency, carbon cost, and carbon loss per cycle. The annual
reactivation O&M cost equation is as follows:
RE = 1008.1 RRTE0-86 CRBC0'67CRBL°•65 (3)
Taking the natural log of both sides of equation 3 yields:
In RE « In 1008.1 + 0.86 In RRTE + 0.67 In CRBC (12)
0 +0.65 In CRBL
To illustrate how CRBC and RRTE vary assuming cost remains
constant, one can take the partial of equation 12 as follows:
•
ln REo _ n a* 9ln RRTE 4. n fi7 . o fic 8ln CRBL ..
din CRBC ~ °'86 Sin CRBC * °*67 * °'65 31n CRBC (13)
If one assumes that change in cost is zero, and activated
carbon loss is constant, then equation 13 can be set equal
to zero and the changes in reactivation frequency vs. carbon
cost studied. This relationship would show the elasticity
of substitution between CRBC and RRTE. These elasticities
are summarized in Table 8.
Table 8
Elasticities for Equation 12 (Vy/Xx)
In RRTE
In CRBC
In CRBL
In RRTE
-
- 1.28
- 1.32
In CRBC
- 0.78
-
- 1.03
In CRBL
- 0.76
- 0.97
-
729
-------
to
o>
35
30
25
r 20
-------
Table 8 shows the changes in independent variables required
to keep costs constant. For example/ if the cost of activated
carbon increased by 10 percent, the reactivation rate (Ibs/day)
has to decrease on the average by 12.8 percent or/ correspond-
ingly/ the period between reactivations has to increase by 12.8
percent. This value will/ of course/ vary depending on the
initial point on the curve in Figure 19. If one wishes to use
a more expensive carbon, the decrease in reactivation frequency
and/or activated carbon loss must be sufficiently great to off-
set the increased cost.
Contact Time and Use Ratio
In design of GAC systems the number of contactors should
be minimized and the volume per contactor maximized. One might
also conclude that/ the greater the total volume of carbon or
Empty ftad Contact time (EBCT) the more efficient the organic
removal process and the lower the unit cost.
Figure 20 shows the relationship between system size, cost/
and empty bed contact time for a 100 mgd (400/000 m ) post
filter adsorption system. From Figure 20/ if the required
EBCT of 18 minutes at a reactivation frequency of 2.4 months
were increased to 22 minutes, the reactivation frequency would
have to be increased to 3.4 months (P2) or greater to achieve
a favorable economic trade-off. Therefore, to maintain the same
cost if the EBCT were increased by 22 percent, the period
between reactivations would have to increase by at least 42
percent.
This result appears to be "counter-intuitive." Conven-
tional wisdom leads one to believe that deeper beds are more
efficient in their removal of organic material (6). If, how-
ever, the use ratio of activated carbon is constant, that is the
"Ibs of carbon per million gallons of water loaded on the bed"
remains constant with increasing bed depth/ then a taallower
bed is more cost effective. The excess activated carbon in the
deeper bed is only so much excess inventory, not contributing
to system performance until the end of a cycle but constantly
costing more.
To properly understand this nonproportionality relation-
ship, one should carefully consider the adsorbent use rate vs.
EBCT. Data from pilot columns and field scale studies can be
used to calculate the use ratio of activated carbon per unit
volume to meet a controlling criterion for any empty bed contact
time. Adsorbent use ratio is calculated by dividing the dry
weight of adsorbent for a given empty bed contact time by
the total volume of water passing through the column until
a performance criterion is exceeded.
731
-------
4 6 8 10 12
Reactivation Frequency in Months
Effect of changing EBCT on Production Cost
•3
Figure 20
*To convert from £/l,000 gal to
-
multiply by 0.26
1.400T
1.200- •
1.000-•
600
400--
200--
10 15 20 26 30 35
Empty Bed Contact Time (min)
40
Figure 21.
Ufie Ratio Versus Contact Time for
Removal of Contaminants
732
-------
If the adsorbent use ratio decreases with increasing empty
bed contact time, then a more than proportional improvement in
performance is gained by increasing the EBCT. Figure 21 shows
some typical use ratios plotted vs. EBCT as obtained from field
studies (7). The decrease in use ratio with increasing EBCT
varies considerably among contaminants to be removed. At some
point, however, the use ratio becomes constant, implying a
point at which increasing the empty bed contact time no longer
increases efficiency of removal.
One can calculate use ratio per million gallons, for
example, by assuming an activated carbon use ratio as described
above and converting to Ibs/million gal. Assume a use ratio of
530 mg/1. The use ratio is converted as follows:
Use ratio » (530 mg/1) x (8.34 Ibs/gal)
= 4,430 Ibs/mil gal
An equation relating EBCT and use ratio has been developed as
follows:
TC = 73184.6 CT°-72URTE°-79 (r2 = 0.99) (14)
where-,
TC = total annual cost for GAC in $/yr for post-filter
adsorbers
C_ * empty bed contact time in minutes
URTE = carbon use ratio in Ibs/mil gal
Table 9 shows the relative elasticities between use ratio
and contact time.
Table 9
Elasticities for Equation 14
3Y - ^x « In CT In URTE
In CT - 1.10
In URTE - 0.91
733
-------
According to Table 9, for cost to remain constant if
contact time increases by 10 percent, use ratio must decrease
by 11 percent. From Figure 21, however/ at some point use
ratio reaches a constant.
Using equation 14 it is possible to find the optimal empty
bed contact time as a function of the relationship between
activated carbon use rate and contact time. For example, assume
the following relationship between use ratio and contact time:
URTE = F(CT) (15)
Taking the partial of equation 14 with respect to C_ yields:
= 52693 C-°«28 URTE0'79 + 57816 C'72 URTE'0'21 «E (16)
Setting equation 16 equal to zero results in the following
relationship:
CT = -0.91 (URTE)
aURTE(17)
Substituting equation 15 into equation 17 yields:
CT = 0.91 F(CT)
3F(C™)(18)
6 CT
The solution to equation 18 (if there is one) will result in the
minimum cost empty bed contact time. If no solution exists to
equation 18, then the situation discussed earlier of the con-
stant carbon use ratio exists and the shallowest possible bed
will result in minimum cost.
OTHER ECONOMIC IMPACTS
There are other economic impacts that are related to the
influence of operating variables on the cost of GAC. These
impacts include general inflation, increased energy costs,
etc.
INFLATION IMPACT
As shown in Table 1 for static economic conditions, sand
replacement systems are slightly less expensive than are post-
filter adsorption systems. Because post-filter adsorption is
734
-------
less labor-intensive than sand replacement, inflation may make
it less expensive some time in the future.. To illustrate this
effect, the cost for a 100 mgd (400,000 m ) sand replacement
system has been assumed as 10.10!/1,000 gal (2.77 jzi/m ).
A filter adsorber system of 100 mgd has been assumed as
11.7 £/l,000 gal (3.1 £/nr) at year 0. Both systems have
been allowed to increase at inflationary rates of 8, 11, and 15
percent over 20 years. The results of assuming the impact of
11 percent inflation on the two types of systems is shown in
Figure 22. The figure shows that the sand replacement system
becomes more expensive than the equivalent post-filter adsorp-
tion system after approximately 10 years. Total increase in
cost caused by inflation in both systems is significant. This
occurs because the capital expenditures for the systems are
assumed fixed over the life of the investment, while operating
costs were increased according to the assumed inflation rate.
To account for total expenditures, a "present value"
analysis was made for the systems listed. Four discount rates
(4, 6, 8, and 10 percent) and three inflation rates (8, 11, 15
percent) were used in the analysis. The results are summarized
in Table 10. For the larger plants at the 11 and 15 percent
inflation rates, for all discount rates, the post-filter
adsorber has a lower present value, indicating that the post-
filter adsorber is the least costly system.
ENERGY COSTS
From the earlier analysis, GAG operating costs are not
highly sensitive to small changes in energy cost. Because the
energy picture is so chaotic, however, energy use should be
examined more intensively.
Tables 11 and 12 show the costs of energy use per Ib of
activated carbon, based on the assumptions in Table 1 for post-
filter adsorption. Fuel and power costs range from 10 percent
to 30 percent of total cost for reactivation. Energy cost is
more nearly constant for contactors. The energy ratios given
in both tables can be used to estimate the cost impacts result-
ing from drastic increases in energy cost.
COMBINATIONS OF UNIT PROCESSES
As all of the previous analysis emphasizes, the reactiva-
tion frequency has an important impact on the cost of GAC
operation. GAC in combination with another unit process that
helps lower reactivation frequency might result in a cheaper
system. Figure 23 shows a hypothetical analysis in which ozone
is assumed in combination with post-filter adsorption. If the
system is initially at point PI (2 months) without ozone, then
735
-------
X
o
70
60
50-
U)
O
U
C 40
O
u
S 30
<0
o
20
10 15
Years After Construction
20
Figure 22.
20._
Effect of 11% Inflation on 100 mgd Sand Replace-
ment and Post-Filter Adsorption Systems
S,
15
10
o
^
o
•o
o
5
2 4 6 8 10 12 14
Reactivation Frequency, (months)
Figure 23. Cost of Ozone and GAC in Combination
•(To Convert fron C/1000 gal to C/m3. Multiply by 0.26)
736
-------
Table 10
Present Value Analysis
Inflation
rate (i)
8
11
15
Discount Rate
(%)
.04
.06
.08
.10
.04
.06
.08
.10
.04
.06
.08
.10
H)_ingd
Post-Filter Adsorption
482.84
397.67
333.06
283.38
613.24
496.41
408,78
342.18
885.16
699.94
562.84
460.37
Present
Sand Replacement
456.75
374.18
311.74
263.88
597.46
480.74
393.45
327.32
890.91
700.39
559.84
454.88
Values
100
Post Filter Adsorption
305.96
251.18
209.71
177.88
395.59
319.06
261.76
218.3
582.52
458.97
367.75
299.55
mgd
Sand Replacement
302.99
246.85
204.52
172.17
408.19
326.60
265.61
219.61
627.65
490.78
390.05
315.01
-------
Table 11
Energy Impact - Contactors*
Reactivation Rate
(Ibs/day)
816.4
3,673.8
7,347.5
73,266.7
109,900.0
System Cost
(5/lb)
0.091
0.062
0.056
0.042
0.042
Energy Cost
($/lb)
0.0045
0.0044
0.0043
0.0040
0.0039
Energy Ratio
(kw hr/lb)
0.15
0.15
0.14
0.13
0.13
Table 12
Energy Impact - Reactivation
Reactivation
Rate
(Ibs/day)
816.4
3,673.8
7,347.5
73,266.7
109,900.0
System
Cost
(S/lb)
0.352
0.180
0.141
0.076
0.070
Energy
Power
0.028
0.011
0.007
0.002
0.001
Cost
Fuel
0.034
0.026
0.025
0.019
0.018
Energy
Power
(kw hr/lb)
0.92
0.36
0.23
0.06
0.05
Ratio
Fuel ,
(BTU x 10 /lb)
0.19
0.15
0.14
0.11
0.10
To convert from Ibs/day to kg/day multiply by 0.4535. To convert from
$/lb to $/kg multiply by 2.21. To convert from kw hr/lb to kw hr/kg
multiply by 2.21. To convert from BTU x 10 /lb to J/kg multiply by 2326.4.
738
-------
the addition of ozone would have to increase the reactivation
frequency to ?2 (4 months) to break even at an ozone dose of 8 mg/1.
If, however, the system without ozone had a reativation frequency of
6 months, no breakeven point exists to justify the addition of more
than 4 mg/1 ozone.
COST IMPACT ON INDIVIDUAL UTILITIES
The Safe Drinking Water Act, as conceived, intended that
the direct costs of water treatment improvements be passed on
to the consumer. This section will discuss some of the hypo-
thetical impacts of GAC systems on the consumer. Cost imoact
analysis for 10 and 100 mgd systems (40 000 and Ann nnn »§/£,
shown in Tables 13 and 14. «uu,UOO m /d) are
Table 13 summarizes the costs that might be incurred when
GAC is added to an existing system. Water bill increases are
assumed based on a family of four consuming 117,000 gal of water
per year. From the table, the costs of adding GAC to small
systems are significant, but become more reasonable at larger
plant sizes.
Table 14 shows the principal and total project cost
(principal plus interest) required for both types of GAC systems
at the 10 and 100 mgd levels.
SUMMARY AND CONCLUSIONS
In this paper, the sensitivity of GAC system costs to
changes in certain operating variables has been examined.
Several variables have been identified as important. These
include choice of system configuration, loading rate and size
of system, reactivation frequency, interest rate and life% of
system, local construction and operating costs, inflation, and
carbon use rate. With the help of the equations and curves
presented in this paper, these sensitivities can be studied.
For a given EBCT, fewer contactors of greater volume
yield minimum cost. Based on activated carbon use rate, how-
ever, bigger is not better. If the activated carbon use rate
decreases with EBCT, then larger volumes or greater EBCT's yield
lower cost. When activated carbon use rate is held constant,
however, deeper beds result in higher cost. The relationship
between activated carbon use rate, activated carbon loss, and
reactivation frequency must be studied carefully. Cheaper
activated carbon could mean higher activated carbon losses and
more frequent reactivation. Regional reactivation may result
in lower costs particularly for small utilities. Inflationary
effects may cause labor intensive sand filtration systems to be
ultimately more expensive than capital intensive post-filter
adsorption systems.
739
-------
Table 13
Granular Activated Carbon Cost**
Treatnent
.il tern itive
Sand
replacement
Post-filter
adsorber
Unit Cost
for GAC
(cA.OOO gal)
44.3
51.5
1-HCD plant
*
Current
Cost Z Increase
(c/1.000 gal) with GAC
121 37
121 43
Unit cost
tor fiAC
(c/1.000 gal)
10.1
12.3
100-MGD plan
Current
Cost
(c/1.000 gal)
59.0
59.0
t
X Increase
with GAC
17
21
Inflated at 62 per year froz 1974 to 1979 from figures shown in reference 9.
To convert from c/1.000 gal to c/ra multiply by 0.26.
Table 14
Capital Costs for GAC
System
1-MGD Plant
Principal Total project cost
(principal & interest)
($) ($)
100-MCD Plant
Principal Total project cost
(principal & interest)
($) ($)
Sanij
replacement 640,000
Post-filter
adsorber 737,000
1,270,000
1,500,000
4,900,000
12,600,00
8,700,000
24,000,000
-------
This analysis emphasizes that although GAG systems are
relatively complex, they can be designed to operate efficiently.
The key to efficient operation is careful study and common sense
which lead to an intelligent understanding of system design and
operation.
REFERENCES
1. Federal Register. 1978. Interim Primary Drinking Water
Standards—Control of Chemical Contaminants in Drinking
Water, Vol. 43, No. 28, Thursday, February 9, 1978,
p. 5756.
2. Journal of the American Water Works Association. 1978.
"News of the Field - Update," 70, (4).
3. Gumerman, Robert C., Russell L. Gulp, and Siguard P. Hansen.
Estimating Cost for Water Treatment as a Function of Size
and Treatment Efficiency; An Interim Report, Municipal
Environmental Research Laboratory, Office of Research and
Development, U.S. Environmental Protection Agency, Cin-
cinnati, Ohio 45268. EPA-600-12-78-182, August 1978.
4. Swindell - Dressier, "Process Design Manual for Carbon
Adsorption," U.S. Environmental Protection Agency,
Technology Transfer, October 1973.
5. Clark, R.M., D.L. Guttman, J.L. Crawford, and J.A.
Machisko. 1977. The Cost of Removing Chloroform and Other
Trihalomethanes from Drinking Water Supplies, U.S. Environ-
mental Protection Agency, Cincinnati, Ohio 45268.
6. Zogorski, John S., and Samuel D. Faust. 1978. "Operational
Parameters for Optimum Removal of Phenolic Compounds from
Polluted Waters by Columns of Activated Carbon," in Carbon
Adsorption Handbook, edited by Paul N. Cheremisinoff and
Fred Ellerbusch, Ann Arbor Science Publishers, Inc., Ann
Arbor, Mich., pp. 753-777.
7. Symons, James M. and Robert M. Clark. Interim Guide for
Controlling Organic Contaminants in Drinking Water Using
Granular Activated Carbon, Water Supply Research Division,
Municipal Environmental Research Laboratory, Office of
Research and Development, U.S. Environmental Protection
Agency, Cincinnati, Ohio 45268, January 1978.
8. Miller, G. Wade and Rip G. Rice. 1978. "European Water
Treatment Practices Promise of Biological Activated
Carbon," Civil Engineering - ASCE, 48, (2) 81-83.
741
-------
9. Clark, R.M., James I. Gillean, and W. Kyle Adams. 1977.
The Cost of Water Supply and Water Utility Management,
Vol. I, Water Supply Research Division, Municipal Environ
mental Research Laboratory, U.S. EPA, Cincinnati, Ohio
45268, EPA-600/5-77-015a, November 1977.
742
-------
COST OF GRANULAR ACTIVATED CARBON
TREATMENT IN FRANCE
F. Fiessinger
INTRODUCTION
To present cost data to an international audience is al-
ways difficult because costs depend upon the economic conditions
particular to each country, as well as the accounting methods
of the companies which must bear the costs. This is certainly
true for amortization, and in this study we have, therefore,
adopted a method that is relatively simple, if somewhat schemat-
ic in nature. We are particularly interested in capital and di-
rect operating costs (reactivation costs, labor, power, etc.),
and it is really their relative significance which we must
establish: an order of importance based upon our finding in
various countries and a comparison with the costs of other
treatments. In the present study we shall examine the incre-
mental costs related to the use of granular activated carbon
(GAC) in addition to an existing treatment process.
All costs are given in U.S. dollars; the rate of exchange
used is 1 U.S. $ = 4.50 FF.
COST OF GAC TREATMENT
Table 1 gives the characteristics of six plants in differ-
ent regions of France which represent a significant, if not
complete, range of GAC treatment in this country. We may make
the following general observations:
• GAC is still used mainly to eliminate tastes and odors,
and the life time of the carbon is generally longer than
1 year.
• The carbon used is generally of high quality and may be
thermally reactivated.
• The carbon is reactivated by the manufacturers; plants
for on-site reactivation do not as yet exist in France.
743
-------
NAME OF THE PLANT IVIGNEUX 1 VIRY-CHATILLCN JMORSANG IHOULLE ILAC MODRISCOT |ST. CHARLES
1 I I I
Area of supply (fferis area I Par is area
Hater Resource (Seine river Seine river
First year of GAC 11971 11973
operation
Max imum production rate 5b 1100
Io3m3/day^
Total 1978 production IH460 117400
103 m3 | |
Average 1979 daily |33 ISO
production rate I
103 m3/day
Paris area IDunkerque
1 (Ground water
1 recharge)
1
Seine river (Houlle river
1
1975 11973
(
1
75 125
1
(
9560 14800
1
1
27 125
1
1
Biarritz
Houriscot laki
1974
8
580
17
(Nancy
fbselle river
11975
100
125500
72
1 IB.P. Prechlori- |B.P. Prechlori- IB. P. Prechlori- IB. P. Pcechlori- IP.B. Prechlorl- |B.P. Prechlori-
| nation (nation (nation (nation (nation (nation
(Coagulation-Floe- 1 Coagulation-Floe- (Coagulation-Floe- (Coagulation-Floe- (Coagulation- ICoagulation-
Iculation (culation (culation Iculation U'locculation | Flocculation
Treatment Process IFloc Blanket (Floe Blanket (Floe Blanket (Floe Blanket (Floe Blanket (Sedimentation
(clarification Iclarif iciation (clarification (clarification (Clarification (GAC Filtration
|«C Filtration |«C Filtration (Sand Filtration (GAC Filtration IGAC Filtration (Ozonation
IC1O2 disinfec- |CK>2 disinfec- (Ozonation 1 |Cl2 disinfection
It ion It ion IOC Filtration 1 1
|C12 disinfection 1 1
1 III
110 filters : US filters :
(of 27m2 x 0.85n> |9f. of 50ra2 x
Contactors 1 0.8m
(Total : 250 m3 |6f. of 36m2 x 0.9m
(of GAC llbtal : 560 m3
(of GAC
(Contact time at
average production 1 10 16
Irate minutes
(Chemviron F 300 Chemviron F 300
(Chemviron F 500 Norit Row 0, 8 S
Type of GAC (Uirgi BS 12
(torit Row 0,8 S
IDuty of GAC (Taste Removal Taste Removal
1
(Possibility of addi- |x
Itional PAC treatment
ILife time of each (1.0 2.5
If ilter year
1
CAC treatment cost |0.79 6.56
1978 estimate - ct/ta3
1 1
4 filters : |3 filters : |2 filters : (22 filters of
60m2 x 1.5m 150m2 x 0.8m 133m2 x 0.8m |60ra2 x 0.8m
1 1 . .
Total : 360 m3 (Total : 120 m3 (Total : 54 m-1 ITbtal : 1100 m3
of GAC (of GAC (of GAC lof GAC
1 1
1 1
1 1
19 17 (45 122
1 1
1 1
Chemviron F 400 ICnemviron F 300 (Qiemviron F 300 IQiemviron F 300
iNorit Row 0,8 S I (Pica NCY 103
1 1
1 1
1 1
COD Removal (COD Removal ITfeste Removal (COD Renoval
1 1
x 1 1
1
1 1
4 (1.0 |3 12.6
1 1 1
1.20 10.85 10.60 |0.£5
!
n
nr
Q)
n
pj
o
ft
n>
CO
n-
H-
o
0)
0)
D)
ft
3
pj
ft
en
C
CO
O
>
O
fa
tT
t-1
ft)
-------
• GAG is almost always used in the first stage of fil-
tration. In this stage, carbon is fed with water which
has been merely clarified, whereas in the second stage,
carbon receives clarified water which has also been sand
filtered. Investment is minimal and the cost of treat-
ment is essentially that of reactivation.
• The cost of GAC treatment is quite low * generally less
than $0.01/m —-even for those plants which are relative-
ly small in size.
THE COST COMPONENTS
Let us now proceed from these general considerations to two
specific examples: the plant at Vigneux-sur-Seine which repre-
sents the first GAC filtration treatment in France and the plant
at Morsang-sur-Seine which, with its ozonation treatment preced-
ing a second stage GAC filtration, constitutes what is presently
considered the best treatment chain for surface water. In our
comparison we have included cost estimates for a project to re-
place the first stage filtration (now used at Vigneux) by a
second stage filtration.
CAPITAL COSTS
Initial Carbon Charge
The cost of carbon depends, of course, on its quality. As
we have seen, most of the plants use high quality carbons whose
prices in France are quite simi'lar. Previous studies (1) indi-
cate that non-reactivatable carbon generally gives poorer re-
sults; thus, it has been almost completely abandoned. Table 2
shows the prices charged in the Paris area by the two major
carbon manufacturers.
TABLE 2
Price of Fresh Carbon for Morsang and Vigneux
(Delivered)
Bulk Density kg/m'
Price in US$/kg
1978
1979
CHEMVIRON F 300
CHEMVIRON F 400
NORIT ROW 0,8 S
420
400
360
1.60 1.71
1.74 1.86
1.99* 1.93*
*converted from the manufacturer's quoted prices of $716/m
(1978) and $695/mJ (1979).
745
-------
From this table we may make the following observations:
• Certain carbon manufacturers give prices by unit of
weight although the contactors in water treatment
plants are characterized by carbon volumes. It is much
easier to work with a price per volume of carbon; the
conversion of weight to volume is complicated by the
variation in the apparent density of the carbon due to
the extent of packing.
• Certain manufacturers attempt to widen their market
by keeping their price increases to a minimum - or
even lowering prices, as in the case of Norit at the
present time. The French GAC market is highly com-
petitive. The prices referred to here are those charged
for smallquantities. For larger supplies of several
hundred m , with competitive bidding, the prices would
be considerably lower.
The amount of carbon used depends directly upon the desired
contact time. Our experience indicates that an optimum time is
about 8 minutes for taste removal and 10 to 12 minutes for COD
removal. Below these values, GAC's efficiency decreases sharply;
above them it increases very slowly. Only a much longer contact
time * higher than 30 minutes » would result in appreciably
improved removal.
The problem, however, is how to calculate the contact time
for a water plant which usually operates with a variable flow
rate. At Morsang and Vigneux an 8' minute contact time was
determined from the average flow during the peak month of June.
Table .^indicates the volumes of carbon and the corresponding
costs for these two plants together with estimated figures for
the project at Vigneux.
Contactors
It is preferable to speak of contactors rather than ad-
sorbers, for in most cases, GAC's efficiency is more closely
related to biological than adsorptive phenomena.
If we are considering only incremental costs, the use
of activated carbon in the first stage - as is the case at
Vigneux - does not represent such a cost. The contactors are
composed of filters, required in any case, which use sand
instead of carbon. The adaptations of backwashing sequences,
the possible modifications of the filter, the installation of
carbon transfer devices, even though crucial for the success of
the treatment, represent investments which are negligible in the
overall cost.
746
-------
Table 3
Capital Costs for GAC Treatment at Morsang and Vigneux
FIRST YEAR OF GAC OPERA-'
TION
MAXIMUM PRODUCTION
Rate 10 J m3/hour
VOLUME OF CARBON
IT. 3
MINIMUM COMTACT
Tiir.e - Minutes
! COST OF CARBON
103 S
CHARACTERISTICS OF
CONTACTORS
n->±i ii ^.w-.-iM^ L/irliAij
COST DUE TO CONTACTORS
103 S
Civil work
Equipment
PUMPING CAPITAL COSTS
103 S
TOTAL CAPITAL COSTS
102 U.S. *
MORSANG
II
1975
.3.3
360
6.5
160
u concrete
rect . -gravity
60m2 x 1.5m
closed
stainless
steel equip.
180
220
1UO
700
_ VIGN'E'o'X
1st Stage
existing
1971
2.<*
230
5.8
90
10 concrete
rectangular
27m2 x 0.85m
open
Steel equip.
-
-
90
Vj.oiii.oA
2c Stage
Project
1S79 ?
2.H
U80
12
250
M- concrete
cylindrical
6.2m diameter
31m2 x 3.80m
Stainless steel
equipment
650
510
200
- 1 610
NOTE: The construction costs given here reflect
price levels during the construction period
corresponding approximately to the year in
the first row of the Table.
747
-------
In the case of Morsang, the second stage represents a more
considerable investment. Table 3 indicates its importance.
There we can also see the cost of the project for second-stage
filtration at Vigneux. Prices are stated in 1971 dollars for
Vigneux, 1975 dollars for Morsang and 1979 dollars for the
Vigneux project.
Pumping
Here again, only the second stage of filtration requires
additional pumping. Table 3 also indicates the cost of the
pumping installation for Morsang and the Vigneux project.
Total Capital Cost
Table 3 shows that, in comparison with Morsang, the project
for second stage filtration at Vigneux is relatively expensive.
Since the second stage at Morsang was included in the original
project, its marginal cost was small. The second stage at
Vigneux would have to be added to an already existing plant,
which would result in a considerably greater additional cost.
Because of the cost of materials in France, the use of
concrete rather than steel, is preferred for works of this size.
In order to keep the cost of contactors as low as possible, it
is also advantageous to construct only a few filters, each de-
signed cylindrically with a deep carbon bed and high filtration
velocity even though this requires additional pumping expendi-
tures. On the other hand, for operational considerations,
particularly regeneration, it is preferable to have more
filters. A good compromise appears to be four filters which will
ensure a minimum contact time of 12 minutes at a maximum flow
rate. Although the minimum acceptable contact time could
probably be slightly reduced, the total cost would not change
appreciably.
In summary, the construction of a second stage of filtration
represents an important investment. The crucial question is
whether the investment improves the quality of the water and/or
whether it is compensated by a reduction in the regeneration
cost.
OPERATING COSTS
Amortization
An interest rate of 9 percent per annum has been'used
throughout. This does not include an inflation allowance and
might be considered rather low. As we have mentioned above, the
calculation of amortization varies greatly from one water de-
partment to another, so that the costs of amortization are only
approximate values. In any case, they are of minor importance.
748
-------
The depreciation time was fixed at 20 years for all works,
as an average figure. Amortization affects the contactors,
pumping installations, and carbon, but the carbon amortization
is arbitrary. One may consider that the carbon is renewed with
each reactivation and does not suffer any depreciation. One may
also consider that the carbon is degraded in the course of
successive regenerations and after a certain time must be
entirely renewed. Our experience at Vigneux, where certain
filters have been reactivated up to seven times since they
have been in use, has shown that the carbon remains efficient
through successive regenerations. However, its physical charac-
teristics are degraded and the carbon losses during washing,
through the formation of fine particles, increase steadily.
Thus, we feel that a depreciation time of 20 years represents
a satisfactory compromise.
