H
United States Environmental
Protection Agency
Association off Metropolitan
Water Agencies
Final Report for
Disinfection By-Products in
United States Drinking Waters
Volume 1 - Report
November 1989
JNM
.ery
Con
Metropolitan Water District ol Southern Calik :
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
ASSOCIATION OF METROPOLITAN WATER AGENCIES
DISINFECTION BY-PRODUCTS IN
UNITED STATES DRINKING WATERS
FINAL REPORT
November 1989
Metropolitan Water District of Southern California
James M. Montgomery, Consulting Engineers, Inc.
-------
PROJECT STAFF
Metropolitan Water District of Southern California
Michael J. McGuire, Ph.D.
Stuart W. Krasner
Kevin M. Reagan
James M. Montgomery, Consulting Engineers, Inc.
E. Marco Aieta, Ph.D.
Joseph G. Jacangelo, Ph.D.
Nancy L. Patania
Karl M. Gramith
-------
ACKNOWLEDGEMENTS
This report has been made possible through the dedicated efforts of a large number of
people. In particular, we would like to acknowledge the the thirty-seven utilities that
participated in this study as well as the following individuals:
United States Environmental Protection Agency
James J. Westrick. Project Officer
Patricia Snyder Fair
Association of Metropolitan Water Agencies
Ken Kirk
Diane Van De Hei
California Department of Health Services
David Spath
Metropolitan Water District of Southern California
Russell Chinn
Eric Crofts
Melissa Dale
Cordelia Hwang
Bart Koch
Ching Kuo
Fong-Yi Lieu
Ted Lieu
Margaret Moylan
Mary Romero
Patricia Rottler
Warren Schimpff
Michael Sclimenti
Suzanne Teague
Randy Whitney
Roniece Widman
James M. Montgomery, Consulting Engineers, Inc.
Martha Frost
Ulises Gil
Yolanda Gutierrez
Seth Jelen
Judy Richardson
Carol Tate
Rhodes Trussell
Jennifer Yarbrough
-------
TABLE OF CONTENTS
Number
VOLUME 1
EXECUTIVE SUMMARY
Introduction ES-1
Utility Selection ES-2
Methodology ES-2
Data Management and Analysis ES-2
Baseline Sampling Results and Discussion ES-3
Treatment Modification Studies -
Results and Discussion ES-6
Ozonation Studies ES-6
Chlorine Dioxide Studies ES-8
Coagulation Studies ES-9
Granular Activated Carbon Study ES-10
Summary and Conclusions ES-11
SECTION 1 - INTRODUCTION
Project Background and Objectives 1-1
Project Objectives 1-1
Target Compounds 1-2
California Department of Health Services/
California Public Health Foundation DBP Study 1-2
Project Description 1-2
Baseline Data Collection 1-3
Treatment Modification Studies 1-3
SECTION 2 - UTILITY SELECTION
Information Request 2-1
Selection Matrix 2-1
Process Trains of the Participating Utilities 2-2
SECTION 3 - METHODOLOGY
Sampling Procedure 3-1
-------
TABLE OF CONTENTS (CONTINUED)
Page
Number
Analytical Methods 3-2
Simulated Distribution System Testing 3-2
Quality Assurance 3-5
On-Site Extraction Studies 3-8
Aldehyde Derivatizations/Extractions On-Site 3-11
Preservation of TOX Samples 3-11
SECTION 4 - DATA MANAGEMENT AND ANALYSIS
Database 4-1
Data Handling Protocol 4-1
Summary Statistics 4-2
Graphical Presentation of Data 4-4
Box-and-Whisker Plots 4-4
Bar Charts 4-5
Star Symbol Plots 4-5
SECTION 5 - BASELINE SAMPLING RESULTS AND DISCUSSION
Overview of Baseline Data 5-1
Star Plot Analyses 5-3
Seasonal Variation 5-4
Influent Water Quality 5-5
Classes of DBP Compounds 5-5
Variation by Treatment Type 5-7
Variation by Source Water Type 5-8
Variation by Disinfection Scheme 5-8
Influent Parameters 5-8
Classes of DBP Compounds 5-9
DBP and Influent Parameter Correlations 5-11
Correlations with THMs 5-12
-------
TABLE OF CONTENTS (CONTINUED)
Page
Number
Correlation with HAAs 5-12
Correlations with Influent Water Quality
Parameters 5-12
Correlations with Bromide and Brominated DBPs 5-13
Correlations with Chloroform 5-14
Correlations with DCAA and TCAA 5-15
Correlations with Formaldehyde 5-15
Correlations with Total Organic Halide 5-15
Special Issues 5-16
Effect of pH 5-16
Effect of Temperature 5-17
Brominated DBPs 5-18
Aldehydes 5-20
Cyanogen Chloride Results 5-21
DBP Levels of Disinfection-Only Utilities 5-22
Removal of TOC During Treatment 5-22
Comparison of THM Levels from USEPA DBP Study
and AWWARF THM Survey 5-23
Comparison of USEPA Study and CDHS DBP
Study Results 5-24
SECTION 6 - TREATMENT MODIFICATION STUDIES - RESULTS
AND DISCUSSION
Ozonation Studies 6-1
Utility 6 6-1
Utility 7 6-3
Utility 19 6-4
Utility 25 6-6
Utility 36 6-8
Discussion 6-10
Chlorine Dioxide Studies 6-12
Utility 16 6-12
Utility 37 6-13
Di scussion 6-14
Coagulation Studies 6-15
Utility 3 6-15
Utility 12 6-17
Discussion 6-18
Granular Activated Carbon Study 6-19
Process Description 6-19
GAC Contactor Operation 6-20
Treatment Study Description 6-20
Simulated Distribution System Testing 6-21
-------
TABLE OF CONTENTS (CONTINUED)
Page
Number
Simulated Distribution System Testing
Results and Discussion 6-21
DBF Levels Produced by Conventional Treatment:
Results and Discussion 6-26
SECTION 7 - SUMMARY AND CONCLUSIONS
Baseline Results 7-1
Process Modification Results 7-3
Research Needs 7-4
REFERENCES
VOLUME 2
APPENDIX A - Questionnaire
APPENDIX B - Sampling Instructions, Sample Information Sheet
APPENDIX C - Analytical/Quality Assurance Methods
APPENDIX D - Correlations
-------
TABLES
Following
Number Page
ES-1 List of compounds targeted in study ES-I
ES-2 Disinfection by-products in drinking water study:
Utility selection matrix ES-2
ES-3 Disinfection by-products in drinking water: Summary
of baseline sampling median values ES-3
1 -1 List of compounds targeted in study 1-2
2-1 Disinfection by-products in drinking water study:
Utility selection matrix 2-1
3-1 Sampling kit contents 3-1
3-2 Sample holding times 3-1
3-3 Utility 11 GAC study: Day 13 DBF results 3-3
3-4 Utility 11 GAC study: Day 25 DBF results 3-3
3-5 Utility 11 treatment study, comparison of GAC
effluent SDS tests: Effect of chlorine dose
and pH on DBFs 3-4
3-6 Minimum reporting levels (MRLs) 3-6
3-7 DBF quality control limits 3-7
3-8 Utility 2: Chlorinated filter effluent on-site
and lab extraction 3-8
3-9 Utility 2: Chloraminated clearwell effluent
on-site and lab extraction 3-8
3-10 Utility 33: Chlorinated effluent on-site and
lab extraction 3-9
3-11 Utility 6: Chloraminated clearwell effluent
on-site and lab extraction 3-10
3-12 Pentane - extractable disinfection by-products:
Matrix holding study 3-10
-------
Tables (continued)
Following
Number Page
3-13 Utility 33: Aldehyde results on-site and lab
extraction 3-11
3-14 Utility 6: Formaldehye results on-site and lab
extraction 3-11
3-15 TOX preservation study 3-12
3-16 On-site TOX preservation studies 3-12
5-1 List of abbreviations used in DBF study 5-1
5-2 Disinfection by-products in drinking water: Summary
of baseline sampling median values 5-1
5-3 Comparison of seasonal DBF levels 5-6
5-4 Comparison of seasonal TTHMs and influent
water quality 5-7
5-5 DBF concentrations at utility with seasonal change
in bromide levels 5-18
5-6 DBF concentrations at inland utilities with high
bromide levels 5-19
5-7 Utility 12: Influence of saltwater intrusion 5-19
5-8 Aldehyde levels in plant influents and effluents 5-20
5-9 Levels of aldehydes and THMs in selected cleanvell
effluents 5-21
6-1 Disinfection scenarios, plant scales and sources
for utilities participating in study 6-1
6-2 Utility 6 treatment study 6-1
6-3 Utility 7 treatment study 6-3
6-4 Utility 19 treatment study 6-5
6-5 Utility 25 treatment study 6-6
6-6 Explanation of abbreviations used in Utility 25
DBF profiles and residence time of distribution
system sampling points 6-7
-------
Tables (continued)
Following
Number Page
6-7 Utility 36 treatment study 6-8
6-8 Disinfection by-products concentrations before
and after ozone addition at Utility 36 6-8
6-9 Utility 16 treatment study 6-12
6-10 Utility 37 treatment study 6-13
6-11 Utility 3 treatment study 6-16
6-12 Percent removal of total organic carbon and
ultraviolet absorbance at 254 nm by various alum
doses at Utility 3 6-16
6-13 Utility 12 treatment study 6-17
6-14 Percent removal of total organic carbon and
ultraviolet absorbance at 254 nm by various alum
doses at Utility 12 6-17
6-15 Concentrations of various disinfection by-products and
total organic halide at varying alum doses at
Utility 12 6-18
6-16 Utility II treatment study sampling plan 6-20
6-17 Utility 11 sampling locations 6-20
6-18 Utility II treatment study 6-21
6-19 Utility 11 treatment study TOX results (in //g/L) 6-25
6-20 Utility 11 treatment study comparison of SDS and
distribution system results 6-26
-------
FIGURES
Following
Number Page
ES-1 Guide to notched box-and whisker plots ES-3
ES-2 Disinfection by-product concentration by DBF class -
four quarters ES-3
ES-3 Percent of sum of DBF class medians by DBF class -
four quarters ES-3
ES-4 Influent total organic carbon by quarter ES-3
ES-5 Influent total organic carbon by source ES-3
ES-6 Trihalomethanes by quarter ES-3
ES-7 Trihalomethanes by influent temperature ES-3
ES-8 Trihalomethanes by disinfection scheme ES-4
ES-9 Correlations with trihalomethanes ES-4
ES-10 Correlations with influent chloride and bromide ES-4
ES-11 Correlations with influent bromide ES-5
ES-12 Correlations with dichloroacetic acid ES-5
ES-I3 Mean total organic carbon removal through
filtering utilities' processes ES-5
ES-14 AWWARF 12-quarter vs. USEPA 4-quarter means ES-5
ES-15 Effect of CI2,NH2C1 and O3,C12,NH2C1 on DBF
formation (Utility 6) ES-6
ES-16 Effect of C12, C12/NH3 & O3/NH3/C12 on DBF formation
(Utility 7) ES-6
ES-17 Effect of various disinfection schemes on
disinfection by-product formation at
Utility 19 ES-7
ES-18 Effects of NH2C1 and and O,/NH2C1 treatment on total
THM formation at Utility 25 ES-7
-------
Figures (continued)
Following
Number Page
ES-19 Effect of various disinfection schemes on
disinfection by-product formation at Utility 36 ES-7
ES-20 Effect of Cl, and C1O2/C12 on DBF formation
(Utility 16) ES-9
ES-2I Effect of Cl, and C1O2 on DBF formation
(Utility 37) ES-9
ES-22 Effect of alum dose on DBF formation (Utility 3,
SDS) ES-9
ES-23 Effect of alum dose on DBF formation (Utility 12) ES-9
ES-24 TOC vs run time for GAC column influent and effluent
Utility 11 ES-10
ES-25 SDS TTHMs vs run time for GAC column influent and
effluent, Utility 11 ES-10
I -1 Structural formulas for disinfection by-products 1-2
2-1 Utilities participating in DBF study 2-2
2-2 Process trains of participating utilities 2-2
4-1 Data-sampling protocol 4-1
4-2 Guide to notched box-and-whisker plots 4-4
4-3 Example of notched box-and-whisker plot for
three data sets 4-5
4-4 Guide to bar charts using median, percentiles and
minimum reporting level 4-5
4-5 Example guide to star plots 4-6
5-1 Disinfection by-product concentration by DBF class -
four quarters 5-1
5-2 Percent of sum of DBF class medians by DBF class -
four quarters 5-1
5-3 Percent of sum of non-THM DBF class medians by class -
four quarters 5-2
-------
Figures (continued)
Following
Number Page
5-4 Trihalomethane concentrations by THM compound -
four quarters 5-2
5-5 Percent of sum of THM compound medians by compound -
four quarters 5-2
5-6 Haloacetic acid concentrations by HAA compound
four quarters 5-2
5-7 Haloketone concentrations by HK compound - four
quarters 5-2
5-8 Haloacetonitrile concentrations by HAN compound -
four quarters 5-2
5-9 Miscellaneous DBF concentrations by compound -
four quarters 5-2
5-10 Aldehyde concentrations by compound -
four quarters 5-2
5-11 Guide to star symbol plots 5-3
5-12 Plant influent by utility ID number - summer
quarter 5-3
5-13 Plant influent by source water type -
summer quarter 5-3
5-14 Plant influent by treatment type -
summer quarter 5-4
5-15 Plant influent by disinfection scheme -
summer quarter 5-4
5-16 Influent total organic carbon by quarter 5-5
5-17 Influent UV absorbance by quarter 5-5
5-18 Influent chloride by quarter 5-5
5-19 Influent bromide by quarter 5-5
5-20 XDBPsum by quarter 5-5
5-21 Trinalomethanes by quarter 5-5
5-22 Chloroform by quarter 5-6
-------
Figures (continued)
Following
Number Page
5-23 Bromodichloromethane by quarter 5-6
5-24 Dibromochloromethane by quarter 5-6
5-25 Bromoform by quarter 5-6
5-26 Haloacetic acids by quarter 5-6
5-27 Dichloroacetic acid by quarter 5-6
5-28 Dibromoacetic acid by quarter 5-6
5-29 Haloacetonitriles by quarter 5-6
5-30 Haloketones by quarter 5-6
5-31 Aldehydes by quarter 5-6
5-32 Chloropicrin by quarter 5-6
5-33 Cyanogen chloride by quarter 5-6
5-34 Chloral hydrate by quarter 5-6
5-35 Influent total organic carbon by treatment 5-6
5-36 Influent total organic carbon by source 5-8
5-37 Influent UV-254 absorbance by source 5-8
5-38 Influent chloride by source 5-8
5-39 Influent bromide by source 5-8
5-40 Influent total organic carbon by
disinfection scheme 5-8
5-41 Influent UV absorbance by disinfection scheme 5-8
5-42 Influent chloride by disinfection scheme 5-9
5-43 Influent chloride by disinfection
scheme (w/o #10) 5-9
5-44 Influent bromide by disinfeciton scheme 5-9
5-45 Influent bromide by disinfection
scheme (w/o #10) 5-9
-------
Figures (continued)
Following
Number Page
5-46 XDBPstim by disinfection scheme 5-9
5-47 Trihalomethanes by disinfection scheme 5-9
5-48 Haloacetic acids by disinfection scheme 5-10
5-49 Haloacetonitriles by disinfection scheme 5-10
5-50 Haloketones by disinfection scheme 5-11
5-51 Aldehydes by disinfection scheme 5-11
5-52 Chloropicrin by disinfection scheme 5-11
5-53 Chloral hydrate by disinfection scheme 5-11
5-54 Cyanogen chloride by disinfection scheme 5-1!