Table 4 shows the amortization figures for Morsang and
Vigneux. The second stage at Morsang is, of course, quite
costly.
TABLE 4
Operating Cost for GAG Treatment at Morsang and Vigneux
1978 Estimate in US$/100m3
Amortization
(contactors + pumps + carbon)
9% over 20 years
CARBON REACTIVATION
LABOUR
POWER
VIAINTAI NANCE
BACKWASH (Air + Water + Power)
CARBON LOSSES
TOTAL TREATMENT COST US$/100 m3
MORSANG
II
0.73
0.30
0.04
0.04
0.01
0.00
0.08
1.20
VIGNEUX
1st Stage
Existing
0.08
0.60
0.05
-
-
-
0.06
0.79
749
-------
Reactivation
As can be seen in Table 4, the cost of reactivation is
a significant portion of the operating costs. This reactivation
cost depends, as we shall later see, on the quality of the
untreated water and the quality objective for the treated water.
The frequency of reactivation, and therefore its cost is con-
siderably lower for a carbon in second-stage filtration than
a carbon in first-stage filtration. At the present time, car-
bon is reactivated off-site by the manufacturers. The prices
of reactivation for the principal carbons used are indicated
in Table 5. If reactivation were carried out on-site, the
saving would be roughly $0.22/kg of carbon. The total amount
of reactivated carbon in the neighboring plants of Morsang,
Vigneux and Viry-Chatillon is now about 200 tons per year. Our
current study of on-site reactivation indicates possible-cost
reductions for water treatment of approximately $0.002/m at
Vigneux and $0.0009/m at Morsang. It would be even more at-
tractive if the frequency of reactivation were to be increased.
The frequency of reactivation is presently about once a year at
Vigneux, where the quality of the untreated water is rather
poor, and about once every four years at Morsang.
TABLE 5
GAC Reactivation Costs at Morsang and Vigneux
us$Ag
CHEMVIRON F-300
CHEMVIRON F-400
NORIT ROW 0.8-S
1976
0.511
1977
0.613
1978
0.664
0.744
0.711*
1979
0.677
0.744
0.703*
Costs include transportation and "make-up" carbon.
*Based on costs of $256/m3 (1978) and $253/m3 (1979) and
a bulk density of 360 kg/m3.
Labor Costs
Labor costs are essentially the costs of handling the
carbon before and after reactivation, and are therefore almost
directly proportional to the frequency of reactivation. An
eductor is used for the procedure, which is relatively rapid;
the loading and unloading of a filter of 60 m can be per-
formed by three men in 8 hours.
750
-------
At Morsang, additional labor is also needed for the back-
washing of the second stage filters, but this is done only once
every 3 to 4 weeks.
Power
Power is the energy required for pumping the water to the
second stage of filtration. At Morsang there is an elevation of
2qm for the entire flow. The energy expended is 5 watts per
m /h raised 1 m. The cost of electricity is $0.045/kWh
(1978).
Maintenance
Maintenance labor, materials, and supplies are included as
5 percent per annum of the equipment capital cost. Cost related
to maintenance of carbon and structures appear negligible.
Backwash
The backwash process uses water and air to scour and wash
the carbon from the second stage of filtration. As before, we
have not shown any additional cost for the carbon in the first
stage, since we assumed that the expenses of the first stage GAC
filtration are no greater than those of sand filtration.
Washing of the GAC filters at Morsang occurs every 3 to 4
weeks.
Carbon Losses
In addition to the depreciation of the carbon, which we
considered above in our discussion of amortization, there must
be compensation for the losses which occur during backwashing
of the contactors. These losses, due mainly to attrition, vary
from year to year and are related to the density, size, and
hardness of the carbon. These characteristics may change
significantly from one reactivation to another. The losses
during the year 1978, were about 3 percent at Morsang and 5
percent at Vigneux* The carbon 'of Morsang is washed much less
frequently than that of Vigneux but is less dense, finer and
softer. Its effective size seems to decrease with time, but
attrition during backwashing and particularly air scouring are
not sufficient to explain this phenomenon. It is possible that
a preliminary oxidation treatment (prechlorination at Vigneux
plus ozone at Morsang) contributes to the degradation of the
carbon. Some carbons are considerably degraded with time, while
others of the same quality seem much more resistant.
The quantity of fine particles is a subject to which many
have given much thought (2). We had originally required car-
bons from which fines had been sieved-out (e.g., Cheraviron
F 300 D), with an effective size of 0.85 mm, a uniformity co-
751
-------
efficient of less than 1.7, and a maximum of 1.5 percent passing
through a U.S. mesh 30 sieve (0.59 mm). However, the cost of
such a carbon is 12 to 15 percent more than unclassified carbon
and makes its use prohibitive. The physical characteristics of
carbon are thus very important. The washing conditions (air and
water or water alone, water velocity, time of washing, etc.) are
equally important. The main point is that costs associated with
carbon loss may vary greatly.
Control of Water Quality and GAC Efficiency
Correctly monitoring the efficiency of the carbon requires
frequent analyses which are both complex and costly. For this
reason, GAC treatment may entail an important additional cost.
The simple methods used in our plants (measurment of taste
threshold numbers, DOC, TOC, UV absorbance) have not created
any appreciable supplementary costs, however, and thus do not
appear in Table 4.
Total GAC Treatment Cost
We conclude that the overall cost of GAC treatment is quite
reasonable - particularly at Vigneux. The selling price of
water in the Paris region is roughly $0.50/m3 ($1.89 per
thousand U.S. g_allons). The cost of GAC treatment therefore
represents approximately 1 to 2 percent of this price. If we
consider only the total treatment cost, without distribution and
other costs, the cost of GAC treatment is about 15 percent of
the total at Morsang and less than 10 percent of the total at
Vigneux.
DISCUSSION
Determination of the Reactivation Rate
The effectiveness of GAC progressively decreases with
its use. This trend, both measured and schematically repre-
sented, is indicated on Figure 1 and 2 for taste removal, figure 3
for COD removal and Figure 4 for TOC removal. The central
plateau undoubtedly corresponds to the development of biological
activity within the carbon bed. These results are for the plant of
Vigneux, but similar data presented in more detail elsewhere (4)
were also found at the plant of Morsang.
After a period of time, however, the carbon no longer pro-
duces the desired results due to saturation and should be regene-
rated. This time depends upon several factors: the quality of
the water to be treated, the quality criteria, and the quality
objectives for the treated water.
752
-------
m 6
SAUO FILTRATION
PAC + SAND FILTRATION
A
GAC
;AC A - 1STSTAGE J]
_,___— —--^.-—...^
GAC K - 2D STAGE
ri —
100
200
500
UOO
15
TIME IN DAYS
20
15
PAC DOSAGE - PPM
Figure x. General Trend of Odor Thresholds with Various
Activated Carbon Treatments in Vigneux-Sur-Seine
Throughput Ratio, m3 water/m3 GAC
Fiyuru 2. General Trend oL Tasto T
of Water Produced by GAC Filtration
753
-------
TYPING GUIDE Cr-.S^T
HUH AI iivAirt, CAMH. n
"5 '.n
70
0 60
O
•rl
•U ',U
U
3
•a
\
• SAND r n~. 4) Reduction with
Activated Carbon Treatment in Vigneux-Sur-Seine
100,
I
•p
I
£
20,
0
Throughput R.itJo, m'* wator/in^ CAC
Pigure H. General Trend of TOC Removal Through GAC Filtration
754
-------
Nature of the Water to be Treated--
The more polluted the water, the more rapidly the carbon
will be saturated. This is not restricted to organic elements
only. Certain inorganic compounds/ such as manganese (at Mor-
sang) or calcium carbonate, may also have an extremely negative
effect on the carbon, particularly if they tend to precipitate.
The presence of an oxidant such as ozone can improve the effec-
tiveness of the carbon by promoting biodegradation.
Quality Criteria Used—-
The plants we have mentioned use GAC to eliminate objec-
tionable tastes in the water. Tastes can be easily reduced by
activated carbon; the lifetime of the filters between two
regenerations in such applications is generally several years.
Other compounds, such as those represented by DOC and TOC,
saturate the carbon more rapidly. For some organics such as the
trihalomethanes (THM), the carbon must be reactivated quite
frequently to ensure efficient removal; no more than 3 months
can be allowed between regenerations to reduce at least 50
percent of the THM at Morsang.
Level of Quality Sought-—
Along with the quality criteria, the quality objective
is a determining factor for the frequency of reactivation. If
a high level of quality is desired, the frequency of reactiva-
tion must increase.
Figure 1 shows the average odor threshold, as a function
of time, for the Vigneux plant. During a single day, the effi-
ciency of the carbon is highly variable, but over the period
of a month or a year the trend of the average values is more
gradual. We can therefore consider that the general trend
is linear. In Figure 5, we see that in order to have an average
odor threshold of 2.5, a second stage filter filled with GAC K
must be reactivated approximately once a year. This conclusion
depends, of course, on the quality of the Seine water, which
varies from year to year. Simple criteria have therefore been
established for the plant operators. The criterion used at
Vigneux is the load of odor threshold numbers (Figure 6) elimi-
nated by the filter, which must be kept smaller than 12,000,000.
This load is equal to the average taste threshold number of the
clarified water minus that of the water filtered through GAC
times the number of cubic meters treated during the considered
period. At Vigneux each filter contains 23 m of carbon;
thus the calculated maximum load is approximately 52,000 per m
of carbon. This figure depends, of course, on the nature of
the carbon, as certain carbons are more efficient than others.
It is, nevertheless, a very practical criterion, which is
also used at other Seine water treatment plants, such as Viry-
Chatillon.
755
-------
tu 6
CO
2:
o
O I*
o
3 3
o
w 2
UJ *
SAND FILTRATION
Duration of the Filter Run in Years
Figure 5. Average Odor Thresholds of Water
Produced with Various Activated
Carbon Treatments in Vigneux-Sur-Seine
3 52
10 Number removed x m3 of water filtered per
m3 of GAC
Figure 6. General Trend of Carbon Odor
Number Removal in Vigneux
756
-------
At Morsang, however, the criterion used at the present time
is the elimination of at least 20 percent of the COD (KMn04)
from the Seine water (Figure 7).
Our experience with GAC treatment of the water at Vig-
neux made possible cost predictions as presented in Figure 8
and Table 6.
TABLE 6
Carbon Treatment Cost* in Vigneux Versus Quality Level
% TOC Removal
GAC 1st Stage
cost in $/m
Gac 2nd Stage
cost in $/m
10
0.003
0.014
20
0.008
0.015
40
0.0029
0.021
1979 Predictions
GAC 1st Stage
+ PAC 10 ppm
THM - CHC13 g/1
0.011
30
0.026
15
*1978 Estimate
Costs in $ per m3 wastewater treated
The Technology of GAC Treatme/nt —
GAC technology also has a considerable influence on the
frequency of reactivation. The most important technological
aspects to be considered are the contact time, the thickness of
the filter bed and the speed of filtration, each of which seems
to have a relatively independent influence (3).
As mentioned earlier, contact times of 8 to 12 minutes
appear optimum. It should also be noted that for a given
contact time, better results are obtained with low filtration
velocities. Tests presently being conducted at another plant
in the Paris region have indicated that a very slow filtration
velocity (0.5 m/h) through very thin carbon beds (10 to 15 cm)
gives excellent results and delays the need for reactivation.
757
-------
-J
V/l
00
CT
c
(0
E
Q)
a:
c
s
100
90
80
70
50
• TREATMENT LINE N* 1
O TREATMENT LINE N° *»
• TREATMENT LINE N* 2
T TREATMENT LINE N* 3
1975
1976
1977
Ifj78
ONDJFMAMJJAS-ONDJFMAMJJASONDJFMAMJ
TIME
Figure 7. Trend of COD (KMnO ) Removal By Different
Treatment Lines at Morsang Sur Seine
-------
It should nevertheless be kept in mind that from the investment
standpoint, it is generally more economic to filter at high
velocity through very deep carbon beds.
The constancy of the flow is also an important parameter.
Most of the above-mentioned plants function with a highly var-
iable flow rate. Such variations seem to have a particularly
harmful effect if tire carbon mainly functions biologically.
Stage of Filtration—
Another very important aspect is the position of the GAC
in the overall treatment process. The carbon in the second
stage of filtration is saturated less rapidly than the carbon
in the first stage. Also, for a given quality objective, the
use of carbon after ozone will delay the need for reactivation,
especially if the carbon is in a second stage of filtration.
The economic advantage of these technologies is not ob-
vious. The simplest solution is first stage filtration » as at
Vigneux • which does not require any additional investment.
However, in this case, the frequency of reactivation is greater.
For the same contact time with the same water, our experience
has shown that reactivation is required from 1.5 to 2 times more
frequently in first stage versus second stage treatment. This
is due/ in particular, to the possibility of using finer carbon
in the second stage. If second stage filtration is chosen, the
investment required must be compensated by the savings on re-
activation. However, the higher the quality objectives, the more
advantageous it is to use a second stage filtration. Figure 8
illustrates this observation with the example of Vigneux. For
commonly required quality levels, first-stage filtration is the
best solution. If high quality water is sought, second-stage
filtration is more economical. For intermediate quality ob-
jectives a combination of GAC with powdered activated carbon
might be optimal.
Figure 8
Total Carbon Treatment Cost
1978 Estimate - Vigneux Plant
• 0.02
.20
Average TOC removal %
759
-------
Combination of Powdered Activated Carbon with GAG
The use of PAC in addition to GAG gives highly interesting
results. During use, GAG often desorbs organic matter. These
desorptions are difficult to predict and are related to varia-
tions in the concentration of the organics in the influent water,
Treatment with even a weak dose of PAC greatly reduces these
variations. The PAC is introduced in a flocblanket clarifier.
The contact time is quite important.
The dosage of the PAC may be adjusted depending on the
quality of the influent water. The goal is to maintain an
almost constant water quality introduced to the GAG. We are
presently attempting this approach at Vigneux, and plan to use
it at Morsang as well.
Treatment with PAC also results in a considerable improve-
ment in the elimination of THM's, if only in reducing the
chlorine contact time.
Table 6 lists predictions of costs versus efficiency for
the various carbon treatments. For the existing quality ob-
jective (10 percent TOG removal), GAG in the first stage still
constitutes the best solution. Economics are further improved
if on-site reactivation is assumed. As the percentage of TOG
desired to be removed increases to a point above 20 percent, GAG
in the second stage becomes the optimum, as is the case if
40 percent TOG removal is required. However, Vigneux Plant
operators are not yet ready to spend more than $0.02/m for
carbon treatment as would be required to achieve a TOG removal
of 40 percent.
The level of THM's, mainly due to breakpoint prechlorina-
tion, remains reasonable. A very high frequency of reactivation
would result in more effective control of THM's, but GAG treat-
ment does not appear to be an economical solution for the
reduction of such a pollutant.
CONCLUSION
When GAC treatment was introduced in France, an attempt was
made in most plants to simplify its use in order to encourage
its development. To this end, the following guidelines were
applied:
• First stage filtration to minimize investment.
• Use of simple parameters with which the plant operators
were already familiar.
• A relatively low frequency of reactivation.
760
-------
• The use of reactivatable carbon whose physical
characteristics are well adapted to filtration.
• The optimal use of existing equipment and careful
adjustment of existing procedures without recourse to
sophisticated technology.
The result has been a relatively modest cost and an almost
universal recognition by French water suppliers of GAC's advan-
tages.
GAG filtration is now accepted as the most reliable means
to control organics in potable water. Based on this conviction,
French water suppliers are now ready for gradual improvement in
its application: better use of biology (BAG), development of
second-stage filtration; use of GAG and PAG in combination; and
introduction of on-site reactivation. However, GAG is probably
not the only solution for the control of specific pollutants
like THM's, and French water suppliers certainly would react
unfavorably to it being considered as such.
REFERENCES
1. Richard, Y. and F. Fiessinger. 1973. "Le traitement des
eaux potables par le charbon actif: aspects techniques
et economiques," Tech. et Sc. Municipales, 2, 43-61.
2. Fiessinger, F. and Y. Richard. 1975. "La Technologie du
traitement des Eaux potables par le charbon actif grajiule,"
Tech et Sc. Municipales, 7, 8, 9, 10: 1-40.
3. Brener, L. and Y. Richard. "Determination on a pilot unit
of technological parameters of granular activated carbon
filtration," American Chemical Society Symposium, Divi-
sion of Environmental Chemistry, Miami Beach, 10/15 Sept.
1978.
4. Fiessinger, F. and L. Brener. "Large Scale Applications of
granular activated carbon for potable water production in
France" American Chemical Society symposium, Division of
Environmental Chemistry, 10/15 Sept. 1978.
761
-------
DISCUSSION
WEDNESDAY, MAY 2
MORNING SESSION
Q PROF. PAUL ROBERTS, Stanford University: I have some ques-
tions about the economics of carbon reactivation in waterworks
practice. (1) Have the Europeans done a cost comparison to
determine what would be the minimum size capacity, in tons per
day, for a reactivation facility that would justify itself eco-
nomically over the cost of fresh makeup carbon at a waterworks;
(2) The reactivation facilities described, which are rather
expensive, complicated units, are used less than 50 percent of
the available hours in the year. Could they be used more or is
it necessary to design excess capacity in case the carbon should
be exhausted by some spill or other accident? Is there a cri-
terion as to how much excess capacity is needed?
A PROF. DR. SONTHEIMER: The lowest amount of carbon reacti-
vated should be in the range of 1-1/2 tons per day to make it
economically feasible. This should be an average figure for the
entire year. Sometimes you may have twice as much reactivation
and sometimes the facility will be idle. It is not important if
the plant is fully automated. The difference is in sizes of the
furnaces, which is very important. There should be lower limits,
especially for. reactivation furnaces, or it stops being economi-
cally feasible.
MR. SCHALEKAMP: It would be best if a reactivation plant
could operate regularly, then there would be little trouble. If
the plant is on standby part of the time, there are problems with
bringing it on line. Therefore, -it shouldn't have too much
excess capacity. I think that's important.
Q PROF. VERN SNOEYIHK, University of Illinois: We've had
different figures presented for losses. If I remember correctly,
Mr. Schalekamp said 3—1/2 percent yesterday. Today we hear 10
to 15 percent. Dr. Klein mentioned that it should be possible
to achieve what Mr. Schalekamp referred to yesterday. I wonder
if the losses that have been discussed at the operating plants
are primarily due to start-up problems or getting to know the
systems. Do the people using the plants think maybe the 3, 4, or
5 percent is something that can be achieved only if the loadings
are as low as achieved at Zurich?
762
-------
A PROF. DR. SONTHEIMER: I would like to comment on this. I
think we should keep something in mind when we spea): about
losses. The prime objective is high quality of the reactivated
carbon. Moreover, if you save time between regenerations, in
some cases, it may be enough of an advantage to compensate for
some losses. The other point is that itVs very difficult to
collect exact data on losses. I thirik'—and I believe my two
colleagues will agree'«-that the only acceptable way is to re-
generate the same carbon five, six, or ten times to accurately
estimate total losses-. Then the effects of backwashing, which
removes small particles (all those less than .5 millimeters)
and could make a difference, would be taken into account. Some-
times some plants don't regenerate frequently, so we have to be
patient. We discussed this problem yesterday too, and we decided
that we would have a special meeting with the people who measured
losses in regeneration to determine the best ways to measure it.
It's not an easy thing to do, but I wouldn't say this is so
important. Some people try to reduce losses, but even for losses
of 10 and 12 percent, it remains economical.
MR. SCHALEKAMP: I agree with Professor Sontheimer. I only
can say that the loss we experienced in Zurich resulted from one
regeneration. If you regenerate three times, there will be much
more loss. I will say we have used carbon from another manufac-
turer and contracted with them to perform reactivation. When it
came back, we put it into our filters and discovered we had a
loss of 20 percent, because it was all powder and we had to get
rid of it.
DR. KLEIN: As I tried to explain in my paper, the loss is
dependent on the kind of the carbon and adsorbate. If the
activity of the carbon is in the same range as that of the
adsorbate, you have a loss. I think you can't avoid it, but if
you have an adsorbate that is more reactive than the carbon
itself, then you can get a minimum carbon loss.
Q FOSTER BURBA, Louisville, Kentucky: My first question is
the factor for converting Deutsche marks to dollars; secondly, I
would like to get an idea of the life of the furnace regeneration
equipment.
A PROF. DR. SONTHEIMER: The conversion factor is 1.9 marks to
one dollar; the life of the furnace, we don't know.
A MR. SCHALEKAMP: From experience in the gas works, I think
it will be about 10 years, but every time it is taken out of
operation, you will have to make minor repairs.
763
-------
A DR. KLEIN: We have an oven used for the regeneration of
carbon which has been working since 1964; that means about 15
years.
Q MR. BURBA: I have one other question. Has there been any
investigation of the possible toxic chemicals formed during the
regeneration process which would be retained on the GAG and per-
haps leached into the water after the GAC is regenerated and put
back into service?
A PROF. DR. SONTHEIMER: We did numerous studies on this
problem because certain compounds under certain temperature
conditions can form toxic compounds. That is the reason why we
say you must exceed a minimum temperature. We are very sure that
if high enough temperatures are maintained, no toxic compounds
will be formed because they and all other compounds are burned.
We found, first, that if we have too low a temperature, organic
compounds will remain on the carbon. We control the minimum
temperature as well as other aspects of carbon regeneration. We
are very sure that all the problems which may arise don't arise
in this plant.
V MR. D. J. OSBORNE: Is the particle size reduction after
regeneration due to the mechanical abrasion of the granular
activated carbon and is this increased due to embrittlement by
the quench after regeneration? Have you done any work with slow
cooling in an oxygen-free environment to reduce this?
A PROF. DR. SONTHEIMER: We didn't do this work and we are
not sure exactly where reduction of grain size occurs. I doubt
that the most important step is the cooling, and for this reason
we didn't study it at all. We also have to get more experience.
Working with such a plant, we found that the largest reduction in
size occurs the first time, but then there is no further change
for a very long time. You have to observe it four, five, six,
seven times, before significant conclusions can be made. We have
not done any studies because we didn't see any need.
A DR. KLEIN: We have regenerated active carbon in a plant 25
times. The distribution of particle size did not differ very
much from that of the first time. The main reductions come in
backwashing and hydraulic transportation, not by regeneration.
V MR. PEARSON: A question was raised earlier about the amount
of time that the Dusseldorf reactivation plant was not in serv-
ice. I have spoken with Professor Sontheimer about the fact that
it was originally designed when the pollution in the Rhine was
very high* Because of the improvement in the waste discharges
from one of the huge industrial complexes—the BASF plant
upstream—the TOC and organics loading has been greatly reduced.
This is one of the reasons that the plant doesn't operate as much
764
-------
as it originally was designed to. Would you care to comment on
that for the benefit of the audience, Professor Sontheimer?
A PROF. DR. SONTHEIMER: I would be very pleased to because
I think all of us working in the drinking water field have to
fight against pollution. In Germany, we have an association of
waterworks along the Rhine river to fight against the pollution
of the Rhine. Mr. Schalekamp is the President. We have achieved
some improvements in the quality of the Rhine River water.
Thanks to this improvement in raw water quality at Diisseldorf,
instead of three furnaces, only two are now needed. There is
some safety factor in this too, instead of 3 months we now range
from 5 to 6 months between regenerations. However, I don't
believe that we will progress with our industrial society to
levels of water quality where activated carbon is no longer
needed. We always will need it as a safety factor.
Q WALTER McCUSKY, U.S. Army Construction Engineering Research
Lab: The first question is for Dr. Klein. Did I understand
your graphs correctly, that you get selective chemical reaction
in the regeneration atmosphere of the carbonized organics up to
a certain point and at that point you begin to get attack on the
carbon matrix? If so, how do you tell where that point is so
that you can get the carbon out of the furnace in time?
DR. KLEIN: We first investigated on a small-scale plant in
the lab to find the optimum conditions for the temperature and
the residence time. Then we applied these conditions in a
large-scale plant. The residence time distribution in the full-
scale furnace is important. Particles which are retained too long
can be destroyed.
MAC McGINNIS, Shirco, Inc.: We have observed a consistent
increase in the ash content of the regenerated carbon, which is
indicative of deposition of some sort of inorganic material on
the carbon. Alum and lime are used in this plant and one would
suspect that they may be sources of the ash increase: a logical
conclusion is that that may preclude complete restoration of
virgin properties.
Q MILLER, Cincinnati: Dr. Love, I have a plea for help.
In the furnaces that you're seeing, it sounds as though some of
the problem is determining the points of loss or attrition. It
would seem as though it is in the transportation system in the
furnace itself. If they're doing this in Germany, or in any of
your projects, I think it would be helpful if we all know about
this.
A
n DR. LOVE: I think Professor Sontheimer addressed at least
part of that this morning; he felt the losses in the furnance
were very low, 1 to 2 percent.
765
-------
NATO CDSM
• OTAM CCMS
NATO-CCMS
766
-------
NEW DEVELOPMENTS IN ADSORPTION TECHNIQUES
767
-------
xvEPA
NATO
COSM
•OTAM CCMS
NATO-CCMS
768
-------
GENERAL CONSIDERATIONS IN ASSESSING THE BENEFICIAL ASPECTS
OF MICROBIAL ACTIVITY ON GRANULAR ACTIVATED CARBON
Francis A. DiGiano
INTRODUCTION
The term "biological activated carbon" has found its way
into the U.S. technical literature in the last year as a result
of reports by Miller and Rice (1) and by Rice, et al. (2) on
European drinking water practice. Professor Sontheimer's
research group (3) at Karlsruhe University was the first to
conduct practical tests of the beneficial aspects of microbial
activity on granular activated carbon (GAC) in pilot plant tests
at Bremen, Federal Republic of Germany from 1970 to 1974. They
proposed that through microbial degradation of organics, the
service time of a GAC bed could be greatly extended. More
recently, research groups led by Professors Benedek (4), Tien
(5), and Weber (6) have focused on formulating and verifying
conceptual mathematic models to describe the interaction between
biodegradation and adsorption on GAC. The intent of this paper
is to discuss the present s'tatus of microbial activity on GAC
as a method of water treatment and to point to further research
needs.
BACTERIAL ATTACHMENT
Attachment of bacteria (and protozoa) to the external
surface of GAC has been shown clearly by electron micrographs
(7) but the mechanisms responsible have not been investigated
as yet. Mechanisms for surface attachment in general have been
recently discussed by Marshall (8). Apparently, attachment can
be either reversible, with desorption produced by the shearing
action of fluid mixing or by mobility of sorbed bacteria, or it
can be irreversible, involving adsorption of extracellular
materials (9).
Irreversible attachment may require time for preparation
of the surface. This was referred to as "seasoning time" by
Pipes (10), in investigation of phenol biodegradation in a sand
filter; about 6 days were required. In pilot plant studies of
water treatment by GAC, van der Kooij (11) noted that 20 to 30
days were needed before the maximum level of 10 colonies per
gram was reached. This could imply either that extracellular
materials were adsorbing or that a build-up of sorbed organics
769
-------
was necessary before microbial activity could increase; in the
latter case, a form of bioregeneration must be assumed. The
evidence is insufficient to conclude which of these possibilities
was in fact the case.
An attempt has also been made to quantify the adsorption
of a mixed population of bacteria on GAC in a batch experiment
similar to that used to measure adsorption equilibrium (12).
While an adsorption "isotherm" was obtained, its meaning seems
lost because "equilibrium" was not reached; i.e., bacterial
death and desorption continued to occur without regard for solid-
liquid phase equilibrium.
The question also remains as to whether or not GAC pro-
vides a more favorable environment for attachment than sand or
non-activated carbon. This uncertainty is shown quite clearly
by comparing surface bacterial counts for these three media in
Figure 1. In this research, van der Kooij (11) claimed that
any advantage shown by GAC was accounted for by its larger
external surface area.
HIGHLIGHTS OF BREMEN STUDY
The Bremen pilot plant study is probably the longest obser-
vation of GAC bed performance without regeneration (3-1/2 years).
Since results of this study have been published only in German
(3), it is worthwhile to present highlights here. At the outset,
it is important to recognize that only practical measures of
organics were used, such as potassium permanganate demand
(KMnO,), total organic carbon (TOC), and UV absorbance. Thus,
it is impossible to cite which specific organics, or even which
broad groups of organics,-were biodegraded. Unfortunately, this
deficiency is common to all pilot plant studies of GAC in which
microbial activity is of interest. Nevertheless, the Bremen
study provides insight into overall efficiency of the process.
Removal of KMnO in a pilot plant GAC bed is compared
with that in the full-scale slow sand filter in Figure 2. As
expected, the performance of the GAC bed was superior during the
first 5 months because of adsorption. However, in the remaining
months, both systems produced the same removal, suggesting that
microbial activity was responsible. Although it is difficult to
state the exact contact time in each system, that of the slow
sand filter is on the order of several hours, while that of the
GAC bed is between 15 and 30 minutes. Based on this rough
comparison, GAC seemed to provide more efficient biodegradation.
As shown in Figure 3, oxygen consumption was much higher in
summer than in winter months; yet nitrification, which accounts
for significant oxygen consumption, was nearly absent in summer
but in evidence in winter. The fact that oxygen consumption
remained high in summer was explained by not only more active
770
-------
20 40 60 80
—•- cum % of samples
100
Figure 1. Comparison of adsorption of colony forming units
(c.f.u.) on activated carbon (AC), non-activated
carbon (NAC) and sand in filter columns (diameter
of 6 cm) operated for one year. Filter columns
were supplied with nonchlorinated tap water having
an organic carbon concentration of 3 mg/1. Incuba-
tign of plates used for counting c.f.u. was at
25°C for 10 days. From van der Kooij. (11)
771
-------
25
20
C?
10
Before GAC and Slow Sand Filter
"••.—i
\
After Slow Sand Filter
•After GAC
MIAIMIJIJ|AISIOINID|JIFIMIAIMIJIJIAISIOINIDLIFIMIAIM'JIJIAISIQINIO|JIFIMIAIMI
1970
1971
1972
1973
Figure 2. Comparison of removal of KMnO. demand by slow
sand filters operating at Bremen and by pilot
plant scale GAC beds (diameter of 14 cm and depth
of 3m). Influent to both the slow sand filter
and GAC beds is product water of full-scale rapid
sand filters. Contact time of GAC beds is between
15 and 30 minutes. From Eberhardt, et al. (3)
772
-------
10 -
1970
1971
1972
1973
Figure 3. Oxygen consumed (uptake) during passage through pilot , ,
- ~ From Eberhardt, et
plant-scale GAC beds at Bremen. From
Upper Half of Bed
20
8.
I ,«!-
1.0 -
§.
O 06 j-
c
| 0.4-
002-
Lo»er Half of Bed
Sept
Oci Nov Dec Jan F«t> Mar Aon) May Jun« July iuq S«pt
1973
Figure 4. Comparison of decrease in organic carbon with increase in inorganic
carbon in the upper and lower halves of pilot plant-scale GAC beds
at Bremen. Carbon type is LSS with particle diameter of 0.5-1 mm.
Total bed depth is 6 m and total contact time is about 36 minutes
(assuming velocity, of 10 m/hr. Data was derived from Table 21 of
Eberhardt, et. ai:3'
773
-------
biodegradation of organics at higher temperatures, but also by
biooregeneration of adsorbed organics. In fact, inorganic carbon
production (as measured by carbon dioxide) was much greater in
summer than in winter, thereby supporting the biaregeneration
concept.
The interplay between adsorption and biodegradation is
clearly seen in Figure 4. Here, the reduction in organic carbon
and the production of inorganic carbon are shown for the upper
one-half and lower one-half of a GAG contact system at the
Bremen pilot plant. Organic carbon removal decreased in the
upper half as adsorption capacity became exhausted. However,
microbial activity then began to predominate, as indicated by
increased production of inorganic carbon. Biooregeneration was
suggested in one summer month when the increment in inorganic
carbon exceeded the decrease in organic carbon. Both adsorption
and biodegradation processes in the lower half of the contactor
lagged that of the upper half. This lag is understandable from
adsorption principles? i.e., adorption capacity must first be
exhausted in the upper half before substrate becomes available
for microbial growth in the lower half of the bed.
The extent of biocregeneration was also examined by recy-
cling water through a GAG bed. In two months of testing, a
steady state production of carbon dioxide was reached which
could be expressed as a bioregeneration rate of 57 g carbon per
m of GAG per day. Given the typical sorptive capacity of
GAG, it was estimated that complete in situ bio-regeneration
could be achieved in 6 months.
Another indirect indication of bioregeneration in the pilot
plant operation was obtained from diojtane and dimethylformamide
extractions of sorbed organics. The total amount of organic
carbon extracted was stated to be far less than that adsorbed by
GAG, based upon measurements of organic carbon in the feed and
product streams of the GAG bed. It was concluded that sorbed
organics were being biodegraded; however, the accuracy of the
extraction procedures and the overall mass balance on sorbed
organics was not discussed.
Based upon all the pilot plant testing at Bremen, it was
concluded that a steady state removal of about 2 mg/1 TOG could
be obtained in a GAG contact time of 30 minutes^ This is equiv-
alent to a biodegradation rate of 4 g TOG per m GAG per hour.
However, it must be noted that this rate was only approached when
a very highly activated carbon (LS-supra) was used; other carbon
types gave values on the order of 2 g TOG per m GAG per hour
which means that in a contact time of 30 mintues, only 1 mg/1 TOG
can be removed. Hence, by comparison to adsorption, the process
of microbial degradation on GAG is relatively inefficient.
Nevertheless, attainment of a steady-state removal offers a
unique advantage over adsorption. The important question left
774
-------
unanswered by this study is: which specific organics can be
removed during steady-state operation?
POSSIBLE EXPLANATIONS FOR ENHANCED MICROBIAL ACTIVITY
Whether or not GAC supports a higher rate of biodegradation
than sand or non-activated carbon has not been proven in a
rigorous fashion. However, in their study, Ying and Weber (6)
proposed three possible explanations for advantages offered by
GAC: enrichment of oxygen by sorption; enrichment of substrate
in the biofilm; and extended adsorption due to bioregeneration
and development of bioforms which can degrade less biodegradable,
but adsorbable, organics.
Although as much as 19 mg oxygen per g GAC can be adsorbed
(13), its availability to the biofilm is questionable if irrever-
sible chemi-sorption occurs; this form of adsorption is supported
by a well established body of literature (14). Conceivably,
chemisorption of oxygen could still be important if it provides
a driving force for transport of oxygen across the biofilm, in
which case aerobic biodegradation can be assured. However, this
seems unimportant in water treatment applications of GAC where
oxygen concentration is high enough in the bulk solution phase
and biofilm thickness is not large.
With respect to the second explanation offered, substrate
enrichment in the biofilm seems likely. This may depend, how-
ever, on the substrate concentration in the bulk solution phase
and on the ability of substrates to adsorb to the biofilm layer
itself.
The third explanation, that of bio-regeneration, is-probably
the most controversial. Because microorganisms have a diameter
of about 1/tm, they are too large to enter the micropores which
have diameters of 1 nm to 100 nm. However, extracellular enzymes
would be able to reach substrate otherwise considered inaccessi-
ble ; it has also been shown that these enzymes can biodegrade
high molecular weight organics (15). Proof of their importance
has been offered in studies of attachment of film-forming marine
bacteria to surfaces; 23 extracellular and intracellular enzymes
were found (16).
There is also evidence which argues against bio-regenera-
tion. For example, in pilot plant studies at the Morsang water
works in France, Benedek (4) withdrew carbon particles from a
microbially active GAC bed and attempted to adsorb more organics.
The fact that no further adsorption was possible argued against
bi©regeneration. In another pilot plant study, van der Kooij
(11) reported that the majority of microorganisms isolated from a
GAC bed were able to grow on simple, non-adsorbing substrates but
not on adsorbable aromatic compounds. However, the data offered
to support this argument were not conclusive.
775
-------
Thus, aside from evidence provided by the Bremen study and
highlighced herein, there is a lack of data, either from labora-
tory or pilot plant studies, to support or refute the hypothesis
of bio/regeneration. If bio-regeneration is unimportant, then
extended service time of GAG beds can only be explained by
removal of biodegradable, weakly adsorbable organics. Should
this be the case, then other organics, perhaps of more health
concern, will escape the bed during extended operation.
EFFECT OF PRE-OZONATION
In some discussions of the so-called biological activated
carbon (BAG) process, pre-ozonation has been proposed as a pre-
requisite for enhanced microbial activity on GAG (1,3). The crux
of the argument is that organics are made more biodegradable by
ozonation. However, there is little direct evidence to support
this contention, and in fact, a study showing the positive
effects of pre-ozonation in wastewater treatment at the Cleveland
Westerly plant drew open criticism in the literature (17, 18,
19). In two other pilot plant studies of wastewater pre-ozona-
tion at Water Factory 21 and Lake Tahoe, some improvement in GAC
performance was measured, but it could not be traced to enhanced
biodegradation (20). At the Morsang water works, pilot plant
operation of pre- and post-ozonation produced almost equivalent
water quality (4). Only a small increase in oxygen uptake and
organic carbon removal was noted through GAC in the pre-ozonation
system. In this case, however, it was reasoned that pre-chlori-
nation probably produced organics which were less biodegradable
even when ozonation was used. On the positive side, extensive
pilot and full-scale testing of pre-ozonation at the Mulheim
water works gave substantial improvement in final wat«jr quality
(21). However, whether or not this involved enhanced microbial
activity needs further discussion.
The Bremen pilot plant study (3) preceded that at Mulheim
(21) and offers an interesting insight into the effect of ozone
dosage on organic removals in the ozone contactor and in the GAC
bed. Table 1 shows these removals and that achieved through
parallel full-scale operation of slow sand filters without pre-
ozonation. By itself, GAC contact did not produce as much removal
as slow sand filtration. However, when taken together with the
ozonation step, total removal exceeded that of slow sand fil-
tration, except at the lowest ozone dosage of 2.3 mg/1. This
suggests that when evaluating the effect of ozonation, both
oxidation and adsorption steps should be considered separately.
Although not directly proven, it appeared that increasing the
GAC contact time increased organic removal by microbial activity.
For comparison, it should be noted that organic carbon removal by
GAC with pre-ozonation was estimated (by means of a correlation
of UV absorbance with TOC) as 1.9 mg/1; this is not significantly
greater than that found in the Bremen studies without ozonation.
776
-------
Table 1. Summary of Results of Pre-ozonation
in Pilot Plant Studies at Bremen
Ozone Period of
Dosage Testing
mg/1 weeks
2.3
8.5-10
5.8
4.3
4.8
6
3
4
9
4
AKMnO. Demand
by Ozonation by GAC
mg/1 mg/1
-1.7
-3.6
-3.2
-2.5
-2.5
-0.4
-1.7
-2.3
-2.2
-2.9
AKMnO. Demand
GAC by Slow Sand
Contact Filtration without
Time Pre-ozonation
min mg/1
14
15
17
30
30
-3.5
-3.8
-3.6
-3.3
-1.9
NOTE: Table 1 is a summary of results originally presented in
Tables 6, 1, 8, 9, and 10 of reference 3. The pilot plant
process scheme consisted of ozonation and GAG contact.
Slow sand filters in the full-scale plant were operated
as controls in parallel with the ozonation-GAC pilot
plants. Influent to both the pilot and full-scale units
was from the full-scale rapid sand filter.
Table 2. Performance of Pilot Plant GAG Beds
at Miilheim with Pre-ozonation
Activated A Inorganic
Carbon A DOC Carbon ANH -N
mg/1 mg/1 mg/1
LSS
LSS
Source :
NOTE:
2.5 m -0.92 +0.83 -1.31
5.0 m -1.69 +0.96 -1.34
Reference 21
LSS is a carbon manufactured by the Lurgi Co
A02
mg/1
-6.32
-6.67
mpany .
of A refer to changes in the first 2.5 m and in the total
depth of 5 m, respectively. Data are mean values for a
6-month operation after a 3-month starting period. Con-
tact time in each bed segment is 15 minutes.
777
-------
Much lower ozone dosages were tried at the Mulheim water
works (about 2 mg/1) and performance improved. However, it is
Important to recognize that the treatment process change at
Mulheim included elimination of pre-chlorination. In a parallel
study of pre-ozonation with and without pre-chlorination, ex-
tended organic removal on GAC occurred only in the latter process
stream. This implies that chlorination produces less biodegrad-
able organics or otherwise inhibits microbial activity. Another
important observation was the noticeable improvement in organic
removal which was brought about by ozone addition prior to
flocculation. This no doubt contributed to extended GAC service
time.
The positive effect of ozonation on flocculation, together
with elimination of a negative effect On GAC performance (i.e.,
the elimination of pre-chlorination) may leave unanswered the
question of whether or not pre-ozonation enhances microbial
activity, per se. If the removal of organic carbon through GAC
contact is compared in the Mulheim (Table 2) and Bremen studies
(Figure 4), the latter obtained without pre-ozonation, it is
difficult to conclude that pre-ozonation actually improved
biodegradability. Table 2 shows that most of the microbial
activity was found in the upper 2.5 m after a contact time of
15 minutes; the amount of inorganic carbon produced is similar
to that shown in Figure 4 at Bremen. As is also evident from
Table 2, nitrification is a very important process at Mulheim.
From stoichiometry, 4.6 mg of oxygen are required to oxidize
1 mg of ammonia nitrogen to nitrate. Accordingly, nearly all of
tne oxygen depletion snown in Table 2 can be accounted for by
nitrification; in fact, some production of carbon dioxide must
also be attributed to this process. Hence, any conclusions
regarding the extent of carbonaceous bio-oxidation and bio-
regeneration should be made cautiously.
Pre-ozonation was shown to improve the removal of precursors
to trihalomethane formation, as measured by the trihalomethane
formation in pilot plant studies conducted by the U.S. EPA in
Cincinnati, Ohio (22). The effect of ozone was to oxidize these
precursors, which are most likely of humic origin: the resulting
products were either more readily adsorbed or biodegraded on
GAC. An additional indicator of improved performance was better
removal of TOC. These results were obtained with an ozone dose
of 1 mg/1 and a GAC contact time of 9 minutes. While effluent TOC
concentrations increased steadily in both the ozonated and
non-ozonated systems, complete breakthrough was not reached in
either system during 10 months of operation. Thus, microbial
degradation was probably occurring in both systems. However,
pre-ozonation gave better performance.
The potential detrimental effect of ozonation on adsorb-
ability of organics must also be considered. Figures 5 and 6
are taken from two independent studies that showed this effect
778
-------
10
7
5
E 2
o>
Without Ozone
UV Wavelength «254nm
Cell Path Length = 10 cm
1 I I 1 I
_L
Figure 5.
01 02 05 10 2345
UV Absorhance (Equilibrium Fluid Phase Concentrationl-m"'
Effect of ozonation on equilibrium adsorption of organics as
measured by UV absorbance. Data taken during pilot plant-scale
operation of GAC beds at the Morsang water Works in France
From Benedek (4) .
I
S 30 -
_v zo -
o
I
10 -
8 5
Ozone Dosoqes
• O g/m3
O 0.45 g/mS
O I I g/m3
A 3.5g/m3
UV Wavelength« H54 nir
0.3 0.4
06 0.8 1.0
1.5
2.0
UV Absorbance (Initio! Pollutant Concentration) -m"'
Figure 6. Effect of ozonation on equilibrium adsorption of organics as
measured by UV absorbance. Abscissa values represent initial
rather than equilibrium concentrations of UV absorbing organics.
Absorbance values are calculated for an equivalent cell path
length of 1 meter. Data taken in laboratory test with pre-
filtered Lake of Constance water. From Kuhn, et al. (23)
779
-------
at dosages as low as 0.45 mg/1 (4,23). Decreased adsorbability
may reasonably be expected because ozone tends to give more
polar organics such as carboxylic acids, catechols, and aldehydes
that are characteristically more weakly adsorbed. Any potential,
though as yet unproven, increase in biodegradability must, there-
fore, be weighed against decreased adsorbability. Benedek (4)
accounted for both effects in predicting the performance of
GAC adsorbers at Morsang. Because this study revealed little
increase in biodegradability, the mathematical model predicted
that ozonation would only be beneficial if extremely long service
times could be used (on the order of 2 years) such that full
advantage was taken of microbial activity; otherwise, the de-
crease in adsorbability was the overwhelming effect.
PREDICTING THE INTERACTION BETWEEN BIODEGRADATION AND ADSORPTION
The objective of developing a predictive, mathematical model
which accounts for microbial activity on GAC, is ultimately to
provide a more sound basis for design and operation of GAC beds.
Clearly, the classical pattern of organic carbon breakthrough
caused by exhaustion of sorption capacity will be altered by
microbial degradation of organics. Simplifications are needed in
the mathematical representation of the adsorption process (e.g.,
competitive adsorption is often ignored) in order to provide a
more useable tool. Simplifications will be even more important
when attempting to "overlay" the effects of microbial activity.
Available models use the simple, and perhaps unrealistic,
assumption that all organics in a complex mixture are biodegrad-
able. Figure 7 shows the resulting general shapes of concentra-
tion gradients in the outer biofilm, the internal pores, and the
sorbed phase of a carbon particle. At time t,, the biofilm has
not developed and therefore adsorption predominates. Growth of
the biofilm depends upon microbial kinetics in response only to
the substrate available external to the carbon particle. At a
later time, t~, biodegradation in the biofilm accounts for most
of the substrate removal and the adsorption capacity becomes
dependent on the concentration at the innermost boundary of the
biofilm, C-, rather than upon the external concentration, C_.
X C
Important characteristics for three available mathematic
models (4,5,6) all of which consider organics to be biodegradable
and adsorbable, are presented in Table 3. These models differ in
description of microbial kinetics, mass transfer, and adsorption
equilibria. However, all yield the same general pattern of or-
ganic component concentration leaving the GAC bed with .service
time. Unfortunately, very few comparisons are available between
predictions and experimental data. Of particular concern is the
lack of verification for organics which are strongly adsorbed and
also biodegradable. This may be because such organics are not
easily found. If such compounds are rare, then the interaction
between adsorption and biodegradation may be less important than the
780
-------
Time, t
Bulk
Liquid
Phase
Activated Carbon
Solid Phase Concentration
Gradient
Pore Concentration
Gradient
Bulk
Liquid
Phase
Activated Carbon
Solid Phase Concentration
Gradient
Pore Concentration
Gradient
Figure 7. Hypothetical concentration gradients existing in
biofilm and solid phase media when biodegradation
and adsorption occur simultaneously. At time t^,
the biofilm is not well developed and adsorption
predominates. At time t2/ the biofilm is developed
and biodegradation predominates.
781
-------
Table 3. Characteristics of Three Available Mathematical Models Describing
the Interaction Between Biodegradation and Adsorption
Biofilm Thickness
Biof Urn Kinetics
Mass Transfer
External Film
Diffusion
through
Biofilm
Internal
Diffusion
Adsorption
Equilibrium
Model
Reactor
Description
Benedek (4)
Fixed
Zero-order
Yes
Yes, Pseudo
Steady State
Lineaj: Driving
d
Force
Freundlich
Plug Flow,
Stationary Bed
Verification
of Model
Pilot Plant
TOC
Tien (5)
Variable
First-order3
No
Yes, Pseudo
Steady State
Linear Driving
Force
Linear
Plug Flow,
Variable Bed
Depth6
None
Available
Ying & Weber (6)
Variable
Non-Linear, Monod
Yes
NO
Surface Diffusion0
Freundlich
Plug Flow-
Stationary &
Completely
Mixed Bed1
Dispersive Flow-
Stationary &
Completely Mixed
Bed1
Laboratory Scale-
Glucose &
Sucrose^
NOTES:
a. Non-linear Monod kinetics approximated by linear function.
b. Gleukhauf's linear diffusional driving force model.
c. Complete profile of solid phase concentration developeffrby finite
difference technique.
d. Earlier model by Benedek considered production of adsorbable, non-
biodegradable organics by microbial activity as a species competing
for adsorption sites but no details were given.
e. Bed depth increases as a result of expansion of bed by growth of biofilm.
f. Two different reactor models were developed.
g. Verification for laboratory data on strongly adsorbable, weakly
biodegradable organics (toluene sulfonate and potassium biphthalate)
was not given.
782
-------
uncoupled removal of biodegradable organics by the biofilm and
non-biodegradable organics by adsorption at internal sites. Such
a conclusion, however, is premature.
The nature of mathematical predictions is best illustrated
by the results given by Ying and Weber (6) for the two extreme
cases of strongly and weakly adsorbed organics. Figure 8 shows
the importance of adsorption in the initial stage of operation
and microbial activity in the later stage. The level of steady-
state removal of organics depends upon the assumed microbial
kinetics/ as seen in Figure 9. Unfortunately, there is a paucity
of information on the GAG system in this regard. In fact, the
initial concentration of attached biomass is also unknown and
this parameter greatly influences the shape of the breakthrough
curve.
Two alternative models, which would be more complex but
perhaps more realistic/ are proposed here. In one model/ two
categories of organics—biodegradable and non-biodegradable—are
considered. As an added complication/ the second model would
include the availability of substrate from internal sorption
sites as well as from the external/ bulk solution; in this case/
the effect of bicnregeneration is accounted for. Tien (5) has
indicated research progress on the first of these models/ but
specifics are unavailable. An earlier model by Peel and Benedek
(24) also considered the production of non-biodegradables by
bacterial action and these were then assumed to adsorb in compe-
tition with biodegradable organics; however/ details were not
provided and later work (4) considered all organics as biodegrad-
able. In both models/ gradients may be very complex and even
reverse in direction as service time increases. This may be
caused by competitive adsorption between the biodegradable and
non-biodegradable fractions and by intermittent decreases in
sorptive loading as a result of bioregeneration.
CONCLUSIONS
The development of microbial activity on GAC is a natural
consequence of treating water containing biodegradable organics.
This effect has been observed in pilot and full-scale studies
in Europe, where limited data are available concerning the over-
all rate of biodegradation and the contact time required. How-
ever/ many questions remain unanswered. Among these are* Which
organics are removed by microbial activity? How can bacterial
attachment be optimized? How important is bio-regeneration of
sorption sites as compared with biodegradation of externally
available substrate? Does pre-ozonation increase the biodegrad-
ability of organics or does it simply improve the treatment
process prior to GAC? Can mathematical models account for the
interaction between microbial activity and adsorption in a
realistic/ yet practical manner?
783
-------
Weak Adsorption without Biodegradation
Figure 8.
Figure 9.
Strong Adsorption without
Biodegradation
Weak Adsorption
with Biodegradation
Strong Adsorption with Biodegradation
500
3500 4000
1000 1500 2000 2500 3000
Number of Mean Residence Times (d)
Predicted effluent concentration as a function of service tine with
the GAG bed behaving as a stationary phase and with plug flow in
the liquied phase. The effect of biodegradation on removal of both
weakly and strongly adsorbable organics is shown. From Ying
Weber (6).
and
I 00
s 80
IT
S .60
o
o
= -40
s
- 20
Monod Kinetics
ond
is dimensionless p.
'0.01
=0.02
0 500 1000 1500 2000 2500 3000 3500 4000
Number of Mean Residence Times (8)
Predicted effluent concentration as a function of service tijte with
the GAC bed behaving as a stationary phase and with dispersive flow
in the liquid phase. Ihe effect of bacterial growth kinetics on
removal of a strongly adsorbed organic is shown. Decreasing the
half-saturation constant, Ks, or increasing the maximum specific
growth rate constant, ^imax, decreases the steady-state effluent
concentration. From Ying and Weber (6).
784
-------
This paper has intentionally avoided the issue of whether
or not microbial activity on GAG can facilitate meeting the pro-
posed new U.S. EPA regulations (Federal Register of February 9,
1978; regarding control of trace organics. It would be specu-
lative to address that question, because not much is known about
the biodegradability of chlorinated organics in extremely low
concentrations. One possible mode by which microbial activity
could extend GAC service time is by removing otherwise adsorbable
organics, thereby lessening the competition for adsorption sites
and enhancing the removal of those less biodegradable organics.
However, the range of adsorbability and biodegradability of
organics present in drinking water sources first must be inves-
tigated more carefully.
The general tenor of this paper has been to raise more ques-
tions than are answerable at present. Clearly, before design
recommendations can be made, far more research under controlled
conditions must accompany pilot and full-scale tests in order to
achieve a better understanding of the influence of microbial
activity on GAC performance.
785
-------
REFERENCES
1. Miller, G.W. and R.G, Rice. 1978. European water treatment
practices-the promise of biological activated carbon. Civil
Engrg. _48( 2) :81-83.
2. Rice, R.G., G.W. Miller, C.M. Robson and W. Kuhn. 1978.
A review of the status of pre-ozonation of granular acti-
vated carbon for removal of dissolved organics and ammonia
from water and wastewater. In P.N. Cheremisinoff and
F. Ellerbusch, ed. Carbon Adsorption Handbook. Ann Arbor
Science Publishers Inc., Ann Arbor, Mich.
3. Eberhardt, M., S. Madsen and H. Sontheimer. 1974. Unter-
suchungen zur Verwendung biologisch arbeitender Aktivkohle-
filter bei der Trinkwasseraufbereitung. Heft 7, Engler-
Bunte-Institut der Universiteft Karlsruhe, Fed. Republic
Germany.
4. Benedek, A. 1977. The effect of ozone on activated carbon
adsorption - a mechanistic analysis of water treatment data.
Presented at 101 Symposium on Advanced Ozone Technology,
Int. Ozone Institute, Toronto, Ontario.
5. Tien, C. 1980. Bacterial growth and adsorption in granular
activated carbon column, Ln Suffet, I.H. and M.J. McGuire,
ed. Activated Carbon Adsorption of Organics from the
Aqueous Phase, Proceedings of the 1978 ACS Symposium in
Miami Beach, FL. Ann Arbor Science Publishers, Ann Arbor,
MI. In Press.
6. Ying, W.C. and W.J. Weber. 1978. Bio-phyBiochemical
adsorption systems for wastewater treatment: Predictive
modeling for design and operation. Presented at the 33rd
Purdue Industrial Waste Conference, Purdue University.
7. Weber, W.J., M. Pirbazari and G.L. Melson. 1978. Bio-
logical growth on activated carbon: An investigation by
scanning electron microscopy. Envir. Sci. and Tech. 12,
817-819.
8. Marshall, K.C. 1976. Interfaces in Microbial Ecology.
pp. 44-47, Harvard University Press,Cambridge, MA.
786
-------
9. Marshall, K.C., R. Stout and R. Mitchell. 1971. Selective
sorption of bacteria from seawater. Cana. J. Microbio. 17,
1413-1416.
10. Pipes, W.O. 1976. Biodegradation of taste and odor com-
pounds. In Shapley and Kaplan, eds. Proceedings of the 3rd
International Biodegradation Symp., pp. 651-657, Applied
Science Publishers, Ltd., London.
11. Van der Kooij, D. 1978. Investigations concerning the
relation between microorganisms and adsorption processes
in granular carbon filters. Presented at the Conference
on Oxidations Techniques in Drinking Water Treatment,
Karlsruhe, 11-15 September 1978. U.S. Environmental Pro-
tection Agency Report No. EPA 57019-79-020.
12. Klotz, M., P. Werner, and R. Schweisfurth. 1976. Investi-
gations concerning the microbiology of activated carbon
filters. In H. Sontheimer, Ed. Translation of Reports on
Special Problems of Water Technology Volume 9 - Adsorption.
pp~;312-330.Municipal Environmental Research Laboratory.
U.S. Environmental Protection Agency Report No. EPA-600/9-
76-030.
13. Prober, R., J.J. Pyhea and W.E. Kidon. 1975. Interaction
of activated carbon with dissolved oxygen. AIChE Jour.
21(6):1200-1204.
14. Weber, W.J. 1972. Physicochemical Processes for Water
Quality Control, pp. 239. Wiley-Interscience, New York, NY.
15. Davies, R. 1963. Microbial extracellular enzymes, their
use and some factors influencing their formation. In
Rainbow and Rose, eds. Biochemistry of Industrial Micro-
organisms. Academic Press, New York, N.Y.
16. Corpe, W.A. and H. Winters. 1972. Hydrolytic enzymes of
some periphytic marine bacteria. Cana. J. Microbiol. 18,
1483-1490.
17. Guirguis, W., T. Cooper, J. Harris and A. Ungar. 1978.
Improved performance of activated carbon by pre-ozonation.
J. Water Poll. Control Fed. 5, 308-320.
18. Weber, W.J. 1978. Discussion of: Improved Performance
of activated carbon by pre-ozonation. J. Water Poll.
Control Fed. 50, 2781.
19. Randtke, S.J. 1978. Discussion of: Improved performance
of activated carbon by pre-ozonation. J. Water Poll.
Control Fed. 50, 2602.
787
-------
20. Gulp, R.L. and S.P. Hansen. 1978. Carbon adsorption
enhancement with ozone. Presented at the 50th Annual Water
Pollution Control Federation Meeting, Anaheim, California.
21. Sontheimer, H., E. Heilker, M.R. Jekel, H. Notte and F.H.
Vollmer. 1978. The Mulheim Process. J. Am. Water. Works
Assoc. 70, 393-396.