5-55 Correlations with trihalomethanes 5-12
5-56 Correlations with haloacetic acids 5-12
5-57 Correlations with influent parameters 5-12
5-58 Correlations with influent parameters 5-13
5-59 Correlations with influent chloride and bromide 5-13
5-60 Correlations with influent bromide 5-13
5-61 Correlations with bromoform 5-14
5-62 Correlations with dibromoacetonitrile 5-14
5-63 Correlations with chloroform 5-14
5-64 Correlations with chloroform 5-14
5-65 Correlations with dichloroacetic acid 5-15
5-66 Correlations with trichloroacetic acid 5-15
5-67 Correlation with formaldehyde 5-15
5-68 Molar XDBPs versus molar TOX 5-15
5-69 XDBPsum by effluent pH 5-16
-------
Figures (continued)
Following
Number Page
5-70 Trihalomethanes by effluent pH 5-16
5-71 Haloacetonitriles by effluent pH 5-16
5-72 Dichloroacetonitrile by effluent pH 5-16
5-73 1.1.1-trichloropropanone by effluent pH 5-17
5-74 Sum of halogenated DBFs by influent temperature 5-17
5-75 Trihalomethanes by influent temperature 5-17
5-76 Haloacetonitriles by influent temperature 5-17
5-77 Utility 12 disinfection by-products
speciation 5-18
5-78 Cyanogen chloride by final disinfectant 5-21
5-79 Mean total organic carbon removal through filtering
utilities processes 5-22
5-80 Frequency distributions of THM survey data 5-23
5-81 AWWARF 12-quarter vs USEPA 4-quarter means 5-24
5-82 Total halogenated DBFs by study 5-24
6-1 Schematic of Utility 6 treatment processes 6-1
6-2 Effect of various disinfection schemes on
disinfection by-product formation at Utility 6 6-2
6-3 Effect of various disinfection schemes on
disinfection by-product formation at Utility 6 6-2
6-4 Effect of C12,NH2C1 and O3,C12,NH2C1 on DBF
formation (Utility 6) 6-2
6-5 Effect of C12,NH2C1 and O3.C12,NH2C1 on THM
formation (Utility 6) 6-2
6-6 Effect of C12,NH,CI and O3,C12,NH2C1 on HAA
formation (Utility 6) 6-2
6-7 Effect of C12,NH2C1 and O3,C12,NH2C1 on HAN
formation (Utility 6) 6-2
-------
Figures (continued)
Following
Number Page
6-8 Effect of C12,NH2C1 and O3,C12,NH2C1 on HK
formation (Utility 6) 6-2
6-9 Effect of CI2,NH2C1 and O3,C12,NH2C1 on misc. DBF
formation (Utility 6) 6-2
6-10 Effect of seasonal temperatures on DBF formation
(Utility 6) 6-2
6-11 Effect of O3.C12,NH2C1 on aldehydes formation
(Utility 6) 6-2
6-12 Schematic of Utility 7 treatment processes 6-4
6-13 Effect of various disinfection schemes on
disinfection by-product formation at Utility 7 6-4
6-14 Effect of various disinfection schemes on
disinfection by-product formation at Utility 7 6-4
6-15 Change in formaldehyde concentration thru
process train, State project water 6-4
6-16 Change in acetaldehyde concentration thru
process train, State project water 6-4
6-17 Effect of C12, C1;/NH3 & O3/NH3/C12 on DBF
formation (Utility 7) 6-4
6-18 Effect of C12, C1,/NH3 & O3/NH3/C12 on THM
formation "(Utility 7) 6-4
6-19 Effect of C12, CK/NH3 & O3/NH3/C12 on HAN
formation (Utility 7) 6-4
6-20 Effect of CI2, C1;/NH3 & O3/NH3/C12 on HK
formation (Utility 7) 6-4
6-21 Effect of C12, CU/NH3 & O3/NH3/C12 on HAA
formation (Utility 7) 6-4
6-22 Effect of Cl,, CK/NH3 & O3/NH3/C12 on misc. DBF
formation "(Utility 7) 6-4
6-23 Schematic of Utility 19 treatment processes 6-4
6-24 Effect of various disinfection schemes on
disinfection by-product formation at Utility 19 6-5
-------
Figures (continued)
Following
Number Page
6-25 Effect of various disinfection schemes on
disinfection by-product formation at Utility 19 6-5
6-26 Effect of O3/C12 & C12 only on DBF formation
(Utility 19) 6-5
6-27 Effect of O3/C12 & C\2 only on THM formation
(Utility 19) 6-5
6-28 Effect of O,/C12 & C12 only on HAA formation
(Utility 19) 6-5
6-29 Effect of O3/C12 & CI2 only on HAN formation
(Utility 19) 6-5
6-30 Effect of O,/C12 & C12 only on HK formation
(Utility 19) 6-5
6-31 Effect of O3/C1, & Cl, only on misc. DBF
formation (Utility 19) 6-5
6-32 Schematic of Utility 25 treatment processes 6-6
6-33 Effect of various disinfection schemes on
disinfection by-product formation at
Utility 25 6-6
6-34 Effect of various disinfection schemes on
disinfection by-product formation at
Utility 25 6-6
6-35 Effects of NH2C1 and O3/NH2CI treatment on
halogenated DBF formation, Utility 25 6-7
6-36 Effects of NH2C1 and O,/NH2C1 treatment on total
THM formation at Utility 25 6-7
6-37 Effects of NH2C1 and O3/NH2C1 treatment on HAA
formation at Utility 25 6-7
6-38 Effects of NH2C1 and O3/NH2C1 treatment on DCAA
formation at Utility 25 6-7
6-39 Effects of NH2C1 and O3/NH2C1 treatment on
chloropicrin formation at Utility 25 6-7
6-40 Effects of NH2C1 and O3/NH2C1 treatment on HAN
formation at Utility 25 6-7
-------
Figures (continued)
Following
Number Page
6-41 Effects of NH2C1 and oyNH2Cl treatment on HK
formation at Utility 25 6-7
6-42 Effects of NH2C1 and O^/NH2C1 treatment on chloral
hydrate formation Utility 25 6-7
6-43 Effects of NH2C1 and CK/NH2C1 treatment on CNC1
formation at Utility 25 6-7
6-44 Effects of NH2C1 and O3/NH2C1 treatment an ALD
formation at Utility 25 6-7
6-45 Effects on NH2C1 and O3/NH2CI treatment on
formaldehyde formation at Utility 25 6-7
6-46 Schematic of Utility 36 treatment processes 6-8
6-47 Effect of various disinfection schemes on
disinfection by-product formation at Utility 36 6-8
6-48 Effect of various disinfection schemes on
disinfection by-product formation at Utility 36 6-8
6-49 Effects of various treatments on halogenated
DBF formation at Utility 36 6-9
6-50 Effects of various treatments on THM formation
at Utility 36 6-9
6-51 Effects of various treatments on HAA formation at
Utility 36 6-9
6-52 Effects of various treatments on HAN formation at
Utility 36 6-9
6-53 Effects of various treatments on HK formation at
Utility 36 6-9
6-54 Effects of various treatments on CH formation at
Utility 36 6-9
6-55 Effects of various treatments on CHP formation at
Utility 36 6-9
6-56 Effects of various treatments on CNC1 formation
at Utility 36 6-9
-------
Figures (continued)
Following
Number Page
6-57 Effects of various treatments on ALD formation at
Utility 36 6-9
6-58 Effects of various treatments on formaldehyde
formation at Utility 36 6-9
6-59 Percent change in DBF and TOX concentrations due
to switch from chlorine only to ozone/chlorine
treatment 6-10
6-60 Change in DBF concentrations due to switch from
chlorine only to ozone/chlorine treatment 6-10
6-61 Percent change in DBF concentrations due to switch
from chloramines only to ozone/chloramines
treatments 6-11
6-62 Change in DBF concentrations due to switch from
chloramines only to ozone/chloramines treatment 6-11
6-63 Percent change in DBF concentrations due to
switch from chlorine only to ozone/chloramines
treatment 6-11
6-64 Change in DBF concentrations due to switch from
chlorine only to ozone/chloramines treatment 6-11
6-65 Schematic of utility 16 treatment processes 6-12
6-66 Effect of C12 and C1O,/C12 on DBF formation
(Utility 16) 6-12
6-67 Schematic of Utility 37 treatment processes 6-13
6-68 Effect of Cl, and C1O2 on DBF formation
(Utility 37) ! 6-14
6-69 Effect of Cl, and CIO, on THM formation
(Utility 37) 6-14
6-70 Effect of CL and CIO, on HAN formation
(Utility 37) 6-14
6-71 Effect of C12 and CIO, on haloketone
formation (Utility 37) 6-14
6-72 Effect of CI2 and CIO, on haloacetic
acid formation (Utility 37) 6-14
-------
Figures (continued)
Following
Number Page
6-73 Effect of C12 and CIO, on misc. DBF
formation (Utility 37) 6-14
6-74 Schematic of Utility 3 treatment process 6-15
6-75 Effect of alum dose on DBF formation (Utility 3,
SDS) 6-16
6-76 Effect of alum dose on THM formation (Utility 3,
SDS) 6-16
6-77 Effect of alum dose on HAA formation (Utility 3,
SDS) 6-16
6-78 Effect of alum dose on HK formation (Utility 3,
SDS) 6-16
6-79 Effect of alum dose on HAN formation (Utility 3,
SDS 6-16
6-80 Effect of alum dose on misc. DBF formation (Utility 3,
SDS) 6-16
6-81 Schematic of Utility 12 treatment process 6-17
6-82 Effect of alum dose on DBF formation (Utility 12) 6-17
6-83 Effect of alum dose on THM formation (Utility 12) 6-17
6-84 Effect of alum dose on HAA formation (Utility 12) 6-17
6-85 Effect of alum dose on HK formation (Utility 12) 6-32
6-86 Effect of alum dose on HAN formation (Utility 12) 6-17
6-87 Effect of alum dose on misc. DBF formation
(Utility 12) 6-17
6-88 Effect of alum dose on ALD formation (Utility 12) 6-17
6-89 Schematic of Utility 11 treatment process 6-19
6-90 TOC vs run time for GAC
column influent and effluent, Utility 11 6-21
6-91 TOC breakthrough profile at Utility 11 6-21
-------
Figures (continued)
Following
Number Page
6-92 TOC breakthrough profiles for three
demonstration-scale contactor runs at Utility 11 6-22
6-93 SDS TTHMs vs run time for GAC
column influent and effluent, Utility 11 6-22
6-94 SDS CHC13 vs run time for GAC
column influent and effluent, Utility 11 6-24
6-95 SDS CHBr3 vs run time for GAC
column influent and effluent, Utility 11 6-24
6-96 SDS HAAs vs run time for GAC
column influent and effluent, Utility 11 6-24
6-97 SDS DCAA vs run time for GAC
column influent and effluent, Utility 11 6-24
6-98 SDS DBAA vs run time for GAC
column influent and effluent, Utility 11 6-24
6-99 SDS HANs vs run time for GAC
column influent and effluent, Utility 11 6-24
6-100 SDS HKs vs run time for GAC
column influent and effluent, Utility 11 6-25
6-101 SDS CH vs run time for GAC
column influent and effluent, Utility 11 6-25
6-102 SDS CHP vs run time for GAC
column influent and effluent, Utility 11 6-25
6-103 Formation of TTHMs in distribution system
(Utility II) 6-25
6-104 Formation of DCAN in distribution system
(Utility 11) 6.27
6-105 Formation of CH in distribution system
(Utility 11) 6_27
-------
LIST OF ABBREVIATIONS
ACETAL Acetaldehyde
ALD Aldehyde
AMWA Association of Metropolitan Water Agencies
AWWARF American Water Works Association Research Foundation
BCAN Bromochloroacetonitrile
Br Bromide
C Concentration
C(, Concentration at time zero
CDHS California Department of Health Services
CE Clearwell effluent
CH Chloral hydrate
CHBr3 Bromoform
CHBr2Cl Dibromochloromethane
CHBrCI, Bromodichloromethane
CHCI3 Chloroform
CHP Chloropicrin
Cl Chloride
CI2 Chlorine
CI,/NH3 Prechlorine/postammonia
CI02 Chlorine dioxide
cm ' Inverse centimeter
/cm Inverse centimeter
CNCI Cyanogen chloride
CONV Conventional treatment
CP Chlorophenol
CPHF California Public Health Foundation
DBAA Dibromoacetic acid
DBAN Dibromoacetonitrile
DBP Disinfection by-product
DCAA Dichloroacetic acid
DCAN Dichloroacetonitrile
-------
List of Abbreviations, Continued
DCP
U-DCP
DF
DIS
EBCT
FE
FI
FRM
FS
GAC
GC
GW
HA
HAA
HAN
HK
HOBr
ID
JMM
LAA
LR
MBAA
MCAA
MCL
Metropolitan
//g/L
mg/L
mgd
MRL
MS
n
NA
ND
NH2CI
2,4-dichlorophenol
1,1 -dichloropropanone
Direct filtration
Disinfection only
Empty bed contact time
Filter effluent
Filter influent
Formaldehyde
Flowing stream
Granular activated carbon
Gas chromatograph
Groundwater
Hydrogen peroxide
Haloacetic acid
Haloacetonitrile
Haloketone
Hypobromous acid
Identification
James M. Montgomery, Consulting Engineers, Inc.
Los Angeles Aqueduct
Lake/reservoir
Monobromoacetic acid
Monochloroacetic acid
Maximum contaminant level
Metropolitan Water District of Southern California
Microgram per liter
Milligram per liter
Million gallons per day
Minimum reporting level
Mass spectrometer
Number of samples
Not analyzed
Not detected
Chloramines
-------
List of Abbreviations, Continued
nm Nanometer
O3 Ozone
PCP Pentachlorophenol
PI Plant influent
r Correlation coefficient
SDS Simulated distribution system
SDWA Safe Drinking Water Act
SFT Softening
SPW State Project Water
t Time
TCAA Trichloroacetic acid
TCAN Trichloroacetonitrile
TCP 2,4,6-trichlorophenol
1.1.1 -TCP I.I.I -trichloropropanone
THM Trihalomethane
TOC Total organic carbon
TOX Total organic halide
TTHM Total trihalomethanes
u Number of utilities
USEPA United States Environmental Protection Agency
UV Ultraviolet
UV-254 Ultraviolet absorbance at 254 nanometers
XDBP Halogenated disinfection by-product
XDBP Sum of measured halogenated disinfection by-products
-------
Executive Summary
-------
EXECUTIVE SUMMARY
INTRODUCTION
The United States Environmental Protection Agency (USEPA) will be developing
regulations to control disinfection by-products (DBFs) in drinking water as a result of
the 1986 amendments to the Safe Drinking Water Act. Although the schedule for
promulgation of regulations under these amendments remains uncertain at this time, the
anticipated regulations require that the presence and control of target DBFs be fully
understood. Thus, in October, 1987, the Association of Metropolitan Water Agencies
(AMWA) entered into a cooperative agreement with the USEPA to perform a study of
the formation and control of DBFs in full-scale drinking water systems. The study was
performed by the Metropolitan Water District of Southern California (Metropolitan)
and James M. Montgomery, Consulting Engineers, Inc. (JMM).
Specific objectives of the project included determining the occurrence of DBFs at 25
drinking water treatment facilities around the nation; determining the seasonal nature of
DBF occurrence as a function of temperature, total organic carbon (TOC), pH and
other water quality parameters; and determining the effect of changes in treatment
processes and/or disinfectants on the production of DBFs at up to 10 drinking water
treatment facilities. The study focused on the identification of DBFs as a function of
source water quality, water treatment process selection and operation, and disinfection
processes and chemicals. The target DBF compounds for the study are listed in Table
ES-1. Target compounds included trihalomethanes (THMs), haloacetic acids (HAAs),
haloacetonitriles (HANs). haloketones (HKs), aldehydes (ALDs), chloropicrin, chloral
hydrate, cyanogen chloride, and 2,4,6-trichlorophenol. A companion study, funded by
the California Department of Health Services (CDHS) through a grant to the California
Public Health Foundation (CPHF), was also conducted by Metropolitan and JMM.
This study evaluated DBF production at 10 drinking water treatment facilities in
California.
The first year of the two-year USEPA study focused on establishing and verifying the
analytical procedures used in the study, selecting utilities to participate in the study,
developing DBF baseline data through implementation of a quarterly sampling program
at the participating utilities, and conducting process modification studies at two utilities.
During the second year of the project, baseline data collection was completed and
process modification studies were conducted at six utilities.
Baseline data collection for the 35 utilities participating in the combined USEPA and
CDHS studies involved sampling of clearwell effluents (after final disinfection but
before distribution) on a quarterly basis for one year. The first sampling quarter (mid-
March through April, 1988) corresponded to the spring season, and subsequent
samplings corresponded to the summer, fall and winter seasons. Utilities were sent
sampling kits containing the sample bottles, detailed instructions for sampling, and a
sample information sheet on which to record plant operating conditions on the day of
sampling.
ES-1
-------
TABLE ES-1
LIST OF COMPOUNDS TARGETED IN STUDY
Compounds
Trihalomethanes
chloroform
bromodichloromethane
dibromochloromethane
bromoform
Haloacetonitriles
trichloroacetonitrile
dichloroacetonitrile
bromochloroacetonitrile
dibromoacetonitrile
Haloketones
1.1-dichloropropanone
l.U-trichloropropanone
Miscellaneous chloro-organics
chloropicrin
chloral hydrate
cyanogen chloride
Haloacetic acids
monochloroacetic acid
dichloroacetic acid
trichloroacetic acid
monobromoacetic acid
dibromoacetic acid
Chlorophenols
2.4-dichlorophenol*
2.4.6-trichlorophenol
pentachlorophenol*
Aldehydes
formaldehyde
acetaldehyde
* These chlorophenols were only analyzed for during the first sampling quarter.
-------
Executive Summary
Treatment modification studies were conducted to identify, in a preliminary manner,
the impact of processes or process modifications on DBF production. Thus, the
following full-scale and/or pilot studies were conducted: five studies involving a change
from chlorine or chloramines to ozone/chlorine or ozone/chloramines; two studies on
alum coagulation for DBF precursor removal; two studies involving a change from
chlorine to chlorine dioxide/chlorine; and one study of granular activated carbon (GAC)
for DBF precursor removal.
UTILITY SELECTION
An information request form was sent to 104 potential participants in the study, and
from the 78 responses received, utilities were selected to complete the matrix shown in
Table ES-2. The matrix was divided into two major categories, treatment type
(conventional, direct filtration, softening and disinfection only) and source water type
(groundwater, lake/reservoir and flowing stream). Within these categories,
geographical location and disinfectant type were also considered. In addition, two
other categories were utilized in developing the selection matrix: population and THM
level.
METHODOLOGY
Grab samples were collected at clearwell effluents (after final disinfection but before
distribution) for DBF and TOC analyses. For those plants which did not have
clearwells. samples were collected at specified points after final disinfection. Additional
analyses were performed on plant influents: TOC, bromide, chloride and aldehydes.
TOC was also measured at filter influents, if filtration was employed as a treatment
process. Holding studies performed by Metropolitan were conducted to identify
preservatives and holding times for the DBF samples.
The THM liquid/liquid extraction (LLE) gas chromatograph (GC) method was modified
by Metropolitan to include THMs, HANs, HKs and chloropicrin. Chloral hydrate was
analyzed by a separate LLE/GC method. HAAs and 2,4,6-trichlorophenol were
analyzed by an acidic salted LLE and GC. Cyanogen chloride was analyzed by a
purge-and-trap gas chromatograph/mass spectrometer (GC/MS) method. Aldehydes
were analyzed by a derivatization/extraction GC method.
DATA MANAGEMENT AND ANALYSIS
The large database developed during the study required a strict data handling protocol
to ensure its accuracy and reliability. Some of the elements of this protocol included
use of signatures by responsible project team members, use of data-sheet reference
numbers for each individual project data sheet, and a tracking system for the status of
each group of data.
Nonparametric statistical methods were used to analyze data from this study.
Nonparametric methods do not require an assumed parametric distribution for the data,
and cases below the detection limits can be incorporated more readily as compared to
parametric methods. A "five-number summary", including the minimum value, 25th
percentile, median. 75th percentile and maximum value of the data set, was used to
present a simple summary of the data. Notched box-and-whisker plots were used also
ES-2
-------
TABLE ES-2
DISINFECTION BY-PRODUCTS IN DRINKING WATER STUDY
UTILITY SELECTION MATRIX
TREATMENT
GROUNDWATER
LAKE/RESERVOIR FLOWING STREAM
CONVENTIONAL
Clermont Co., OH 1
Long Beach, CA 2
Norwich, CT* IB
MWD, CA -Mills 2a,5b
Arlington, TX* 2A
Hackensack, NJ 2e
MWD, CA -Weym. 2
San Francisco, CA 1
Big Spring, TX* 5
Shreveport, LA 6a,8b
Cape Girardeau, MO* 2A
Cincinnati, OH 1
Contra Costa WD, CA 2A
Sacramento, CA 1
Santa Clara Valley, CA 2
Newport News, VA IA
DIRECT
FILTRATION
East Bay MUD, CA I
Las Vegas, NV I
Little Rock, AR 1
Aurora, CO laA,5bA
Los Angeles DWP, CA 3
SOFTENING
Palm Beach Co., FL 2
Wausau, WI* 1
Minot, ND 1
Santa Monica, CA I
Macomb, IL* 1A
Galveston.TX 7c,4d
Louisville, KY 2
Ft. Meyers, FL* I
Emporia, KN* 2A
Omaha, NB 1A
DISINFECTION
ONLY
Mesa Consol., CA 3
North SkagitCo., WA* I
New York City, NY 1
Newark, NJ I
Note: Utilities participating in the California Public Health Foundation study are listed in bold type.
* Population under 50,000; all others over 50,000.
Key for chemical addition:
1 - chlorine only
2 - chlorine + chloramines
3 - ozone + chlorine
4 - chlorine + chlorine dioxide
5 - chloramines only
6 - chloramines + chlorine dioxide
7 - chlorine -f chloramines -I- chlorine dioxide
8 - ozone + chloramines
A - powdered activated carbon
B - potassium permanganate
a - first quarter only
b - second through fourth quarters
c - first through third quarters
d - fourth quarter only
e - clearwell effluent sampled
before ammonia addition
-------
Executive Summary
used for presentation of project data. This type of plot is presented and described in
Figure ES-1.
BASELINE SAMPLING RESULTS AND DISCUSSION
Table ES-3 summarizes the baseline median DBF values for each quarter as well as for
all four sampling quarters combined. It should be noted that these data represent
clearwell effluent samples or samples collected at specified points after disinfection for
those plants without clearwells. Some distribution system sampling was performed for
the process modification studies (discussed in more detail below). Results of the
process modification studies indicated that some DBFs, such as THMs, increased in the
chlorinated distribution systems of some utilities, while their concentrations did not
change in the chloraminated systems of other utilities. In addition, it is important to
note that the disinfection practices of some of the participating utilities, such as the use
of chloramines. are utilized to meet the current TTHM regulation and not to meet
requirements of the proposed Surface Water Treatment Rule (SWTR). Thus, some
utilities would produce different DBF levels if their current disinfection practices
required modification in order to meet proposed concentration-time (CT) requirements
of the SWTR.
Figure ES-2 is a summary of the four-quarter median concentrations of each DBF class
measured in this study. The median value of total THMs (TTHMs) was 36 /ug/L, and
the median value of HAAs was 17 //g/L during the four quarters of baseline data
collection. On a weight basis, THMs were the largest class of DBFs detected in this
study (54.5 percent of the total measured DBFs), and HAAs were the second largest
fraction (25.4 percent of the total) (see Figure ES-3). A running annual average of
TTHMs is utilized to determine compliance with the TTHM maximum contaminant
level (MCL) of 0.10 mg/L. When the running annual average (i.e., the mean of (he
four baseline TTHM values) was computed for each utility, the 35-utility median
TTHM concentration was 39
TOC analyses were performed on plant influent samples after the first sampling
quarter. Figure ES-4 is a box-and-whiskers plot of the plant influent TOC data for the
summer, fall and winter quarters. Note that the "notches" in the plots of the three
quarters overlap, indicating that there is not a statistically significant difference, at 95
percent confidence, between the medians of any two quarters. In Figure ES-5, influent
TOC values are plotted by source water type. Again, there is no statistically significant
difference between the medians of any two sources.
Figure ES-6 is a plot of TTHMs by sampling quarter. As would be expected, TTHM
levels were highest in the summer quarter when water temperatures were the highest,
lower in the fall, and lowest in the winter and spring quarters, although these
differences were not statistically significant at a 95 percent confidence level. Figure
ES-7 shows TTHM levels plotted as a function of the influent water temperatures
measured during baseline data collection. In this plot, the median TTHM level is
significantly higher in the highest temperature range than in the lower three ranges,
illustrating the dependence of TTHM production on water temperature.