22. U.S. Environmental Protection Agency Report. Interim Treat-
ment Guide for Controlling Organic Contaminants in Drinking
Water Using Granular Activated Carbon. (Compiled and Edited
by James M. Symons). Water Supply Research Division, U.S.
EPA, Cincinnati, Ohio (Jan. 1978).
23. Kuhn, W., H. Sontheimer, L. Steiglitz, O. Maier and R. Kurz,
1978. Use of ozone and chlorine in water utilities in the
Federal Republic of Germany. J. Am. Water Works Assoc. 70,
326-331.
24. Peel, R. and A. Benedek. 1975. The modelling of acti-
vated carbon adsorbers in the presence of bio-oxidation.
Presented at the AIChE 68th Annual Meeting. The Design
of Carbon Adsorption Systems Session. Los Angeles, CA.
Nov. 16-20.
788
-------
CONSIDERATIONS IN THE SEQUENCING OF
CHEMICAL OXIDATION AND ACTIVATED CARBON ADSORPTION
A. Benedek
INTRODUCTION
Chemical oxidation and adsorption are the two primary
methods of organic removal in current water treatment practice.
More specifically, in many newer plants, the flow scheme incor-
porates an ozonation step followed by a granular activated carbon
adsorption step. It is difficult to predict, on a theoretical
basis, whether the combination of ozonation followed by activated
carbon adsorption is indeed the optimum way of combining these
two methods.
The advantages and disadvantages of ozonation prior to acti-
vated carbon contacting have been recently reviewed by Benedek
(1979). He listed the advantages of pre-ozonation as: precipi-
tation and subsequent removal by filtration of reduced iron and
manganese ions when present, and increased organic removal and
activated carbon run length due to the increased biodegradability
of ozonated organics (assuming that biodegradable organics are
removed by biodegradation and not adsorption in granular
activated carbon contactors).
The disadvantages of pre-ozonation are as follows: lower
adsorptivity on activated carbon of ozonated molecules; loss of
adsorptive capacity due to the effect of residual ozone on acti-
vated carbon surface; and potential problems with bacterial slugs
in the activated carbon contactor effluent due to the bacterial
growth promoted by pre-ozonation.
The combination of granular activated carbon treatment
followed by ozonation would eliminate all of the above noted dis-
advantages of pre-ozonation and, furthermore, would ensure disin-
fection of the activated carbon effluent. On the other hand, it
would not aid biological activity in the contactor and may pre-
cipitate inorganic hydroxides at a stage in the process where no
further filtration is likely to occur.
In the case of groundwater or bank-filtered water, the pre-
cipitation of iron and manganese is an important requirement in
the treatment. Thus, in the Federal Republic of Germany, where
789
-------
bank filtration is a common method of pretreatment, the combina-
tion of ozonation followed by activated carbon adsorption is the
accepted treatment combination. In France, on the other hand,
surface waters are commonly pretreated by coagulation and sedi-
mentation similar to the common North American practice. In
such cases, the choice between pre-ozonation and post-ozonation
is much more complex. furthermore, pretreatment requirements
such as prechlorination also become important variables
THE WATER TREATMENT PLANT AT MORSANG-SUR-SEINE, FRANCE
Motivation for Pilot Studies
In recognition of the above factors, the operators of the
Morsang-sur-Seine, France, water treatment plant, the Societe
Lyonnaise des Eaux et de L'Eclairage (SLEE), in conjunction with
the designers of the plant (the Degremont Company), decided to
undertake long term studies on the optimum combination of ozona-
tion and activated carbon adsorption. As shown in Figure 1, four
parallel treatment sequences are operated at this plant. These
four processes are chosen in such a way that they enable an eval-
uation of the relative cost of powdered versus granular carbon,
the relative merits of pre-ozonation versus post-ozonation, and
the effect of prechlorination.
The first three sequences are full scale, operations and have
been operated in the indicated mode over the last three years.
The fourth sequence has been started up in late 1977 and, there-
fore, has only little over one year of operational data at the
time of writing.
An initial summary of operating results has been presented
by Richard and Fiessinger (1977). Subsequently, the author spent
some time in France working with Messrs. Richard and Fiessinger
on a mechanistic study of activated carbon contactor behavior of
treatment sequences 2 and 3. The results of this work are de-
scribed in detail by Benedek (1979). This paper serves to review
previous work, update the earlier data with more recent results
(particularly in connection with the fourth treatment sequence),
and to recommend, on the basis of the Morsang results, the optimum
combination of ozone and activated carbon contacting.
Description of Full-scale Treatment Plant
The Morsang plant draws its water from the Seine, about
30 km upstream (south) of Paris. At this point, the Seine is
still free of most of the pollutants generated by human activiti-
ties in and around Paris.
The plant is constructed in an aesthetically pleasing tri-
pointed star (or winged) pattern (See Figure 2). The first
"wing" of the plant was constructed in 1970 and has always been
790
-------
RAW WATER
MACROSCREENING
• CHLOR1NATION
COAGULANT ADDITION
J
FLOCCULANT ADDITION
TO WING 1
PAC ADDITION
CLARIFICATION WITH
POWDERED
ACTIVATED CARBON
SAND FILTRATION
TO WING 2
TO PILOT
PLANT
I
COAGULANT
ADDITION
I
FLOCCULANT
ADDITION
I
CLARIFICATION CLARIFICATION
WITHOUT WITHOUT
ACTIVATED CARBON ACTIVATED CARIiON
I 1
SAND FILTRATION SAND FILTRATION
.-^
*^^
^^
OZONATION
FILTRATION ON
OZONATION GRANULAR
ACT I BATED CARBON
\
\
TREATMENT TREATMENT
SEQUENCE 1 SEQUENCE 2
1 .
rf
1
FILTRATION ON
GRANULAR
ACTIVATED CARBON
OZONATION
t
TREATMENT
SEQUENCE 3
1
\
GRANULAR
ACTIVATED
CARBON
FILTRATION
1
OZONATION
|
TREATMENT
SEQUENCE 4
1
• POST CHLORINATION
TO DISTRIBUTION RESERVOIR
Figure 1. Treatment Schematics of the Morsang Water Treatment Plant
Wing 1
Wing 2
Pulsator
Sand Filters
Ozcne Chamber
Suporpulsator
Sand Filters
Carbon Filters
Ozone Monitoring
Central Operating Bldg.
i i
i 1
Future Wing 3 •
Figure 2. The Three-Painted Star Layout of the Morsang-sur-Seine Plant
791
-------
operated according to process sequence 1 in Figure 1. The second
wing went on stream in October 1975. Initially, the entire wing
was operated in process sequence 2 mode (see Figure 1); however,
6 months later, half of the wing was switched to process sequence
3. As already indicated, sequence 4 was started up in the fall
of 1977. The various sequences are designed primarily to deter-
mine the optimum treatment sequence for the future wing 3, as
well as a second future three-pointed star.
Wings 1 and 2 process an average of 40,000 m /d of water
each. Thus, approximately 20,000 m /d of water is treated by
processing sequence 2 and 20,000 m /d is treated by processing
sequence 3. Processing sequence 4 treats only 5.2 m /d of water
in separate pilot scale equipment.
The average chemical dosages at the plant during the past
year are summarized in Table 1.
Table 1
Chemical Dosage at the Morsang Plant
ChemicalUseDose
Chlorine Prechlorination 2.0 mg/1 as C12
Alum Coagulation 40 to 80 mg/1 as
(with alum) A12(S04)|18H20
Activated Silica Flocculant 2 mg/1 as SiO_
Powdered Activated Adsorbent 7 mg/1
Carbon (PAC)
Ozone Chemical to a residual of
oxidation 0.4 mg/1 (typically
the applied dosage is
1.2 mg/1).
The ozone is provided by the 6.6 kg/h electric discharge
units (Degremont Company). A contact time of 10 minutes is pro-
vided for ozonation.
^Full scale activated carbon contacting is accomplished in
Deqremont Model T Aquazur filters. These filters are specially
modified to allow air/water backwash of the 1.3 m deep layer of
granular activated carbon (GAC). The 34 tons of activated carbon
placed in each filter is type F400 (Chemviron Ltd., Brussels,
792
-------
Belgium). The average superficial contact time in these filters
is 16 minutes at a feed velocity of 5 m/h.
The pilot scale activated carbon contactor contains 26.8 kg
of F400 activated carbon. The pilot plant operating conditions
are similar to those of the full-scale contactors (superficial
contact time of 17 minutes at a feed velocity of 4.5 m/h).
DATA COLLECTION AND ANALYSIS
Water samples have been collected on a monthly basis since
1975, after each contacting mode. On each of these samples,
water quality analyses have been done in the laboratories of the
Degre'mont Company, Rueil, France, and/or in the laboratories of
SLEE at Le Pecq, France. Activated carbon samples were usually
removed from the filters by direct insertion of a cup sampler.
Four separate analyses were used to determine the organic level,
namely ultraviolet absorbance, chemical oxygen demand (COD),
total organic carbon (TOC), and threshold odor number (TON).
Ultraviolet adsorbance was determined at 253.7 nm wavelength in
a 10 cm cell using a Beckman Acta III spectrophotometer. COD was
measured by KMnO. titration in acidic medium, and TOC was
measured by a TOCSIN analyzer (manufactured by Phase Separation
Ltd., United Kingdom). TON was determined by successive dilu-
tions of the original sample until taste of any kind disappeared.
The required number of times the original sample is diluted is
then expressed as the TON. Additional experimental details are
available in the articles by Richard and Fiessinger (1977), and
Benedek (1979).
ORGANIC REMOVAL RESULTS
Table 2 details the overall organic removal observed at this
plant during the first 9 months of operation of the four parallel
treatment sequences. As shown in Table 2, all four treatment
sequences produced a high caliber water, low in taste and organic
content as well as trihalomethanes. Treatment sequences 2 and 4,
using pre-ozonation, produced the highest odor and collective
organics parameter removal, with sequence 3 very close behind.
The poorest quality water is produced by sequence 1 using PAC in-
stead of GAC. The greatest overall organic removal was observed
in sequence 4. The removal here, however, cannot be directly
compared to other processes because the granular activated carbon
in process sequences 2 and 3 has served over 2 years longer than
the GAC in sequence 4. In terms of trihalomethanes, sequence 4,
without prachlorination, produced the lowest level, as expected.
The other three sequences averaged 27+4 ug/1 of trihalo-
me thanes.
The removal of organics by the different processing steps
in the four treatment sequences is summarized in Tables 3 and 4
for COD and TON, respectively. As shown in Table 3, the bulk of
793
-------
Table 2. A Comparison of the Organic Removal Effectiveness
of the Four Treatment Sequences
Parameter
TOCa
Absorbance
CODKMn04
TONC
Trihalome thanes
Units
mg/1
-
mg/1
-
pg/i
Raw
Water
3.0
0.953
3.80
12
0.8
1
1.32
0.198
1.6
2
26
Sequences
2 3
1.12
0.142
1.23
NDe
31
1.15
0.174
1.26
1.5
24
4
0.95
0.03
0.68
NDe
1
aTOC is total organic carbon
Absorbance is ultraviolet absorbance determined at 253.7 nm
wavelength in a 10 cm long cell.
c
TON is threshold odor number
Trihaloraethanes is essentially the chloroform level in this
plant.
!ND me<
a lack of odour in the undiluted sample.
eND means not detected, i.e., in the case of TON, ND implies
the organic substances are removed in the processing steps
intended primarily for turbidity removals-coagulation, clarifi-
cation, and sand filtrati6n-*-and precnlorination (in sequences 1
to 3). Interestingly, the greatest COD removal during the tur-
bidity removal steps occurs in sequence 4, possibly due to the
fact that lack of prechlorination allowed some biological degra-
dation to proceed during the process steps. Ozone appears to
remove very little COD at the low applied dosages of this study.
Granular activated carbon contacting also removes relatively
little COD with the exception of sequence 4 where, as discussed
earlier, the carbon has had a low service time.
In the case of odor removal (Table 4), once again the first
four, primarily turbidity removal steps, are credited with elimi-
nating the bulk of the odor in sequences 1 to 3. In sequence 4,
the lack of prechlorination means that only 50 percent of the
odor was eliminated by the first three steps. On the other hand,
in sequence 4 ozonation more than makes up for the odor removed
by break point chlorination in the other processing steps.
Activated carbon plays an important role in taste removal as in
794
-------
Table 3. Removal of ?Mn0J4 *n m9/l Across Various Types of
Process Steps in tne Different Treatment Sequences
Process
Coagulation, Clarification
plus Sand Filtration
Ozonation
Carbon Adsorption
Treatment
1 2
1.6* 1.44
-0.1 0.13
0.33
Sequence
3
1.43
0.21
0.23
4
2.0
0.0
1.12
*Includes the effect of powdered carbon adsorption as well
Table 4. Reduction of Threshold Odor Number Across various
Types of Process Steps in the Different Treatment
Sequences
Treatment Sequence
Process
Coagulation, Clarification
plus Sand Filtration
Ozonation
Carbon Adsorption
123
9*f 8.5f 8.5f
1 0.5 -1.5
X5 . ^3.5
4
6
4
>2
*Includes the effect of powdered carbon adsorption as well
as prechlorination-
"^Includes the effect of prechlorination
795
-------
every case, after GAG contacting, the water became tasteless.
Interestingly, ozone reintroduced some taste to the water in
sequence 3.
On the whole, the above data indicate that sequences 2 and
4, employing pre-ozonation, seem to produce the best treated
water quality, especially when odor removal is of key importance
to consumer acceptance (as is the case for this particular
plant).
ORGANIC REMOVAL MECHANISMS DURING CARBON CONTACTING
The mechanism of organic removal during carbon adsorption
has been examined in detail by Benedek (1979) for this plant.
The cumulative removal curves for the granular carbon contactors
of sequences 2 and 3 are plotted in Figure 3. The slope of such
cumulative plots is equal to the percent removal attained in the
contactor. According to this figure, the pre-ozonated carbon
contactor of sequence 2 removes slightly more than the post-
ozonated carbon contactor of sequence 3.
As indicated in the introduction, ozonation is expected to
decrease adsorptivity. This is clearly illustrated in Figure 4,
where the isotherms of ozonated and non-ozonated water samples
are compared. Thus, it can be hypothesized that biological, as
well as adsorptive mechanisms, must be involved in the removal of
organics in these contactors. A small program was subsequently
undertaken to examine bacterial activity in the contactors.
Figure 5 shows that high levels of viable cells were present in
both contactors; however, the contactor treating ozonated water
harbored nearly ten times as many bacteria. Interestingly, the
highest bacterial levels were found near the surface of the
contactors in spite of the fact that some residual bactericide
(chlorine or ozone) may be present in the feed. This was ex-
plained in a recent review article by Benedek (1978) in terms of
bacterial habitats having been established within the activated
carbon macropores, wherein the bacteria were expected to be
protected from chemical bactericides of an oxidizing nature.
Additional research described by Benedek (1979) indicates
that bacterJlal respiration rates further confirm the viable cell
bacterial distributions data of Figure 5. Thus, significantly
higher oxygen uptake rates were observed in the preozonated con-
tactor and the oxygen uptake rates generally declined with depth.
On the basis of the data in Figures 3 to 5, bacterial oxidation
is the most likely explanation for the operating results during
the final straight line region of the results reported in
Figure 3.
796
-------
.12
.11
.10
o-09
* ^ X
/A/
^A/
•
/* • WITH Os
« A WITHOUT Os
»'
/
/
' i i i i 1 i i i i 1 i i i i 1
1 .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24 .26 .28 .30
gm TOC APPLIED / gm AC
Figure 3. The Effect of Ozone on Activated Carbon Adsorption at
Morsang-sur-Seine, France. (From Benedek, 1979.)
30
20
10
°c
10 r-
(•) TOC (b) COD KMnQ4 7
/ 3°!- 9
/ I'
' / "' / 1*
r r /
• 7 / .= • / /
I./- . 4/^*-
¥ P
> 1 2 "0 1 2 .3
mgKMn04-COD/l
. WITH 03 |
(c) uv-Absorbance
; S
*s/^/^
-/*
.iOcm
i i i i i i i i i i i i i
31 .02 .05 .10 .2 .3 .4 .9
Figure 4. Morsang Isotherms in Terms of a) TOC, b) COD-KMnO. and
c) UV Absorbance. (Q-Carbon Equilibrium Loading; CE~
Equilibrium Liquid Concentrations) (From Benedek, 1979).
797
-------
_J
u
u
ui
u
4
00
10° P
10'
I08
10"
WITH 0,
WITHOUT 03
40
80
120
CARBON BED DEPTH - cm
Figure 5. The Effect of Ozone on Bacterial Distributions in Activated
Carbon Contactors Treating Prechlorinated, Coagulated, Settled
and Prefiltered Seine Water. (From Benedek, 1979).
MATHEMATICAL MODELLING OF BIOLOGICALLY ACTIVE CARBON CONTACTORS
A biologically active carbon contactor can be modelled through
the use of a differential mass balance in the column, whereby
where R = biological reaction rate per unit surface
area of the carbon particle
d = carbon particle diameter
p = mass of carbon per unit volume
1-e - volume fraction of carbon in contactor
v = superficial velocity
s
c = organic substrate concentration
z = contactor bed depth
q = average weight of substrate per weight of GAC in
the carbon particle
The above mass balance is only applicable to thin biological
films. In the case of water treatment plant contactors, the
films are extremely thin, and the carbon surface is believed to
798
-------
be only partially covered (Klotz, et al.). Thus, the effect of
film thickness in changing particle size is neglected in equation
1 and the reaction rate is based purely on the external surface
area of the carbon particles.
To solve the contactor mass balance, as outlined by Maqsood
and Benedek (1977), additional equations, such as equilibrium and
mass transfer kinetic expressions, are needed. The major experi-
mental model inputs are: adsorption isotherm(s), batch adsorp-
tion kinetic data, and biological reaction rate (R_).
For Morsang, the collection and evaluation of the input
data was discussed by Benedek (1979). The value of R, at
Morsang, as obtained from thegconstant slope region or Figure 3,
was evaluated to be 6.5 x 10"" gTOC removed • cm", »hr for non-
ozonated water, versus 7.6 x 10~ g TOC«cm~ «hr~ for ozonated
water. The relative magnitude of this biological rate, as com-
pared to other types of water, is revealed by an examination of
Table 5. In general, the rates at Morsang are similar to other
potable water treatment data, and water treatment contactor rates
are generally well below the rates obtained with the more
concentrated and degradable organics found in wastewater.
The operation of the non-ozonated (sequence 2) contactor
can now be compared to that of the pre-ozonated contactor
(sequence 3) through the'model. As shown in Figure 6, the con-
tactor without pre-ozonation initially removes a greater fraction
of the applied organics. This greater removal primarily occurs
in the range of 0.02 to 0.05 g applied TOC/g GAG, where adsorp-
tion continues to occur in the contactor without preozonation.
Subsequently, however, the ozonated contactor removes slightly
more of the organics by biological action and the pre-ozonated
contactor appears superior. Figure 6 points out that post-
ozonation is favored at high carbon usage or short regeneration
time systems. Pre-ozonation is, of course, favored in systems
with long carbon regeneration times. Up to now, most of our
contactors were regenerated at periods greater than t«*o years.
Thus, pre-ozonation appears to be favored from an oifganic stand-
point in such systems. On the other hand, post-ozonation may be
best for the near complete synthetic organic compound removal
that may be required for health protection, as in such cases,
regeneration frequencies may turn out to be as low as once per
month.
CONCLUSIONS
The following conclusions can be drawn from this study:
(i) Pre-ozonation of GAC contactors is favored for odor
removal, especially when prechlorination is not
practiced.
799
-------
Table 5. Observed Biological Reaction Rates, Expressed on the Basis of
External Activated Carbon Surface Area (From Benedek, 1978)
Type of Water
Presence
of
Oxygen
Specific Reaction
Rate
g TC
Chemically clarified and filtered
domestic wastewater
Anoxic
2 x 1Q-6 at 25°C
1 x IQ"6 at 5 3C
00
o
o
Chemically clarified and filtered
domestic wastewater
Chemically clarified domestic
wastewater with nitrate addition
Chlorinated, chemically clarified
and sand filtered Seine water
(at Morsang)
Chlorinated, chemically clarified,
sand filtered and ozonated Seine
water (at Morsang)
Chemically clarified, sand filtered
and ozonated Rhine water
Synthetic phenol solution
Synthetic phenol solution
Anoxic
Anoxic
Aerobic
io"6
Aerobic 7 x if)""8
8 x io~8
Aerobic 3-10 x io~8
Aerobic 2 x 10~4
Anoxic 5 x 10~5
-------
00
o
.20
.It
CJ
CE
.1?
UJ
UJ
a:
u
o
.OS
CO
21
CD
.04
.02
0.00
0.00
Figure 6
CMS TOC RPPLIEC / GM flC
. A Comparison of Ozonated and Non-Ozonated Adsorbers with
Biological Activity (Ozonated Rf = 7.6 x 10~ , Non-
Ozonated 6.5 x 10""8 g TOC/cnT hr.) (From Benedek, 1979)
-------
(ii) In terms of organic removal, pre-ozonation is favored
at low carbon regeneration frequency, but at high
frequencies, post-ozonation is probably favored.
It also appears that prechlorination tends to reduce overall
organic removal; however, this conclusion will require further
long-term pilot studies for substantiation.
ACKNOWLEDGEMENTS
Many people have been instrumental in making this paper
possible. Mr. Richard of Degre'ment and Mt. Fiessinger of SLEE
gave encouragement and advice. In addition, at Degremont,
Messrs. Brener, Conan and Le Baron; at SLEE, Mmes. Rizet and
Dalga; and at McMaster, Messrs. Banesi and Peel helped to carry
out different aspects in the study. Financial support for the
author's European stay came from McMaster University and the
Department of Health and Welfare, Government of Canada.
REFERENCES
Benedek, A. 1978. Simultaneous Biodegradatioh and Activated
Carbon Adsorption - A Mechanistic Look. Paper presented at
the ACS Symposium on "Activated Carbon Adsorption of Organics
from Aqueous Phase," Miami Beach.
Benedek, A. 1979. The Effect of Ozone on Activated Carbon
Adsorption - a Mechanistic Look. Ozonews, 6^, (1), Part 2
Klotz, M., Werner, P. and Schweisfurth, R. 1976. Investiga-
tions Concerning the Microbiology of Activated Carbon
Filters. In Translation of Reports on Special Problems of
Water Technology, Sontheimer, H. ed. , U.S. Government Report
EPA 600/9-76-030.
Maqsood, R. and Benedek, A. 1977. "The Effect of Low Tem-
perature on Organic Removal and Denitrification in Activated
Carbon Columns", J. Wat. Pollut. Contl. Fed., 49, 2107.
Richard, Y. 1979. Optimisation des Chaines de Traitement,
Rapport 1574 R, Societe Degremont, Rueil, France.
Richard, Y. and Fiessinger, F. 1977. Emploi Complementaire
des Traitements Ozone et Charbon Actif. Proceedings of the
Paris Meeting of the International Ozone Institute.
802
-------
QUESTIONS AND ANSWERS
(after Dr. Benedek)
DR. RIP RICE: I would like to comment on the detrimental
effects of ozonation on the adsorption of the total organics,
measured as DOC. First, one should say that ozone will decrease
adsorbability of those organics which it does oxidize; but for
those that it does not oxidize, such as some of the halogenated
compounds, it does not decrease adsorbability. Secondly, there
are some data from the full scale plant operation at DUsseldorf
showing that both adsorption and biological degradation occur
s imultaneously.
f*
V PROF. WALTER MAIER, University of Minnesota: Please com-
ment on the temperature in the columns that you have examined.
A PROF. ANDREW BENEDEK: We didn't make an attempt to study
temperature in this particular case, but we had an earlier paper
which was published in JWPCF about 2 years ago. The purpose of
that particular paper was to look at the effect of temperature:
we found that decreasing temperature increases adsorptivity but
decreases biological degradation. Initially, the adsorptivity
was important. Subsequently, since biological degradation
became the key mechanism, the separation and slope corresponded
to what you would expect from biokinetics.
Q MR. MAIER: I am curious whether or not you would look at
temperature in your modeling. Obviously, in filter operations,
temperature had been used to establish the weight limitations
that accrued from diffusion limitations as opposed to the bio-
logical, because the effect of temperature is greater on the
biological system. I just wondered whether the same analogy
could be applied here.
A PROF. BENEDEK: The model has every physical property and
the biokinetic coefficient, as well as temperature, modulated.
Q PROF. DR. SONTHEIMER: A very silly question, maybe. I
want to ask Dr. Benedek if he really knows enough about BAC to
make modeling worthwhile?
803
-------
A PROF. BENEDEK: I don't know how to answer that. The
problem is that we know so much about it. We are trying to come
to grips with it, but I have no illusions of understanding it
fully. I know how to describe it mathematically, and we're
making improvements in our description.
Q PROF. MICHAEL SEMMENS, University of Minnesota: I had a
question concerning the length of time it takes to establish the
biological community on the activated carbon. If you put a
freshly reactivated granular carbon into a filter and put it in
service, what length of time might you expect before the bio-
logical activity reaches its peak?
A PROF. BENEDEK: That's a very good question. I didn't have
time to show data on the bacteria that build up in the carbon
column as a function of time. There have been data published by
us and various people in Europe, which indicates that this is
very rapid. From our data in France, when you put a fresh new
generative carbon contactor in service virtually all the bacteria
initially adsorb. It is then strictly a function of the amount
of bacteria in the feed and the final equilibrium that you have,
adsorptive equilibrium if you like, between the feed and what's
on the carbon.
Q PROF. FRANCIS DiGIANO: May I just add a little bit to that.
I think probably just as important, Mike, is what are those Bac-
teria doing? I mean, they may be there as Dr. Benedek mentioned;
they're adsorbed from the feed stream, but if the substrate isn't
available in sufficient amounts, how effective are they in the
system?
A PROF. BENEDEK: Fran, I agree wholeheartedly. I would add
that it is difficult to know when biodegradation becomes impor-
tant if you have adsorption and biodegradation going on simul-
taneously. It's hard to separate the two.
804
-------
HUMIC ACID REMOVAL WITH MACRORETICULAR ION EXCHANGE RESINS
AT HANNOVER
Walter Kolle
THE FUHRBERG WATERWORKS PLANT
The Fuhrberg waterworks plant of the Hannover waterworks
covers about 30 percent of the drinking water supply for the city
of Hannover, corresponding to an annual output of about 18 mil-
lion cubic meters. At Fuhrberg, a high-color groundwater is used
which contains many high-molecular-weight components, especially
humic acids. The conventional treatment methodology used to
handle this water quality problem was: aeration, flocculation
(precipitation of iron and of calcium carbonate by adding calcium
hydroxide and activated silica), potassium permanganate addition,
and filtration. Chlorination was carried out at the flocculators
to support flocculation, at the filters to keep them sterile and
at a proper pH, and at the clearwell to eliminate ammonia by
exceeding the breakpoint. The total chlorine dosage amounted to
about 19 mg/1. Typical quality criteria of the raw and finished
water are summarized in Table 1.
The disadvantages of the finished water include a high
concentration of dissolved organic carbon, 6 mg/1, and the simul-
taneous presence of 10 mg/1 of oxygen. This enabled a rapid
aerobic bacterial growth which was characterized by a doubling
time of the bacterial counts during the logarithmic growth phase
of 5 hours (temperature = 10°C). Dealing with this problem is
difficult because organic carbon causes a substantial chlorine
demand.
Keeping the finished water sterile during distribution was
difficult and resembled a running match between chlorine consump-
tion and bacterial growth on the one hand, and control measures
like flushing or subsequent chlorination in the distribution
system on the other hand (1). In addition, the total organic
chlorine and haloforms resulting in the finished water were a
hygienic disadvantage which could not be tolerated (2,3,4). The
haloforms formed during treatment consisted mainly of chloroform.
Attempts to reduce the humic acid content by conventional,
economically acceptable processes (e.g., additional flocculation)
were unsuccessful. Only the use of a strong basic macroreticular
805
-------
ion exchange resin offered a solution to the problem (5-8). Ex-
tensive tests were carried out with Lewatit MP 500 A (Bayer
Chemical Co.)/ which has been used in the full scale operation
of the plant since July 1978.
Table 1. Analytical Characterization of the Conventional
Water Treatment at the Fuhrberg Water Works Plant
Odor
Oxygen (mg/1)
Iron (mg/1)
Manganese (mg/1)
Ammonia (mg/1)
Sulfate (mg/1)
Chloride (mg/1)
Dissolved org. carbon,
DOC (mg/1)
Chemical oxygen demand,
COD (mg/1)
Total organic chlorine,
TOC1 (yg/1)
Total haloforms lyg/1)
Raw Water
H2S
not detect.
15
1.1
1.1
130
45
9
27
not detect.
not detect.