Figure ES-8 illustrates TTHM levels as a function of the disinfection scheme in use by
the utilities on the day of sample collection. Only chlorination, prechlorination/
ES-3
-------
GUIDE TO NOTCHED BOX-AND-WHISKER PLOTS
D
EXTREME OUTLIER
any value outside 3 interquartile ranges
measured from the 25 and 75 percentiles
OUTLIER
any value outside 1.5 interquartile ranges
measured from the 25 and 75 percentiles
MAXIMUM VALUE
this is the largest value (excluding outliers)
INTERQUARTILE RANGE
contains the data between
the 25 and 75 percentiles
95% CONFIDENCE INTERVAL
for the median (including outliers)
75 PERCENTILE
75% of the data are less than or
equal to this value (including outliers)
MEDIAN
50% of the data are above or
below this value (including outliers)
NOTE: Horizontal width of box is
proportional to square root of sample size (N)
1 -^— 25 PERCENTILE
25% of the data are less than or
equal to this value (including outliers)
MINIMUM VALUE
this is the lowest value (excluding outliers)
FIGURE ES-1
-------
TABLE ES-3
DISINFECTION BY-PRODUCTS IN DRINKING WATER
SUMMARY OF BASELINE SAMPLING MEDIAN VALUES
1st 2nd 3rd 4th All
Disinfection By-Products Quarter Quarter Quarter Quarter Quarters
(/fg/L) (Spring) (Summer) (Fall) (Winter) Combined
Trihalomethanes
Chloroform 15 15 13 9.6 14
Bromodichloromelhane 6.9 10 5.5 4.1 6.6
Dibromochloromethane 2.6 4.5 3.8 2.7 3.6
Bromoform 0.33 0.57 0.88 0.51 0.57
Total Trihalomethanes 34 44 40 30 36
Haloacetonitriles
Trichloroacetonitrile <0.012 <0.012 <0.029 <0.029 <0.029
Dichloroacetonitrile 1.2 1.1 1.1 1.2 1.2
Bromochloroacetonitrile 0.50 0.58 0.70 0.59 0.57
Dibromoacetonitrile 0.54 0.48 0.51 0.46 0.50
Total Haloacetonitriles 2.8 2.5 3.5 4.0 3.3
Haloketones
I.l-Dichloropropanone 0.52 0.46 0.52 0.55
1,1,1-Trichloropropanone 0.80 0.35 0.60 0.66
Total Haloketones 1.4 0.94 1.0 1.8 1.2
Haloacetic adds
Monochloroacetic acid < 1.0 1.2 < 1.0
Dichloroacettc acid 7.3 6.8 6.4
Trichloroacetic acid 5.8 5.8 6.0
Monobromoacetic acid <0.5 <0.5 <0.5
Dibromoacetic acid 0.9 1.5 1.4
Total Haloacetic acids 18 20 21 13 17
-------
Table ES-3
Disinfection By-Products In Drinking Water
Summary of Median Values, Continued
Aldehydes
Formaldehyde
Acetaldehyde
1st
Quarter
(Spring)
NA
NA
2nd
Quarter
(Summer)
5.1
2.7
3rd
Quarter
(Fall)
3.5
2.6
4th
Quarter
(Winter)
2.0
1.8
Ail
Quarters
Combined
3.6
2.2
Total Aldehydes
Miscellaneous
NA
6.9
5.5
4.2
5.7
Chloropicrin
Chloral hydrate
Cyanogen chloride
2 ,4,6-TrichIorophenol
Halogenated DBPsum
Total Organic Halide
Plant Influent Characteristics
Total Organic Carbon, mg/L
Ultraviolet absorbance, cm"1
Chloride, mg/L
Bromide, mg/L
0.16
1.8
0.45
<0.3
64
150
NA
NA
NA
NA
0.12
3.0
0.60
<0.4
82
180
2.9
0.11
28
0.07
0.10
2.2
0.65
<0.4
72
170
2.9
0.1 1
32
0.10
0.10
1.7
0.80
<0.4
58
175
3.2
0.13
23
0.07
0.12
2.1
0.60
<0.4
70
170
3.0
0.11
29
0.08
NA = Not Analyzed
Note (1): Total class median values are not the sum of the medians of the individual compounds, but
rather the medians of the sums of the compounds within that class.
Note (2): The halogenated DBPsum median values are not the sum of the class medians for all
utilities, but rather the medians of the halogenated DBPsum values for .all utilities. This
value is only the sum of halogenated DBPs measured in this study.
-------
5
U
Disinfection By-Product Concentration
by DBP Class Four Quarters
7O
6O
50
40
3O
20
1O
0
THM
HAN
HK
HAA
ALD
OP
CH
CNCI
TCP
Trihalome thanes
Haloatstomtnies
HBtokeTones
Hotoacetic Acids
Aldenydes
Chtoropicrln
Chloral Hy*ate
Cyanogen Chloride
TrlcNorocHenol
7S»1 I
VJSMWffA
-X--X--X-
THM HAN HK HAA ALD CHP CH CNCI TCP
DBP Class
FIGURE ES-2
Percent of Sum of DBP Class Medians
By DBP Class Four Quarters
HAA 25.4%
ALD 8.5%
MISC 5.O%
HAN 4.9%
HK 1.8%
THM 5
ThM
HAN
I-K
HAA
ALD
MSC:
CHP
CH
CNCI
TCP
Tr ina tome rnanes
Haloacetom tr i tes
Haloketones
naioacetic Acids
Aioenyoes
OHoropicrin
Crvocai Hyoate
Cyarogeo Crtonde
Tricnloropnenol
FIGURE ES-3
-------
Influent Total Organic Carbon
Bg Quarter
Influent Total Organic Carbon
By Source
T
_J
a
E
0
.0
o
0
•-H
c
01
o
*J
0
_L
_L
•4
J
a
o
o
u
•H
c
a
o
•P
a
T
T
J_
SUHHCR
35
35
Quart i
UIHTER
35
FS
33
It
21
7
Sourci
LR
51
17
RGURE ES-4
FIGURE ES-5
-------
Trihalamethani
Bg Quarter
Trihalomethan«»
Bg Influ«nt Temperature
T
38*
T
J
O
•P
V
0
1.
h-
_L
M*
J
o
•P
M
E
0
••4
L
T
SPRINQ
35
35 35
Quarter
UINTER
35
8.8-18.1
33.B-31.*
21 1(7 'I* 27
17 27 26 17
Temperature Rang* <°C)
FIGURE ES-6
FIGURE ES-7
-------
Executive Summary
postammoniation, and chloramination are shown since other disinfection schemes (such
as preozonation/postchlorination) had very small sample sizes. All other factors being
equal, it would be expected that the chlorinating utilities would produce a higher
median TTHM level than utilities using the other two disinfection schemes. However,
Figure ES-8 indicates that the prechlorinating/postammoniating utilities had a higher
median TTHM level than utilities employing the other two disinfection scenarios, and
this difference was significant at a 95 percent confidence level. The same trend was
observed for the median level of the sum of halogenated DBFs (XDBP ) measured in
this study. The reasons for this trend are not immediately apparent from the median
influent TOC levels for the three disinfection schemes, which occurred within the
narrow range of 2.8 to 3.2 mg/L. However, the median values of ultraviolet
absorbance at 254 nanometers (UV-254) for the prechlorinating/postammoniating
utilities, indicate that these utilities had a higher UV-254 level than utilities employing
either of the other two disinfection schemes, although the difference was not statistically
significant at a 95 percent confidence level. The higher UV-254 levels may indicate
higher levels of DBF precursors even though higher precursor levels were not reflected
in the TOC results. Thus, the chlorinating utilities participating in this study may have
been able to use free chlorine for oxidation/disinfection and still meet the TTHM MCL
of 0.10 mg/L, while other utilities have had to employ free chlorine for
oxidation/disinfection and then use ammonia at some point in their process trains in
order to meet the TTHM MCL.
In all. over 300 correlations were determined for the baseline data collected in this
study. Correlations of non-THM DBFs with THMs were performed in order to explore
the potential of using TTHMs as a surrogate for other DBFs such as HAAs or HANs,
since performing analyses for several different DBFs is beyond the scope of a utility's
routine monitoring program. Figure ES-9 illustrates the correlations of THMs with
XDBPsum. the sum of non-THM DBFs, and HANs. As illustrated in the figure, there
was a strong correlation between TTHMs and XDBP (r=0.96). Since THMs
represent the largest DBF fraction detected in this study, the data were re-evaluated by
correlating TTHMs with the sum of non-THM DBFs (XDBPs - THMs). In this
instance, r decreased to 0.76. However, this lower correlation coefficient does not
mean that THMs cannot be used as a surrogate or predictor of the sum of all XDBPs.
It should be noted, though, that correlations between some classes of compounds were
low (for instance, comparing TTHMs to haloketones yielded an r of only 0.06),
whereas for TTHMs and HANs, r was equal to 0.78.
Figure ES-10 illustrates the correlation of influent chloride levels with influent bromide.
There was a strong correlation between these two parameters (r=0.97), and these
results indicate that chloride may be used as a predictor for bromide. Using all of the
data points collected in the baseline sampling program (excluding three outlier values
from a utility with atypically high bromide levels), linear regression analysis yielded the
following equation:
[Br-] = -0.0071 + 0.0034[C1-]
It is significant to note that high bromide levels were detected not only at utilities
impacted by tidal influences or saltwater intrusion, but at inland utilities as well.
ES-4
-------
Trih«J.om«th«n««
By Disinfection Schemi
J
a
3 «••
M
•
+J
tt
0
•H
L
I
eta
n - 78
u - 21
CL2NH3
dO
1 I
II
14
Disinfection Schsms
FIGURE ES-8
-------
Correlations with Trihalomethanes
XDBPsum us THMs
J
a
a
x
400
380
280
188
r-0.96
XDBPs-THMs us THMs
168
128
a
j
a.
S
x
40
S8 188 158 288 288 300
n-UO
THMS
r=0.76
60 100 150 288 250 300
0=1^0
THMS (ugxL)
HANS us THM!
-I
a
01
-------
Correlations with Influent
Chloride and Bromide
Br- us CL-
CHBr3 us Br-
2.5
J
\
a
a
a
1.5
8.5
r=0.97, 0.86*
a
o
68
48
38
X '
„«•
, r=0.57, 0.69-
280 480 688
n-105, 102*
Influ«nt Chiorid*
-------
Executive Summary
Figure ES-ll shows the relationship of influent bromide to levels of chloropicrin,
1,1,1-trichloropropanone (1,1,1 -TCP) and trichloroacetic acid (TCAA) measured during
baseline data collection. In each case, an exclusion relationship is demonstrated, i.e.,
the presence of bromide appeared to exclude the presence of the particular DBF.
Correlations of the four THM compounds with dichloroacetic acid (DCAA) are
illustrated in Figure ES-12. The best correlation was found with chloroform (r=0.86).
However, as the THMs shift to the more brominated species, the correlation
coefficients decrease until an exclusion relationship is observed between DCAA and
bromoform (r = -0.33). This progression is consistent with the finding that bromide and
various chlorinated DBFs were related by exclusion.
Of the 35 utilities included in this study, only three employed ozone, yet almost all had
detectable levels of formaldehyde and acetaldehyde in the clearwell effluent samples.
These aldehydes were also found in some plant influent samples, and chlorination alone
was found to produce these compounds.
Chloramines are recognized as an effective control strategy for THMs and other DBFs.
However, for most waters studied in this project, concentrations of cyanogen chloride
were found to be significantly higher in chloraminated waters as compared to
chlorinated waters. Moreover, it was possible to statistically divide the distribution of
cyanogen chloride into two groups, depending on whether the final disinfectant was
chlorine or chloramines.
Figure ES-13 illustrates the removal of TOC within the filtering plants included in the
baseline sampling program. Overall, TOC removal within these plants averaged
approximately 24 percent. It should be noted that the treatment practices of the
utilities participating in the baseline sampling program most likely focused on turbidity
control and were not optimized with respect to TOC removal. Results of the two
treatment modification studies on improved coagulation (described in more detail
below) indicated that these two utilities were able to achieve higher TOC removals than
those indicated in Figure ES-13; however this ability may be source-water specific.
The TTHM data from this study of 35 utilities was compared to that of the THM
survey conducted by the American Water Works Association Research Foundation in
1987. which involved 727 utilities around the nation. Since compliance with the
TTHM MCL of 0.10 mg/L is based on a running annual average, mean TTHM values
were computed for each of the 35 utilities in this study for the 4 sampling quarters.
The AWWARF survey utilized the means of three years of quarterly data. The TTHM
means for bolh projects are plotted in Figure ES-14, which illustrates that the two
frequency distributions were very similar.
TREATMENT MODIFICATION STUDIES - RESULTS AND DISCUSSION
Ozonation Studies
Because of the increasing use of ozone in the United States for disinfection and control
of DBFs, five treatment modification studies focused on the use of ozone at various
water treatment plants.
ES-5
-------
Correlations with Influent Bromide
CHP us Br-
1,1,1-TCP us Br-
2
1.6
i
1.2
C
>«J
U
•4
* e.e
i
a
I
a
o
' a
«
J.
e.4 r •
t
fa r=-0.22, -0.25*
Uf •&*«,-. . "^
8
1
? *
s
y
L
§• *
L
0
£
U
•H
1.
H 2
Jj
A
*• ^
a
a
B
i
i
•
B
A i
•<* t^0.° r=-0.26, -0.35-
ffi-*E»-q.* S - N*V>1
2.8 3
n«10l», 101*
Influent Browid*
n-105, 102*
Influ.nt Bromid. (mgxL)
TCAA us Br-
CHC13 us Br-
au
i ,.
J
u
H 28
w
5
a
a
H
fi "
a
a
l'.
>,
i a
a
a
[• °
o
a
L ti •
1- I'- •
»• . r— 0.27, -0.35*
C
0 C^^^SEL! 2 i5j
158
128
i .
?
0
5 68
0
•^
5
38
8
a
' o
a
a
a
a
o
M
^ !
e
Jt\.m r— 0.25, -0.29*
«• °^-rt= . . v*>>'>>
8 8.5 1 1.5 3 2.5 3
n-105, 102*
Influent Bronid* (mg/L)
* Excludes indicated outliers
« 9.5 !
1.5 2 2.5
n-105, 102*
Influent Bromide (mgxL)
FIGURE ES-11
-------
<
u
J
a
a
Correlations with Dichloroacetic Acid
CHC13 us DCAA
se
68
38
•«i
•/*:• '•--
°.f
r»0.86
188
68
CHCl2Br us DCAA
0
|
Q
48
28
88
»
»
28
18 28 38 48 88
Oichloroccvtic Acid
-------
(13
O
0)
DC
U
O
\-
(D
O
4^
-------
AUUARF 12-Quarter us. USEPA
4-Quarter Means
499
J
01
3
in
01
c
ID
4J
01
E
0
-H
10
•H
L
309
299
let
AUIUIAW
727
USBPA
35
Study
FIGURE ES-14
-------
Executive Summary
The study at Utility 6 involved sampling before and after ozonation was incorporated
into the treatment process of a 52-mgd plant treating a reservoir water. A conventional
treatment process was used before ozonation was installed, with chlorine added to the
plant influent, flocculator influent, and filter influent and effluent. Ammonia was
added after filtration. When ozonation was installed (in combination with the rapid mix
process) the plant was modified to direct filtration, with flotation and skimming
replacing the flocculation and sedimentation processes. As a result of ozone
implementation, the plant was able to reduce chlorine doses and free chlorine contact
time. The "before" and "after" samples (before and after ozonation was placed on-
line) were collected approximately five months apart. Figure ES-15 summarizes the
impact of ozonation on DBF levels measured in the utility's distribution system after a
residence time of approximately 7 hours. The figure illustrates that reductions of 56 to
66 percent occurred for all XDBP classes after the implementation of ozonation, with
the exception of HKs which increased slightly. Chloral hydrate, chloropicrin and
cyanogen chloride concentrations also decreased. However, from this study, it was not
possible to attribute the lower levels of most DBFs directly to ozone since changes in
treatment process, and lower chlorine doses and shorter contact times were employed.
It cannot be determined if ozone caused a decrease or modification of DBF precursor
material or if lower DBF levels can be attributed solely to the decreased use of
chlorine. From a full-scale perspective, however, it is notable that the application of
ozone decreased the dependence on chlorine for oxidation and disinfection, with the
overall result of decreased DBF concentrations.
Three different treatment scenarios were studied at Utility 7, which operates a 400-mgd
conventional treatment plant. Chlorine-only (using chlorine doses of 2.3 mg/L at the
plant influent and 1.1 mg/L at the filter influent) and prechlorination/postammoniation
(chlorine doses of 2.3 mg/L at the plant influent and 0.6 mg/L at the filter influent;
ammonia dose of 0.49 mg/L as ammonia-nitrogen at the filter effluent) were studied at
full scale and a 6-gpm pilot plant was utilized to study preozonation/postchloramination
(ozone dose of 2.0 rng/L before rapid mix; and 0.5 mg/L ammonia and 1.5 mg/L
chlorine at the filter effluent). Samples were collected for each scenario after 2-hour
and 24-hour simulated distribution system (SDS) tests were conducted. In these tests,
samples were dosed with disinfectants and held under conditions representative of
Utility 7's distribution system to provide an estimate of the levels of DBFs that would
have been produced under realistic environmental conditions. The effects of ozonation
on DBF levels after the 24-hour SDS test are illustrated in Figure ES-16. Preozonation
followed by concurrent addition of ammonia and chlorine after filtration decreased the
levels of THMs. HAAs, HANs, and chloral hydrate as compared to chlorine-only or
prechlorination/postammoniation treatment. Concentrations of 1,1,1-TCP were lowest
for ozone/chloramines treatment, higher for prechlorination/postammoniation and
highest for chlorination treatment, while 1,1-dichloropropanone (1,1-DCP)
concentrations exhibited the opposite trend. Very little difference in chloropicrin levels
were observed between the three treatment schemes; however, all levels were less than
I yug/L. Cyanogen chloride levels were highest for the ozone/chloramines treatment
after 2 hours of holding time, but after 24 hours, there was very little difference in
cyanogen chloride concentrations between the three treatments. Aldehydes were not
measured at this utility in this study.
Utility 19 operates a large (600-mgd) preozonation/direct filtration facility. Chlorine-
only (1.8 mg/L of chlorine added to the plant influent and 0.3 mg/L added to the filter
ES-6
-------
_t
•&»
o
U
§5
Effect Of CI2.NH2CI and O3.CI2.NH2CI
on DBP Formation (Utility 6)
15O
100
50
0
CI2NH2CI
(11/21/86)
ESS) O3.CI2.MH2a
(5/15/89)
Dlst. Sys, Residence Time (t)
t = 7 hrs
• TCMM MOW l
«-» MBAA b*low
NA
XDBPsum THM
HAN
HK
HAA
ALD
DBP Class
FIGURE ES-15
Effect Of CI2. CI2/NH3 & O3/NH3/CL2
on DBP Formation (Utility 7)
200
ES3 CIS
CO
h_
*-•
$
U
5
U
Q
150
1OO
50
0
12
XD8PSUH
Diet. Sys. Residence Time (t)
1: t = 2 hrs
2: t = 24 hrs
* Below
12 12 12 12
HAN t-K HAA 2.4-OCP
DBP Class
FIGURE ES-16
-------
Executive Summary
effluent) and preozonation/postchlorination (ozone dose of 1.7 mg/L, and chlorine dose
of 1.5 mg/L added to the filter effluent) treatments were studied. Samples were
collected at the plant's clearwell effluent and in the distribution system at residence
times of 4.3 and 11 hours. Figure ES-17 shows concentrations of DBFs after 11 hours
of residence time in the distribution system. Decreases of 13 //g/L and 8.7 /ug/L were
observed for TTHMs and HAAs, respectively, after implementation of ozonation with
subsequent chlorination. A 2.3 //g/L increase in chloral hydrate was observed. After
ozonation, HANs were decreased by 1.2 //g/L, and small increases were observed for
HKs and chloropicrin. The cyanogen chloride analysis was not performed during the
ozonation trial, and this compound was only slightly detected during the chlorine-only
trial.
Sampling at Utility 25 was conducted at full scale at a 90-mgd conventional treatment
facility. Samples were collected before the plant went on-line with two-stage
ozonation. Chloramines-only treatment was utilized before ozonation was
implemented, with disinfectant addition limited to the concurrent addition of ammonia
(1.6 mg/L) and chlorine (8.0 mg/L) at the rapid mix. After the ozone system was
placed on-line, ozone was applied to both the raw and settled water at 4.0 mg/L per
stage, and chlorine (5.0 mg/L) and ammonia (1.0 mg/L) were added concurrently prior
to filtration. Samples were collected at various points in the plant: 2nd-stage ozone
contactor influent (2OI), filter influent (FI), filter effluent (FE), clearwell effluent (CE);
and at four distribution system locations: at 4 to 5 hours of residence time (LI), 8 to 9
hours of residence time (L2), 9 to 10 hours of residence time (L3), and 18 to 20 hours
of residence time (L4). Figure ES-18 presents the profile of TTHM concentrations
through the plant and distribution system. The data show that for both treatment
scenarios, THMs were formed immediately after chlorine and ammonia addition and
remained stable through the plant and into the distribution system, but TTHM
concentrations were significantly lower after ozone was incorporated into the plant's
treatment process. The same trend was observed for levels of XDBPsum and HAAs.
The unexpectedly high levels of DBFs produced by chloramines-only treatment at
Utility 25 may have been due to poor mixing upon concurrent addition of chlorine and
ammonia, and/or to the high influent TOC concentration of this utility's source water
(7.7 mg/L). Concentrations of chloropicrin and ALDs were substantially higher in the
plant and distribution system during the ozonation trial compared to the chloramines-
only experiment.
At Utility 36, a 5-gpm pilot plant was employed to evaluate the effect of five different
treatment scenarios on the formation of DBFs. Conventional treatment was employed
with options to add preozonation (2 mg/L), with and without hydrogen peroxide
addition (0.67 mg/L), and free chlorine (6.5 mg/L) or chloramines (2.1 mg/L chlorine
and 0.5 mg/L ammonia) as disinfectants in the rapid mix. Samples were collected from
the filter effluent and held for 24 hours at ambient temperature in order to simulate the
residence time in a distribution system. Figure ES-19 illustrates the effect of the five
different disinfection schemes on various DBF classes and chloral hydrate. The highest
levels of TTHMs. HAAs. HANs and chloral hydrate were observed with those
disinfection schemes which employed chlorine as the final disinfectant.
Preozonation/postchlorination appeared to slightly increase the level of TTHMs and
slightly decrease the concentration of HAAs; chloral hydrate was increased. Large
decreases were observed under any disinfection scheme which employed chloramines as
a disinfectant. For example, as compared to chlorine-only treatment, chloramines-only,
ES-7
-------
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY-PRODUCTS FORMATION
AT UTILITY 19
NA: Not Analyzed
FIGURE ES-17
-------
s
•i—'
(0
g
o
u
16
*—'
O
70
60
5O
40
30
20
10
O
Effects of NH2CI and O3/NH2CI Treatment
on Total THM Formation at Utility 25
V777A NH2CI
H^B O3/NH2CI
NA = Not analyzed
*-**• = No lime addition
-*-*-
RAW 2OI Fl FE CE L1 L2 l_3 L4
Treatment & Residence Time
FIGURE ES-18
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY—PRODUCT FORMATION AT UTILITY 36
FIGUREES-19
-------
Executive Summary
o/one/chloramines and ozone/hydrogen peroxide/chloramines decreased TTHM levels
by 96. 97 and 98 percent, respectively.
At Utilities 19 and 36. the use of preozonation/postchlorination resulted in a shift to the
brominated THMs and HAAs. Ozone can react with bromide ions in the raw water
causing the formation of hypobromous acid (HOBr). Reactions of HOBr with natural
organic material can produce bromoform, dibromoacetic acid, and other brominated
DBFs. This was found to occur even at a utility with low levels of influent bromide,
such as Utility 19, which typically treats water having an influent bromide
concentration of only 0.04 mg/L.
Ozonation was found to increase aldehyde concentrations at Utilities 6, 25 and 36.
Aldehydes were not measured at Utilities 7 and 19 during the treatment modification
studies. However, subsequent pilot testing at Utility 7 indicated that aldehydes were
formed in the ozone contactors. When the pilot filters at this utility were operated
without the upstream addition of a secondary disinfectant, aldehyde concentrations in
the filter effluent were lowered to detection limits or below, suggesting that biological
activity within the filter bed was responsible for aldehyde removal. Utilities 6, 19, 25
and 36 employed secondary disinfection prior to filtration and, thus, may have
precluded or reduced the potential for aldehyde removal within their filters.
Furthermore, when Utility 7 applied chloramines before the pilot filters, aldehyde
concentrations were not reduced through the filtration step.
Chlorine Dioxide Studies
Chlorine dioxide studies were conducted at two utilities. Both studies evaluated the
combination of chlorine dioxide/chlorine as a DBF control method compared to free
chlorine.
Utility 16 operates a 400-mgd direct filtration treatment system with free chlorine for
both preoxidation and final disinfection, but periodically switches to chlorine dioxide
preoxidation to control THMs and taste and odor. Samples were collected in the plant
and at two distribution system locations (approximate residence times of 45 minutes and
7 days). On the chlorine-only sample date, chlorine doses were 2.0 mg/L at the plant
influent and 1.0 mg/L at the filter effluent. For the chlorine dioxide test, the chlorine
dioxide dose was 0.5 mg/L at the plant influent and the chlorine dose was 1.9 mg/L at
the filter effluent. Figure ES-20 illustrates levels of DBFs measured in the distribution
system. There were very little differences observed in DBF levels produced by the two
different oxidation/disinfection schemes.
Utility 37's 30-mgd treatment facility provides for two separate treatment trains, one of
which has provision for chlorine dioxide preoxidation, and one which employs free
chlorine. Both treatment trains have free chlorine as a final disinfectant. For this
study. 0.9 mg/L of chlorine dioxide was added to the flocculator effluent of one
treatment train, and 2.25 mg/L of chlorine was added to the other treatment train for
preoxidation. Although detectable levels of free chlorine were measured in the chlorine
dioxide treatment train, and vice versa, the preoxidant in one treatment train was
predominantly chlorine dioxide (referred to as "chlorine dioxide" treatment), and in the
other train, the preoxidant was predominantly chlorine (referred to as "chlorine"
treatment). Samples were collected at the plant's sedimentation basin effluent, after a
ES-8
-------
Executive Summary
residence time of approximately 2.5 hours and before chlorine was added for residual
disinfection. Figure ES-21 shows the effect of the two different preoxidants on the sum
of measured halogenated DBFs and the DBF classes. From this figure, it is apparent
that even in the relatively short detention time in the sedimentation basins, the use of
chlorine dioxide preoxidation resulted in lower levels of all measured DBFs compared
to chlorine treatment. XDBPSIM11 was almost 50 percent lower with chlorine dioxide
treatment.
Coagulation Studies
Coagulation studies were conducted at two utilities to evaluate the effect of coagulant
dose on DBF formation. These utilities were selected because they had the capabilities
to adjust alum doses at full scale without severely compromising the quality of their
finished water.
Utility 3 operates a conventional treatment facility with a 10.5-mgd capacity. Alum
doses were varied from low (10 mg/L) to medium (19 mg/L) to high (40 mg/L) for the
study, and coagulation pH was held constant at pH 5.5. Chlorine was not added until
the clearwell influent: therefore, samples were collected at the filter effluent and
24-hour SDS tests were conducted using a chlorine dose of 3.5 mg/L. Removal of
TOC increased from 25 to 50 percent in the filter influent as the alum dose increased
from 10 to 40 mg/L. At the low alum dose, a greater percent of TOC was removed
through filtration (20 percent) than at the medium and high alum doses (15 and 9
percent, respectively). This was most likely due to the better settling characteristics of
the floe at the higher doses. Figure ES-22 illustrates the effect of increasing alum dose
on XDBP and the DBF classes. In general, DBFs decreased with increasing alum
dose. XDBPsum was lowered from 150 to 94 /vg/L, and TTHMs from 86 to 55 /ug/L as
alum doses increased from 10 to 40 mg/L.
The study at Utility 12 was conducted at a conventional treatment facility with a
capacity of 72 mgd. Chlorine was added at two locations in the plant: before the
rapid mix (1.8 mg/L dose) and before filtration (1.3 mg/L dose). The total chlorine
contact time was approximately 100 minutes. Ammonia was added approximately 4
minutes after filtration, prior to the clearwell. Samples were collected at the plant
influent, sedimentation basin effluent, filter effluent and clearwell effluent. Alum doses
were varied from low (24.6 mg/L) to medium (45.7 mg/L) to high (73 mg/L) for the
study. Since the plant had no capability to control pH before or during the
sedimentation process, the pH values decreased as the alum doses increased. The low
alum dose removed 33 percent of the influent TOC, while the medium and high alum
doses removed 47 and 46 percent, respectively, as measured in the filter effluent. Most
of the TOC removal occurred in the sedimentation basins, with little or no additional
removal occurring through filtration. Figure ES-23 illustrates the effect of alum dose
on DBF concentrations by class. For XDBP , concentrations decreased from 87 to 69
//g/L as the alum dose increased from 25 to 75 mg/L. For TTHMs, the levels
decreased from 53 to 39 x/g/L. In general, individual THMs and HAAs decreased
slightly with increasing alum dose; little or no change was observed for HKs, HANs or
ALDs. That greater DBF removal was not observed with increasing alum dose was due
to the utility's prechlorination practices. Approximately 1.8 mg/L of free chlorine was
added to the raw water, with 75 minutes of contact time from the point of addition to
ES-9
-------
o
s
u
Effect Of CI2 and CIO2/CI2
on DBP Formation (Utility 16)
6O.O
5O.O
40.0
3O.O
20.0
10.0
O/-\
.o
-
.
.
-
X
^
2
/
/
1
oe
2
PSLfl
?
£
/
1
n
^
TVtv/
7
/
/"
^
X
?
2
El
^^ CIO2/CI2
Dist. Sya Residence Time (t)
1: t = O.75 hrs
2.- t = 16O hrs
MA - Not Analyzed
.fl >- ll 11.
12 12 12 12
HAN HK HAA ALD
DBP Class
RGURE ES-20
_J
01
O
-------
_J
^
g
•*—'
10
Effect of Alum Dose
on DBP Formation (Utility 3. SOS)
200
150 -
100 -
Low Alum Dose
Medium Alum Dose
High Alum Dose
Ixtaw »*».
~ TON BCAH mo COM Mo» !>*»-•
•» MBAA and OBAA
• OfUA bM»r »*«.
XDBPsum THM
HAN
HK
HAA
ALD
DBP Class
FIGURE ES-22
_j
s
*-t
0)
o
U
Effect of Alum Dose
on DBP Formation (Utility 12)
Low Alum Dose
Medium Alum Dose
High Alum Dose
» TCAN below M3L
** ACBIataerryOo batow
»•» Acataldehyda not analyzed
XDBPsum THM
HAN
HK
HAA
ALD
DBP Class
FIGURE ES-23
-------
Executive Summary
the sedimentation basin effluent. Consequently, DBF formation occurred before and
during TOC removal processes.
As discussed previously, mean overall TOC removal through the filtering plants
participating in the baseline sampling program was 24 percent. It is important to note
that for those utilities capable of increasing TOC removal by increasing applied alum
doses, the enhanced precursor removal represents only an incremental increase over that
achieved under normal operation. For example, at Utility 12, TOC removals of 24 to
35 percent were observed from plant influent to clearwell effluent during baseline data
collection (normal operation), and TOC removals up to 47 percent were achieved by
increasing alum doses in the treatment modification study.
Granular Activated Carbon Study
The granular activated carbon (GAC) study involved the collection of samples in Utility
11's 235-mgd conventional treatment facility and 2-mgd GAC demonstration plant over
a period of approximately 4 months. Unchlorinated water was diverted from the
conventional plant after presedimentation (alum and polymer addition followed by
lamella separation and 3-days of off-line storage), and through a rapid sand filter before
application to the GAC column. The GAC was Filtrasorb 400 (12 x 40 mesh) and the
column was operated with an empty bed contact time of 15 minutes. Samples were
collected at the GAC column influent and effluent and 3-day SDS tests were performed
using 4.5 mg/L of chlorine in order to evaluate DBF production. Samples were
collected on GAC column Run Days 0.2, 13, 25, 54, 82 and 95. The TOC removal
performance of the GAC column is illustrated in Figure ES-24. The GAC was very
effective for TOC removal at Utility 11, although research at other utilities indicates
that this technology has site-specific benefits. For the first 25 run days, the column
effluent TOC remained the same as that measured on Run Day 0.2 (O.I mg/L), which
represents the non-adsorbable fraction of the TOC. On Run Day 54, the column
effluent TOC had only increased to 0.2 mg/L, and it was not until the 82nd run day
that a substantial increase in column effluent TOC was observed (0.6 mg/L). The
impact of TOC removal on levels of SDS THMs is illustrated in Figure ES-25. This
figure shows that SDS TTHMs were extremely low (approximately 6 //g/L or less) in
the column effluent for the first 54 run days, indicating very effective removal of THM
precursors. After 82 run days, SDS TTHMs had increased to 34 //g/L and after 95 run
days. SDS TTHMs were 47 /ug/L. Comparing levels of SDS TTHMs in the column
influent and effluent. GAC treatment led to reductions in SDS TTHM levels of >97,
>92, >97 and >96 percent on each sampling day through the 54th run day, and 71
and 65 percent on the 82nd and 95th run days, respectively.
Despite the low bromide levels in Utility 11 's raw water (less than or equal to 0.06
mg/L), GAC effluent SDS samples on the 82nd and 95th run days had levels of
bromoform and dibromoacetic acid that exceeded levels in the GAC influent SDS
samples. The increase in the percentage of brominated THMs and HAAs may be due,
at least in part, to the increased ratio of bromide to precursor material after significant
levels of TOC have been removed in the GAC contactor.
ES-10
-------
8
O
TOG vs Run Time for GAG
Column Influent and Effluent. Utility 1 1
-+GAC Inf A A GAG Eff
Immediate Immediate
A ---- A
_J_J L I II I __l J j_A. t..-l 1 t t
20
40
6O
Run Days
FIGURE ES-24
80
.A
1OO
SDS TTHMs vs Run Time for GAG
Column Influent and Effluent. Utility 1 1
+ +GAC Inf A AGAC Eff
3-day SDS 3-day SDS
2OO
15O
1OO
5O
O
CI2 = 4.5 mg/L
pH = 8.2
25 C
'"•
20
4O
6O
Run Days
FIGURE ES-23
** All Tt-tM
..A-"
80
1OO
-------
Executive Summary
SUMMARY AND CONCLUSIONS
THMs. on a weight basis, represented the largest class of DBFs measured in this study
and the median TTHM value for the four quarters of baseline sampling of clearwell
effluents was 39 A/g/L (computed as the median of running annual average TTHMs for
each individual utility). HAAs were the second largest fraction detected, and aldehydes
were the third largest. Little difference was observed in the concentrations of influent
water quality parameters and concentrations of DBFs on a seasonal basis when
considering the overall medians for all 35 participating utilities; however, seasonal
variations were observed for individual utilities. In addition, levels of XDBPsum and
TTHMs were found to depend on water temperature.
The numerous correlations conducted for this study indicated that TTHMs correlated
well with the sum of halogenated DBFs measured in this study and with some DBF
classes (e.g., HANs), while correlations between THMs and other classes of DBF
compounds (e.g., HKs) were low; that bromide levels could be predicted from chloride
concentrations; and that bromide present in raw waters impacted the speciation of
THMs, HANs and HAAs. In addition, the shift in speciation to brominated DBFs
occurred even in inland utilities not impacted by saltwater intrusion, and in waters low
in bromide after preozonation or GAC treatment.
Ozone in conjunction with chlorine or chloramines as final disinfectants was generally
effective in lowering concentrations of classes of halogenated DBFs. The extent to
which halogenated DBF levels were decreased or increased after implementation of
ozonation depended primarily on the final disinfectant which was employed. Aldehyde
levels were found to increase upon ozonation. although it was found in one case
possible to remove aldehydes within the treatment plant when filters were operated
without a residual disinfectant.
Chlorine dioxide was effective for controlling DBFs at one utility, while at another, no
difference from DBF levels resulting from the utility's normal chlorination practices was
observed. Coagulation was effective in removing DBF precursors at the two utilities
studied, as long as chlorine was not added before the DBF precursors were removed in
the coagulation, flocculation, sedimentation and filtration processes. GAC was an
effective technique for controlling levels of DBFs by removing precursors at the one
utility studied.
ES-11
-------
Section 1
Introduction
-------
SECTION 1
INTRODUCTION
The United States Environmental Protection Agency (USEPA) will be developing
regulations to control disinfection by-products (DBFs) in drinking water as a result of
the 1986 amendments to the Safe Drinking Water Act (SDWA). Under these
amendments, the USEPA is required to develop a priority list of chemicals that may be
present in drinking water and to develop maximum contaminant levels (MCLs) for those
compounds. Included on this list are disinfectants, trihalomethanes (THMs) and other
DBPs. Although the schedule for promulgation of regulations under the SDWA
amendments remains uncertain at this time, the USEPA anticipates proposing DBP
regulations in September, 1991 and finalizing the regulations by September, 1992. If
the provisions of the SDWA are to be met within the required regulatory timetable, the
presence and control of the target DBPs must be fully understood.
PROJECT BACKGROUND AND OBJECTIVES
In October. 1987. the Association of Metropolitan Water Agencies (AMWA) entered
into a cooperative agreement with the USEPA to develop information on the formation
and control of DBPs in full-scale drinking water treatment systems. AMWA contracted
with the Metropolitan Water District of Southern California (Metropolitan) to perform
the study. Engineering services for the study were provided by James M. Montgomery,
Consulting Engineers, Inc. (JMM) through a subcontract with Metropolitan.
Project Objectives
The principal objective of the study was to collect data from representative water
utilities in the United States on the occurrence and control of DBPs in drinking water.
Specific objectives of the project included:
o Determine the occurrence of DBPs at 25 drinking water treatment facilities
around the nation. Facilities were selected to provide a broad range of source
water qualities and treatment processes.
o Determine the seasonal nature of the occurrence of DBPs as a function of
temperature, total organic carbon (TOC), pH, and other water quality
parameters.
o Determine the effect of changes in treatment processes and/or disinfectants on
the production of DBPs at bench, pilot and/or full scale at up to 10 drinking
water treatment facilities.
The study focused on the identification of DBPs expected in United States drinking
waters as a function of source water quality, water treatment process selection and
operation, and disinfection processes and chemicals. Previous studies or those currently
underway at the USEPA have been designed to define the occurrence and levels of
DBPs in a broad sampling of water treatment systems. This study, however, focused
1-1
-------
Introduction
on the relationships hetween source water quality, processes employed at water
treatment facilities, and the level and frequency of occurrence of DBFs.
Target Compounds
The DBFs of interest to the project are listed in Table 1-1, and include THMs,
haloacetonitriles (HANs). haloketones (HKs), haloacetic acids (HAAs), chloropicrin,
chloral hydrate, cyanogen chloride. 2,4.6-trichlorophenol, and aldehydes (ALDs). The
chemical structure of each of the target compounds is illustrated in Figure 1-1.
All of the halogenated DBF compounds selected for study appear on the USEPA's
Drinking Water Priority List of contaminants expected to be regulated by 1991. The
aldehydes (formaldehyde and acetaldehyde) were added to the study since ozonating
utilities were included in the study and these compounds have been identified as by-
products of the ozonation process (and speculated to be chlorination by-products as
well). In addition, the USEPA is considering the regulation of these aldehydes as part
of an ozone DBF priority list.
California Department of Health Services/California Public Health Foundation DBF
Study
The USEPA DBF project was conducted in conjunction with a similar study being
funded by a grant from the California Department of Health Services (CDHS) and the
California Public Health Foundation (CPHF). Metropolitan and JMM also conducted
the CDHS study. The CDHS study involved 10 utilities around the State of California
selected to provide a broad range of source water qualities and treatment processes.
The CDHS study focused on utilities with source waters representative of supplies used
by the majority of consumers in California.
Conducting the two studies simultaneously has been beneficial for both the USEPA and
the CDHS. Data for all 35 utilities involved in the combined studies has been available
to both funding agencies by providing each agency with copies of progress reports
prepared for the other agency's study. In addition, the combined USEPA/CDHS
studies provided the opportunity to study a larger number of utilities, allowing a more
representative selection of utilities. Data in this report reflect results from the 35
utilities participating in the combined USEPA/CDHS studies. A separate report was
prepared for the CDHS study (Metropolitan, 1989), focusing on the 10 California
utilities.
PROJECT DESCRIPTION
The USEPA project has been conducted over a period of two years. The first year of
the project (October, 1987 through September, 1988) focused on establishing and
verifying the analytical procedures at Metropolitan's Water Quality Laboratory,
selecting utilities to participate in the study, developing DBF baseline data through
implementation of a quarterly sampling program at the 25 utilities, and performing
process modification studies at two utilities. During the second year of the project
(October. 1988 through September, 1989), baseline data collection was completed and
process modification studies were conducted at six utilities. Project status has been
summarized in quarterly progress reports submitted to the USEPA.
1-2
-------
TABLE 1-1
LIST OF COMPOUNDS TARGETED IN STUDY
Compound*?
Trihalomethanes
chloroform
bromodichloromethane
dibromochloromethane
bromoform
Haloacetonitriles
tricliloroacetonitrile
dichloroacetonitrile
bromochloroacetonitrile
dibromoacetonitrile
Haloketones
I .l-dichloropropanone
1,1, l-tricliloropropanone
Miscellaneous chloro-organics
chloropicrin
chloral hydrate
cyanogen chloride
Haloacetic acids
monochloroacetic acid
dichloroacetic acid
trichloroacetic acid
monobromoacetic acid
dibromoacetic acid
Chlorophenols
2.4-dichlorophenol*
2.4.6-trichlorophenol
pentachlorophenol*
Aldehydes
formaldehyde
acetaldehyde
* These chlorophenols were only analyzed for during the first sampling quarter.
-------
TmHALOMCTHAMES
Cl
Cl
1
- C - H
i
Cl
Cl
1
Cl - C - H
i
Br
Cl
i
Br - C - H
i
Br
Br
i
Br - C -
i
Br
H
CHLOROFORM
DICHLOROBROMO-
METHANE
DIBROMOCHLORO-
METHANE
BROMOFORM
HALOACETONITRILES
Cl
Cl - C - C S N
i
Cl
Cl
I
Cl - C - C = N
i
H
Br
i
Cl - C - C = N
i
H
Br
i
Br - C - C = N
H
TRICHLORO-
ACETONITRILE
DICHLORO-
ACETONITRILE
BROMOCHLORO-
ACETONITRILE
DIBROMO-
ACETONITRILE
HALOKETONE3
Cl O H
i M i
CI-C-C-C-H
i i
H H
Cl
O
H
H
CI-C-C-C-H
i i
Cl H
1.1 -DICHLOROPROPANONE
1,1,1-TRICHLOROPROPANONE
MISCELLANEOUS
Cl
i
Cl - C - NOj
i
Cl
Cl H
i i
Cl - C - C - OH
i i
Cl OH
Cl - C 5 N
CHLOROPICR1N
(TRICHLORONITROMETHANE)
CHLORAL HYDRATE
CYANOGEN CHLORIDE
Structural Formulas for Disinfection By-Products
FIGURE 1-1
-------
KALOACrnC ACIM
Cl O
I II
H - C - C - OH
H
MONOCHLORO ACETIC
ACID
Cl O
I H
Cl - C - C - OH
i
H
DICHLOROACETIC
AGIO
Cl O
i n
Cl - C - C - OH
i
Cl
TRICHLOROACETIC
ACID
Br O
i n
H - C - C - OH
i
H
MONOBROMOACETIC
ACID
Br O
i n
Br - C - C - OH
i
H
DIBROMOACETIC
ACID
CHLOftOPHENOLS
Cl
Cl -/ O V OH
2,4-DICHLOROPHENOL
Cl
Cl Y O V OH
Cl
2.4,6-TRICHLOROPHENOL
Cl Cl
s /
Cl -/ O V OH
) (
Cl Cl
PENTACHLOROPHENOL
ALDEHYDES
H H
I I
H - C ~ C =O
I
H
ACETALDEHYDE
H - C =: O
FORMALDEHYDE
Structural Formulas for Disinfection By-Products
FIGURE 1-1 (Continued)
-------
Introduction
For (he CDHS study, (he first "year" of Ihe project (February, 1988 through June,
1988) included utility selection, baseline data collection at the 10 utilities, and
performance of process modification studies at two utilities. During the second year of
the study (July. 1988 through June, 1989), baseline data collection was completed.