Finished Water
not detect.
10
<0.05
<0.02
<0.05
130
64
6
17
400
69
THE EFFECT OF RESIN TREATMENT
Strong base anion exchange resins for humic acid removal are
generally used in their chloride form and are regenerated with a
solution containing 10 percent sodium chloride and 2 percent
sodium hydroxide. The resin is able to remove about 85 percent
of the humic acids from the Fuhrberg drinking water. However,
according to the normal behavior of anion exchangers in the
chloride form, chloride exchanges for sulfate. The rate at which
the exchanger transforms into the sulfate form depends on the
sulfate content of the drinking water to be treated. In the case
of the Fuhrberg drinking water, this transformation is finished
after a throughput of about 300 bed volumes. After that, the
mean humic acid removal is scarcely better than about 50 percent,
but this efficiency decreases only slightly during the remaining
operation cycle. After a throughput of about 5,000 bed volumes,
regeneration of the resin is necessary. At this point, the resin
is loaded with humic acids to the extent of 15 g of DOC per liter
of resin. The behavior of the resin is evident from Figures 1
and 2.
806
-------
6
5
o
c
o
o
DOC input •
Cl"(meq./l)
SO* (meq./l)
0 100 200 300
Throughput (bed volumes)
400
Figure 1. Changes of Chloride, Sulfate and DOC in the
Filtrate of the Resin, Immediately after a new Cycle has Started
(preceding regeneration with unused regeneration solution)
KMnO, -consumption
0 2000 4000
Throughput (bed volumes)
6000
Figure 2.
Decrease of the Humic Acid Removal (as DOC and as
KMnO^-consumption) during an Operation Cycle.
Throughput: 54 bed volumes per hour
807
-------
Numerous tests indicated that the resin's efficiency for
humic acid removal is almost independent of the filtration rate.
This held true for filtration rates up to about 60 bed volumes
per hour or 70 meters per hour.
Laboratory tests indicated that the following improvements
of the water quality could be expected:
1. Decrease of organic carbon and organic chloride.
2. Increase of the bacterial doubling time from 5 to about
12 hours.
3. Decrease of the consumption of free chlorine, approximately
proportional to the decrease of organic carbon and strongly
dependent on the decrease of chemical oxygen demand.
THE RESIN TREATMENT PLANT
The results obtained in extensive tests at a pilot plant and
in the laboratory were the basis for constructing an additional
treatment unit in the full scale range at the Fuhrberg waterworks,
This project is sponsored by the German government. The design
data of the plant are: 4 filters with 12.5 m resin, each;
throughput 625 m /h each3equivalent to 50 bed volumes per hour;
operation cycle 62,500 m , corresponding to 4 days. The total
throughput amounts to 2,500 m /h, which is 60 percent of the
total design throughput of the Fuhrberg waterworks. Thus, during
peak water demand situations, the surplus of water beyond design
capacity by-passes the resin filters.
Strong base anion exchange resins may detach trimethy1amine
from the production process into the water. This compound is
hygienically dubious because of the nitrosamine problem. Gas
chromatographic analyses showed that the previously unused resin
is leached according to an exact exponential law with a "half-
life" of the trimethylamine concentration of a few hours, depend-
ing on the operating conditions. This behavior of the resin
could be verified down to the detection limit of the analytical
method, which was 0.5 yg/1. After reaching this level, tri-
methylamine remained below the detection limit at all operation
conditions and is, consequently, no problem at all.
REGENERATION OF THE RESIN
The regeneration of the resin is carried out with 2 bed
volumes of a solution containing 10 percent sodium chloride and
2 percent sodium hydroxide. The used solution is adjusted in
its content of sodium chloride and hydroxide and reused seven
times. The DOC of the solution increases to about 25 g/1 during
808
-------
reuse. Naturally, sulfate is also enriched in the solution,
reaching approximately the same level as the chloride concentra-
tion. Then the solution has to be discharged. It is interesting
that the sulfate enrichment in the regeneration solution affects
the behavior of the resin immediately after starting a new
operating cycle according to Figure 1. The efficiency during the
main period of a cycle seems to be quite independent from the
sulfate content.
The relation between the volume of treated drinking water
and the volume of spent regeneration solution is 20,000: 1.
Thus, the volume of the solution is small enough to be trans-
ported by truck for disposal. Resin treatment clearly has less
environmental impact than conventional treatment, even if the
solution is discharged into the sewage plant; the chloride impact
originating from resin regeneration is less than 25 percent of
the impact caused by the excessive chlorination that was previ-
ously necessary. In regard to the environmental burden of humic
acids, the only change occurs with the route by which they reach
the sewage plant.
ADDITIONAL MEASURES AT THE FUHRBERG WATERWORKS
Special attention was directed to reducing the formation of
haloforms. The first step of chlorine-saving was the replacement
of activated silica as a flocculation aid by a polyacrylamide
polymer. This guarantees excellent flocculation without the
additional need of chlorination. As a consequence, the filters
could also be operated chlorine-free. For pH-adjustment (origi-
nally an effect of chlorination) sulfuric acid had to be intro-
duced. Under these conditions, biological nitrification of
ammonia could start, although very slowly. Thus, an additional
portion of chlorine could be saved. In addition, the chlorine
content of the finished drinking water could be reduced because
of the treatment effect of the ion exchange unit. Every step of
lowering the chlorine dosage decreased the formation of halo-
forms. Table 2 illustrates how the haloform concentration was
influenced by the changes in chlorination.
Table 2 shows that the haloform problem no longer exists for
the Fuhrberg water, provided that the dosage of combined chlorine
can be maintained without resulting bacterial difficulties.
PRACTICAL EXPERIENCES
The installation of the new treatment plant was carried out
as a research program because the effects to be realized in prac-
tice cannot be anticipated by laboratory tests in a reliable way.
The interdependence among chlorination, organic substances, bac-
terial growth, manganese precipitation, and corrosion phenomena
within real distribution systems is extremely complicated.
809
-------
Indeed, extensive research was carried out in this context for
corrosion phenomena in the old portions of the Hannover distri-
bution system (9,10). In the case of the Fuhrberg water, no
significant effect of the humic acid removal on corrosion pro-
cesses has been observed. Apart from the remaining questions to
be answered by long term observation, the experiences with the
macroreticular resin treatment so far are quite positive.
Table 2. Haloform Formation as a Function of Chlorination
in the Fuhrberg Waterworks Plant
Date Changes in Chlorination Haloforms
(yg/D
Aug. 76 Conventional operation; Chlorination of 69
four flocculators; finished water:
0.45 mg/1 free chlorine; finished water
quality as in Table 1.
Feb. 78 Half of the flocculators and half of the 55
filters chlorine-free; finished water:
0.45 mg/1 free chlorine
Nov. 9, 78 All flocculators and filters chlorine-free; 36
resin treatment; finished water: 0.45 mg/1
free chlorine
Nov. 23, 78 Reduction of chlorine dosage; finished 20
water: 0.25 mg/1 free chlorine
Feb. 5, 79 No longer breakpoint Chlorination; 0.4
finished water: 0.30 mg/1 combined chlorine
ACKNOWLEDGEMENT
The program is sponsored by the Bundesministerium fu'r
Forschung und Technologie; the Hannover waterworks are grateful
for their support. In addition, the author wishes to thank
Dr. Richard Heck and Prof. Dr. Heinrich Sontheimer for their
intensive support and Dr. Ludwig Stieglitz for carrying out the
trimethy1amine analyses. Last but not least, the author thanks
his many colleagues of the Hannover waterworks who are cooperat-
ing in this program.
810
-------
REFERENCES
1. Heck, R.: Deterioration in the Physical and Chemical Quality
of Water in the Distribution System. International Water
Supply Association/ Tenth Congress, Brighton, 1974.
2. Kolle, W.: Anwendung makroporpser lonenaustauscher in der
Trinkwasseraufbereitung. Veroffentlichung des Bereichs und
des Lehrstuhls fur Wasserchemie, Heft 9, 345-353, Karlsruhe
1975.
3. Kolle, W.: Anwendung von makroporosen lonenaustauschern
und/oder Aluminiumoxid zur weitgehenden Entfernung von
Huminstoffen und Erprobung der Auswirkung derselben auf das
Verteilungsnetz. Neue Technologien in der Trinkwasserver-
sorgung, Statusseminar Hannover 1978, DVGW-Schriftenreihe
Wasser Nr. 101, 179-191.
4. Kolle, W. 1979. Resin Treatment Improves High Color
Groundwater. Water and Sewage Works, 126, 68-69.
5. Martinola, F., Richter, A.: Macroporous Ion Exchange Resins
as Regeneratable Adsorbents of Organics in Water and Solu-
tions. Proceedings 31. Internat. Water Conference, Pitts-
burgh, USA, 157-165 (1970).
6. Kiihne, G. 1977. lonenaustauscher fur die Behatidlung von
Trinkund Brauwasser. Brauwelt, 117, 1832-1834.
7. Ruffer, H., M8hle, K.A., Schilling, J. 1973. Versuche zur
Aufbereitung huminstoffhaltigen Oberflachenwassers. Vom
Wasser, 41, 243-276.
8. Schilling, J., Ruffer, H., Mohle, K.A. 1976. Versuche zur
Aufbereitung huminsaurehaltigen Oberflachenwassers. Vom
Wasser, 46, 199-220.
9. Kolle, W., Sontheimer, H. 1977. Untersuchungen zur
Schutzschichtbildung in Gussrohren. Vom Wasser 49,
277-294.
10. Kolle, W.: Corrosion in Drinking Water Systems. Interna-
tional Water Supply Association, Twelfth Congress, Kyoto
1978.
811
-------
INVESTIGATIONS CONCERNING THE MICROBIOLOGY OF GAC-FILTRATION
FOR DRINKING WATER TREATMENT
P. Werner, M. Klotz, and R. Schweisfurth
INTRODUCTION
In granular activated carbon (GAC) treatment of drinking
water, populations of microorganisms always develop on the
filter granules and are found in the filtrate. Prior to the
investigations reported here, understanding of microbiological
activity during activated carbon treatment was very limited.
This was because microbiological investigations in this field
were carried out by the waterworks using standard methods for
water examination, which can describe neither quantitatively nor
qualitively, the populations of microorganisms present (1).
The studies described here concerning the microbiological
activity on activated carbon used for drinking water treatment
were carried out for the most part at a water treatment plant on
the Rhine River. The water is taken from the river, aerated,
settled, chlorinated, and flocculated. The water then follows
rapid filtration through sand and activated carbon before it is
infiltrated into the ground. Using German Standard Methods
for the determination of colony counts, the water of the effluent
of the GAC-filters usually satisfies the drinking water quality
requirement.
QUANTITATIVE DETERMINATION OF THE POPULATION OF MICROORGANISMS
The determination of the number of bacteria in the water of
the activated carbon filters can be conducted using media either
rich or poor in nutrients. The only important condition for
high colony counts is the incubation of the samples for 7 days
at a temperature of 27°C (1). Ford observed that an incubation
period of 3 days was not sufficient; he therefore suggested to
extend it to 7 days (2).
812
-------
Figure 1 shows mean values of colony counts of some
characteristic stages in this waterwork. The colony counts of
raw water (approximately 2 x 10 /ml) are almost completely
eliminated by chemical pretreatment (high chlorination and
flocculation) before the GAC-filter. Without high chlorination,
the colony counts in the water at this sampling point decrease
only to 10 /ml. In the activated carbon filters, there is an
increase of bacteria to values of about 5 x 10 /ml water in the
filtrate. In contrast, the values generally are below 100/ml
using the method, normally applied, for the determination of
colony counts (German Standard Method: 2 days incubation at a
temperature of 22°C).
106 4 COLO NY COUNT/ml WATER
103-
102-
RAW
WATER
GAC-FILTER
INLET
W|TH-UITU
OUT]WITH
CHLORINB
OUTLET
Figure 1. Colony Counts at Different Stages of Treatment
(SPC-Agar, 7 Days Incubation at 27°C)
813
-------
On the activated carbon, one can find about 10 bacteria
per g wet weight. These colony counts can be obtained only by
the special method mentioned above; no standard method is
available for this application. Applying this special method,
one can get 20 percent of the living cell number and about 5
percent of the total cell number, including all dead and living
bacteria in the water of the activated carbon filters (Figure 2)
100 °/o
TOTAL CELL NUMBER
5%
COLONY COUNT
/SPECIAL METHOD!
20°/o
LIVING BACTERIA
COLONY COUN
{STANDARD METHOD)
Figure 2. Percentage Colony Counts in Effluent from
GAC Filter by Different Methods
Freshly filled activated carbon filters will reach the
mentioned high colony counts within a short time (Figure 3).
.The speed at which the populations of microorganisms develop
depends on the temperature. The higher the temperature, the
earlier the maximum colony counts are achieved. The fresh
filter material, which is initially sterile, is fed into the
empty vessel with water from the effluent of the activated
carbon filters; therefore, the GAC is contaminated by already
adapted microorganisms. This is the reason why there is no lag
phase at the beginning of the curves in Figure 3. The filter
velocities also have an influence on the speed of the develop-
ment of populations of microorganisms in the GAC filters.
The slope of the curves in the initial phase decreases with
increasing velocity. Moreover, it can be seen in Figure 3 that
the largest increase of colony counts takes place in the upper
part of the carbon column; therefore, the elongation of the
GAC filters (e.g., by connecting two filters in series) causes no
further increase of colony counts.
After the initial phase, the colony counts remain at a level
of about 5 x 10 /ml water if no manipulation takes place in the
pretreatment of the raw water.
814
-------
COLONY COUNTIml WATER
-300cm
-170cm
30
TIME (DAYS)
Figure 3. Colony Counts as a Function of GAC
Service Time and Bed Depth
(SPC-Agar, 7 Days Incubation @ 27°C)
815
-------
Investigations show that the magnitudes of colony counts
discussed above depend on the pretreatment of the water. If the
mass and degradability of the organic substances in the raw
water decrease because of enhancement of the biological activity
occurring in the pretreatment step, colony counts in the effluent
of the GAC filters decrease. In another water treatment plant
where slow-sand filtration of the raw water takes place before
activated carbon filtration, extremely low colony counts of the
water of the GAC filter effluents were found (3). The success
of slow-sand filtration can be explained in terms of degradation
of organic substrate that otherwise would support biological
activity during activated carbon treatment.
The observed maintenance of constant colony counts in an
activated carbon filter suggests the analogy to a chemostat. A
chemostat is a well-mixed vessel into which nutrient broth flows
at a constant rate. The quantity of inflowing nutrient broth is
responsible for the density of the suspension of bacteria leaving
the vessel. If the quantity of nutrient broth is constant, the
density of suspension of bacteria is also constant. However,
the nutrient concentration in an activated carbon filter is not
constant, because of the changing quality of the raw water, but
the organic substances are adsorbed by and desorbed from the
activated carbon in the steady state and can therefore be uti-
lized by the bacteria.
This interplay of adsorption and desorption results in a
relatively constant concentration of nutrients in the water of
GAC filters; hence, an activated carbon filter can be compared,
in a crude sense, with a chemostat.
Figure 4 is a typical picture of loaded activated carbon
from the waterwork. In order to investigate the distribution of
Figure 4. Bacteria on Activated Carbon
816
-------
microorganisms on the carbon granules, electron-scan microscopy
can be used after special treatment of the GAC (4,5). Only
single bacteria could be found on the coal; the area available
for bacteria (carbon surface including that of pores of diameter
> 1 urn) is only partially covered (less than 1 percent).
Electron-scan microscopic investigations of GAC from other
waterworks confirm this result.
QUALITATIVE DETERMINATION OF THE POPULATION OF MICROORGANISMS
As described earlier (Figure 1), the bacterial content of the
raw water and the GAC effluent is nearly the same. Between these
two treatment steps there is high chlorination which kills almost
all bacteria. Studies to characterize the two populations of
microorganisms from the raw water and from the GAC effluent were
carried out by methods of numerical taxonomy to find qualitative
differences between them. For this purpose a large number of
bacteria strains were collected from both sampling points to
determine, in as many as possible, their morphological and bio-
chemical properties. With the characterization of these prop-
erties, it was possible to determine the genera of the bacteria.
The properties of the various strains were compared statisti-
cally; strains with a high degree of similarity were grouped into
clusters. The expression "population" is defined as the sum of
all bacteria in a certain biotope.
It has been found that the bacterial population of the raw
water is more diversified than that of the filter effluent. This
can be seen not only in the number of different species of micro-
organisms, but also in the biochemical activities of the various
strains of bacteria. The cluster analysis shows that the
clusters formed by the sirains of the filter effluent are more
similar than those of the raw water (Figure 5). The fraction of
bacterial strains belonging to the genus Pseudonomas increases
from 55 percent in the raw water to 80 percent in the filter
effluent. It is obvious that, although both populations express
themselves in nearly equal colony counts, the population of
microorganisms found in the water of the GAC filters is sub-
stantially different from that of the raw water.
Furthermore, the results of the numeric taxonomical studies
showed that bacteria of different genera may indeed have similar
biochemical properties; consequently the classification of the
bacteria gives no reliable indication of the behavior in the
biotope.
The qualitative microbial tests were completed by the
determination of the microorganisms most common in the water of
the GAC filters at the waterworks. In Table 1 the species of
bacteria are shown: 26 species of bacteria out of 11 genera
were found; 9 species belong to the genus Pseudomonas and 6
817
-------
WORKS INLET
COLLECTING OUTLET(GAC)
55% Pseudomonas
80% Pseudomonas
SIMILARITY (%)
Figure 5. Similarity of Bacterial Clusters in
Raw versus Treated Water
Note: Clusters are described as triangles whose baseline is
proportional to the number of strains per cluster. The
baseline of each cluster is situated on the level of the
average similarity of all strains forming the cluster.
The connecting lines between the clusters describe the
similarity among each other.
818
-------
Table 1
Bacteria in the Effluent From Activated Carbon Filters
of the Rhine Waterworks
Pseudomonas alcaligenes
Pseudoroonas cepacia
Pseudomonas facilis
Pseudomonas fluorescens
Pseudomonas lemoignei
Pseudomonas mendocina
Pseudomonas ruhlandii
Pseudomonas stutzeri
Pseudomonas spec.
Gluconobacter oxidans
Azomonas agilis
Azomonas ins ignis
Azomonas macrocytogenes
Chromobacterium violaceum
Neisseria sicca
Acinetobacter calcoaceticum
Micrococcus Tuteus
Staphylococcus saprophyticus
Bacillus cereus
Bacillus circulans
Bacillus licheniformis
Bacillus megaterium
Bacillus pumulis
Bacillus thuringensis
Corynebacterium spec.
Micromonospora spec.
species to the genus Bacillus. All 3 known species of the genus
Azomonas were detected"!No human-pathogenic bacteria could be
found in the water of the GAC filters of this waterwork. In
none of six investigated German waterworks were bacteria of
the family Enterobacteriaceae detected, nor did other authors
find a proliferation of bacteria of this family in GAC filters
(6,7).
Investigations to enumerate and to determine microorganisms
in activated carbon filters were carried out in 19.78 by McKeon
and McElhaney (8).
If biological activated carbon filters (BAG) are to be used
for drinking water treatment in the future, the question will
arise as to the hygienic significance of bacteria and in particu-
lar as to their toxins in the filter. Endotoxins (lipopoly-
saccharides) are produced by Enterobacteria; the endotoxins are
located in the cell wall and are released during the (auto) lysis
of these bacteria. In the GAC-filte/rs no Enterobacteria could be
found. On the other hand, no information is available concerning
other Gram-negative bacteria, which also produce endotoxins.
Moreover, it is uncertain whether the endotoxin released by the
decaying bacteria remains on the activated carbon. Furthermore,
it is generally accepted that the oral use of endotoxins is
absolutely harmless, at least for the healthy person, as these
substances do not penetrate the intestinal walls (9).
Yeasts and filamentous fungi were found only occasionally,
and seem to be of less quantitative and qualitative importance
for drinking water treatment (Table 2). None of these detected
microorganisms is human-pathogenic.
819
-------
Table 2
Yeast and Filamentous Fungi in the Effluent From
Activated Carbon Filters of the Rhine Waterworks
Phialophora hoffmannii Cryptococcus uniguttulatus
Phialophora mutabilis Candida guilliermondii
Taphrina spec. Hansenula anomala
Rhodotorula minuta
INTERACTION BETWEEN BACTERIA AND ACTIVATED CARBON
Activated carbon adsorbs organic substances that can be used
by bacteria as a substrate and it also adsorbs bacteria. This
result was found in batch tests and the adsorption of bacteria by
GAC agrees with the Freundlich isotherm (Figure 6).
With increasing ion concentration (phosphate buffer pH 7.2),
the bacterial loading on the carbon increases. No dependence on
temperature was noted in the range of 5 to 37°C and on the
pH-value in the range of 5 to 8. Dead bacteria are slightly
better adsorbed than living bacteria.
Bacteria and nutrient molecules are of different size;
therefore, bacteria and nutrient molecules are separated, selec-
tively, due to the pore structure of the activated carbon.
Figure 7 shows the influence of activated carbon on the
metabolic rate of bacteria. In this case phenol was the sub-
strate. Phenol is highly adsorbable and at low concentrations
easily degradable, but in higher concentrations it is toxic to
bacteria. Using 0.3 g/1 phenol as a substrate, activated carbon
has a negative influence on the bacterial utilization of phenol;
this means that carbon hinders the metabolism. Using 2.5 g/1
phenol, there is no decomposition in the absence of activated
carbon because of the toxicity at this concentration. However,
in the presence of activated carbon the following is observed:
the carbon adsorbs phenol and renders it harmless for the
bacteria; in the steady state, phenol desorbs in non-toxic
concentrations and bacteria can utilize this substance as a
substrata. Hence, GAC shows a positive influence on the metabo-
lization of phenol by bacteria. One can imagine a similar effect
in a GAC filter for drinking water treatment. Moreover, GAC
increases the retention period of slowly degradable substances,
thereby providing more time for their decomposition by bacteria.
Furthermore, GAC may accumulate organic substances present
in small concentrations in the water and thereby make bacterial
decomposition possible. The biodegradation depends on the
820
-------
12-
10-
8-
6-
LOADING Q
(log colony count/g GAC!
Figure 6.
5 6 7 8 9 10 11 12
c CONCENTRATION OF
ADSORPTIVE
(log colony count7200ml)
Adsorption Isotherm of Bacterial Loading on Activated Carbon
i • OXYGEN CONSUMPTION
(ml) ^002,69/1 phenol with GAC
15-
10-
0.3g/l phenol without GAC
0.3 g/l phenol with GAC
lout GAC
50
100
150
200
Figure 7.
TIME (HOURS)
Influence of Activated Carton on the
Metabolic Rate of Bacteria
821
-------
steady state concentration of substances in the liquid phase.
The substrate is released from the GAC into the liquid phase by
diffusion. Diffusion is a time-dependent step and therefore has
a decisive influence on the rate of decomposition of the sub-
strate. This is one reason why faster decomposition occurs if
the diameter of the granules decreases. The other reason is the
increasing specific surface area with decreasing diameter which
affords more opportunity for bacterial attachment.
Using pumice stone as granulated material/ the same sub-
strate enhancement of the metabolic rate is observed because
there is no separation between bacteria and substrate caused by
the structure of pumice stone. In this case, bacteria, sub-
strate, and enzymes accumulate on the exterior surface. Investi-
gations showed a positive correlation between available surface
and rate of decomposition; this means that the size of the
granules is a decisive factor. But if phenol is used in high
concentrations there is no protection from its toxicity (as
there is in the presence of activated carbon) and the micro-
organisms will be destroyed.
The pH-value influences not only the adsorptive properties
of activated carbon but also the metabolic rate. Using various
buffered pH-values, in a range of 5.8 to 7.8, there is no notice-
able influence on the metabolism in the presence or absence of
activated carbon.
ACTIVITY OF THE BACTERIA IN THE GAC FILTER
Through their metabolization of organic substrate, micro-
organisms in a GAC»filter provide a valuable contribution to
drinking water treatment. The question of the extent to "which
the microorganisms are responsible for the transformation of
organic substances or their removal during GAC treatment was
addressed by comparing a sterile and a non-sterile pilot plant.
The raw water of the pilot plant was highly chlorinated and the
efficiency of the GAC during these investigations was still very
high. Under these conditions the contribution of the micro-
organisms to the decrease of organics during filtration is small
(approximately 5 percent), as measured by 0V-»absorption, DOC and
KMnO -consumption (4,5). On the other hand/ the contribution of
the microorganisms to the decrease of easily degradable organics
is high (70 percent). BOD investigations demonstrated this. This
observation is of special importance because it implies that the
regrowth of bacteria which usually occurs in the distribution
system, now occurs in the waterwork itself and can be better
controlled.
It has been concluded from the comparison between a sterile
and a non-sterile pilot plant that the contribution of the micro-
organisms to oxygen consumption and carbon dioxide production
during filter passage is substantial, approximately 60 percent
822
-------
for both parameters. The remaining 40 percent is caused by
abiotic processes on the activated carbon. Observations indi-
cated that microorganisms increase the efficiency of activated
carbon filtration and prolong the running time for GAC by
continuous regeneration. Other authors (10) previously have
reported evidence of continuous regeneration of loaded activated
carbon by biological processes, but these demonstrations have
been limited to the field of sewage treatment.
CONCLUSIONS
Some years ago, drinking water treatment with sterile
GAC filters (11) was tried with little success. On the other
hand, the activity of bacteria in slow-sand filtration has been
used for almost a century. Why should we not use the activity
of microorganisms in activated carbon filters? Bacteria must
be tolerated because there are no means or methods which are
successful or applicable for the sterile operation of large
GAC-fliters.
Sporadically, people try to enhance the activity of bacteria
in GAC filters by replacing breakpoint chlorination by ozonation
(12). As a rule, breakpoint chlorination changes the organic
substances .in water to substances difficult to degrade while
ozonation has the opposite effect.
It might be proposed to replace activated carbon by another
porous but not activated material, which might be less expensive
and equally effective. This material could serve, as well as
activated carbon, as an attachment surface for the microorgan-
isms. Nevertheless, it must be pointed out that the superiority
of adsorption and especially the combination of adsorption and
microbial processes has considerable advantages. On the one
hand, bacteria metabolize the easily degradable organics which as
a rule are difficult to adsorb; on the other hand, the slowly
degradable substances are easily adsorbed by activated carbon.
Moreover, there is a buffering effect and protection of the
system by activated carbon if the water contains toxic sub-
stances. These substances will be adsorbed by activated carbon
and thus will be separated from the microorganisms, thereby not
destroying the population.
These are the reasons why we should not combat bacteria in
activated carbon filters in the future as has been done in the
past. On the contrary, we should make use of the activity of
microorganisms for drinking water treatment in activated carbon
filters.
823
-------
REFERENCES
1. Werner, P. Mikrobiologische Untersuchungen von Trink-
wasserversorgungsanlagen unter besondererBerucksichtigung
der Aktivkohlefiltration. Diplomarbeit, Universitat des
Saarlandes, Saarbrucken, 1978.
2. Ford, D.B. The Use of Granular Carbon Filtration for
Taste and Odor Control. Activated Carbon in Water Treat-
ment. Papers and proceedings of a Water Research Associ-
ation Conference, held at the University of Reading,
England, April 1973, Paper 12, p. 263-278.
3. Werner, P. Okologisch-mikrobiologische Untersuchungen an
Aktivkohlefiltern in Zusammenhang mit der Trinkwasserauf-
bereitung. Dissertation, Universitat des Saarlandes,
Saarbrucken, 1979.
4. Klotz, M.; Werner, P.; Schweisfurth, R* Untersuchungen
zur Mikrobiologie der Aktivkohlef liter. Verb'f f entlichungen
des Bereichs und des Lehrstuhls fur Wasserchemie der
Universitat Karlsruhe, Heft 9, 270-282 (1975).
5. Klotz, M.; Werner, P.; Schweisfurth, R. Investigations
Concerning the Microbiology of Activated Carbon Filters.
Translation of Reports on Special Problems of Water
Technology, 312-330, EPA 600/9-76-030 (1976).
6. van der Kooij, D. Einige bakteriologische Aspekte der
Trinkwasseraufbereitung mittels Aktivkohlefiltration.
Veroffentlichungen des Bereichs und des Lehrstuhls fur
Wasserchemie der Universitat Karlsruhe, Heft 9, 302-308
(1975).
7. Johnston, P.R.; Burt, S.C. Bacterial Growth in Charcoal
Filters. Filtration and Separation, 240 (1976)
8. McKeon, W.R.; McElhaney, J. Enumeration and Identifica-
tion of Bacterial Populations in GAC-Columns and their
Effluents. Presented at the AWWA Technology Conference,
Louisville, Kentucky, 6 Dec, 1978.