The major components of the project, the baseline data collection and the process
modification studies, are described in more detail below. The utility selection process
is described in Section 2 of this report.
Baseline Data Collection
A sampling schedule for the 35 utilities participating in the USEPA and CDHS studies
was developed which provided for sampling of clearwell effluents (after final
disinfection but before distribution) in the selected plants on a quarterly basis for one
year. Ihe first sampling date was March 14, 1988 and the fourth sampling was
completed in February, 1989. The first sampling quarter (i.e.. mid-March through
April) corresponded to the spring season. The subsequent sampling quarters
corresponded to the summer, fall and winter seasons (i.e.. July through August.
October through November, and late December through February, respectively).
Furthermore, the utilities were sampled each quarter in the same order where possible.
In this manner, each utility was sampled four times, each time representing different
seasonal conditions in terms of temperature and water quality.
Utilities were sent coolers containing the sample bottles, packing material, blue ice,
detailed sampling instructions and a sample information sheet. Utility personnel
collected the grab samples at the designated sampling locations on the designated
sampling dates. The utilities were asked to fill in the sample information sheet with
plant operating information such as chemical dosages and locations of chemical addition
on the day of sampling. After sampling, the utilities returned the sampling kits via
overnight mail to Metropolitan for analysis.
The results of the baseline data collection achieved project objectives by providing
information on the occurrence of DBFs on a seasonal basis at the selected utilities.
These results are reported and discussed in detail in Section 5 of this report.
Treatment Modification Studies
Treatment modification studies are described in detail in Section 6 of this report. The
utilities at which these studies were conducted were selected based on the representative
nature of their normal treatment process, the flexibility offered by their treatment plant
design or other facilities (i.e., pilot plant facilities), and their willingness to contribute
resources to the study.
The principal goal of conducting the treatment modification studies was to identify, in a
preliminary manner, the processes or process modifications having the greatest impact
on DBF production. To achieve this goal, the following treatment modification studies
were conducted:
o Utility 3: Full-scale study of effects of alum dose on DBF precursor removal
by coagulation at pH 5.5.
1-3
-------
Introduction
o Utility 6: Full-scale study of effects of changing oxidant/disinfectant
(chlprine/chlpramines and ozone/chloramines).
o Utility 7: Pilot and full-scale study of effects of changing oxidant/disinfectant
(chlorine only, chlorine/chloramines, and ozone/chloramines).
o Utility II: Demonstration-scale study of granular activated carbon (GAC)
adsorption for removal of DBF precursors.
o Utility 12: Full-scale study of effects of alum dose on DBF precursor
removal by coagulation.
o Utility 16: Full-scale study of effects of changing oxidant/disinfectant
(chlorine only and chlorine dioxide/chlorine).
o Utility 19: Full-scale study of effects of changing oxidant/disinfectant
(chlorine only and ozone/chlorine).
o Utility 25: Full-scale study of effects of changing oxidant/disinfectant
(chloramines and ozone/chloramines).
o Utility 36: Pilot study of effects of changing oxidant/disinfectant (chlorine
only. chloramines only. ozone/chlorine, ozone/chloramines and
ozone/hydrogen peroxide/chloramines).
o Utility 37: Full-scale study of treating parallel trains with different
preoxidants/disinfectants (chlorine and chlorine dioxide).
Eight of the treatment studies were performed for the USEPA study and two were
performed for the CDHS study. Utilities 36 and 37 were selected for process
modification studies but did not participate in the baseline sampling program. The
results of all ten studies are reported and discussed in Section 6.
1-4
-------
Section 2
Utility Selection
-------
SECTION 2
UTILITY SELECTION
Twenty-five utilities around the United States were selected for participation in the
USEPA project. The first task of the selection process was the development of a
criteria matrix and an information request to be filled in by potential participants.
Then the potential utilities were screened and the utilities were selected for participation
based on the criteria matrix developed for the study.
INFORMATION REQUEST
An information request packet was sent to 104 potential participants in the study, along
with a cover letter introducing the study and providing some background information
which would familiarize the recipients with the study and emphasize the study's
importance to the water utility industry. The letter also indicated the level of
commitment from each participant that would be required for the study. In some
instances, a particular plant owned/operated by the utility was identified in the
information request and in other cases where the utility had more than one treatment
plant, they were asked to respond for the plant producing the highest levels of THMs.
Copies of the information request form and cover letter are provided in Appendix A of
this report.
The information request was to be filled in by each recipient, providing data on the
plant's treatment processes, chemical doses and contact times, and the quality of the
plant's raw water, finished water and distributed water. Responses were received from
78 utilities, a response rate of 75 percent.
Not only did the completed information requests provide the basis of the selection
process, but as the study progressed, the information was a valuable reference for
interpreting baseline data as they were collected.
SELECTION MATRIX
The 35 utilities (25 for the USEPA study and 10 for the CDHS study) were selected for
participation in the study from those responding to the information request based on the
following criteria:
o Utility willingness to participate,
o Disinfectant/oxidant(s) in use or available,
o Source water type.
o Source water quality.
o Geographical location.
o Treatment process configuration,
o Ability to alter or segregate treatment processes, and
o On-site pilot plant capabilities.
The resulting matrix of utilities is presented in Table 2-1. The matrix is divided into
two major categories, treatment type (conventional, direct filtration, softening and
2-1
-------
TABLE 2-1
DISINFECTION BY-PRODUCTS IN DRINKING WATER STUDY
UTILITY SELECTION MATRIX
TREATMENT
GROUNDWATER
LAKE/RESERVOIR FLOWING STREAM
CONVENTIONAL
ClermontCo., OH I
Long Beach, CA 2
Norwich, CT* IB
MWD, CA -Mills 2",5h
Arlington, TX* 2A
Hackensack, NJ 2e
MWD, CA -Weym. 2
San Francisco, CA 1
Big Spring, TX* 5
Shreveporl, LA 6a,8b
Cape Girardean, MO* 2A
Cincinnati, OH I
Contra Costa WD, CA 2A
Sacramento, CA 1
Santa Clara Valley, CA 2
Newport News, VA IA
DIRECT
FILTRATION
East Bay MUD, CA 1
Las Vegas, NV I
Little Rock, AR I
Aurora, CO l"A,5bA
Los Angeles DWP, CA 3
SOFTENING
Palm Beach Co., FL 2
Wausau. WI* I
Minot. ND I
Santa Monica, CA 1
Macomb, IL* IA
Galveston,TX 7c,4l1
Louisville, KY 2
Ft. Meyers, FL* I
Emporia, KN* 2A
Omaha, NB IA
DISINFECTION
ONLY
Mesa Consol., CA 3
North SkagitCo., WA* I
New York City, NY I
Newark, NJ 1
Note: Utilities participating in the California Public Health Foundation study are listed in bold type.
* Population under 50.000; all others over 50,000.
Key for chemical addition:
I - chlorine only
2 - chlorine + chtoramines
3 - ozone + chlorine
4 - chlorine + chlorine dioxide
5 • chloramines only
6 - chloramines + chlorine dioxide
7 - chlorine 4- chloramines +
8 - ozone + chloramines
A - powdered activated carbon
B - potassium permanganate
chlorine dioxide
" - first quarter only
h - second through fourth quarters
c - First through third quarters
*' - fourth quarter only
e - clear well effluent sampled
before ammonia addition
-------
Utility Selection
disinfection only) and source water type (groimdwater, lake/reservoir, and flowing
Stream). Within these categories, geographical location and disinfectant type (free
chlorine, chloramines, chlorine dioxide, and/or ozone) were also considered.
population was also used as a category in developing the selection matrix (<50,000
0nd >50.000). as was TTHM level (<25 x/g/L, 25 to 50 /ug/L, and >50 //g/L).
The selection process was aided by data from the THM survey conducted for the
American Water Works Association Research Foundation (AWWARF) in 1987
(McGuire and Meadow. 1988). Using the AWWARF data plus the questionnaire
responses, the selection process was aimed at filling the utility selection matrix as
completely as possible based on the criteria described above.
To illustrate the nation-wide distribution of the selected utilities, a map showing their
locations is presented in Figure 2-1. It should be noted that 10 of the 11 participating
utilities in California were involved in the CDHS study.
PROCESS TRAINS OF THE PARTICIPATING UTILITIES
Simplified schematics of the process trains employed at each participating utility are
shown in Figure 2-2 by utility identification number. The figure indicates that of the
35 utilities participating in the combined USEPA and CDHS studies:
o Sixteen utilized a conventional treatment process with coagulation,
flocculation. sedimentation and filtration;
o Five employed a direct filtration treatment process;
o Ten softened; and
o Four employed disinfection only.
The schematics, in addition to the information provided in Table 2-1, indicate the
following disinfection schemes for the participating utilities (during the majority of
baseline data collection):
o Nineteen of the utilities used only free chlorine throughout their treatment
trains and as a final disinfectant;
o Nine utilities used free chlorine and provided for some free chlorine contact
time before the addition of ammonia (either with or without further chlorine
addition) to form chloramines as the final disinfectant;
o Three utilities used only chloramines (concurrent addition of chlorine and
ammonia, with no free chlorine contact time) throughout their treatment
trains and as a final disinfectant;
o One utility employed chlorine with subsequent ammonia addition to form
chloramines. in addition to chlorine dioxide;
o Three utilities used ozone (two utilities employed ozone and free chlorine,
and one utility utilized ozone, chloramines, and chlorine dioxide).
Several utilities had changes in their disinfection schemes during the course of the
baseline sampling. For instance, Utility 4 changed from prechlorination/
postammoniation to concurrent addition of chlorine and ammonia after the first quarter.
Utility 18 changed from chlorine to chloramines after the first quarter. Utility 25
switched from chloramines to ozone/chloramines after the 1st quarter. Utility 27, a
2-2
-------
UTILITIES PARTICIPATING IN DBP STUDY
PALM
BEACH
FIGURE 2-1
-------
AERATION
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
Chlorine
UTILITY 1
t
Aeration
v^ mum ic i
m^^
V '•-.''
Fluoride
Phosphate
UTILITY 2
t
Polymer
t
Retention
, te-
t
\
\.
N
V,
Polymer
Chlorine
Chlorine'
Ammonia
t
Alum
Caustic Soda
Potassium Permanganate
Polymer
UTILITY 3
Fluoride, Chlorine
Caustic Soda, Phosphate
UTILITY 4
t
Ammonia
Chlorine
Alum
Polymer
t
d
o
Polymer
Chlorine
1
* Configuration during last 3 quarters: during 1st quarter same as Utility 7
Sodium Hydroxide
FIGURE 2-2
-------
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
T
Powdered
Activated
Carbon
T
Alum
Polymer
UTILITY 5
Fluoride
Ammonia
Sodium Hydroxide
UTILITY 6
t
Alum
Polymer
Chlorine
I
o
o
Chlorine -
Chlorine Ammonia, Chlorine
Caustic Soda
* Clearwell effluent sampled before ammonia addition
1
t
Chlorine
Alum
Polymer
UTILITY 7
Ammonia
Sodium Hydroxide
Chlorine
Alum
UTILITY 8
Chlorine
Polymer
Chlorine
Sodium Hydroxide
Fluoride
FIGURE 2-2 (Continued)
-------
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
1
Lime
Chlorine
Polymer
Alum
Powdered Activated
Carbon (Intermittent)
UTILITY 9
Ammonia
Fluoride
Chlorine, Lime
T
Ammonia
Chlorine
Lime (Intermittent)
Alum
1
O
o
UTILITY 10
Alum
Polymer
UTILITY 11
1 .
o
o
*v
s
\
Lamella
Settler
.
'\
Reservoir
Storage
(3 days)
Chlorine
Lime
Fluoride
Ferric Sulfate*
I
Hydraulic
Jump
••'<••••„
• Fciric Sulftte added intermittently (in use on 5/8/89.6/12/89 and 7/17/89)
FIGURE 2-2 (Continued)
-------
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
T
Chlorine
Alum
Powdered Activated
Carbon (Intermittent)
UTILITY 12
o
o
Lime, Fluoride
Chlorine, Polymer
Caustic Soda
Ammonia
UTILITY 13
I
Chlorine
Alum
f
O
O
Polymer
(Intermittent)
Chlorine
Lime
Polymer
Chlorine
Alum
UTILITY 14
Polymer
Chlorine
Ammonia
FIGURE 2-2 (Continued)
-------
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
Alum
Chlorine
Fluoride
UTILITY 15
Chlorine
Lime
1
UTILITY 16
t
Chlorine
T
Polymer
Ferric Chloride
Sodium Hydroxide
Sodium Silicate
Chlorine
1
UTILITY 17
Chlorine
Alum
o
o
Lime, Phosphate
Fluoride, Chlorine
1
FIGURE 2-2 (Continued)
-------
OZONE
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
Polymer, Fluoride
Ammonia
UTILITY 18
t
Polymer
Alum
Pr«i/H*»n"H Artivatft/1 farlvm
~1
o
1
Chlorine, Polymer
Powdered Activated Carbon.
Ammonia
I
Chlorine
Ammonia
Fluoride
* Used Chlorine only during 1st quarter
UTILITY 19
1
000000
f\ f\ /\ S\
0000
0000000
000000000
feme imonae
Polymer
Chlorine
1
T »
\
X.
^s
s
Chlorine- intermittent
i
i
i
f
•
•
K ',',',
U^ t f
Chlorine
Powdered Activated Carbon
(Intermittent)
Aiu
UTILITY 20
urn
Lime
oo
Chlorine
Chlorine, Fluoride
Phosphate, Lime
1
FIGURE 2-2 (Continued)
-------
AERATION
FILTRATION
t
Chlorine
Lime
Polymer
UTILITY 21
Solids
Contact
Clarifier
Chlorine Ammonia
I I
__ Alum, Lime,
V Activated Silica
UTILITY 22
Solids
Contact
Clarifier
Fluoride.
Carbon Dioxide
Chlorine
UTILITY 23
Lime
Sodium Aluminate
Polymer
Solids
Contact
Clarifier
Fluoride
Phosphate
Chlorine"
Carbon Dioxide
1
Chlorine
UTILITY 24
Aeration
Ion
Exchange
Caustic Soda
Chlorine
i
FIGURE 2-2 (Continued)
-------
OZONE
RAPID MIX FLOCCULATION SEDIMENTATION OZONATION
FILTRATION
Chlorine Dioxide
UTILITY 25
pooooo
0000
0000 OCX)
ooooooooo
Chlorine**, Ammonia **
Alum
Polymer
Chlorine
Ammonia'
»*. " >l 1 > :
Mf.=^^^'
000000
0000
0000000
ooooooooo
; Fluoride
Lime Chlorine Dioxide
* No Ozonation during 1st quarter
** 1st quarter only.
UTILITY 26
Alum
Polymer
Lime
Powdered Activated
(Intermittent)
Solids
Contact
Clarifier
Chlorine
Recarbon-
ation
Carbon Dioxide
Chlorine
UTILITY 27
t t
Chlorine Dioxide Lime
Polymer Ferric Sulfate
Solids
Contact
Clarifier
Polymer
Carbon Dioxide
Chlorine
' Ammonia off-line during 4th quarter
Phosphate, Fluoride
Chlorine Dioxide
Ammonia
FIGURE 2-2 (Continued)
-------
PRESEDIMENTATION RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
UTILITY 28
1
Alum
Polymer
Chlorine
/•?' ^^^'^
< "•*" ^MtV^
.
o
o
>ss_
S.
k>
s
Sf^^^
•v^AU-S,-, ^f-
W-^> ** »"*• ffv. •"<
Fluoride
Carbon Dioxide
Ammonia
Lime
Soda Ash
Chlorine, Ammonia
UTILITY 29
Solids
Contact
ClariFier
Solids
Contact
Clarifier
Alum
Polymer
Chlorine
Alum
Polymer
Lime, Chlorine
Ammonia
Lime. Alum
UTILITY 30
Polymer
Chlorine
^"VvfS
I.
o
o
Ammonia
Chlorine
Fluoride
Carbon Dioxide
I
Carbon Dioxide
Powdered Activated Carbon
FIGURE 2-2 (Continued)
-------
OZONE
PRESEDEVtENTATION
SEDIMENTATION FILTRATION
I
Polyr
l
ner
t
fe* : -•*,.
\j « »
Retention
Fluori
t
de, Chlorine —
'
-^-
Powdered Activated Carbon
(Intermittent)
Polymer
Lime, Silica
Chlorine
i
000000
0000
0000000
ooooooooo
UTILITY 32
Chlorine
i
UTILITY 33
Chlorine
i
UTILITY 34
Chlorine
i
UTILITY 35
Chlorine
Lime
FIGURE 2-2 (Continued)
-------
Utility Selection
prechlorine/postamrnonia/chlorine dioxide utility, had their ammonia off-line on the 4th
quarter sampling date. Additionally, although Utility 6 employed prechlorination/
poslammoniation. the clearwell effluent samples were collected before the ammonia
addition point, therefore it is classified as a chlorinating utility.
In all. 35 utilities were each sampled four times during the course of baseline data
collection, yielding 140 total data points. Of the 140 disinfection schemes in use
during baseline data collection. 78 (or 56 percent) were free chlorine only, 40 (29
percent) were prechlorination/postammoniation. 11 (8 percent) were chloramines only.
8 (6 percent) were ozone/chlorine, and 3 (2 percent) were ozone/chloramines.
The schematics indicate the following coagulants were used by the utilities:
o Seven of the utilities employed alum only;
o Ten used alum with a polymer as a coagulant aid;
o One utility used polymer as a primary coagulant;
o Two used ferric chloride and polymer; and
o Six utilities used no coagulant.
Additionally, the softening utilities used a variety of coagulants in several different
combinations:
o Lime plus polymer;
o Lime, alum and polymer (three utilities);
o Alum/polymer in one solids contact clarifier for color removal and
lime/polymer in a second unit for softening;
o Lime, alum and activated silica;
o Lime, sodium aluminate and polymer;
o Lime, ferric sulfate and polymer; and
o Lime. alum, ferric sulfate. activated silica and polymer.
Eight of the 35 utilities utilized powdered activated carbon, at least on an intermittent
basis. Only one utility added potassium permanganate as a preoxidant. One utility
employed ion exchange for softening.
2-3
-------
Section 3
Methodology
-------
SECTION 3
METHODOLOGY
This section briefly describes the sampling and analytical methodology of the study.
References are provided for further information. In addition, detailed analytical
protocols are provided in Appendix C of this report.
SAMPLING PROCEDURE
The sampling instructions (Appendix B) included in each sampling kit provided detailed
information to the utilities on how to properly sample for the study. A project
engineer contacted each utility to arrange the sampling date for each quarter. At that
time, if water quality conditions or treatment plant operations were deemed
unacceptable (non-representative), the sampling date was postponed. Typically, on the
sampling date, all sample collection was completed by a utility within 1 to 2 hours.
Evaluation of plug flow conditions in each plant was not within the scope of the study.
Grab samples (see Table 3-1) were collected by utility personnel at clearwell effluents
(after final disinfection but prior to distribution). Starting with the second sampling
quarter. TOC samples were collected each sampling date at plant influents and filter
influents. Also starting with that quarter, ultraviolet (UV) absorbance measurements
were made at all three sampling points, the bromide and chloride levels of plant
influents were determined, and formaldehyde and acetaldehyde analyses were made of
the clearwell effluents. Some plant influents were analyzed for these aldehydes during
the third sampling quarter; aldehyde determinations were made for all plant influents
during the fourth sampling quarter.
Plant influent samples were collected at the head of the treatment plants and
corresponded to "raw" water before the addition of disinfectant/oxidant, coagulant,
lime, etc. This sampling location was important in enabling the characterization of the
DBF precursors (i.e., as measured by TOC and UV-254) prior to the addition of any
treatment chemicals. In addition, bromide measurements were made on plant influent
water to determine the influence of this ion on the production of brommated DBFs.
Since strong oxidants such as chlorine or ozone can convert bromide to hypobromous
acid, bromide samples were collected upstream of the preoxidation point.
Samples were dechlorinated and/or preserved as outlined in Table 3-1 and were
analyzed as soon as possible within established holding times. Holding times are
described in Table 3-2. The dechlorination agents and preservatives for the DBF
fractions were evaluated in holding studies to ensure that analyte concentrations held to
+ /- 20 percent of their initial values. Total organic halide (TOX) samples were
dechlorinated with a sodium sulfite solution. Because sampling kits were prepared and
shipped to field locations at least 2 weeks prior to sampling, the instability of the
dechlorination solution required that TOX samples be dechlorinated and preserved upon
receipt at the laboratory. Standard Methods (1985) establishes a 28-day maximum
storage for bromide sampledThis holding time is only applicable to unchlorinated
samples, as an unquenched free chlorine residual can lower the level of bromide. In
this study, bromide samples were collected upstream of chlorine addition points.
3-1
-------
TABLE 3-1
SAMPLING KIT CONTENTS
Analytical
Fraction
Sample Bottles
Number Size
Dechlorination Agent
and/or Preservative*
Plant influent sample bottles:
Formaldehyde/acetaldehyde 3, 40 mL
Bromide/chloride 1, 60 mL
TOC/UV 3, 60 mL
Filter influent sample bottles:
TOC/UV 3, 60 mL
Clearwell effluent sample bottles:
PE-DBPs< 3. 40 mL
Chloral hydrate 3. 40 mL
Haloacetic acids'1 4, 40 mL
Cyanogen chloride 4, 40 mL
Formaldehyde/acetaldehyde 3. 40 mL
TOX 2, 250 mL
TOC/UV 3. 60 mL
Additional bottles in kit:
Travel blank--NH4CI 2, 40 mL
Travel blank-ascorbic acid 2, 40 mL
Travel blank—formaldehyde 2, 40 mL
None; HgCI2 solution + NH4C1 crystals'1
None
None"
None"
65 mg NH4CI crystals
20 mg ascorbic acid
65 mg NH4C1 crystals
20 mg ascorbic acid
None; HgCl2 solution + NH4C1 crystals'1
None3
None8
65 mg NH4C1 crystals
20 mg ascorbic acid
None; HgC!2 solution + NH4CI crystals'1
HFor the following analytical fractions, additional reagents are added after
sampling as soon as possible after receipt at Metropolitan:
TOC/UV: Acidify with reagent-grade phosphoric acid to pH < 2.
TOX: Dechlorinate and acidify, respectively, by addition of l/i mL of 50%
sulfuric acid and 3 drops (utilizing a 3V4" pasteur pipette) of a fresh
saturated sodium sulflte solution.
''No dechlorination agent and preservative used during second and third sampling
quarter; 40 ^uL of a 10 mg/mL mercuric chloride solution plus 65 mg ammonium
chloride crystals used during fourth sampling quarter.
cPE-DBPs (pentane-extractable disinfection by-products) are THMs. HANs.
haloketones. and chloropicrin.
'Includes 2.4.6-trichlorophenol.
-------
TABLE 3-2
SAMPLE HOLDING TIMES
Analytical Sample Holding Extract Holding
Fraction Limit Limit
PE-DBPs Extract immediately 2 weeks
Chloral hydrate 21 days 15 days
Haloacetic acids 9 days 7 days
Cyanogen chloride Analyze immediately Not applicable
Formaldehyde/acetaldehyde Extract immediately 14 days
TOX 2 weeks Not applicable
TOC/UV 28 days Not applicable
Bromide/chloride 28 days Not applicable
-------
Methodology
ANALYTICAL METHODS
A method was developed for the analysis of THMs, HANs, haloketones. and
chloropicrin that employed modified THM liquid/liquid extraction (LLE) (Koch et al.,
1988). For this analytical fraction, pentane was used as the extraction solvent, sodium
sulfate was added to improve the partitioning from the aqueous phase to the solvent,
and a capillary gas chromatograph/electron capture detector (GC/ECD) was utilized for
adequate resolution of the analytes. The more polar chloral hydrate required a similar
LLE method, but methyl t-butyl ether was used as the extraction solvent (Krasner et al.,
1989). HAAs and 2,4,6-trichlorophenpl were analyzed by an acidic, salted ether LLE,
and they required esterification with diazomethane prior to GC/ECD analysis (Krasner
et al., 1989). Cyanogen chloride was analyzed by a purge-and-trap gas
chromatograph/mass spectrometer (GC/MS) method (Krasner et al., 1989).
Formaldehyde and acetaldehyde were analyzed by a derivatization/extraction GC/ECD
method (Yamada and Somiya, 1989; Glaze et al., 1989b)
The TOC measurements were performed according to the persulfate-ultraviolet
Standard Method 505B (1985). TOC samples were acidified with reagent-grade
phosphoric acid to a pH of less than 2 after receipt of the samples at Metropolitan's
laboratory. This preservation technique was also used on UV samples. UV was
measured at 254 nanometers (nm) with a UV-visible spectrophotometer (Lambda 5,
Perkin-Elmer Corp., Norwich, CT) and a 1-cm quartz cell. The UV was also measured
at 800 nm to provide a correction for the presence of turbidity or suspended solids.
Because some plant effluent samples were chloraminated, monochloramine (which has
UV absorbance at 243 nm) could have presented an interference problem. However,
the acid preservation converted monochloramine to dichloramine in approximately 2
hours. The latter species has a UV peak at 293 nm, so it did not present an
interference problem. To evaluate the effect of the phosphoric acid on the UV results,
some unpreserved parallel effluent samples were also analyzed, and these yielded
comparable results to the acidified samples.
Bromide and chloride analysis was conducted with an ion chromatograph (Model 2000,
Dionex Corp., Sunnyvale, CA), a 20-^uL or 50-//L sample loop, a high-performance ion
chromatography analytical column (AS4A, Dionex Corp.), an anion micromembrane
suppressor, and a conductivity detector. The eluant was a solution of 2 millimolar
Na2CO3 and I millimolar NaHCO3.
The TOX analysis was performed using USEPA Method 9020 from SW 846 Test
Method for Evaluating Solid Waste, Physical/Chemical Methods (Second Edition).
SIMULATED DISTRIBUTION SYSTEM TESTING
Some samples in this study were analyzed using a simulated distribution system (SDS)
test protocol (Koch et al.. 1989) developed at Metropolitan. In this protocol, samples
are chlorinated and incubated under conditions which simulate actual full-scale
conditions in Metropolitan's distribution system. Typically, samples are chlorinated at
a level which will leave a chlorine residual of at least 0.5 mg/L at the end of the
incubation period. Firstly, this mimics typical distribution system practices of
providing some minimal chlorine residual to consumers, and secondly, some DBFs are
unstable when the chlorine residual has gone to zero. Samples are buffered to a pH of
3-2
-------
Methodology
8.2. which is consistent with many utilities' practice of corrosion control. Some DBFs
will undergo some hydrolysis at this pH, as would be experienced in actual distribution
systems operating at such a pH. The samples are incubated at 25°C, which will
produce maximum levels of THMs and other DBFs, as would be experienced in
summer months. The samples are incubated for one to three days, to provide a
measure of DBF levels that would be expected in a distribution system with such a
detention time. Most importantly, the use of standardized SDS conditions enables a
consistent set of data to be produced that often cannot be realized in a distribution
system which may be influenced by other confounding variables. Also, the use of SDS
testing allows evaluation of expected distribution system DBF levels when studying pilot
plant and demonstration-scale facilities (see Section 6).
For the GAC study at Utility 11 (see Section 6), SDS tests were performed. The SDS
test was selected to provide a standard set of conditions by which to evaluate DBF
precursor removal at Utility I I's GAC demonstration plant. For example, Utility I I's
ambient water temperature ranged from 11 to 27°C during the course of the study.
However, with SDS testing, all samples were incubated at 25°C. During the day 13
and day 25 samplings, some additional SDS tests were set up at ambient water
temperature. The results are shown in Tables 3-3 and 3-4.
An SDS test can only mirror a limited number of actual operating conditions. For
example, routine SDS samples at Utility 11 received a 4.5 mg/L chlorine dose, in order
to provide adequate chlorine for a 3-day incubation period. However, the actual plant
dosed different amounts of chlorine, depending on current demand requirements. In
addition, the actual plant applied chlorine in three locations, two before conventional
filtration and one after. The pH of SDS samples are buffered to a pH of 8.2 to 8.3,
whereas the pH ranged from 8.4 to 8.8 in Utility 1 I's actual distribution system during
the course of this study.
For some DBFs, the distribution value is between that detected in the two SDS tests at
Utility 11 (Tables 3-3 and 3-4). For example, for the Day 13 testing, chloroform was
42. 31. and 55 //g/L in the distribution, ambient SDS, and standard SDS samples,
respectively. At that same time, 1.1,1-trichlpropropanone had values of 0.55, 1.2, and
0.28 x^g/L. respectively. Since this ketone is an intermediate DBF (chloroform is the
stable endproduct of 1.1.1-trichloropropanone), it appears as if the elevated
temperature of the standard SDS test drove the reaction of 1,1,1-trichloropropanone
hydrolysis the furthest. During the Day 25 testing, this ketone was detected at levels of
0.18. 1.5. and 0.36 /ug/L in the distribution, ambient SDS, and standard SDS samples,
respectively. The lowest ketone value was found in the distribution system where the
elevated pH (i.e.. 8.8) probably resulted in base-catalyzed hydrolysis of this DBF.
In general, though, the SDS tests yield comparable DBF levels as that detected in the
actual distribution system. Unless noted. SDS tests were performed at the standard
SDS condition of 25r>C.
Another SDS consideration was that the GAC effluent would not have as high a
chlorine demand as the GAC influent. Thus. GAC effluents were analyzed by two
different SDS tests, one at a chlorine dose of 4.5 mg/L and another at 2.0 mg/L. The
latter dose is a more realistic chlorine level for Utility 1 I's GAC effluent when TOC
removal is high. Since the chlorine dose may influence the level of DBF production,
3-3
-------
TABLE 3-3
UTILITY 11 GAC STUDY: DAY 13 DBF RESULTS
Parameter
Temperature. °C
pH
Detention time/
incubation, days
Cl: dose*. mg/L
C\2 residual. mg/L
DBFs, x/g/L:
CHC1,
CHBrCI,
CHBr,Cl
CHBr*
TCAN
DCAN
BCAN
DBAN
1.1 -DCP
l.l.l-TCP
MCAA
DCAA
TCAA
MBAA
DBAA
Chloropicrin
Chloral hydrate
Distribution
System
15
8.7
3 (approx.)
0.7
42
15
6.0
0.49
<0.03
1.8
0.55
0.64
0.34
0.55
1.6
15
9.4
<0.5
1.4
0.60
9.9
Ambient SDS Standard SDS
GAC Inf.* SDS Blank GAC Inf.* SDS Blank
12
8.41
3
4.5
2.2
31
13
4.8
<0.91
<0.03
2.0
0.59
-------
TABLE 3-4
UTILITY 11 GAC STUDY: DAY 25 DBF RESULTS
Parameter
Temperature. °C
PH
Detention time/
incubation, days
Cl, dose*. mg/L
CI-, residual. mg/L
DBFs, uglL:
CHCI,
CHBrCI,
CHBrXl
CHBr;
TCAN
DCAN
BCAN
DBAN
I.I-DCP
I.I.I-TCP
MCAA
DCAA
TCAA
MBAA
DBAA
Chloropicrin
Chloral hydrate
Distribution
System
16
8.8
3 (approx.)
2.7
0.7
52
16
6.1
0.27
<0.03
0.75
0.49
0.20
0.25
0.18
1.7
17
8.4
<0.5
I.I
0.65
II
Ambient SDS Standard SDS
GAC Inf.* SDS Blank GAC Inf.* SDS Blank
14
8.22
3
4.5
0.90
33
14
5.0
<0.9I
<0.03
1.6
0.57
-------
Methodology
the GAC effluent was dosed at the same concentration as the GAC influent in order that
comparisons could be made where the TOC level was the only variable.
A comparison of SDS data for the GAC effluent is made for the two SDS protocols
used at Utility 11 (see Table 3-5). The standard SDS conditions for the GAC influent
of a 4.5 mg/L chlorine dose and buffering to pH 8.2 were followed on one set of GAC
effluents. The other set of GAC effluents received 2.0 mg/L chlorine and were not
buffered (pH approximately 7.5). On Day 82, there was 0.60 mg/L of TOC in the
GAC effluent, and the chlorine demand was 0.7 and 1.4 mg/L for the 2.0 and 4.5
mg/L dosed tests, respectively. On Day 95, there was 0.85 mg/L of TOC in the GAC
effluent, and the chlorine demand was 1.0 and 1.5 mg/L for the 2.0 and 4.5 mg/L
dosed tests, respectively. The difference in chlorine demands between the two SDS
protocols may, in part, be related to the differences in pH of the two tests. On both
davs. the THMs were higher in the 4.5 mg/L chlorine dose tests; however, the
difference was less pronounced when there was higher TOC present (i.e., 38 and 47
u%IL TTHMs in the 2.0 and 4.5 mg/L chlorine dose tests, respectively). The HANs
and the 1,1.1-TCP were lower in the 4.5 mg/L chlorine dose tests. As these DBFs are
reactive intermediates, the higher chlorine dose and pH probably resulted in their
degradation (and resultant formation of other DBFs as stable endproducts).
SDS tests involve preparation of blanks (i.e., distilled water plus a bromide spike at the
level occurring in the raw water). There are low levels of DBFs detected in SDS
blanks. Whether these are formed due to DBF precursors in the distilled water, SDS
bottle, reagents used, or a combination of these is uncertain. If the blank levels are
due to precursors in the bottle and/or reagents, then SDS blanks should be subtracted
from the SDS sample results. For example, the ambient SDS test performed on Day
13 samples from Utility 11 had 0.80 and 0.35 yug/L 1,1-dichloropropanone in the SDS
sample and blank, respectively. If blank subtraction is performed, the difference of
0.45 0g/L compares well with the 0.34 //g/L found in the distribution sample. Thus,
data reported in Section 6 reflect SDS blank subtraction.
SDS blank subtraction was performed in a previous GAC study performed by
Metropolitan (McGuire et al., 1989). In the latter study, SDS samples of virgin GAC
effluents each typically had levels of 1 to 2 /yg/L of dichloroacetic acid (DCAA), when
different empty bed contact times were evaluated in parallel (i.e., 7.5, 15, 30, and 60
minutes). One explanation was that this DBF formation was due to a nonadsorbable
fraction of TOC. However, SDS blanks tended to have, within analytical error, the
same values for DCAA, which tended to support that these levels were background
levels in all samples. To resolve this issue for the current study, minimum reporting
levels (MRLs) for SDS testing were calculated based upon the analysis of seven SDS
blank samples (see discussion in Quality Assurance section below). Thus, relatively
high MRLs were obtained for SDS samples as contrasted to instantaneous samples (see
Table 3-6). SDS data that were less than these values were not reported, thus
precluding the need for SDS background subtraction in many instances. However,
when higher blank levels were detected, they were subtracted from SDS sample results.
For example, the standard SDS tests at Utility 11 for the Day 13 samples (see Table
3-3) for trichloroacetic acid (TCAA) in the GAC influent and blank yielded values of 19
and 2.1 */g/L, respectively. Thus, a blank-subtracted result of 17 /ug/L was reported,
As in this example, the blank subtraction often resulted in a number which did not
differ from the uncorrected value within analytical error. The evaluation of SDS data is
3-4
-------
TABLE 3-5
UTILITY 11 TREATMENT STUDY
COMPARISON OF GAC EFFLUENT SDS TESTS:
EFFECT OF CHLORINE DOSE AND pH ON DBFs
Day 82 GAC Effl. SDS* Day 95 GAC Effl. SDS*
Chlorine Dose: 2.0 mg/L 4.5 mg/L 2.0 mg/L 4.5 mg/L
pH
C12 residual, mg/L
DBFs, //g/L:
CHCI^
CHCUBr
CHClBr,
CHBr3
TCAN
DCAN
BCAN
DBAN
1.1 -DCP
U.I-TCP
MCAA
DCAA
TCAA
MBAA
DBAA
Chloropicrin
Chloral Hydrate
7.55
1.30
4.0
7.9
8.5
1.8
<0.029
0.40
0.47
-------
Methodology
more complex, though, when SDS results approach the same value as that detected in
SDS blanks. For this reason, MRLs are set at higher values for SDS testing, since
quantification can only be reliable when SDS values are significantly higher than blank
levels.
QUALITY ASSURANCE
Metropolitan is certified by the CDHS for the analysis of THMs and volatile organic
compounds. Furthermore, Metropolitan obtained permission to utilize the
THM/HAN/haloketone/chloropicrin method described above for THM compliance
monitoring. For the list of DBFs under investigation, no officially-approved USEPA
methods exist. To validate all the DBF methods used for this study, Cincinnati tap
water and Metropolitan's Weymouth plant were both sampled as is and a separate set
of bottles was spiked with all the target-compound DBFs. These samples were split
between Metropolitan, Montgomery Laboratories (Montgomery), and the Technical
Support Division (TSD) and the Environmental Monitoring Support Laboratory (EMSL)
of the USEPA (not all laboratories ran each analytical fraction).
Metropolitan's, Montgomery's, and TSD's laboratories all agreed on THM, HAN,
haloketone, and chloropicrin results (Fair, 1988a). Metropolitan and TSD obtained
comparable results for chloral hydrate and cyanogen chloride (Fair, 1988a).
Metropolitan and TSD had comparable TOC results, and Montgomery and TSD had
comparable TOX data (Fair, 1988a). In addition, Metropolitan, TSD and EMSL, in
both this initial cross-calibration and a subsequent one which also included
2.4.6-trichlorophenol, obtained comparable HAA results (Fair, 1988a; Fair, I988b).
Finally, Metropolitan submitted a 16-point Quality Assurance (QA) Project Plan for the
USEPA study, which was approved by the USEPA QA officer Audrey D. Kroner
(McGuire, 1988). The QA protocol covers accuracy, precision, independent
verification and the use of an internal standard. Accuracy is dependent on many
factors, but the most important is the calibration curve. Accuracy was monitored by
calculating the recoveries of samples which had been enhanced with known
concentrations of the compounds of interest. Precision is another parameter that is
dependent on more than one factor. The precision of a method was monitored by
analyzing samples in duplicate and calculating the difference between the two analyses.
Independent verification of a method was done by interlaboratory calibration (discussed
above). The internal standard was used to insure that there were consistent injections
into the GC of samples and standards.
All of the above mentioned parameters were important in assuring that good quality
data were produced. It is important to note that all portions of the QA program met
the established standards in order for an analysis to be considered in control.
Initial calculations of the method detection limits (MDLs) were made according to the
Code of Federal Regulations 40 Part 136, July 1, 1987. A set of 7 standards were
prepared in organic-pure water at 1 to 5 times the estimated detection limit. Each
standard was analyzed according to the method and the standard deviation of the 7
replicate measurements for each analyte was determined. These MDLs were used as
minimum reporting levels (MRLs), except where the instrumental detection limit has
proved to be higher. Often, the MRLs corresponded to the lowest level standard on the
3-5
-------
Methodology
calibration curve. The MRL for each analyte is shown in Table 3-6. Since some DBFs
were detected in the SDS blanks, a higher MRL was used for SDS samples, the value
based on a calculation utilizing the results from 7 SDS blank samples.
At the beginning of this study, many MRLs corresponded to the MDL-calculated
values, even though the calibration curves did not include a standard at that low of a
level. Thus, MRLs were either raised to correspond to the lowest level standard
analyzed (e.g., chloroform MDL was 0.021 /yg/L, but lowest level standard run was
0.102) or the calibration curve was expanded to include a lower level standard at the
same level as the calculated MDL (e.g., dichloroacetonitrile MDL was 0.025 /vg/L, but
lowest level standard had initially been 1 /7g/L; thus a 0.027 j/g/L level standard was
added to the curve). As was discussed in the previous section on SDS testing, SDS
samples require a higher MRL to remove the uncertainty in reporting SDS data that are
comparable to those levels detected in SDS blanks.
Quantitation was done using an external standard calibration curve. Standards were
prepared in organic-pure water spiked with the appropriate DBFs and extracted,
derivatized. or purged with the samples. The extracted or purged standards were used
to compensate for the varying extraction or purging efficiencies of the different
compounds in the analysis. Samples with analytes outside the range of the calibration
curve required re-analysis with an appropriate dilution.
An internal standard was spiked into each sample or extract. The purpose of the
internal standard was to monitor injections made into the GC. A sample injection was
deemed acceptable if the area counts of the internal standard peak did not vary more
than +/- 10 percent from other samples which were extracted or purged on the same
date.
Sample spikes were analyzed to monitor the extraction or purging efficiency of specific
analytes in sample matrices. This measured the accuracy of the method in a natural
matrix. The spiked samples were analyzed at a frequency of at least 10 percent of the
samples. The samples were spiked at the levels that are typically found in samples.
Data were entered into the quality control (QC) table directly after the analysis. The
QC charts were reviewed by the analyst and the immediate supervisor. All spike
recoveries must fall within the upper and lower control limits to be acceptable. If a
spike recovery was not acceptable, then the samples were re-analyzed from the point
where the last spike was in control and a sample was re-spiked and re-analyzed only for
those analytes that were out of control.
Initial control limits were defined by calculating the mean percent recovery from the
most recent 50 sample spike data points. The 99 percent confidence interval is +/-
three times the standard deviation, which defined the control limits. Warning limits
were defined as +/- two times the standard deviation. If a sample recovery was above
or below the warning limit this indicated there was a potential problem. The problem
was determined and corrected before the analysis was out of control. Control limits
and warning limits were re-calculated on a semiannual basis using the most recent 50
spiked sample percent recovery values. Data points that were out of control were not
included in the re-calculation of new control limits.