Westphal, O. Endotoxine - eigenschaften und Vorkommen.
9. Brief an Prof. Dr. R. Schweisfurth (Homburg/Saar), 1978.
10. Weber, W.J. Jr; Hopkins, C.B. ; Bloom, R. Jr. 1970.-
Physico-chemical Treatment of Wastewater. JWPCF 42, 83-89.
824
-------
11. Bernhardt, H. 1970. Entkeimung von Aktivkohlefiltern durch
Erwarroung. Schriftenreihe WaBoLu, Heft 31, 165-169.
12. Jekel, M.; Sontheimer, H. 1978. Biologisch-adsorptive
Trink-wasseraufbereitung in Aktivkohlefiltern. DVGW-
Schriftenreihe Wasser 101, 241-250.
825
-------
QUESTIONS AND ANSWERS
(after Dr. Werner)
0 UNIDENTIFIED QUESTIONER: It seems that on every GAG filter,
there is biological activity. Calling something biological acti-
vated carbon is really calling something GAC, as we now refer to
it. That's what you defined for us?
A DR. WERNER: We found that in every GAC filter examined,
bacteria could be found everywhere. There was no sterile acti-
vated carbon filter. I think it's impossible to devise a method
to maintain large activated carbon filters in a sterile condi-
tion. For small carbon filters in the laboratory you can use
membrane filtration, but it's not practical for large filters.
Contamination is everywhere.
Q BILL BATCHELOR, Texas A&M University: I was interested in
your technique on how you measured total bacterial counts and
living bacterial counts. Could you elaborate a little bit on
that?
A DR. WERNER: Total bacterial counts were found by accumula-
tion and staining on membrane filters. They were counted on the
microscope. The living bacteria have active electron transport
systems and dead bacteria don't. We used a stock that is color-
less when oxidized; when it is reduced, it is colored red.
Q MR. BATCHELOR: Formalin?
A DR. WERNER: Yes.
826
-------
USE OF POWDERED ACTIVATED CARBON TO REDUCE
ORGANIC CONTAMINANT LEVELS
J.E. Singley, B.A. Beaudet,
A.L. Ervin, and W.C. Zegel
INTRODUCTION
In September, 1977, the City of North Miami Beach, Florida,
received numerous taste and odor complaints regarding their
drinking water. The problem was tracked to the East Drive Well
Field. Analysis of the treated water from this field by a local
laboratory showed the presence of pesticides at levels much
above the Primary Drinking Water Regulations. The city immedi-
ately discontinued use of the well field and sought confirmatory
analyses of the water. Subsequent analyses by other private and
public laboratories showed the report to be false. However, 40
synthetic organic compounds were found in both the raw and
treated well water.
These compounds could account for the taste and odor
complaints. Many of them also exceeded the Maximum Contaminant
Levels (MCL) proposed by the EPA based on the report, "Drinking
Water and Health" (National Academy of Sciences, 1977) and
published in the Federal Register, February 9, 1978. These
limitations are, in summary form:
^ The total trihalomethane (TTHM) content in the distribution water
supply will not exceed 100 micrograms per liter (ug/1)
for supplies serving populations greater than 75,000
• The concentration of volatile halogenated organic
compounds (other than trihalomethanes) in the effluent
shall not exceed 0.5 yg/1.
• The removal of influent total organic carbon (TOC) with
fresh activated carbon shall be at least 50 percent.
• The effluent TOC may not exceed the value obtained from
fresh activated carbon by more than 0.5 mg/1.
827
-------
In addition, suspected carcinogens such as vinyl chloride
should be reduced to concentrations that consider the toxic
effects and the economics of treatment. The organics in the
finished water that exceeded these limits are presented in
Table 1.
Table 1
Organic Analysis of Finished Water
(which exceed EPA proposed MCL's)
Concentration (yg/i)
Compound Sample 12345
Vinyl Chloride
Vinylidene Chloride
Trans-1,2 dichloroethene
1,1 dichloroethene
1,1,1 trichloroethane
Trichloroethylene
Bromodichlorome thane
Dibromochlorome thane
Chlorobenzene
TTHM • s
3.3
1.7
11.7
5.0
16.7
21.7
6.7
0.8
^^«B
2.1
1.7
2.9
5.3
1.5
4.4
0.8
115
2.2
1.1
3.0
5.0
15
1.9
1.5
5.9
0.9
__ .—
4.0
3.3
4.4
15
2.4
0.9
0'.9
_•• v
1.7
4.2
5.0
17
3.2
1.3
3.3
0.9
«.«
Although granular activated carbon was considered to be the
best broad spectrum solution to the problem, the appropriate
studies for the development of design criteria would have taken
from 6 to 12 months to complete, and the city had no reasonable
alternatives for a water supply. Thus, a short term solution was
needed to allow use of the well field while granular activated
carbon studies were undertaken. Environmental Science and
Engineering Inc. developed such a short-term solution in the form
of a powdered activated carbon system. The powdered activated
carbon could be mixed with the raw water, then the carbon could
be removed in the treatment plant; thus little additional equip-
ment would be needed to treat the water. Since the general
applicability of activated carbon to this problem has been
established by EPA, a plant test of powdered activated carbon
was designed and conducted.
828
-------
DESCRIPTION OF THE TREATMENT PLANT
The water from the East Drive Well Field is processed in
the Sunny Isles Water Treatment Plant. This is a conventional
lime softening plant with a nominal capacity of 12.8 mgd (See
Figure 1). The raw water is pre-chlorinated and then dis-
tributed to three Hydrotreaters where calcium hardness removal
is accomplished by reaction with pebble lime. One of the
Hydrotreaters is rated at 9.0 mgd, while the other two are rated
at 1.9 mgd each. Pebble lime is fed to a single slaker, con-
verted to hydrated lime slurry, mixed with return sludge, and
pumped to a splitter box. The splitter box distributes lime to
all three of the Hydrotreaters at the center drum of each unit.
Anionic polymer (Nalco 8173) is blended with the raw water to
assist settling in each Hydrotreater unit. The water is then
recarbonated and distributed through nine high-rate filters
prior to final chlorination and distribution.
Recarbonation is accomplished through three submerged,
gas-burning recarbonation units. Further water stabilization is
achieved by the addition of polyphosphate (Nalco 918) to each
recarbonation unit.
Six of the filters are single media, containing anthracite,
and are rated at 7.2 m/h (3.0 gpm per square foot). The other
three are dual media containing both sand and anthracite and are
rated at 12 m/h (5.0 gpm per square foot). Each has a filter
surface-area of 26 m (280 square feet), giving a capacity of
0.053 nr/s (1.2 mgd) each for filters 1-6, and 0.088 nr/s
(2.0 mgd) for filters 7-9.
PLANT TEST PROCEDURE
In performing the plant test, the system requiring the
minimum time and additional equipment as well as minimum disrup-
tion to the operation of the treatment plant, was to introduce a
slurry of the powdered activated carbon into the raw water at
the well field. This would also maximize the contact time
between the raw water and the activated carbon as it flows
through a 0.61 m (24 inch) diameter, 0.5 km (16,500 foot) long
line. Experience has indicated a minimum velocity of 1.5 m/s (5
feet per second) for maintaining a suspension of the carbon in
water. This corresponds to a flow of 0.4 m/s (9 mgd), below
normal flow rates. The contact time would be 30 to 40 minutes
at normal flows. The injection point was selected to be as
close to the wells as possible and convenient to the apparatus
for producing the slurry.
The powdered activated carbon (PAC) mixing and feed equip-
ment was set up in the space aroufcd an existing emergency
generator. The apparatus is shown in Figure 2. The system was
designed to meet a wide range of PAC feed doses for test
829
-------
TOSJJOGSWr
Figure 1. Sunny Isles Water Treatment Plant Schematic
830
-------
3/4"
LINE
OOU8LS
• CHECK VALVE
I 030
DIAPHRAGM MHTSWG
PUMP 0-1 iSgpn
1-sucncNUNS
T BBCWCJLATTCN LINE ^*
FROM DISTRIBUTION
SYSTEM
aao g«. SLURRY TANK
100
Fiqure 2. Process Schematic of the East
Drive Well Field PAC Feed Site
831
-------
purposes. A smiliar backup system was provided to insure
continued utilization of the East Drive Well Field. PAC was
added to a 3.3 m (880 gallon) slurry tank to a design slurry
mixture of 132 pounds carbon per m (1.1 pounds per gallon).
The tank was agitated by a dual-prop mixer that extended to
two-thirds of the tank depth, and by a 100 gpm recirculation
pump which drew slurry from the bottom of the tank and recircu-
lated it to the top of the tank. The slurry was delivered to
the injection line by a Wallace and Tiernan (Model 44-115)
metering pump which was adjustable through a range of 0 to
0.43 m /h (0 to 115 gph). The suction of this metering
pump was drawn from the discharge of the recirculation pump to
provide a constant suction head. In the 19 mm (3/4-inch)
injection line, the slurry was diluted with treated water. The
injection point was through five 63 mm (1/4-inch) holes located
at the center of well header and directed against the flow of
raw water.
The PAC used in the test was ICI Hydrodarco-B, a lignite-
base carbon exhibiting a preponderance of mid-range 5 to 500 nm
pores as compared to micropores (less than 5 nm).
During the plant test, three levels of carbon dose were
tested together with a control run. In addition, the recarbona-
tion units were fed air and were not burning gas. The
chlorination point was changed from the raw water receiving tank
to the recarbonation unit during a portion of the tests. The
average daily raw water flow to the plant was 10.4 mgd during
the test period. This corresponds to a theoretical 47 minute
contact time between the raw water and the carbon in the trans-
mission line, and a 71 to 79 minute retention time in the
Hydrotreaters. Hence, true contact time in these units is judged
to be below 70 minutes. This corresponds to almost 2 hours total
contact time. Each test run lasted for 24 hours.
Raw water samples were collected for analysis of volatile
and non-volatile organic compounds. Dual samples for the
non-volatile organic analysis were collected over a 12-hour
period in XAD-4 resin cartridges. Flow rates through the resin
cartridges were checked prior to, during, and at the completion
of the sample run. Raw water samples were collected daily and
the dual sample was reported as a single, average, daily result.
Samples for volatile organic analyses were taken in
organically clean VOA bottles at the initiation of a resin
cartridge run and at the termination of a cartridge run. The
dual volatile sample was reported as a single, average, daily
result.
The finished water samples were taken from a transfer pump
on the discharge side of the clearwell at 1030 hours each day
and terminated at 2200 hours each night. This provided a
832
-------
sufficient lag time for PAC dose changes to stabilize through
the raw water line and in the process equipment of the plant.
These samples were collected in the same manner as the raw water
samples.
Two TTHM samples were collected in VOA bottles at the plant
clearwell. One TTHM sample was collected as an instantaneous
sample and the second/ as a 3-day formation potential sample.
Immediately after the instantaneous sample was collected, sodium
thiosulfate was added to prevent further reaction with chlorine.
The 3-day formation potential samples were allowed to stand for
72 hours before dechlorination.
RESULTS OF THE PLANT TESTS
The plant tests were conducted in July 1978 over a 10-day
period. Table 2 summarizes the test conditions for each day of
the test. No PAC was added the first day. The PAC feed was
Table 2
Total Organic Carbon for Various Doses of
Powdered Activated Carbon
AC Chlorination Raw Water Finished Water
(mg/1) Point* TOC (mg/1) TOC (mg/1)
0
7.9
7.9
14.3
14.3
26.6
26.6
7.1
7.1
R
R
R
R
R
R
R
C
C
(15.2)**
10.9
13.7
12.7
16.5
12.8
(14.5)
(16.0)
9.2
13.0
7.1
9.0
11.2
10.2
9.1
11.0
3.6
5.2
Percent
Reduction
14.5
34.9
34.3
11.8
38.2
28.9
24.1
77.5
43.5
*R = Raw Water, C = Recarbonation Unit (also have post-
chlorination in both cases)
**( ) = interpolated data based on measured non-volatile organics
833
-------
started on the second day and an average concentration of 14.3
mg/1 maintained for 2 days. r two days. After a 1 day purge of the
system, the carbon feetf was restarted to maintain an average of
7.9 mg/1 for the next 2 days, and 7.1 mg/1 for the following
2 days. The chlorination point was changed from the raw water
intake to the recarbonation units for these two days to deter-
mine whether the chlorine was inhibiting the adsorption of the
organics.
TRIHALOMETHANES
The total trihalomethane (TTHM) data show no definite trend
with increasing dose of PAC (see Figure 3). Apparently varia-
tions in the levels of precursor organics in the raw water
occurred which masked the effects of PAC. In an effort to
normalize these data, the isomers of dichloroethene and 1,1,1
trichloroethane were used as internal standards. These were
selected because they represent chemicals that are present in
the raw water at relatively high levels, and may be indicators
of the level of synthetic organics in the raw water. Normal-
izing the TTHM data to the base data (zero carbon dose), yields
Figure 4. This indicates that PAC may have a minor effect on
reducing TTHM, and was successful in keeping the levels below
100 ug/1 for all but one of the test days.
The TTHM levels tend to increase in the finished water with
time due to slow reaction of precursors with residual chlorine.
Data were also gathered on 3-day terminal TTHM. A comparison of
the 3-day terminal TTHM levels with PAC dose shows a definite
decrease with dose (see Figure 5).
These results indicate that PAC is not very effective in
reducing TTHM, but does remove the precursors which will slowly
react with residual chlorine to produce TTHM in the distribution
system.
VOLATILE HALOGENATED ORGANICS
Nine volatile halogenated organics were detected in the
finished water at levels exceeding 0.5 yg/1 during the test
period. These are the trihalomethane group, the dichloroethene
group, chlorinated ethanes group and chlorobenzene. We have
already examined the trihfclometfo«mes, which contained chloro-
form, bromodichloromethane and dibromochloromethane in our test
runs. The dichloroethene group contains cis-and trans-1,2
dichloroethene, and vinylidene chloride, with the cis isomer
predominating. The variation in concentrations for this group
with PAC dose is shown in Figure 6. This figure shows no clear
trend, although some reduction with dose is possible. Normal-
ization of the data, as was done for the total halogenated
834
-------
120-
100-
1 M
1
20-
o
"a
9
o
o
10
18
OOSC (mg/l)
20
Figure 3. Total Trihalomethane Levels in the Finished Water
at Different Doses of Powdered Activated Carbon
During the Plant Test
so-
t -
20-
10-
O
O
o
o
I 10 IS 20 29
MC OOM (mg/l)
Figure 4. Total Trihalomethane Levels in Finished Water
Normalized by Possible Indicator Species at
Different Doses of Powdered Activated Carbon
During the Plant Test
835
-------
S 120-
100-
a
g »'
M
5J 60'
40-
20-
I
10
18
t
20
PAC ooae
Figure 5. Three Day Total Trihalonethane Formation Potential in the Fin-
ished Water at Different Doses of Powdered Activated Carbon
During the Plant Test
a 90-
1
40-
O
I
30-
20.0
10-
\
\
\
\
Q
10
OOSC (mg/l)
20
2S
Figure 6. Levels of Dichloroethene Isomers in Finished Water at Various
Doses of PAC for the Plant Test
836
-------
methane, yields the results shown in Figure 7. This indicates
some reduction with increasing dose of carbon.
The chlorinated ethanes group consists of 1,1 dichloro-
ethane, and 1,1,1 trihalomethane, with the latter predominating.
It also shows little trend in levels with increasing dose of
PAC, and after normalization it indicates a slight decrease may
occur with the addition of PAC (See Figures 8 and 9).
Chlorobenzene was present at low levels (0.5 ug/1) in the
control run, and showed good removal by the addition of PAC. It
was reduced to 0.1 yg/1 or less by all dosages, except for-on«
day in which there appeared to be higher than normal organics in
the raw water (Figure 10).
Non-Volatile Synthetic Organics Compounds
Non-Volatile Synthetic Organics (NVSOC) are of high molecu-
lar weights such that special precautions need not be taken in
their sampling or analysis to prevent their loss by exposure to
the atmosphere. The response of NVSOC to the addition of PAC in
the plant was immediate and obvious as shown in Figure 11. They
are most effectively removed by the addition of even the lowest
levels of PAC tested.
Total Organic Carbon
The total organic carbon (TOC) data obtained are fragmen-
tary and no definite conclusions can be drawn except that TOC
is reduced in all cases. The available TOC data show a close
correlation with the total of the measured non-volatile organics.
Using this relationship to fill in missing data yields the
results shown in Table 2.
It is apparent from this Table and Figure 12 that the PAC
reduces the TOC. However, the most dramatic reduction in TOC
occurs when the chlorination point is changed and carbon is
added.
CONCLUSIONS
PAC was most effective in removing organics with low
volatilities at all doses tested. Some reduction in
volatile halogenated organics was observed, particu-
larly for the halogenated organics of relatively
higher molecular weight and lower volatility. It is
suspected that the high percentage distribution of
pores larger than 5 nm in the test carbon, may limit
the effectiveness of removal for low molecular weight,
highly volatile species. It is hoped that future
tests of carbons which include a greater percentage of
837
-------
2
M
3
o
o
Figure 7.
Levels of Dichloroethene Isomers in Finished
Water (Normalized) at Various Doses of PAC
for the Plant Test
838
-------
so-
2 30 •
O 20'
•a
3 10-
X
X
X
X
X
X
X N
\
\
\
O
o
O
o
o
o
10
13
25
PAC DOSE (mg/l)
Figure 8. Levels of Chlorinated Ethanes in Finished
Water at Various Doses of PAC for Plant Test
-. 30 -
1
i
w
I »H
la 10-
I
\
\
\
x
X
\
10
13
OOSC (mg/l)
I
20
i
25
Figure 9. Levels of Chlorinated Ethanes in Finished
Water Normalized by Raw Water Indicator
Species at Various Doses of PAC for the
Plant Test
839
-------
3 i.o H
i
a
X *»».
X
oo
10
15
PAC OOSC (mg/l)
20
2S
Figure 10. Levels of Chlorbbenzene in Finished Water At Various Doses of
PAC for Plant Test
-\
i-
•\
k \
\ \
10
18
OOS8 (mfl/l)
T
20
I
2S
Figure 11. Levels of Non-Volatile Synthetic Organic Compounds with Various
Doses of Powdered Activated Carbon During the Plant Test
840
-------
-
/ CHLOWNATION AT RfCAMONATION UMT
"
g
g 40-
20-
/
/
o
CHLOfNMATION OF MAW WATCT O
\ I I i i
5 10 18 20 2S
PAC OO«B (mg/1)
Figure 12. Percent Reduction in Total Organic Carbon
with Various Doses of PAC and with a
Change in the Point of Chlorination
841
-------
micropores, will demonstrate more effective removal of
these compounds.
PAC was quite effective in controlling the TTHM that
would be formed by "slow" reactions occurring in the
distribution system, presumably by removal of the pre-
cursors. This conclusion coupled with the conclusion
above demonstrates that the precursors participating
in the "slow" reactions are low volatility organics.
Changing the chlorination point from the raw water
feed to the recarbonation units improved the effec-
tiveness of the PAC in removing TOC, presumably by
eliminating competition by chlorine for adsorption
sites, forty to fifty percent removal of TOC would be
expected with PAC using this chlorination point. This
removal of TOC is probably due to the removal of the
low volatility component of the synthetic organics
present.
PAC slightly reduced instantaneous TTHM levels and the
levels of volatile organics/ but was not particularly
effective in controlling these levels irrespective of
the location of the chlorination point.
842
-------
APPENDIX
(All levels yg/1 unless otherwise noted)
00
**
00
•-El'-JLl'M ««ill
itle leu fin
...J
D.r
Tekej leu fin
1 Oey 1 Oey
l.hed leu rtuUbcd lieu fin
Vinyl Cklorl4e ^* * * * * *
Vinyl lee.. Cklorlee
I.I Plcklocoelkene
Cle 1.1 Oickloroelkene
Ckluroforei
I.I.I Tclch lococlkene
1,1 Blckleroelkene
Gefkon T.I f .cklor le.
Tr Ickloroetkyl CMe
IrOBMMllcfcloroettlkeoe
l»l hKOeMMklorowel fcene
ChlorokcMeene
IciMMiforei
Btkyl lenzene
Toluene
lyl«ne
• ielkyUleulflee
Tetreckluroetkeue
Helkyl llycene
..e/o. Ineene
Helkyl lwle.e
•ephlkelene
Methyl Neplheleiie
C] Alkyl Neplkelcne
AceneptheHe
Octyl Pkenol
2.4 Olterlktityl-p-cieeol
Flitoreiie
D 1 kencul k lopliene
Hcfkyleue Pfceiietilhiecene
f yfene
BlphenyleoUe
C) Alkyl leu cue
Ct Alkyl len «««
Ck 1 urot u 1 ucn
Telrecklorne hew*
Pklkelelee
Iftickloi-okenc ue
Hclhylbentof fen
Elkyl llyiene end Ind.n.
TTMH Inet
1.1 I.I O.S 1. O.I
14.* 4.2 12. i 1. 12.1
11.1 1.0 21.1 1. 1S.1
14. S
1
.1
.1
.01
.1
.01
.01
.01
.OS
.4
.01
.It
.14
.01
.01
.01
.04
01
.04
.14
.01
.01
1.1 4
II
.0
.1
.J
.2
.1
.04
.1
.Ol
.01
.0*
.0)
.Ot
.01
.01
.01
01
.01
.11
k
1. 01
1.1
4.0
.0 12.1
.1
.1
.01
.2
.02
.02
.001
.01
.11
.04
.1
.11
.01
.01
.01
.OS
.01
.0)
.11
.61
r
1.0 1
4
.1
.0
.1
.1
14. S
.1
.1
.01
.2
.02
.02
.00)
.01
.It
.01
.1)
.11
.01
.01
.01
.01
.01
.04 .01
0.14
1 0.01
A 0.01
S.t 1.0 1
1.0 - *
.1
.1
.1
.01
1.4
b.O
.
.
.4
.1
.1
.01
.2
.01
.01
.00)
.0)
.14
.01
.1
.24
.02
.01
.01
.01
.01
.01
.11
.01
.01
.» 4
»
r- - "J • -2*1-
[.licj leu rinlekee' leu Flu
.*
1.0
S.4
M
.1
.0
.1
.01
1.0
(. |
.
1
1
.)
.4
.1
.01
.1
.01
.01
.00)
.0)
.It
.01
.1)
.14
.01
.01
.01
.04
.02
.04
.14
.01
.01
.1 1
•>
1
.1
.1
.4 1
.1
.1 1
.0
.1
.1
S.4
i.O
I.O 1
V
1
4
.4
1
.S
.1
.01
.4
.01
.01
.02
.24
.OS
.14
.14
.01
.01
.01
.04
0)
A
.01
.Ot
.01
.01
I.I II
1)
II.*
S
•keJ '
Key 4 Be. 1
leu rl.l.k.4 leu rUi.ked ]
leu ri.ilK,*
1. 1.' 1. II 1. 1.1
.) S. 4.1 1. 1.1 1. l.S
.4 II. 14. t
.S 1. 40. S
.*
.2
. 1
.$
!s
.01
.4
,„
.004
.01
.01
r
i.O
s.o
I.I .'
11. t
.1
1
. J
.S
.01
.4
.01
.01
.01
.01
.01
.0)
.14
.01
.01
.01
.M
.02
.0)
.11
.01
.01
.4 11
11
1.1 i
.1
.4
•1
.4
.01
.1
.04
.)
.0
.0
11.1
1S.O
14.4
A
.1
.1
.01
.1
.01
.01
.004
.01
.44
.01
.11
.11
.01
.11
.01
.01
.M
.01
.IS
.01
.01
.1 41
-
.4
• *
.1
.1
.01
.01
.01
(
.1 1
.0
.4 i
II. S
14.1
11.4
A
A
T
1.1
1.1
.1 T
A
A
.1 0.1
.01 0.01
A
.1 A
.01 T
.01 A
.01 T
A
.04 A
.SI T
.1 t
.31 T
A
.11 T
.01 t
.001 A
.01 A
.01 T
.OS A
*
A
A
A
A
.01 0.01
.11 0.01
.01 A
.01 T
.1 11.1
111.0
.1 S.I
TTMH 1« -no 11. 1 H. » '••» '° 2 "•*
KM: lu«/l)
-------
REMOVAL OF ORGANIC CONTAMINANTS FROM DRINKING WATER
USING TECHNIQUES OTHER THAN GRANULAR ACTIVATED CARBON ALONE'
A PROGRESS REPORT
James M. Symons, J. Keith Carswell,
Jack DeMarco, and 0. Thomas Love, Jr.
INTRODUCTION
In the proposed Regulation (1) for the control of organic
contaminants in drinking water published in the Federal Register/
February 9, 1978, the use of granular activated carbon adsorption
was proposed for water utilities serving more than 75,000 whose
raw water source was vulnerable to organic contamination. In
this proposed Regulation, the U.S. Environmental Protection
Agency (U.S. EPA) included the use of granular activated carbon
(GAC) adsorption because it was the best broad-spectrum organic
removal unit process available at the time. In an effort to
continue to expand the knowledge related to this drinking water
unit process, the Drinking Water Research Division (DWRD) of the
U.S. EPA has maintained an active research program on the subject
of granular activated carbon adsorption. In addition to this
research, the DWRD has also been performing research on other
organic removal unit processes. These are: synthetic resin
adsorbents; aeration; oxidants plus adsorbents; coagulation and
clarification; powdered activated carbon; and reverse osmosis.
The purpose of this paper is to report on the progress to
date (May, 1979) on the first three of these unit processes. The
last three unit processes have not yet been studied extensively
and will be reported on in a future paper(s). The studies on the
unit processes of synthetic resin adsorbents and aeration have
been carried out mainly on contaminated groundwaters, while the
studies on the use of oxidants plus adsorbents have been made on
both contaminated surface and groundwaters.
GROUNDWATER CONTAMINATION
In the past groundwater was thought in general to be largely
free from organic contamination. However, recent surveys in New
Hampshire, Connecticut, New York, and New Jersey have revealed
that many groundwaters in this part of the United States are
contaminated with various organic chemicals. Future surveys will
reveal whether or not this problem is widespread throughout the
remainder of the United States. Potential sources of groundwater
844
-------
contamination are: pump lubricants (minor); industrial dis-
charges; individual households; and groundwater recharge. The
possibility of groundwater pump lubricants contributing major
organic contamination to water has been discounted through calcu-
lations of contaminant potential. This leaves contamination from
industrial discharges (either through spreading on the land or
the improper disposal of industrial wastes in dump sites), the
use of solvent septic tank cleaners and similar products in
individual households/ and the use of treated waste for ground-
water recharge as the major threats to groundwater quality.
The groundwater protection research program of the U.S.
Environmental Protection Agency located in Ada, Oklahoma, pre-
pared a list of organic contaminants they judge troublesome in
groundwaters (Table 1). The surveys mentioned above have con-
firmed the presence of four of the 12 compounds on the list and
have added three more. Several of these compounds have appeared
in more than one location and frequently are found in water whose
total organic carbon (TOC) concentration is relatively low. Many
of these organic chemicals have proven adverse health effects,
making their removal important.
Table 1. Problem Groundwater Contaminants Suggested
by Groundwater Research Program
Listed by USEPA GroundwaterProtection ResearchProgram
Chloroform Toluene
1,2-Dichloroethylene* Xylenes
Tetrachloroethylene* Chlorobenzene
Trichloroethylene* Dichlorobenzene
1,1,2-Trichloroethane Phthalates
Benzene Methyl Chloroform*
Additional Compounds Found in Surveys
1,1-Dichloroethane
1,2-Dichloroethane
1,1-Dichloroethylene
"Identified in Specific Locations
845
-------
SYNTHETIC RESIN ADSORBENTS
Ambersorb XE-340*
Adsorption
Recently, the Rohm and Haas Company has developed a synthe-
tic resirt adsorbent reported to be highly active for the adsorp-
tion of low molecular weight chlorinated organic compounds. The
performance of this material, called Ambersorb XE-340, has been
investigated in six locations on a total of 10 different organic
contaminants. Table 2 presents the performance of this resin for
removing the ten different halogenated pollutants under the
various operating conditions described. In many cases 10 percent
of the influent concentration of the compounds had not yet
appeared in the effluents at the time of preparing this paper.
In other cases the service time at which 10 percent breakthrough
occurred is shown. Although these data show that often adsorp-
tion of these ten organic substances was excellent (Summary
Tables 3 and 4), removal was better in waters with lower TOG
concentrations (Table 3) than those with higher concentrations of
TOC and other contaminants (Table 4). In all of the locations in
Table 4 the water was chlorinated and contained high concentra-
tions of trihalomethanes. The poorer performance of XE-340 under
these conditions shows the influence of competitive adsorption.