3-6
-------
TABLE 3-6
MINIMUM REPORTING LEVELS (MRLs)
Analyte
DBFs, //g/L:
Trihalomethanes
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Haloacetonitriles
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
Haloketones
1 . 1 -Dichloropropanone
1.1,1 -Trichloropropanone
Haloacetic Acids
Monochloroacetic acid
Dichloroacetic acid
Trichloroacetic acid
Monobromoacetic acid
Dibromoacetic acid
Chlorophenols
2 ,4-Dichlorophenol
2.4.6-Trichlorophenol
Pen t achlorophenol
Aldehydes
Formaldehyde
Acetaldehyde
Miscellaneous DBFs
Chloropicrin
Chloral hydrate
Cyanogen chloride
Minimum Reporting Levels
1st
Qtr
0.021
0.020
0.017
O.OJ3
0.012
0.025
0.035
0.076
0.030
0.013
1.0
0.6
0.6
0.5
0.6
2.0
0.3
0.4
NA
NA
0.010
0.2
0.02
2nd
Qtr
0.021
0.020
0.017
0.013
0.012
0.025
0.035
0.076
0.030
0.013
1.0
0.6
0.6
0.5
0.6
NA
0.4
NA
1.0
1.0
0.010
0.2
0.1
3rd
Qtr
0.102
0.103
0.108
0.101
0.029
0.027
0.027
0.028
0.026
0.029
1.0
0.6
0.6
0.5
0.6
NA
0.4
NA
1.0
1.0
0.026
0.050
0.1
4th
Qtr
0.102
0.103
0.108
0.101
0.029
0.027
0.027
0.028
0.026
0.029
1.0
0.6
0.6
0.5
0.6
NA
0.4
NA
1.0
1.0
0.026
0.050
0.1
Spring/
Summer
1989
0.102
0.103
0.108
0.101
0.029
0.027
0.04
0.08
0.026
0.029
1.0
0.6
0.6
0.5
0.6
NA
0.4
NA
3.0
3.0
0.026
0.050
0.2
ai,
SDS
Tests
0.52
0.22
0.48
0.91
0.03
0.04
0.19
1.20
0.09
0.03
1.7
2.1
1.5
0.5
0.6
NA
0.4
NA
1.7
1.8
0.04
0.43
O.I
-------
Table 3-6, Continued
Minimum Reporting Levels (MRLs)
Analyte
Other Analytes
Total organic carbon, mg/L
Total organic halide, //g/L
Ultraviolet absorbance. cm '
Chloride, mg/L
Bromide. mg/L
Minimum Reporting Levels
1st
Qtr
0.05
10
NA
NA
NA
2nd
Qtr
0.05
10
0.005
O.I
0.01
3rd
Qtr
0.05
10
0.001
0.1
0.01
4th
Qtr
0.05
10
0.001
O.I
0.01
Spring/
Summer
1989
0.05
10
0.001
O.I
0.01
SDS
Tests
NA
10
NA
NA
NA
NA = Not Analyzed
SDS = Simulated Distribution System
-------
Methodology
Prior to the calculation of initial control limits for spike recoveries, control limits of
+ /- 20 percent were used for all DBF fractions except for cyanogen chloride. For the
latter analyte, control limits from EPA Method 525.2 for volatile organic compounds
were preliminarily used (i.e., +/- 40 percent; however, warning limits were set at +/-
20 percent). As the data in Table 3-7 indicate, calculated control limits were consistent
with the preliminary assumptions. For the HAAs, the use of +/- three standard
deviations created control limits that were relatively high due to some outlier values
affecting the statistics. In fact, virtually all spike recoveries fell within +/- two
standard deviations; therefore, the control limits for these analytes were based on the
latter statistics.
Sample duplicates were analyzed in order to monitor the precision of the method.
Duplicates were analyzed on randomly selected samples at a frequency of at least 10
percent of the samples. Data were entered into the QC table after the analytical run
was completed. The QC charts were reviewed by the analysts and the immediate
supervisor. Control limits were determined by calculating the range as a function of
the relative standard deviation (coefficient of variation) as specified in Standard
Methods proposed Method 1020B. The normalized range (R,,) was calculated as
tollows:
+ x2)/2J
where x, and x2 are the duplicate values.
A mean normalized range (Rin) was calculated for 50 pairs of duplicate data points:
n
where n = number of duplicate pairs.
The variance (s2) of the normalized ranges was calculated:
S* = ~
n-1
The standard deviation (s) was calculated as the square root of the variance. The upper
and lower control limits were defined as R,,, + 3s and zero, respectively. All duplicates
must fall within the control limits to be acceptable. If a duplicate was not acceptable,
then the samples were re-analyzed from the point where the last duplicates were in
control and a duplicate was re-analyzed only for those analytes that were out of control.
The upper warning limit was defined as R + 2s. If an Rn was outside the warning
limit, this indicated there was a potential problem. The problem was investigated
before the analysis was out of control. Control limits were recalculated on a semi-
annual basis using the most recent 50 points. Data points that were out of control were
not included in the recalculation of new control limits.
3-7
-------
TABLE 3-7
DBF QUALITY CONTROL LIMITS
Analyte
CHC13
CHCI2Br
CHCIBr,
CHBr,
J
TCAN
DCAN
BCAN
DBAN
I.I-DCP
Period
Covered
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Spike Quality Control Limits
UCL UWL LWL LCL
Percent Recovery
120
121
119
120
123
119
120
125
126
120
120
120
120
122
116
120
126
134
120
131
141
120
134
135
120
123
126
113
111
116
112
117
117
113
112
114
110
119
124
122
129
124
124
116
119
83
81
86
84
86
82
87
81
77
85
91
84
87
82
82
84
88
87
80
75
73
80
79
77
80
78
73
80
81
73
80
68
78
80
84
74
80
78
71
80
71
74
80
81
79
Duplicate Limits
UCL UWL
Normalized Range
0.2
0.2
0.14
0.2
0.2
0.10
0.2
0.2
0.13
0.2
0.2
0.13
0.2
0.2
0.2*
0.2
0.2
0.21
0.2
0.2
0.20
0.2
0.2
0.19
0.2
0.2
0.20
0.10
0.076
0.099
0.095
0.15
0.15
0.14
0.15
-------
Table 3-7, Continued
DBF Quality Control Limits
Analyte
I.I.I-TCP
Chloropicrin
Chi Hydrate
MCAA
DCAA
TCAA
MBAA
DBAA
TCP
CNCI
Period
Covered
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Recalculated
Preliminary
Initial
Preliminary
Initial
Spike Quality Control Limits
UCL UWL LWL LCL
Percent Recovery
120
121 113 82
127 118 81
120
123 115 81
120 113 86
120
124 113 72
130 119 74
120
1200
1190
120
1290
1290
120
1370
1340
120
1220
1200
120
1210
1230
120
1330
140 120 80
138 126 75
80
75
72
80
73
79
80
62
63
80
870
760
80
740
740
80
810
780
80
820
820
80
820
740
80
700
60
63
Duplicate Limits
UCL UWL
Normalized Range
0.2
0.2
0.17
0.2
0.2
0.15
0.2
0.2
0.074
0.2
0.2
0.21
0.2
0.2
0.16
0.2
0.2
0.16
0.2
0.2
0.20
0.2
0.2
0.19
0.2
0.2*
0.2
0.28
0.13
0.11
0.057
0.16
0.12
0.12
0.15
0.15
0.21
-------
Table 3-7, Continued
DBF Quality Control Limits
UCL = Upper control limit
UWL = Upper warning limit
LWL = Lower warning limit
LCL = Lower control limit
* Not enough points for calculation of control limits.
# The calculated control limits were relatively high due to some outlier values affecting the
statistics; therefore, the calculated warning limits were used as the control limits in order to
keep closer control over the analysis.
-------
Methodology
Prior to the calculation of initial control limits for the precision of replicate analyses,
control limits of 0.2 for a normalized range were used for all DBF fractions. As Table
3-7 indicates, calculated control limits were determined later in the study; however,
these values were consistent with initial assumptions.
ON-SITE EXTRACTION STUDIES
For the DBF studies, samples were collected from 35 utilities throughout the United
States. These samples were collected in bottles containing a dechlorination agent
and/or preservative to ensure the integrity of the sample during shipping and storage
prior to analysis. To evaluate how effective the preservation methods were in different
matrices, two of Metropolitan's laboratory staff extracted samples on-site at Utility 2.
In addition, split samples were preserved and brought back to the laboratory for
analysis. The latter samples were stored at 4°C for 24 hours to simulate typical
overnight shipping conditions. Samples were analyzed in triplicate, both on-site and at
the laboratory, to provide adequate statistics. For the Metropolitan laboratory
extractions, the pentane-extractable DBFs were extracted upon receipt (after the
24-hour shipping period). The chloral hydrate and HAA samples were extracted in the
laboratory within the established holding period (see Table 3-2).
Both the chlorinated filter effluent and chloraminated clearwell effluent were sampled in
order to evaluate the effects of the dechlorination agents/preservatives on both
disinfectants. Ammonium chloride is added to samples to convert free chlorine to
chloramines as a means of preserving pentane-extractable DBFs (i.e., THMs, HANs,
haloketones, and chloropicrin). Since the clearwell effluent is already chloraminated,
that location was sampled on-site both with and without the addition of ammonium
chloride for this analytical fraction. Other analytical fractions evaluated included
chloral hydrate (dechlorinated with ascorbic acid) and HAAs (preserved with
ammonium chloride).
A concern in this special study was the effect of the dechlorination agents/preservatives
on the pH of samples. The table below shows the pH drop brought about by these
reagents:
Sample Actual + Ammonium Chloride + Ascorbic Acid
Filter effluent 7.48 7.27 3.78
Clearwell effluent 7.55 7.16 4.42
The data are presented in Tables 3-8 and 3-9; the findings are as follows:
I. The THMs, chloropicrin, choral hydrate, and the HAAs were stable to + /- 20
percent between extractions on-site and in the laboratory in both samples.
Note (in Table 3-9) that mpnochloroacetic acid was 27 percent higher in the
clearwell sample analyzed in the laboratory. However, if one examines the
number of significant figures in the results (i.e.. 2) and the relative standard
deviation of the on-site results (i.e.. standard deviation divided by the mean,
3-8
-------
TABLE 3-8
UTILITY 2: CHLORINATED FILTER EFFLUENT
ON-SITE AND LAB EXTRACTION
On-Site Extraction
Compound
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Trichloroacelonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
1 . 1 -Dichloropropanone
I.I.I -Trichloropropanone
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Chloropicrin
Chloral Hydrate
Mean*
Owg/L)
60.3
15.6
3.41
0.15
0.32
13.1
3.08
0.63
0.83
2.26
1.2
15.5
24.6
ND
0.6
0.10
7.02
Std.
Dev.
2.1
0.31
0.05
0.003
0.01
0.35
0.05
0.04
0.02
0.06
NA
0.67
0.61
NA
0
0.002
0.35
* Some mean data reported to three digits
however, data only
# Percent Stability =
NA = Not analyzed
ND = Not detected
good to two
100 x (mean
Lab Extraction
Mean*
(x/g/L)
58.6
14.6
3.00
0.13
0.14
10.9
2.03
0.24
0.66
1.91
1.3
14.2
23.3
ND
0.6
0.09
6.3
to enable percent
Std.
Dev.
3.2
0.73
0.25
0.007
0.006
0.47
0.17
0.05
0.06
0.11
O.I
0.45
1.0
NA
0
0.008
0.26
stability
Percent
Stability*
97
94
88
87
44
83
66
38
80
85
108
92
95
NA
100
90
90
calculations;
significant figures.
of lab extraction)/(mean of on-site extraction)
-------
TABLE 3-9
UTILITY 2: CHLORAMINATED CLEARWELL EFFLUENT
ON-SITE AND LAB EXTRACTION
On-Site Extraction
Compound
Chloroform
Bromodichloromethane
Dihromochloromethane
Bromoform
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonilrile
Dibromoacetonitrile
1 . l-Dichloropropanone
1 . 1 . l-Trichloropropanone
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Chloropicrin
Chloral Hydrate
Unpreserved*
Mean* Std.
0/g/L) Dev.
56.2
14.7
3.00
0.14
0.21
12.7
2.66
0.47
1.43
1.71
NA
NA
NA
NA
NA
0.13
NA
1.0
0.30
0.11
0.005
0.004
0.37
0.07
0.01
0.1 1
0.06
NA
NA
NA
NA
NA
0.04
NA
Preserved"
Mean*
0/g/L)
56.0
14.7
3.03
0.14
0.19
12.6
2.69
0.47
0.98
1.80
1.1
13.6
22.9
ND
0.5
0.12
6.74
Std.
Dev.
0.46
0.03
0.03
0.001
0.004
0.17
0.03
0.009
0.02
0.01
0.21
0.10
0.47
NA
0.05
0.00
0.07
Lab Extract.
Preserved"
Mean*
0/g/L)
49.8
13.0
2.47
0.12
0.097
9.90
1.84
0.18
0.58
1.46
1.4
13.7
21.7
ND
0.6
0.10
6.2
Std.
Dev.
3.4
1.1
0.29
0.008
0.005
0.62
0.23
0.21
0.04
0.14
0.12
0.06
0.35
NA
0
0.005
0.02
Percent
Stability**
89
89
82
86
51
79
68
38
59
81
127
101
95
NA
120
83
92
*
**
NA
ND
Results from an unpreserved sample (i.e., no ammonium chloride).
Results from a preserved sample (ammonium chloride for all but chloral hydrate,
which was ascorbic acid preserved).
Some mean data reported to three digits to enable percent stability calculations:
however, data only good to two significant figures.
% Stability = 100 x ((mean of lab extraction)/(mean of preserved on-site extraction)!
Not analyzed
Not detected
-------
Methodology
thus 19 percent), the calculation of the stability is affected by the uncertainty
in the former data.
2. Trichloroacetonitrile (TCAN), as in Metropolitan water in previous studies
(Koch et al.. 1988). degraded to 44 to 51 percent of its initial value in 24
hours. In Metropolitan's previous studies, the other HANs were stable to at
least 80 percent of their initial value; however, the same situation was not
experienced at Utility 2. Bromochloro- and dibromoacetonitrile (DBAN)
degraded to 66-68 and 38 percent, respectively, of their initial (on-site) values
in 24 hours; although, dichloroacetonitrile (DCAN) did remain stable (i.e., to
79-83 percent). Note that DBAN experienced a high standard deviation in
the clearwell samples.
3. Both haloketones were stable in the filter effluent samples; however,
1,1-dichloropropanone (1,1-DCP) degraded to 59 percent of its on-site value
in 24 hours in the clearwell sample. Also note that clearwell samples
collected with and without preservative yielded similar on-site results for all
pentane-extractable DBFs, except for 1,1-DCP. The unpreserved sample had
1.43 /wg/L 1,1-DCP; whereas, the ammonium-chloride-preserved sample
yielded an on-site 1,1-DCP value of 0.98 //g/L.
To continue to evaluate the preservation techniques, a laboratory analyst from
Metropolitan performed on-site extractions at Utility 33. This utility was chosen for its
low alkalinity and lack of natural buffering capacity. This matrix is ideal for studying
the pH change and effects caused by the addition of these preservatives. Samples that
were extracted in the field included the pentane-extractable DBFs and chloral hydrate.
The pH of the water was measured in the field as is and at the laboratory after field
preservation and overnight shipment. The results are as follows:
pH
Sample Actual + Ammonium Chloride + Ascorbic Acid
Effluent 6.96 6.31 3.46
The data on the chlorinated effluent are presented in Table 3-10; and the findings are
as follows:
I. There was little in the way of brominated DBPs, due to the low bromide level
in the source water. Thus, it was difficult to evaluate those data, as on-site
results were close to the minimum reporting levels. However, data for the
other DBPs indicate that chloroform, bromodichloromethane, the haloketones.
chloropicrin, and chloral hydrate were stable to + /- 20 percent of their on-
site values.
2. The major HAN present was DCAN; although, the levels detected were less
than 1.0 //g/L. The laboratory result was higher by 0.24 //g/L.
During the field study, an attempt was made to analyze some samples where the
chlorine residual was not quenched or converted to chloramines. This utility practices
3-9
-------
TABLE 3-10
UTILITY 33: CHLORINATED EFFLUENT
ON-SITE AND LAB EXTRATION
On-Site Extraction
Compound
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
1 . 1-Dichloropropanone
I.I.I -Trichloropropanone
Chloropicrin
Chloral Hydrate
# Percent stability = 100
* On-site result too close
stability
** Tentative result due to
NA = Not analyzed
ND = Not detected
Mean
(x/g/L)
22
1.3
ND
0.16
ND
0.61
0.07**
0.09
1.8
3.3
0.50**
1.7
x [(mean
Std.
Dev.
1.1
0.06
NA
0.01
NA
0.02
0.0
0.12
0.04
0.14
0.02
0.04
Lab Extraction
Mean
(A/g/L)
23
1.4
ND
-------
Methodology
disinfection only, so the effluent sample is located, in time, very close to the point of
chlorine application. Thus, a high free chlorine residual was present in the effluent
sample (i.e.. 2.5 mg/L). The field laboratory where on-site extractions were performed
was located approximately two hours away from the effluent sample tap. Thus,
unpreserved samples continued to react with free chlorine yielding high DBF levels
(e.g.. 62 /yg/L of chloroform was detected; whereas, preserved samples yielded 22
/ug/L). These results emphasize the need for a dechlorination agent or preservative to
stop the reaction of free chlorine in forming DBFs after a sample is collected.
One final utility (Utility 6) was selected for an on-site study. The pH results are as
follows:
Sample Actual + Ammonium Chloride + Ascorbic Acid
ClearwelI effluent 8.00 7.19 4.66
At Utility 6. chloraminated clearwell effluent was sampled for pentane-extractable DBFs
and chloral hydrate. In addition, water was collected with no preservatives. The data
from Table 3-1 I indicate the following:
1. The THMs. chloropicrin, and chloral hydrate were stable to +/- 20 percent.
2. TCAN was not detected in these samples. The other HANs, unlike at Utility
2, were all stable. The laboratory results were higher than the on-site values;
however absolute differences were j<0.24 /ug/L for each HAN.
3. The haloketones were stable between the on-site and laboratory analyses of
preserved samples (j<0.25 /ug/L for each compound). As observed at Utility
2. the unpreserved sample had 1.6 /ug/L 1,1-DCP, while the preserved sample
had 0.91 //g/L.
Overall, the on-site experiments confirm the findings of Metropolitan's holding studies
(Koch et al. 1988; Krasner et al, 1989), that the chosen dechlorination agents and
preservatives for the pentane-extractable DBFs, chloral hydrate, and HAAs do result in
laboratory results representative of the DBF levels at the time of sampling.
Furthermore, in continuing to evaluate the stability of the pentane-extractable DBFs,
nine different utilities' clearwell effluents were sampled and shipped (overnight) to
Metropolitan. They were extracted upon receipt, plus a second set of aliquots were
extracted four days after the initial sampling. The results were essentially the same
(within analytical error) on both days (see Table 3-12). TCAN. though, did degrade
between the two extraction dates, which is consistent with previous results on this
unstable HAN. Similar experiments have been performed with the HAAs and chloral
hydrate, and re-extractions have yielded comparable data. More holding studies need
to be performed in other matrices in order to better define the validity of these
dechlorination agents and preservatives. The data, however, do support the idea that
samples can be shipped to the laboratory for analysis.
3-10
-------
TABLE 3-11
UTILITY 6: CHLORAMINATED CLEARWELL EFFLUENT
ON-SITE AND LAB EXTRACTION
On-Site Extractions
Unpreserved*
Mean Std.
Compound O^g/L) Dev.
Chloroform 6.9 0.17
Bromodichloromethane 6.4 0.21
Dibromochloromethane 3.5 0.12
Bromoform 0.45 0.01
Trichloroacetonttrile ND NA
Dichloroacetonitrile 0.81 0.09
Bromochloroacetonitrile 0.50 0.03
Dibromoacetonitrile 0.57 0.03
!.l-Dichloropropanone 1.6 0.17
l.l.l-Trichloropropanone 0.87 0.1
Chloropicrin 0.20 0.01
Chloral Hydrate 3.2 0.04
* Results from an unpreserved sample (i.e
Preserved"
Mean
(x*g/L)
6.5
6.2
3.4
0.43
ND
0.79
0.50
0.63
0.91
0.95
0.20
2.6
Std.
Dev.
0.06
0.06
0.1
O.I
NA
0.02
0.03
0.10
0.07
0,06
0
0.02
. , no ammonium
** Results from a preserved sample (ammonium chloride for
hvdrate. which was ascorbic acid preserved).
ft Percent stability = 100 x ({mean of lab
NA = Not analyzed
ND = Not detected
extraction)/(mean
Lab Extract.
Preserved"
Mean Std.
0/g/L) Dev.