Figure 1 compares the ability of XE-340 to remove trihalo-
methane precursors as measured by trihalomethane formation poten-
tial (THMFP) in both raw and finished waters. The upper portion
of Figure 1 shows that XE-340 is partially effective for removal
of THMFP from raw water, but the lower portion of Figure 1 shows
that when treating water that had been pretreated by softening,
the resin could no longer remove any precursor material. This
implies that the type of precursor that is adsorbable on XE-340
is also amenable to removal by coagulation and sedimentation, and
the type of precursor that remains following lime softening is
not adsorbable on XE-340.
Desorption
Studies were also performed to determine whether or not
XE-340 would permit the desorption of previously adsorbed mate-
rial if the influent concentration were lowered, similar to the
situation experienced when using granular activated carbon as the
adsorbent. To investigate this possibility, a column of XE-340
that had been loaded with a particular organic contaminant was
placed in series following a fresh bed of resin. In this way,
the influent concentration to the previously loaded column was
now drastically reduced because of adsorption on the first column
*Mention of commercial products does not imply endorsement
by the U.S. Environmental Protection Agency.
846
-------
Table 2. Removal of Halogenated Pollutants by XE-340
COMPOUND SITE
Tetrachloroethylene A
B
C
E
1, 1-Dichloroethylene C
1, 1-Dichloroe thane B
00 C
^ 1*1, 1-Trichloroethane
(Methyl Chloroform) A
B
C
cis-1 , 2-Dichloro-
ethylene A
B
D
1,2-Dichloroethane C
Trichloroethylene A
B
C
CONCENTRATION
yg/L
<1-4
<1-17
60-205
597-2500
41-74
<1-9
3-7
<1-2
11-214
172-286
3-9
<1-4
5-28
1-2
120-276
<1-10
1-2
LOADING
gpm/ft
0.8
0.9
1.5
0.9
1.5
0.9
1.5
0.8
0.9
1.5
0.8
0.9
1.2
1.5
0.8
0.9
1.5
EBCT
mln
9
8.5
5
8.5
5
8.5
5
9
8.5
5
9
8.5
6.2
5
'9
8.5
5
BED
VOLUMES
20,160
58,108
42,048
17,788
42,048
58,108
42,048
20,160
58,108
42,048
20,160
58,108
4,646
42,048
20,160
58,108
42,048
TIME IN
SERVICE
DAYS
>126f
>343f
>_146f
105*
>146f
>343f
>146f
>126f
>343f
>146f
>126f
>343f
20
5^146*
>126*
>343f
>146+
ORGANIC
CARBON
mg/L
0.5
-
0.5
2-4
0.5
-
0.5
0.5
-
0.5
0.5
-
6.1
0.5
0.5
-
0.5
-------
Table 2. Removal of Halogenated Pollutants by XE-340 (continued)
COMPOUND
Carbon Tetrachloride
Bromodichlorome thane
00
4^
00 Chloroform
SITE
X
X
D
X
X
D
X
X
CONCENTRATION
Hg/L
NF-87
NF-87
27-56
14-39
26-39
55^140
39-95
64-95
LOADING
gpm/ft3
0.75
1.5
1.2
0.9
1.5
1.2
0.9
1.5
EBCT
min
10
5
6.2
10
5
6.2
10
5
BED
VOLUMES
15,120
7,560
11,615
14,688
12,097
5,342
11,088
6,048
TIME IN
SERVICE
DAYS
105*
52*
50*
102*
42*
23*
77*
21*
ORGANIC
CARBON
mg/L
1.5-2.0
1.5-2.0
6.1
1-2
1-2
6.1
1-2
1-2
Effluent concentration did not reach 10% of influent value within the number of days shown,
which corresponds to the run length for which data were available when this paper was written.
*Time at which effluent concentration reached 10% of influent value.
-------
Table 3. Summary of Performance of XE-340
« 1 mg/1 TOC in water)
Two Locations
Compound
Tetrachloroethylene
1, 1-Dichloroethylene
1, 1-Dichloroethane
Methyl Chloroform
cis-1, 2-Dichloroethylene
1 , 2-Dichloroe thane
Trichloroethylene
EBCT
Min.
5
5
5
5
9
5
9
Maximum
Cone. Range
yg/i
60-205
41-74
3-7
172-286
3-9
1-2
120-276
Time in Service
to 10% Leakage
Days
> 146**
> 146**
> 146**
> 146**
> 126*
> 146**
> 126*
*No breakthrough at end of test.
**Test still continuing. These data collected at the time the
paper was prepared (May, 1979).
Table 4. Summary of Performance of XE-340
(>1 mg/1 TOC in water)
Trihalomethanes Present at three locations
Compound
Tetrachloroethylene
cis-lf 2-Dichloroethylene
Carbon Tetrachloride
Chloroform
Bromodichlorome thane
EBCT
Min.
8.5
6.2
5
5
5
Maximum
Cone. Range
yg/i
597-2500
5-28
NF-87
64-95
26-39
Time in Service
to 10% Leakage
Days
105
20
52
21
42
849
-------
O)
g
H-
<
u
O
O
1000
900
800
700
r= 600
O
a.
Z
O
h-
I
O
u. 500
500
600
INFLUENT
XE-340
RAW WATER
FINISHED WATER
400
300
80
TIME, DAYS
100 120
Figure 1. Removal of Trihalomethane Precursors
by XE-340
850
-------
in series. Figures 2 and 3 show that for chloroform and cis-1,2-
dichloroethylene, desorption did occur. Another attempt was made
to evaluate this phenomenon under different circumstances. In
this case, a sudden increase in carbon tetrachloride concentra-
tion appeared in the water being treated with XE-340 resin.
Shortly after the appearance of this high concentration of carbon
tetrachloride, its concentration again fell to near zero. Figure
4 shows the adsorption pattern for carbon tetrachloride and a
slight, but steady desorption following the disappearance of the
carbon tetrachloride from the influent.
Regeneration
Another reported advantage of Ambersorb XE-340 is its
ability to be regenerated with steam. This was investigated at
one of the DWRD research project sites where chlorinated surface
water was being treated. With respect to chloroform removal,
steaming the XE-340 partially restored its ability to adsorb
chloroform, but breakthrough times were shortened. The possi-
bility of regenerating XE-340 with steam will be further investi-
gated at another location starting in the summer of 1979.
Amberlite IRA-904
Another synthetic resin adsorbent manufactured by Rohm and
Haas Company is IRA-904. This material, a strong-base anion
exchange resin, is used as an organic scavenger in some indus-
trial processes. Its ability to remove general organic contami-
nants as measured by the parameter TOC was compared to XE-340.
Figure 5 shows that XE-340 had almost no capacity for removing
general organic compounds as measured by TOC, while IRA-904
removed approximately one-third of the TOC from the influent.
On the other hand, Figure 6 shows that for the organic compound
cis-l,2-dichloroethylene, XE-340 showed nearly complete removal
throughout the test while IRA-904 removed little, if any, of
this organic compound. Finally, Figure 7 shows that for trihalo-
methane precursor material in raw water, as measured by THMFP,
IRA-904 was initially quite effective, but, similar to XE-340,
IRA-904 was unable to remove any precursor material from water
that had been pretreated by lime softening. The left hand side
of Figure 7 also shows that an unadsorbable fraction of precursor
material existed in the raw water, as the initial adsorber
effluent concentration for THMFP was the same even when the bed
depth was doubled from 2.5 ft (75 cm) to 5 ft (150 cm), 9 and 18
minute Empty Bed Contact Time, respectively. Therefore, although
synthetic resins are capable of removing organic contaminants
from water, different products have different adsorption charac-
teristics; some are likely to be very pH dependent.
851
-------
3!
a.
o
c
o
O
E
5
"o
3
XE-340 Resin
C hlocofOf in
6.2 Min. EBCT
30 «0
TIME, DAYS
Figure 2. Desorption of Chloroform From XE-340
15
10--
OJ
C
o
C
93
U
C
o
U
V
C
0)
V
2 5-|-
_O
O
O
CM
U)
-------
100
Desorption
XE-340 Resin
Cincinnati Tap Water
Carbon Tetrachloride
5 Min. EBCT
TIME, WEEKS
Figure 4. Desorption of Carbon Tetrachloride
from XE-340
853
-------
10
e
o
c
o
u
c
o
u
c
o
.o
c
a
9)
2
6
2-
Effluent
10 20 30 4O SO
TIME. DAYS
60
Figure 5. Removal of Organic Contaminants as Measured
by TOC, by XE-340, and IRA 904 Resins
so
TIME. DAYS
Figure 6. Removal of cis-l,2-Dichloroethylene
by TOC, by XE-340, and IRA 904 Resins
854
-------
00
-------
AERATION
Since many of the compounds found in contaminated ground-
waters can be concentrated prior to gas chromatographic separa-
tion by purging from the liquid with an inert gas onto an
adsorbent trap, aeration was investigated as a possible effec-
tive unit process for removing these compounds from water.
Figure 8 demonstrates that for the trihalomethanes, good removal
can be obtained, provided a sufficient air to water ratio is
maintained. On the other hand, the removal of general organic
contaminants as measured by the non-purgeable organic carbon
concentration was largely unaffected by aeration in this diffused
air system.
In one laboratory and one field experiment, the ability to
remove eight chlorinated organic compounds was investigated.
These studies were carried out in a diffused-air aeration system
with a 10 minute contact time and an air to water ratio (volume
to volume) of 4 to 1. The removal percentages for these eight
compounds ranged from a high of 96 percent to a low of 20 per-
cent. Plotting the percent removals vs. the solubility in water
at 25°C in milligrams per liter, Figure 9 shows a general rela-
tionship between these two parameters. Other characteristics
such as boiling point, molecular weight, and partition coeffi-
cient did not exhibit this relationship. Further study will
indicate whether an approximate judgement can be made for removal
of this type of organic contaminant by knowing its solubility in
water.
Figure 10 shows that a relationship exists between percent
remaining of an organic compound and the air to water ratio.
Therefore, if aeration data were collected at one air to water
ratio, and a different percent removal was required to meet a
given effluent target concentration, this type of relationship
could be used to determine what air to water ratio would be
needed in order to reach this other treatment efficiency.
Finally, studies by McCarty, et al. (2) have indicated
success using cascade type aerators. Studies of this type will
be performed on contaminated groundwaters to determine whether
or not the high air-to-water ratios in this type of tower would
improve treatment as reported.
OXIDANTS PLUS ADSORBENTS
First Experiment
Several reports (3, 4, 5) have indicated that inserting
ozone prior to an adsorption step in a drinking water treatment
train will improve the performance of the combination of the two
processes over the performance of either process alone. To
856
-------
O)
^120
C
o
2: 100--
c
05
u
§ 80
(J
V
c
60--
40--
20--
(0
-O
Cincinnati Tap Water
1 0 Minute Contact Time
--10
: NPOC
TTHM
--06
--04
--0.2
C
o
o
u
u
'£
O5
(C
Cb
O)
51
10 1
15 1
20 1 25 1
30 1
35 1
40 1
45 1
50 1
Figure 8.
AIR TO WATER RATIO
Influence of Aeration on Trihalomethane and
Non-Purgeable Organic Carbon Concentrations
120-)- rTetrachloroethylene
/ /-1, 1-Dichloroethylene
100
•o
0)
o
E
0)
C
V
u
h»
0)
Q.
80--
60--
40--
20--
r-
/ / /-Carbon Tetrachloride
I 1 I u.ab.)
If I /- Methylchloroform
-------
10 Min. Contact Time
Diffused Air
Lab. Data
Chloroform
0.44 um/L
Trichloroethylene • 0.83 um/L
Carbon Tetrachloride •
0.54
Tetrachloroethylene
0.64 um/L
0 2/1 6/1 10/1 14/1 18/1 22/1
AIR TO WATER RATIO
Figure 10. Removal of Four Organic Contaminants
at Various Air to Water Ratios
858
-------
investigate this possibility, a 75 gallon per day (290 liters
per day) pilot column system was set up to treat coagulated and
settled Ohio River water. Two parallel columns were used; in
one, settled water was applied directly onto a granular activated
carbon bed; in the other, an oxygen plus ozone* mixture was
added to the water prior to the filter/adsorber such that the
ozone dose was approximately 1.5 to 2.5 milligrams per liter.
Figure 11 shows that total organic carbon concentrations in
the effluent of the column receiving oxygenated and ozonated
water were always lower than in the effluent from the granular
activated carbon column receiving untreated settled water. A
similar "beneficial effect" was noted on the removal of disin-
fection byproducts formation potential (DB-PFP) as measured by
total trihalomethane formation potential (see Figure 12). This
"beneficial effect" was presumed to result from biodegradation of
some organic compounds.
Second Experiment
This led to a second experiment in which coagulated and
settled Ohio River water, 150 gal/d (580 1/d), was fed to four
parallel treatment trains constructed of stainless steel, Teflon,
and 1.5-inch (4 cm) diameter glass columns. Each treatment train
consisted of a dual-media (anthracite coal over sand) filter fol-
lowed by three granular activated carbon columns in series, each
with a 10 minute empty bed contact time. One train received
untreated settled water, a second train received water that was
oxygenated only, at a rate equivalent to the oxygen in the "low
ozone" system to be described below. A third train received
oxygen plus ozone with the ozone dose approximately 5 mg/1 ("high
ozone"), and a fourth train received oxygen plus ozone with the
ozone dose approximately 0.5 mg/1 ("low ozone"). All four gas
contactors are unpacked countercurrent flow glass columns with a
stainless steel diffuser. The contact time is 18 minutes.
This experiment had four objectives: (1) to determine if
the previous experiment could be confirmed; (2) to determine
what minimum ozone dose was needed to produce the "ozone effect";
(3) to determine if oxygen alone would produce a similar
"effect"; and, (4) to determine if the biological growths could
be contained within the system.
*Because pure oxygen gas was used to generate the ozone, the gas
fed into the gas contactor was an oxygen-ozone mixture. For
accuracy, therfore, the term "oxygen plus ozone" is used for
the system with oxidant added.
859
-------
Effluent-GAC + Ozone & Oxygen
3 456 78
TIME IN SERVICE. MONTHS
Figure 11. Influence of Ozonation Prior to Adsorption
on Organic Carbon Removal
100
2 DAY, 25;C TOTAL TRIHALOMETHANE
FORMATION POTENTIAL
ADSORBER
INFLUENT
EFFLUENT-GAC
ONLY /
EFFLUENT-GAC + OZONE & OXYGEN
45678
TIME IN SERVICE, MONTHS
10
Figure 12. Influence of Ozonation Prior to Adsorption
on Trihalomethane Formation Potential Removal
660
-------
First Objective
Figure 13 shows that comparing the TOC removals, measured by
"fraction remaining" (C /r), from the settled water to a point
in the treatment train after the first of the three granular
activated carbon columns in series/ indicate that pretreating the
water with oxygen plus ozone (5 mg/1) had some beneficial effect
when compared to the control, as in Figure 11. This satisfied
part of the first objective.
Note that in Figure 13, as in Figure 11, both treatment
systems approached "steady-state" conditions toward the end of
the test, and the TOC being removed reached some reasonably con-
stant value greater than zero. Further, although this "steady-
state" removal value is higher in the system with oxygen plus
ozone pretreatment (the fraction remaining is lower), a "steady-
state" condition is also reached in the control, indicating that
biodegradation is also occurring in the control, even without any
stimulation by pretreatment.
To investigate the reason for this effect, Figure 14 com-
pares TOC "fraction remaining" for the oxygen plus 5 mg/1 ozone
system and the control at an intermediate point in the treatment
train, between the dual-media filter and the granular activated
carbon absorbent. Here again, a beneficial effect was noted, as
indicated by a lower TOC "fraction remaining" through the dual-
media filter in the system where the water had been pretreated
with oxygen and 5 mg/1 ozone. The top half of Figure 15 is a
repeat of Figure 14 for comparison purposes. The lower half
of Figure 15 shows that on a "fraction remaining" basis, the two
parallel granular activated carbon columns were behaving simi-
larly, regardless of the pretreatment. Therefore, although the
combination of the dual-media filter and granular activated
carbon adsorber in series showed a beneficial effect from oxygen
plus 5 mg/1 ozone as a pretreatment, this benefit manifested
itself in improved removal of TOC in the dual-media filter,
probably through biological action, and the performance of the
adsorber itself was not enhanced by the pretreatment of oxygen
plus 5 mg/1 ozone. Note the evidence of biodegradation in both
granular activated carbon adsorbers after about 175 days as shown
by the lack of increase in the "fraction remaining" from then to
the end of the test; see the lower portion of Figure 15.
Table 5 presents example calculations for the removal of
organic carbon as measured by TOC for the control and the 0~ +
5 mg/1 03 systems on two selected days. Note that although ^
the percent passing (fraction remaining) is nearly the same for
the two adsorbers, column 9 and Figure 15 (lower part), the load
on the control adsorber and the quantity removed (column 8) is
higher because of the lack of removal of TOC in the prior fil-
tration step. Therefore, unless sufficient biological activity
861
-------
I JULY
50 100 150
TIME. DAYS
AUG I SEPT | OCT | NOV I DEC
200
JAN
250
Figure 13. Removal of Organic Carbon By Granular Activated
Carbon With and Without Pretreatment
12
o
as
0.4--
Dual Media Filter On I v«
Data
50
I JULY I AUG
100 150 200
TIME, DAYS
SEPT I OCT I NOV I DEC I JAN
250
FEB
Figure 14. Removal of Organic Carbon By Dual Media
Filtration With and Without Pretreatment
862
-------
u
LU
O
O
O
I-
O)
c
'c
<0
E
c
o
o
CD
1.2
1.0--
0.8 --
0.6--
0.4--
0
0.8
•= 0.6 --
0.4--
0.2--
Dual Media Filter
Control Only
02 + 5 mg/LOa
FirstGAC Column Data
ONLY, 10 Min EBCT
® '
• •
•
• 02 + 5 mg/LO3
• Control
50
200
100 150
TIME, DAYS
JULY l AUG i SEPT i OCT i NOV i DEC I JAN I FEB
250
Figure 15.
Removal of Organic Carbon Individually by
the Dual-Media Filter and the First Adsorber
with and without Pretreatment
863
-------
Table 5. Concentrations of Organic Carbon at Various Points
in the Treatment Train for Selected Days
00
System Day Influent
TOC, mg/L
(1) (2) (3)
Control 91 1.9
02 + 5 mg/L 03 91 1.9
Control 133 2.1
02 + 5 mg/L 03 133 2.1
Contactor Dual-Media
Effluent Filter Effluent
TOC, mg/L TOC, mg/L
(4) (5)
2.0 2.0
1.7 1.2
2.0 1.9
2.0 1.4
% Passing
DM Only*
(6)
100
71
95
70
First
Granular Activated
Carbon Adsorber Effluent
TOC,
mg/L
(7)
0.8
0.4
1.1
0.8
ATOC,
by <;AC
mg/L
(8)
1.2
0.8
0.8
0.6
% Passing
GAC Only**
(9)
40
33
58
58
Percent Removal
Overall System***
(1-CE/C1)X100
(10)
58
79
48
62
See Upper part Figure 15
See Lower Part Figure 15
Sue Figure 13
-------
develops in the control adsorber to biodegrade this additional
adsorbed organic carbon, as may have been happening after the
200th day (lower part of Figure 15), a given effluent concentra-
tion should be reached sooner in the control adsorber as compared
to the O- + 5 mg/1 03 system. Planned adsorption equilibrium
studies should help determine whether the difference in the quan-
tity of TOC adsorbed by the test system adsorber is a concentra-
tion effect or because the dual-media filter effluent contains a
less adsorbable mixture of organic compounds.
Investigations were also carried out on the performance of
this pilot column system for the removal of DB-PFP as measured
by trihalomethane formation potential. The top section of
Figure 16 shows that when oxygen plus 5 mg/1 of ozone was added
to the system as pretreatment, the net effect through the gas
contactor, the dual-media filter, and the first granular acti-
vated carbon adsorber was beneficial. The effluent from that
system had a consistently lower "fraction remaining" of THMFP
than the control. This confirms the data from the previous
experience shown in Figure 12 and satisfies the remainder of the
first objective.
To investigate which unit process was responsible for the
improved performance, the "fraction remaining" of THMFP in each
of the unit processes, the gas contactor, the dual-media filter,
and the granular activated carbon adsorber, were compared indi-
vidually to their respective controls (lower three sections of
Figure 16). In contrast to the situation that occurred witn TOC
removal, in this case oxygen plus 5 mg/1 of ozone itself had some
influence on THMFP, as shown by a lower "fraction remaining" in
the gas contactor effluent as compared to the control (see second
section of Figure 16). Note, the equivalent data for TOC removal
were not plotted in Figure 15, as inspection of the data showed
no removal of TOC occurred through the gas contactor.
The third section of Figure 16 shows that, similar to the
situation with TOC removal, THMFP was being removed in the dual-
media filter during the latter portion of the test, probably
because of biological activity. Finally, as with TOC removal,
little difference was shown in the performance of the granular
activated carbon adsorber in spite of the addition of oxygen plus
5 mg/1 of ozone (see fourth section of Figure 16). THMFP was
evident in the control granular activated carbon adsorber as well
as in the test system.
Second Objective
To determine what minimum ozone dose was required to
produce the effect noted, the TOC removal data through the dual-
media filter portion of the treatment train for the "low ozone"
system, oxygen plus 0.5 mg/1 ozone, was compared to the control.
865
-------
z
o
5°"
1^
5
-—O"
O2 + 5mg/l O3
AUGUST
80
SEPTEMBER
100
OCTOBER
120
NOVEMBER
TIME, D*YS
Figure 16.
Removal of Trihalomethane Precursors by
Various steps in the Treatment Train
with and without Pretreatment
866
-------
Throughout the first 175 days of the test an average of 87 per-
cent TOC passed the dual-media filter in the "low ozone" system
as compared to 93 percent passing in the control, indicating
little difference. On day 175 the ozone dose was raised to
1 mg/1 and on day 191 to 1.5 mg/1. Inspection of the TOC data at
these increased doses showed some improvement in performance, but
the data are too sketchy to draw firm conclusions as to the
minimum dose required to produce some "effect" on TOC removal
through the dual-media filter.
Although the TOC removal data were not conclusive, an
attempt was made to determine the influence of raising the ozone
dose on the system by examining the Standard Plate Count (SPC) in
the effluent of the gas contactor. Table 6 compares the geome-
tric mean SPC for the three periods. These data show that nearly
complete elimination of the SPC in the gas contactor effluent did
not occur until an ozone dose of 1.5 mg/1 was reached. Although
this study was halted in late February, 1979, it will be repeated
in an attempt to determine what minimum ozone dose is necessary
to produce a favorable effect on the removal of TOC and THMFP.
Table 6. Effect of Ozone Dose on Microbiological
Content of Gas Contactor Effluent
Control Low Ozone System
Sample
Period
Days
0-174
175-189
190-238
Geom. Mean
SPC
#/ml
2,720
1,580
1,380
No. of
Samples
n
24
2
7
°3
Dose
mg/1
0.5
1.0
1.5
Geom. Mean
SPC
#/ml
3,560
280
6
NO. Of
Samples
n
23
2
7
Third Objective
The treatment train in which "oxygen only" was added to the
gas contactor was designed to determine whether any of the bene-
ficial effect in the oxygen plus ozone system was caused merely
by the presence of excess oxygen. For the first 175 days of the
test, oxygen was added to this system at a rate equivalent to the
oxygen in the treatment train receiving the lower ozone dose. On
the 175th day, the oxygen gas flow rate was raised to match the
quantity of oxygen being added to the gas contactor in the higher
(5 mg/1) ozone system. Figure 17 suggests that before the change
in the oxygen gas flow rate, with respect to TOC "fraction
867
-------
remaining," the "oxygen only" system was behaving the same as the
control. After the increase in the oxygen flow rate, however,
the TOC "fraction remaining" data indicated an improvement in the
performance of the "oxygen only" system to a point where it
equalled the performance of the system to which oxygen plus 5
mg/1 of ozone was being added.
1 2
U
\
LLJ
O
U
1 0--
c' 08
C
ra
£
cu
QC
06--
c
O
04-
Dual Media Filter
Data
Control
02 + 5 mg/L03
— Change 02 Cone.
•4-
0 50 100 150 200 250
TIME, DAYS
I JULY i AUG SEPT I OCT '• NOV I DEC ! JAN ! FEB I
Figure 17. Removal of Organic Carbon by Dual-Media
Filtration With and Without Pretreatment
Since this potentially important finding was only suggested
by these data, a new experiment was begun in March, 1979 that
consists of three parallel systems. In the control, settled Ohio
River water is directly applied to a dual-media filter. In a
second unit, settled water is oxygenated prior to filtration, and
in a third system an equal amount of oxygen plus 5 mg/1 of ozone
is added to the water prior to filtration. Comparison of TOC and
THMFP removals in these three systems should help demonstrate
whether or not the presence of ozone is needed in the gas to pro-
duce an improvement in performance, at least in Ohio River water.
As previously shown, much of the improvement in performance
of some of these systems was attributed to biological activity
occurring in the dual-media filter. Therefore, if the suggested
improved performance shown in Figure 17 did occur when the oxygen
content was raised, the increased microbiological activity in the
dual-media filter should have produced an increase in the SPC in
the filter effluent. Table 7 showed that this did not occur.
The reason for this is not known at this time (Spring, 1979) but
may have been: (1) because only four samples were collected
868
-------
following the change in oxygen dose, or (2) because the incuba-
tion temperature of the SPC test was so much higher than the
temperature of the water in the test that low recoveries of
organisms occurred/ or (3) because the higher numbers of organ-
isms were attached to the media and did not appear in the efflu-
ent. Detailed microbiological analyses will be made during the
new experiment described above; this should help clarify this
situation.
Table 7. Effect of Oxygen Dose on Microbiological
Content of Dual-Media Filter Effluent
Control Oxygen Only System
Sample
Period
Days
0-175
175-283
Geom. Mean
SPC
#/ml
673
470
No. of
Samples
n
27
8
°2
DOse
Low
High
Geom. Mean
SPC
#/ml
615
288
No. of
Samples
n
23
4
Finally, efforts to assess the impact of increasing the
oxygen gas flow rate on THMFP removal were thwarted because,
during this cold weather period, the THMFP concentration was
declining so rapidly in the influent to all of the systems, that
any change in performance which might have been caused by increas-
ing the oxygen gas flow rate was masked by this other occurrence.
Again, the new experiment should help resolve the issue of the
necessary presence of ozone in the gas to enhance DB-PFP removal.
Fourth Objective
In an effort to determine whether or not the expected bio-
logical growths could be contained in the filtration-adsorption
system, standard plate counts were determined for the influent
settled water and for samples taken at each intermediate point
in the treatment train. For the summer (from the start of the
experiment through September 21, 1978), these data show that
5 mg/1 of ozone reduced the geometric mean SPC from 2,900/ml in
settled water to 16/ml in the gas contactor effluent (see
Figure 18) while essentially no change occurred in the control.
Following the dual-media filter, however, the geometric mean
SPC had rebounded to 25,500/ml in the system receiving oxygenated
and ozonated water (5 mg/1), while the geometric mean SPC actu-
ally declined somewhat through the dual-media filter of the
control. This bacterial growth gives support to the contention
that the organic removal occurring in the dual-media filter
869
-------
10
O ,.3
u
LU
10
< 2
Q 10
z
<
CO
CD
O i
O 10
10'
oc
LU
H-
<
LU
O
H-
U
-
O
u
DUAL MEDIA C=CONTROL
FILTER
0V
GAC1
I
GAC2
GAC3
1
I
c o3 c o3 c o3 c o3 c o3
SUMMER
Figure 18. Standard Plate Count at Various Stages
of Treatment
870
-------
portion of the system is because of biodegradation. Measurement
of the dissolved carbon dioxide content in the dual-media filters
usually showed a higher concentration in the oxygen plus 5 mg/1
ozone system as compared to the control, providing further
evidence of biological activity.
In the oxygen plus 5 mg/1 ozone system, sampling of the
bacterial quality following each of the three granular activated
carbon adsorbers showed a steady decline in the geometric mean
SPC, indicating that microorganisms are being trapped in the
adsorbers and, therefore, that the microorganisms are being con-
tained in the system. These data do, however, reiterate the
necessity for postidisinfection following a process such as this,
and may indicate that a single granular media bed with a 10-
minute empty bed contact time may not be sufficient to contain
the microorganisms if growth is going to be so prolific in the
first unit process.
Continued monitoring of the standard plate count in the
effluent at various points in these two treatment trains through
the fall (September 21, 1978 to December 21, 1978) and winter
(December 22, 1978 through the end of the experiment in late
February, 1979), indicated that temperature had a profound
influence on the SPC in the various effluents. Table 8 shows
that, although the geometric mean SPC did not decline much in
the settled water through the seasons, a sharp drop occurred in
these effluent values from the dual-media filter and various
granular activated carbon adsorbers. These two systems, control
and oxygen plus 5 mg/1 dzone, will be continued through the
spring and summer of ]/979 and should indicate whether or not
the microbiological content of these processes will rebound in
the warmer weather.
Finally, some preliminary work on bacterial speciation
indicates that although the number of bacteria in the effluents
from the system treated with oxygen and 5 mg/1 ozone increases
in comparison to the control, the number of species of organisms
tends to decrease. If this finding can be confirmed with further
study, it would tend to indicate that certain species of organ-
isms can predominate in the treatment processes and that these
organisms are responsible for the major portion of the biodegra-
dation that is occurring in these systems.
Future Studies
Future studies on these unit processes will involve both
exploratory projects with Ohio River water and three field eval-
uation projects at other locations. Chlorination of coagulated
and settled Ohio River water will determine whether or not the
removal of organic halogenated compounds is influenced by the
presence of oxygen or oxygen plus ozona as a pretreatment, which
has been seen with other parameters such as TOC and THMFP
871
-------
Table 8. Influence of Water Temperature on SPC in
Effluents at Various Stages of Treatment
Standard Plate Count (SPC)
Geometric Mean (N/ml)
Control
Sunnier
Fall
Winter
Ozonated*
Summer
Fall
Winter
N
15
12
9
15
13
8
Inf.
(Settled Water)
2,900
1,500
1,440
2,900
1,500
1,440
Gas
Contactor
5,140
1,370
1,420
161°
7f
5*
Dual Media
Filter
1,230
350
469
25,550
10,440
408
GAC 1
2,170
202
67
10,590
595
44
GAC 2
650
165
41
3,810
330
10
GAC 3
420
134
48
1,360
205
17
Ozonated with 5 mg/1 0_ (together with excess 02).
°Positive values only, 9 were 0/ml
tPositive values only, 2 were below detection limit
Positive values only, 6 were below detection limit
Ground Water
100 GPM (560 mVd) Pilot Plant
*- Filter
Air + 03 or
Filtered
Water
•^ Adsorber
•+* Adsorber
High TOC and precursor, synthetic organics source
data collection begins December, 1979.
Figure 19. Oxidant + Adsorbent Studies in Miami, Florida
872
-------
(Figures 11, 12, 13, and 16). Additionally, oxidants other than
ozone, such as hydrogen peroxide, will be investigated to deter-
mine whether or not similar effects to those described above will
occur.
Figure 19 is the schematic diagram of a project being
planned in Miami, Florida. Here, the raw water has high concen-
trations of TOC and THMFP, as well as the presence of synthetic
organic contaminants. The major objectives are:
1. To compare the effectiveness of oxygen plus ozone
before GAG vs. GAC alone for the removal of TOC, THM
precursors and specific trace organic compounds of
health concern from a groundwater supply.
2. To study the bacterial population types that exist in
the various modes of treatment operations investigated.
At this time (Spring, 1979) no decision has been made
as to whether the ozone will be generated from oxygen
or air. If air is used the test streams will be "air
plus ozone" rather than "oxygen plus ozone" as in the
pilot plant studies described above and the influence,
if any, of oxygen itself will be reduced.
Figure 20 is a schematic diagram of the experimental set-up
in Philadelphia, Pennsylvania. Here the TOC and THMFP concentra-
tions of the river water are reasonably similar to those in the
Ohio River, but synthetic organic contaminants are in somewhat
higher concentration. In this case the ozone will be generated
from air. The major objectives are:
1. A bench- and pilot-scale study to determine the effec-
tiveness of oxygen plus ozone before GAC to remove
trace organic compounds of health concern from drinking
water.
2. Compare the cost effectiveness of oxidant plus GAC vs.
GAC only at this location.
Finally, Figure 21 is a schematic diagram of the study to
be conducted in Shreveport, Louisiana. In this case, synthetic
organic contaminants are nearly absent, but very high concentra-
tions of TOC and THMFP are present. Here the ozone can be gener-
ated from either bottled oxygen or air. The major objectives are
to:
1. Compare the efficiency of oxygen plus ozone before GAC
vs. GAC alone for the removal of THM precursors in a
high TOC concentration surface water.
2. Compare the efficiency of ozone only vs. ozone and
ultra-violet radiation (if current studies show it to
873
-------
Filtered
Water
Riverwater
20 GPM (112 m3/ d) Pilot Plant
i Adsorber
AIR » 03
2 Series Filters
Plant
Filtered-
Water
> 2 Series —
Adsorbers
Diverted from Full-Scale Plant
i ' 2 Series Adsorbers
Filter
AIR + 03
2 Series Adsorbers
Medium TOC and precursor, synthetic organics source
data collection begins summer, 1979.
Figure 20. Oxidant + Adsorbent Studies in
Philadelphia, Pennsylvania
Impoundment
10 GPM (56 m3, d) Pilot Plant
20 Minute
Retention
x Minute
Retention
*»2 Series Adsorbers
2 Series Adsorbers
2 Series Adsorbers
Air+0
-2 Series Adsorbers
High TOC and precursor, low synthetic organic source
data collection begins Fall, 1979
Figure 21. Oxidant + r^sorbent Studies in
Shrevepor,, Louisiana
874
-------
be cost effective) as the preliminary oxidation step
in the operation of the oxidant plus adsorbent columns
for the removal of THM precursors in a high TOG
concentration surface water.
3. Study the influence of contact time in the gas
contactor on system performance.
These three field evaluation projects/ combined with the
continued exploratory research using the pilot plant treating
Ohio River water, should provide a wide range of data under a
variety of operating circumstances related to the possibility
of combining the unit processes of oxidation and adsorption.
SUMMARY
Considering the three organic contaminant removal unit
processes reported in this paper, adsorption by synthetic resins,
aeration, and a combination of oxidant and granular activated
carbon adsorbent, along with what is already known about the per-
formance of granular activated carbon alone as an adsorbent,
leads to the conclusion that no one unit process is effective for
the removal of all categories of organic contaminants. In a
theoretical sense, all of the above named unit processes remove
some fraction of all organic compounds, but in a practical sense
many of these removals are so small as to be insignificant.
Figure 22 is an attempt to demonstrate the variability of
the different unit processes"for the removal of various types of
organic contaminants. The first bar (I) shows that only a small
portion of all organic compounds can be analyzed by gas chroma-
tography and mass spectrometry. The second bar (II) breaks the
analyzable portion of all organic contaminants down into seven
categories that will be discussed below in relation to their
treatability by the four unit processes under discussion.
Bar III, representing those organic contaminants adsorbed
by granular activated carbon alone, shows that this unit process
removes a wide range of organic contaminant types because of its
ability to remove non-GC analyzable compounds measured typically
as total organic carbon, AREA H, trihalomethane precursors,
AREA G (only partially GC-analyzable, AREA G is smaller in
Bar II), as well as a wide variety of synthetic organic chemi-
cals, AREA F, such as polychlorinated biphenyls, polynuclear
aromatic hydrocarbons, pesticides, and taste and odor producing
compounds. Further, some chlorinated solvents, AREA E, such as
trichloroethylene and tetrachloroethylene, and more soluble
chlorinated solvents, AREA D, such as chloroform, are well re-
moved by granular activated carbon adsorption when the adsorbent
is fresh. Finally, granular activated carbon also adsorbs some
heavy metals as indicated by AREA I.
875
-------
All Organics
GC-Analyzable j All || Others)
GC Analyzable
I [AJTJCMEI
£j
Adsorbed by GAC
Adsorbed by XE-340 Resin
|B|C|0|E|F|
Removed by Aeration
Removed by Oxidant + Adsorbent
AREA EXAMPLES
A Vinyl Chloride
B 1, 1- & 1, 2-Dichloroethane
C Methyl Chloroform
D Chloroform
E Tri- and Tetrachloroethylene
F PCB, PAH, Pesticides, Taste and Odor Compounds
G THM Precursors
H Compounds Measured Typically as TOC
I Heavy Metals
J Reduced Iron & Manganese, Sulfide
K Ammonia
Note: Bar Length
somewhat
Related to
Effectiveness
Figure 22. Summary of Treatment Effectiveness
876
-------
Bar IV shows that the synthetic resin Ambersorb XE-340
removes some of the same materials that are reasonbly well
adsorbed by granular activated carbon/ AREAS D and E, but is not
effective for all compounds represented by AREA F (smaller than
in Bar III) and not effective for organic compounds measured as
TOC, AREA H. Also AREA G is smaller in Bar IV indicating that
less trihalomethane precursor material is adsorbed by XE-340 than
by granular activated carbon. Finally, heavy metals, AREA I, are
not adsorbed by XE-340. On the other hand, XE-340 is effective
for removing compounds represented by AREA Bf for example, 1,1-
and 1,2-dichloroethane/ and AREA C, for example, methyl chloro-
form!/ which are compounds poorly adsorbed by granular activated
carbon but well removed by XE-340. AREA D is larger than in
Bar III showing the improved performance of XE-340 for chloroform
removal.
Bar V demonstrates that aeration can remove a class of
organic compounds that cannot be well removed by adsorption;
AREA A, as represented by vinyl chloride. In addition, aeration
removes some compounds that are well adsorbed on XE-340, AREA C,
and adsorbed both by XE-340 and granular activated carbon, AREA E,
and some taste and odor compounds from AREA F. In addition to
removing these organic contaminants^ aeration can remove two
categories of inorganic contaminants; AREA J, representing iron,
manganese, and sulfide; and AREA K, representing ammonia nitrogen.
Bar VI shows the demonstrated improvement of the combined
treatment of oxidation and adsorption on granular activated
carbon as compared to granular activated carbon adsorption alone.
In Bar VI, AREA G and AREA H are larger than in Bar III indicat-
ing that this combination treatment is more effective for remov-
ing trihalomethane precursor compounds and compounds measured
typically as total organic carbon than granular activated carbon
adsorption alone. An economic analysis must be made, however, to
determine whether or not the increase is great enough to justify
the expense of the oxidant step (6). In addition, this combina-
tion treatment can remove ammonia nitrogen, AREA K, and reduced
inorganics, AREA J.
Figure 22 shows that the broadest spectrum of organic
contaminants can be removed by using granular activated carbon
alone or granular activated carbon plus pretreatment with an oxi-
dant, but if contaminants are present that are poorly removed by
adsorption or biodegradation, then the alternative of aeration or
adsorption on synthetic resins either alone or in combination
with granular activated carbon adsorption may be a more effective
method for controlling these specific compounds, depending on
individual circumstances.
877
-------
ACKNOWLEDGEMENTS
The authors wish to thank several people who contributed to
the collection of data upon which this paper was based. These
include: the local personnel at each site where data were
collected, Mr. Bradford Smith/ Mr. Dennis Seeger, and Ms. Clois
Slocum who provided much of the organic analysis data;
Mr. Kenneth Kropp who built much of the experimental apparatus
and who/ along with Ms. Dianne Hinderberger, operated the pilot
plant; and Mr. Raymond Taylor and Prof. John Caruthers who pro-
vided the microbiological data. Finally the authors wish to
thank Mr. Gordon G. Robeck, Mr. Alan A. Stevens and Mr. Edwin E.
Geldreich for their valuable reviews and Ms. Patricia Pierson
who typed the manuscript.
878
-------
REFERENCES
1. Control of Organic Chemical Contaminants in Drinking Water,
Federal Register, 43, No. 28, 5756-5780, February 9, 1978,
Part II.
2. McCarty, P.L., Reinhard, M., Dolce, C., Nguyen, H., and
Argo, D.G. 1978. "Water Factory 21: Reclaimed Water,
Volatile Organics, Virus, and Treatment Performance,"
EPA 600/2-78-076, U.S. Environmental Protection Agency,
Cincinnati, Ohio. 88 pp.
3. Rice, R.G. "Biological Activated Carbon," Proceedings of
a Seminar on Control of Organic Chemical Contaminants in
Drinking Water, U.S. Environmental Protection Agency,
Washington, D.C. In Press.
4. Sontheimer, H., Heilker, E., Jekel, M., Nolte, H., and
Vollmer, F.H. 1978. "The Mulheim Process," JAWWA, TQ_,
393-396
5. Eberhardt, M., Madsen, S., and Sontheimer, H. 1974.
"Untersuchungen zur Verwendung biologisch arbeitender
Aktivkohlefilter bei der Trinkwasseraufbereitung" Engler-
Bunte-Institute der Universitat Karlsruhe, Heft 7,
Karlsruhe, F.R.G. 86 pp.
6. Clark, R.M., and Dorsey, P. "Influence of Operating
Variables on the Cost of Granular Activated Carbon
Adsorption Treatment," pp. 705-742 in this Symposium.
879
-------
A CONTRIBUTION TO THE DISCUSSION ON BIOLOGICAL PROCESSES IN
GRANULAR ACTIVATED CARBON FILTERS
D. van der Kooij
Granular activated carbon (GAC) filters have been intro-
duced and still are applied in drinking water treatment for the
removal of taste and odor compounds and residual chlorine. More
recently, the removal of specific toxic organic substances
originating from wastewaters, as well as from chlorination
processes, has become important. The specific applicability of
GAC filtration for these purposes is based on the adsorptive
capacity of GAC toward many organic compounds.
In the GAC filters, biological activity has generally been
observed (1-5, 7-9, 13, 14). Many of the investigators suggested
that the adsorption processes enhance the microbial degradation
(oxidation) of organic compounds, by increasing both the sub-
strate concentration in the filter bed and the contact time
between the substrate and the organisms. In recent papers (3,8)
it was emphasized that ozonation of water prior to GAC filtration
results in an improvement of the biological oxidation processes
in the GAC filters.
The general attention paid toward the biological activities
in GAC filters might suggest that these processes are considered
as being more important than the adsorption processes. In this
context, the following remarks can be made:
1. An enhancement of biological degradation of organic com-
pounds by adsorption processes in GAC filters used in drink-
ing water treatment has never been directly proved. On the
contrary, evidence has been presented (6, 10, 11), that
microorganisms do not have the adsorbed compounds at their
disposal. This evidence is based on the observation that
the die-off rates of different types of micr©organisms•*—
coliforms, pseudo-monads and the bacteria assessed by the
colony count—present on GAC (Norit ROW 0.8 Supra) and a
non-activated carbon (NAC) taken from filters, which had
been supplied with rapid sand filtered river water during a
period of one year, were equal for identical groups on both
types of the carbon (10).
This observation indicated that the organic load of GAC did
not influence the microorganisms. Moreover, it has been
880
-------
shown (11) that the increase in the number of microorganisms
on sand, NAC, and GAC in filters supplied with tap water was
not influenced by the type of filter material. The final
levels of the numbers of microorganisms on these materials
were only slightly different, which is in contrast with the
large difference in adsorption capacity between GAC on one
hand and NAC and sand on the other. Experiments also
revealed that microbiological activity on exhausted carbon
does not restore the adsorption capacity (6), whereas
phenol, which is both adsorbable and biodegradable, was
oxidized only partially or not at all by bacteria in the
presence of GAC (15). Based on these observations, it is
concluded that the so-called biological regeneration of GAC
in filters used for the preparation of drinking water is
unlikely to occur. Moreover, it may be concluded that
adsorption of biodegradable compounds results in an unneces-
sary load on the carbon and prevents the complete removal of
these compounds by a biological process. For these reasons,
GAC filtration should always be preceded by another filtra-
tion stage in which biological processes remove biodegrad-
able compounds from the water.
The level of biological activity in GAC filters is influ-
enced by easily biodegradable compounds coming onto the
filter. Many of these compounds have low molecular weights
and adsorb very poorly. A high filtration velocity results
in a relatively large supply of these compounds per unit
of filter volume. This partially explains why the bio-
activity and numbers of microorganisms on GAC filters are
larger than in slow sand filters (11). On the other hand,
the biological activity in a GAC filter may be larger than
in other filters in similar conditions because in the GAC
filters a relatively large surface area per unit of filter
volume is available for the attached growth of micro-
organisms. Indications of this aspect were obtained in
experiments in which the performances of a GAC filter and a
NAC filter were compared. In one experiment, these filters
were supplied with river water pretreated by rapid sand
filtration.
Bacteriological investigations (Figure 1) revealed that
the numbers of bacteria measured as colony-forming units
per milliliter (c.f.u./ml) in the effluents did not differ.
However, the numbers of bacteria on the filter materials
(c.f.u./ml of carbon) sampled from the effluent end of the
filters were two to three times higher on GAC than on NAC.
Assuming a certain equilibrium between the number of
microorganisms in the water (c.f.u./ml) and the number of
microorganisms per unit of surface area of filter material,
these observations may indicate that the specific surface
area (cm /cm ) available for attachment of microorganisms
881
-------
10
10
3[
10-
ef.fly.ent
o-/
L.
, as--*"
,. . *• ^ ^-
• AC
o NAC
20 40 60
cum.% of samples
80
100
Figure 1. Cumulative Frequency Distribution of the
Colony Counts on GAC and NAC and of the
Effluents of Filters Supplied with Rapid
Sand Filtered River Water
882
-------
to GAC is two to three times higher than the corresponding
value for NAC.
3. The stimulation of biological activity in the GAC filter by
preozonation is not specific for these filters. Experiments
performed by the author (12) showed that the amount of
assimilatable (biodegradable) organic carbon (AOC) produced
by ozonation was effectively reduced by a simple dual-media
(anthracite-sand) filtration (v=15 m/h; h-2.2 m). An addi-
tional GAC filtration only resulted in a small additional
AOC reduction (Table 1).
Table 1
The Effect of Ozonation and filtration
(Waterworks of Rotterdam) on the Organic Substance
in the Water as Measured by AOC, DOC, ancj uv
Concentrations of Organic Substances
DOC UV ADsorbanoe*
( g org. C/l) (mg org. C/l) (m-1)
Raw water from storage reservoir 39 5.2 3.32
After ozonation 270 2.9 1.92
After double-layer filtration 41 2.9 1.68
After granular activated carbon 15 2.4 1.16
filtration
tAssimilatable (degradable) organic carbon
^Measured as absorbance at 254 nm.
Note: Average water temperature = 16°C.
4. Combination of ozonation and GAC filters results in an in-
creased operating time, when the TOC concentration in the
final effluent is used as a criterion. However, this in-
crease in the allowable running time before TOC breakthrough
may be irrelevant when the breakthrough criterion is keyed
to specific organic compounds that are not influenced by
ozonation.
883
-------
REFERENCES
1. Eberhardt, M., 1971. Untersuchungen zur optimalen
Kombination von Adsorption, Filtration and biologischer
Reinigung. Veroffentl. Ber. Lehrstuhl Wasserchem., 5 :
358-379, Karlsruhe.
2. Eberhardt, M., Madsen, S. und Sontheimer, H., 1975.
Untersuchungen zur Verwendung biologisch arbeitender
Aktivkohlef ilter bei der Trinkwasserairfbereitung.
GWF-Wasser/Abwasser, 116 : 245-247.
3. Guirguis, W., Cooper, T. , Harris, j., and Ungar, A., 1978.
Improved performance of activated carbon by preozonation
Jour. Wat. Poll. Contr. Fed., 50 : 308-320.
4. Kolle, W. and Sontheimer, H., 1974. Experience with
activated carbon in West-Germany. In: Activated carbon in
water treatment; Papers and Proceedings of a Water Research
Association Conference; Reading, April 1973.
5. Love, O.T., Robeck, G.C., Symons, J.M. and Buelow, R.W.,
1974. Experience with activated carbon in the USA. In:
Activated carbon in water treatment; Paper and Proceedings
of a Water Research Association Conference; Reading, April
1973.
6. Maqsood, R., Benedek, A./ 1977. Low-temperature organic
removal and denitrification in activated carbon columns.
Jour. Wat. Poll. Contr. Fed., 49; 2107-2117.
7. Sontheimer, H., 1974. Use of activated carbon in water
treatment practice and its regeneration. Int. Water Suppl.
Ass. Congress, Brighton.
8. Sontheimer, H., Heilker, E., Jekel, M.R., Nolte, H. and
Vollmer, F., 1978. "The Mulheim process." Jour. Am. Wat.
Wks. Ass., July 1978 : 392-396.
9. Symons, J.M., 1978. Interim treatment guide for control-
ling organic contaminants in drinking water using granular
activated carbon. EPA Rep. keg. nr. 43-28, 57faO-5780.
884
-------
10. van der Kooij, D., 1978. Some investigations into the
presence and behavior of bacteria in activated carbon
filters. Transl. Rep. Spec. Probl. Wat. Technol.
EPA/600/9-76-030 : 348-354.
11* van der Kooij, D., 1978. Investigations concerning the
relation between microorganisms and adsorption processes in
granular activated carbon filters. Contribution to Conf.
Oxidation Techniques in Drinking Water Treatment,
Karlsruhe; September 1978. EPA/570/9-79-020:689-701
12. van der Kooij, D., 1978. Determination of the aftergrowth
potential of tap water. Proceedings of the seminar on
development of methods for determining water quality. Nov.
1978. Rotterdam.
13. Weber, W.J., Hopkins, C.B. and Bloom, R., 1970; Physico-
chemical treatment of wastewater. Jour. Wat. Poll. Contr.
Fed. , j42 : 83-99.
14. Weber, W.J., Friedman, L.D. and Bloom, R., 1972.
Biologically-extended physicochemical treatment. Adv. Wat.
Poll. Res., Proc. 6th International. Conf. 641-656,
Jerusalem.
15. Werner, P., M. Klotz and R. Schweisfurth. 1978.
Mikrobiologische Unfcersuchungen zur Aktivhohlefiltration.
Contribution to the Conf. on Oxidation Techniques in
Drinking Water Treatment, Karlsruhe, September 1978,
EPA-570/9-79-020.
885
-------
&EPA
NATO
CMM
• OTAN CCMS
NATO-CCMS
886
-------
CLOSING REMARKS
887
-------
NAT*
••TAN
NATO-CCMS
888
-------
CLOSING REMARKS
PROF. DR. HEINRICH SONTHEIMER
For someone who has proposed such a conference, as I have done,
who indicated some points of interest on which papers could be
presented, who has helped to choose speakers, and so on, this
meeting can be viewed in a very different manner than by someone
who is only a participant. I have to ask, was it really worth-
while to have this meeting? Did people get enough information,
particularly new information? Has the experience from pilot
plants and large waterworks been presented in such a way that
others really could use it? Or, since it is a NATO-CCMS meeting,
did the papers presented for discussion really help to overcome
the challenges to our modern society and help to foster a better
understanding between nations and their people? I can only base
my response to all these questions on my own impressions and on
things I have learned myself during these 3 days.
There is a great deal of research going on in many countries on
the best way to use granular activated carbon filters or other
adsorption processes. Pilot plants are being used for this
purpose and although methods and study goals are similar in some
ways, there are enough dissimilarities so that it is possible to
learn how one should proceed further in such studies to provide
more reliable research.
I have derived many new ideas from the variety of very interest-
ing reports presented, and it was interesting to realize that, in
the long run, the best results are obtained by a sound combina-
tion of very practical studies and fundamental research. Some
very good examples of this were given in one paper. For me, it
confirms that this combination will be the best for the future.
I have to say that besides the effect of such studies, I am very
impressed when real practical conclusions are drawn from such
pilot studies, leading to real changes in existing plants, and
through these to more safe and better drinking water. One
example of this, in addition to many others, was demonstrated
in the report of Mr. Schulhof from France. This special example
also shows that we often don't need much money to rearrange our
waterworks and to utilize new experience to optimize drinking
water treatment. One could learn from Mr. Schulhof's, as well as
other papers, that new techniques are not always more expensive
than traditional treatment. This confirms my hope that the new
889
-------
techniques really can help to solve some of our modern water
quality problems; I think we should continue in this direction.
We all have learned during these 3 days that activated carbon
filters and their optimum use has to be seen in context with all
the other treatment methods. Best possible use of the activated
carbon can also be achieved by promoting biological treatment
within these filters. There will be many solutions for this
combination in order to find the best solution in each case.
I have found that we can also learn much from the experience
gained in advanced wastewater treatment. I think this has been
made very clear in Perry McCarty's paper on Water Factory 21.
We all know that much research has been done in the U.S.A.,
especially on advanced waste treatment technologies. I think
we can apply much of the knowledge from this type of waste water
treatment to drinking water treatment. I was especially happy
to have Professor McCarty's paper confirm many of the findings
of a research study we started a few months ago/ using counter
current aeration in conjunction with activated carbon filters for
the total removal of organic trace contaminants, both volatile
and nonvolatile.
I could add many more points to my concluding remarks, but I
don't want to repeat information already mentioned and take up
too much of your time. All of these papers will be published
soon, in addition to the papers from the Karlsruhe conference, so
you can read them and can see what is useful for each particular
case.
Finally, I have to thank all of you for taking part in this
meeting. I was really astonished to see so much interest in our
lengthy program, and it speaks for this program that so many
people are still here. I also want to thank all participants in
the discussions; the only regret is that we did not have enough
time for proper discussion. I apologize for that.
I especially have to thank all of the speakers, who did a very
fine job involving a lot of preparation for their excellent
slides and papers. It was worthwhile to listen to all of them.
Everything worked out well; I wish to thank all the people that
prepared this conference. Today, I especially want to thank one
person, Joe Cotruvo, who did a lot of work to assure that this
conference would be a success. I think he has succeeded. I
surely hope that this won't be the last time that we arrange such
an international conference, and with this hope in mind I want to
finish my remarks.
I hope you all had a good time and I wish all of you success in
all of your undertakings and work. I'm very grateful that you
have been here.
890
-------
DR. COTRUVO
I thank you, Prof. Sontheimer, for the kind words, but as you
know, there were many many people that put this symposium
together. All of the people from the EPA's Office of Research
and Development, the Municipal Environmental Research Laboratory,
who put in so much time, Jim Symons, Gordon Robeck, their staffs,
and all of the people from my office, including Craig Vogt and
Bruce Capner. They have been with us throughout these past few
days, and they deserve that credit.
I would like to turn the mike over to Victor Kimm for a few
minutes; he wanted to say some words of farewell to you.
•
DR. VICTOR KIMM
I feel that there ought to be some kind of endurance award for
the people who remain.
There are two or three specific things I'd like to mention. I
feel compelled to say that, in a job like mine, you find one foot
in the world of politicians and one foot in the science world,
making you unqualified to address either world correctly. I find
it much more productive to talk about science to the politicians
and policy to the scientists. Perhaps it's best for me to stay
in that mode to survive.
First I'd like to say something by way of a special vote of
thanks to our international visitors who participated in the pro-
gram. Few, if any of us, could attempt a technical presentation
in a language other than our own. I suspect the preparations
brought many anxious moments to our visitors. I thank them; we
are indeed indebted.
One of the purposes of the conference is to open a dialogue on
this new challenge to the water supply profession. Much of the
discussion I have seen here has been along those lines. It
appears that the conference has been very successful in intro-
ducing people with similar interests to one another, for which
we are most thankful.
My final point, essentially directed more to my U. S. waterworks
colleagues, was to try to return somewhat to our original theme.
That is, if we are to remove trace organic contaminants from our
891
-------
drinking waters, it is going to take a collaborative effort. It
is something that will involve the water utilities, the academic
community, consulting engineers, and regulatory officials at the
Federal and State levels. I'm very pleased that all of those
groups were represented here at this meeting and I hope that we
can continue the dialogue that we have begun here as we face this
challenge in our domestic regulatory framework.
DR. COTRUVO
I believe the theme of the program would have to be that, when
one encounters contamination problems in drinking water, one has,
indeed, a tremendous number of tools available to try to meet
those problems. There really are no limits because there are
choices among combinations of biological, chemical, and physical
treatment processes which, when applied to the given problem, can
be arranged to produce the optimum solution. This means that the
present moment is an extremely exciting time to be a water
supply engineer or scientist.
I think there is an opportunity to bring together all of the
different technical tools that we presently have, chemistry,
engineering, etc., to produce a product that is the best possible
within the economic limitations that constrain us. There is an
opportunity to let our imaginations range widely, to innovate and
to see the products of our efforts actually brought to fruition
to benefit the public.
I hope that the conference has been enjoyable. It's been an
opportunity for you to let your imagination wander a bit, to meet
many new people, to learn or at least to be introduced to many
new concepts, and an opportunity to take those ideas home with you
and let them ripen a bit before beginning to apply them in your own
system.
I thank all of the participants, our foreign visitors, our
American friends, and particularly Professor Sontheimer. Until
the next time, I wish you farewell.
892
•D.S. OOVEfOKENT PRDHIBG OFFICE I 1984 Q-tZl-082/537
------- |