7.2
6.6
3.8
0.43
ND
1.0
0.63
0.87
1.1
1.2
0.21
2.7
chloride).
all but chloral
0.4
0.4
0.36
0.03
NA
0.06
0.06
0.23
0.12
0.06
0.01
0.09
of preserved on -site
Percent
Stability'
III
106
1 12
100
NA
127
126
138
121
126
105
104
extraction))
-------
TABLE 3-12
FOIMB-D
THACTABL
Utility/Martwr of
Days after Sailing CBClj CBCljBi
Utility 1
1 Day
4 Days
Utility 3
1 Day
4 Days
Utility 23
1 Day
4 Days
Utility 24
1 Day
4 Days
Utility 26
1 Day
4 Days
Utility 27
1 Day
4 Days
Utility 29
1 Day
4 Days
Utility 31
1 Day
4 Days
Utility 35
1 Day
4 Days
0.80
1.1
2.3
2.4
1.9
1.8
0.81
0.76
43
40
6.9
7.2
117
114
36
34
20
19
1.5
2.2
1.2
1.2
3.8
3.6
0.85
0.81
10
9.9
19
18
56
50
15
14
2.4
2.3
X DZSZHTE
CHClBr2
2.3
3.4
0.26
0.27
5.3
5.0
4.5
4.3
2.5
2.4
26
23
18
17
6.5
6.2
0.11
0.10
UJ.UB HI
CHBrj
1.3
2.0
0.080
0.074
2.3
2.1
23
21
0.10
0.097
7.2
7.2
0.76
0.74
0.61
0.62
HD
ND
-fiaXJUCXS: mXXIX HOUJlBG STBDI
Concentration (pq/L)
TCMT DCM BCMI DBftB
ND
HD
0.009
ND
HD
ND
ND
ND
0.017
ND
HD
HD
0.055
0.027
RD
HD
ND
ND
0.15
0.16
0.67
0.68
0.97
0.99
ND
HD
3.6
3.6
0.94
1.0
10
9.8
1.4
1.3
0.85
0.85
0.36
0.45
0.12
0.12
1.3
1.3
0.31
0.27
1.2
1.1
2.0
2.0
4.8
4.5
1.2
1.0
0.046
0.043
0.90
1.2
0.060
0.054
1.9
2.0
5.2
4.1
0.29
0.28
3.7
3.5
1.2
1.1
0.62
0.59
ND
ND
1,1-DCF
ND
0.11
1.1
1.0
0.26
0.25
ND
ND
0.52
0.51
0.41
0.48
1.5
1.5
0.35
0.33
1.1
1.1
1,1,1-TCP CHP
0.067
0.087
1.6
1.5
0.097
0.12
HD
ND
2.0
1.8
0.66
0.50
4.4
4.3
0.41
0.40
2.0
1.9
0.010
0.015
0.082
0.083
0.043
0.053
ND
ND
0.86
0.78
0.058
0.13
0.38
0.41
0.29
0.28
0.16
0.17
HD = Not detected
-------
Methodology
ALDEHYDE DERIVATIZATIONS/EXTRACTIONS ON-SITE
When aldehyde analysis was added to the scope of work, a new preservation concern
developed. Formaldehyde in particular is a very unstable compound. Thus, samples
were analyzed upon receipt at the laboratory. Another investigator performed
derivatizations and extractions in the field for aldehyde analyses of four utilities (Glaze
et al. 1989b); however, such an action is cost prohibitive when 35 utilities nationwide
are being analyzed on a quarterly basis. Therefore, on-site evaluations were performed
at Utilities 33 and 6 for the aldehyde fraction.
A combination of two preservatives for aldehyde fractions was evaluated. Mercuric
chloride was added as a biocide, in order to inactivate microorganisms capable of either
producing or degrading aldehydes. In addition, ammonium chloride was added to
convert free chlorine to chloramines, as it was suspected that a free chlorine residual
could continue to form additional aldehydes. At Utility 33 (see Table 3-13), it is clear
that the addition of mercuric chloride was essential for the influent sample, as
unpreserved samples had no detectable levels of the measured aldehydes after the
24-hour shipping period. Levels of aldehydes at this utility, which utilizes free
chlorine, were low (<:5 pgIL of each aldehyde), and the coefficient of variation was
high for some of the analyses.
However, when Utility 6 implemented ozonation, it was expected that higher levels of
aldehydes would be produced, which would yield more conclusive results on the use of
preservatives and the ability to ship aldehyde samples back for laboratory analysis. In
this experiment, unpreserved samples on-site were compared to fully preserved samples
analyzed after a 24-hour shipping period. As Table 3-14 shows, formaldehyde levels
were identical between on-site and laboratory analyses (e.g., the ozone contactor
effluent had 16 and 15 yug/L of formaldehyde for on-site and laboratory analyses,
respectively). Most of the acetaldehyde results were comparable as well (e.g., 7.4
versus 5.7 //g/L for the plant effluent analyzed on-site and in the laboratory,
respectively). However, the on-site extraction blank contained 3.2 /wg/L of
acetaldehyde, while none was detected in the laboratory extraction blank, so it was
difficult to compare the two sets of results due to a possible acetaldehyde contamination
problem on-site. However, these data do indicate that aldehyde samples can be
preserved and shipped to the laboratory for analysis of these analytes upon receipt.
PRESERVATION OF TOX SAMPLES
For this study, total organic halide (TOX) samples were collected with no
dechlorination agent or preservative in the field and shipped iced overnight to
Metropolitan's laboratory. After receipt, the samples were dechlorinated with a fresh
sodium sulfite solution and preserved with sulfuric acid. Several studies were
performed to evaluate to what extent TOX formation occurred during the 24-hour
shipping period. TOX bottles were shipped with (1) no preservative reagents and (2)
only sodium sulfite. Contrary to expectations, the latter samples yielded TOX results
which were generally much higher than those dechlorinated in the laboratory. For
example, the TOX results for Utility 33 were 680 /ug/L for a laboratory-dechlorinated
sample and 880 pgIL for the sample shipped with sulfite. These data suggested to
some USEPA scientists (Sorrell and Brass. 1988) that the sodium sulfite solution was
unstable and, perhaps, samples collected in bottles shipped with this chemical were not
3-11
-------
TABLE 3-13
UTILITY 33: ALDEHYDE RESULTS
ON-SITE AND LAB EXTRACTION
Location/Compound
On-Site Extractions
Unpreserved Preserved*
Mean Std. Mean Std.
0/g/L) Dev. (jig/L) Dev.
Laboratory Extractions
Unpreserved Preserved*
Mean Std Mean Std.
0/g/L) Dev. (//g/L) Dev.
Influent/
Formaldehyde
Effluent/
Formaldehyde
Influent/
Acetaldehyde
Effluent/
Acetaldehyde
2.3
3.4
1.6
4.0
2.9
O.I
0.2
0.3
NA
1.8
NA
2.2
NA
0.1
NA
0.3
<1.0 NA
5.1 0.58
-------
TABLE 3-14
UTILITY 6: FORMALDEHYE RESULTS
ON-SITE AND LAB EXTRACTION
On-Site Extract.* Lab Extract.*
Location
Plant Influent
O, Contactor Eff.
Plant Effluent
Mean
0/g/L)
0.63**
16
16
Std.
Dev.
0.06
0
0
Mean
0/g/L)
0.83**
15
16
Std.
Dev.
0.12
0.58
0.58
Percent
Stability*
NA**
94
100
* On-site samples contained no preservatives; laboratory samples contained mercuric
chloride and ammonium chloride.
# Percent Stability = 100 x [(mean of lab extraction)/(mean of on site extraction)]
** Levels detected below minimum reporting level of 1.0 /ug/L
NA = Not analyzed
-------
Methodology
being dechlorinated in the field. When the next set of samples was received, TOX
bottles shipped with the sulfite solution were analyzed for chlorine residual and, indeed,
a residual remained. Because sampling kits were prepared and shipped to field
locations at least two weeks prior to sampling, the instability of the dechlorination
solution required that TOX samples be dechlorinated and preserved upon receipt at the
laboratory. Samples dechlorinated at Metropolitan's laboratory were treated with a
fresh sodium sulfite solution each time in order to ensure that the samples were
dechlorinated.
To test the importance of timing of the addition of preservation agents, the following
experiments were performed. Metropolitan's chlorinated filter effluent and
chloraminated plant effluent were sampled and stored at 4°C for 24 hours to simulate
overnight shipping conditions. Then the samples were preserved with those reagents
not originally present in the TOX bottle at time of sampling. These data (Table 3-15)
indicate that in Metropolitan's matrix, results were the same when fully dechlorinated
and preserved at the time of sampling or after 24 hours at 4°C. However, samples only
dechlorinated at the time of sampling had a slight loss of TOX when not acidified
immediately.
When on-site extractions were performed at Utility 2, another experiment was
performed with TOX samples as well (Table 3-16). Samples were evaluated by on-site
dechlorination (with fresh sodium sulfite) and preservation (with sulfuric acid) versus
dechlorination/preservation at the laboratory (after 24-hour storage at 4°C). The
chloraminated water had the same TOX when dechlorinated/preserved on-site or in the
laboratory. However, the chlorinated water did appear to increase by 16 percent
during shipment/storage with no dechlorination at time of sampling.
When on-site analyses were conducted at Utility 33, special TOX sampling was
included (Table 3-16). The sample dechlorinated and acidified in the field contained
390 /yg/L of TOX, while the sample dechlorinated/preserved 24 hours later contained
610 //g/L. There was a 56 percent increase in TOX during the shipping period. These
results reflect the fact that even during refrigerated shipping, the free chlorine was able
to react with the sample matrix and produce additional TOX during the 24-hour period.
Since Utility 33 practiced disinfection only and the sample tap was located near the
point of chlorination. the large amount of free chlorine residual (i.e., 2.5 mg/L) and
short contact time before sampling resulted in the sample containing sufficient chlorine
and precursors to produce the additional TOX during snipping.
Further TOX experiments were carried out at Utility 6 (Table 3-16). Utility 6's filter
influent was sampled, since it represented a point close to initial chlorination (at the
time of this sampling). TOX was 60 /yg/L when dechlorinated on-site and 110 //g/L
when dechlorinated in the laboratory. The filter effluent samples had a longer contact
time with chlorine prior to sampling and there was a smaller difference in TOX results
between the on-site and laboratory-dechlorinated samples (77 //g/L on-site versus 110
/jg/L in the laboratory). The chloraminated plant effluent samples were less affected by
the timing of dechlorination (mean results of 78 //g/L of TOX for on-site dechlorination
and 89 /yg/L for the laboratory dechlorination). Samples collected from Utility 6's
distribution system (water temperature 6°C) had 76 to 89 /yg/L of TOX (dechlorinated
in the laboratory), suggesting that the TOX did not increase over time in this utility's
cold, chloraminated distribution system.
3-12
-------
TABLE 3-15
TOX PRESERVATION STUDY
Preservatives at
Sample1 Time of Sampling
Filter Effluent2 Sulfite'/acid
" " None
Sulfite
Plant Effluent3 Sulfite/acid
None
Sulfite
Preservatives Added
after 24 hr @ 4°C
None
Sulfite/acid
Acid
None
Sulfite/acid
Acid
# Percent difference = 100 x [(mean TOX value - "control" mean
"control" mean TOX value]
where "control" sample = sample dechlorinated and acidified at
* Fresh sodium sulfite solution used.
1 Samples from Metropolitan's Weymouth filtration plant.
: Chlorinated water
3 Chloraminated water
TOX, //g/L
(Replicates)
170, 160
160, 170
130, 150
150, 150
130. 170
130, 140
TOX value)/
time of sampling.
Percent
Diff/
Control
0
-15%
Control
0
-10%
-------
TABLE 3-16
ON-SITE TOX PRESERVATION STUDIES
Sample Location
Cl, Dose Nli, Residual
(mg/L) Dose Cl,
(mg/L) (mg?L)
TOX,
On-Site*
MWDSC Lab*
Percent
Diff.*
Utility 2
Filler Effluent
Clearwell Effl.
Utility 33
Plant Effluent
Utility 6
8.8
NA
0.5
2.67 total
2.2 total
2.5 free*
310. 320
330, 270
390
380, 350
310, 295
610
-f 16%
+ 1%
+56%
Filter Influent
Filter Effluent
Plant Effluent
Distrib. Loc. ff\
Distrib. Loc. #2
Distrib. Loc. #3
2 0.5 free
0.12 free
1 0.4 0.95 total
0.70 total
0.80 total
0.40 total
60, 61
76, 78
80, 75
NA
NA
NA
NO. 110
110, 110
98. 80
82
76
89
-1-82%
+43%
-f-15%
p Replicate analyses when available,
* Fresh sodium sulfite solution + acid added on-site.
ft Sodium sulfite solution 4- acid added at MWDSC lab.
i Percent difference = 100 x |(mean MWDSC lab TOX - mean on-site TOX)/
mean on-site TOX))
X Very short free CI2 contact time before sample site.
NA = Not analyzed.
-------
Methodology
The TOX experiments to date suggest that for chloraminated waters, dechlorination
after receipts of samples was not problematic. However, the results indicate that
chlorinated waters would continue to produce additional TOX, which appears to be
related to the amount of chlorination that occurred prior to sampling. That is, if
samples were collected shortly after chlorination, TOX values would be off by a high
degree, whereas samples collected after sufficient chlorine contact time had already
occurred in the treatment plant probably would not continue to produce a significant
amount of TOX during refrigerated shipping. Since 14 of the 35 utilities in this study
used chloramines as a final disinfectant, those TOX data are probably representative of
values at the time of sampling. The largest error in interpreting the TOX data will
probably be for disinfection-only utilities; however, only four of these were included in
the study. The on-site TOX experiment at Utility 33, though, provides some data for
interpreting the TOX results of a disinfection-only utility.
Experiments on dechlorination of TOX samples are very difficult, in that specific DBFs
cannot be isolated for study as was performed for the pentane-extractable DBF fraction
(see discussion above). In fact, Croue and Reckhow (1988) demonstrated that sodium
sulfite (the TOX dechlorination agent) readily destroys many DBFs, including some of
particular health concern (including a reduction in the mutagenicity of a sample). For
this study, it was found to be impractical to have the individual utilities perform the
proper dechlorination/preservation at time of sampling (e.g., the sulfite solution is
unstable and has been shown to be difficult to ship to participating utilities). For field
surveys, it appears that the dechlorination and preservation of TOX samples requires
more study.
3-13
-------
Section 4
Data Management and
Analysis
-------
SECTION 4
DATA MANAGEMENT AND ANALYSIS
This chapter discusses the organization of the database used in the study in terms of the
individual database files, the fields contained in each, and their interrelationships; the
data handling and tracking protocol; and the statistical and graphical methods by which
data were analyzed and summarized.
DATABASE
The conceptual design of the database involved two main data files. The first data file,
called UTILITIES, contained information about each participating utility. Each utility
record (one per utility) had fields to store information such as utility identification (ID)
number, name, water source, treatment type, population size, geographic location,
coagulant type, and disinfectant type. The second data file, called RESULTS, stored
all sampling results. Each measured value was stored in a separate record. The fields
of the record included utility ID, location of sampling point, analysis code, sampling
quarter, detection limit flag, and the measured value of the analysis compound or
parameter. In addition to the two main data files, minor files were used to translate the
various abbreviation codes into their full-length descriptions. This complete set of data
files and their interlinking key fields is referred to as a "relational database". The
database was implemented using the commercial software package dBASE HI PLUS™
(Ashton-Tate) on an IBM compatible PC.
DATA HANDLING PROTOCOL
The basic elements of data handling protocol for this study were: 1) use of signatures
by responsible team members, 2) data-sheet reference numbers for each individual
project data sheet, and 3) tracking the status of each group of data. Figure 4-1 shows
the various protocol steps performed at Metropolitan and JMM.
The first component of protocol involved the sign-off procedures used by Metropolitan
in transferring data from laboratory analysts' notebooks to the data-entry sheets sent to
JMM. Data sheets were signed or initialed by the transcriber and subsequently by the
person performing a transcription check. The second protocol step took place at JMM
when data was entered into the database. Every individual piece of paper (data sheet)
was assigned a sequential number referred to as the Reference Number. As data were
entered into the computer data-entry screens, the data-entry clerk typed in the
Reference Number for each screen/sheet (screens and sheets correspond one-to-one) of
data, and all data from that sheet were stored in the database tagged with that
Reference Number. The purpose of the Reference Number is to allow any data point
printed from the database into a report, table or graphics file to be traced back to the
original paper data sheet from which it came, and from there back to the individual lab
analyst's notebook, if necessary. A Data Entry Status Log kept at the computer work
area was continually updated to prevent use of duplicate Reference Numbers. The
data-entry clerk signed off on each data sheet as it was entered.
4-1
-------
MWD Inorganics and Organics Lab Analysts' Notebooks
I
Lab analysts enter data onto data sheets
JMM TOX Lab Reports
Supervisor reviews data sheets for errors
Data sheets forwarded to JMM for entry into database
Engineer transcribes data onto JMM
TOX Worksheets
Data-entry clerk enters data sheets and records data-sheet Reference Numbers
I
Engineer produces first verification printout and notes corrections
Engineer makes corrections in database and checks off corrections on printout
Engineer produces final verification printout and checks against first printout
Engineer files first printout and sends final printout to MWD
DATA-SAMPLING PROTOCOL
FIGURE 4-1
-------
Data Management and Analysis
The next protocol step was to create a data verification computer printout for the range
of newly-entered Reference Numbers. A single dBASE software program was created
to produce the printout for all project data, referred to as the Verification Report. The
output format of the report was virtually identical to the original data-entry sheets,
allowing rapid visual comparison of the printout to the originals. An engineer familiar
with the project performed the data-verification step. The printout was compared to
the originals, and any needed corrections were marked on the printout with a colored
pen. As this verification took place, the engineer signed off on each of the original
data sheets.
Following verification, the engineer went into the computer database files with the
dBASE data-file editor and manually performed the necessary corrections. As each
correction was made, it was checked off on the verification printout using a second pen
of a different color. The final protocol step was the production of a final Verification
Report, which was compared to the original, marked-up verification printout to ensure
that all necessary corrections were made. Completion of each of the above steps was
duly noted on the Data Entry Status Log.
The use of signatures by all responsible parties and the careful checking of data at each
step resulted in an extremely reliable database.
SUMMARY STATISTICS
The data from this study consist of plant influent and filter influent (where applicable)
water quality characteristics and clearwell effluent DBFs from 35 water utilities'
treatment plants. As described in preceding chapters, each plant was sampled once
every quarter for a total of four quarters (although not all analytes were determined
every quarter). For each DBF compound or water quality parameter, it is of interest to
examine and compare the set of values (one per plant) for particular quarters, and the
complete annual set of values (from all plants for all quarters sampled). This large
amount of data created the need for a way to summarize the various data sets.
For conveying the essential features of any set of data, it is desired to express both a
measure of central tendency and a measure of variability. The traditional parametric
approach consists of reporting the sample mean and sample standard deviation.
However, this approach suffers from several drawbacks:
I. Assumed Distribution: In the water quality field, it is frequently assumed that all
water quality data is either normally or log-normally distributed. This assumption
is perhaps based on a limited understanding of statistics, and on studies involving a
few conventional water quality parameters or compounds which were found to be
adequately modeled by normal or log-normal distributions. Thus, a wholesale
tendency has emerged to apply parametric techniques based on those distributions,
frequently without testing the validity of the assumption. It is not always
recognized that the common practice of computing the mean and standard
deviation is rooted in the statistical methodology of estimation, in this context the
estimation of the population parameters of a normal distribution from a limited
sample assumed to be drawn from the population. In association with the
computation of the sample mean, the 95% confidence interval for the mean is also
frequently reported as a measure of the accuracy of the population estimate.
4-2
-------
Data Management and Analysis
Wilh these classical statistical methods (in the early phase of this project), the
sample means of compounds and class totals across the 35 utilities, and their
associated 95% confidence intervals based on the normal distribution, were
computed. Subsequent testing of the individual DBF compounds and chemical
class totals indicated that they followed different distributions, and that neither the
normal nor log normal distribution seemed to apply in some cases. Furthermore.
upon careful reflection it was realized that a major objective of the study was to
explore, describe, and summarize the sample data, not to estimate the parameters
of an underlying distribution for a larger population, such as the population of all
United States water utilities, for example. Even if this were a major objective, it
is possible that Ihe non-randomized sample of 35 utilities with different selected
source waters and treatment practices would be inadequate to do so.
2. Sensitivity to Outliers: A single or several extremely outlying data values can have
a substantial adverse impact on both the sample mean and sample standard
deviation. Especially for exploratory data analysis, it is often advantageous to use
simpler summaries based on sorting and counting which are more resistant: that is.
an arbitrary change in a small part of the data set can have only a small effect on
the summary.
3. Loss of Information: Information about the characteristics of a data set can be
obscured by reducing the data to one or two summary numbers such as the mean
and standard deviation. For example, the minimum and maximum values cannot
be discerned, and questions about how many of the values in the sample are above
or below a specified level cannot be answered.
4. Censored data: Data for compounds which are reported as below detection limits
(in this study. Minimum Reporting Levels (MRLs)) complicate the issue of how to
compute summary statistics such as the mean and standard deviation. Various
approaches for handling such "left-censored" data (below detection limits) are
encountered in practice, including treating these values as equal to zero, equal to
the detection limit, or equal to one-half the detection limit. However, each of
these techniques bears conceptual difficulties and objections (Helsel and Cohn,
1988).
Consideration of the above discussion led to a decision to utilize nonparametric
measures to analyze the data from this study. Nonparametric statistical methods do not
require an assumed parametric distribution for the data, and cases below the detection
limits can be incorporated. In the area of applied statistics called exploratory data
analysis, (he data in a sample set are sorted in ascending order of magnitude, and then
characterized by counting up from the lowest value and determining percentages called
percenliles. In particular, the "five-number summary" (Tukey. 1977) is frequently
used to summarize a data set.
The five-number summary consists of the minimum value, 25th percentile, median.
75ill percentile, and maximum value from the data set. The "nth" percenlile is the
value below which n percent of the data lie. The most familiar percentile is the
median, the 50th percentile, above and below which 50 percent of the data lie.
Because the 25th. 50th, and 75th percentiles divide the data set into four quarters, the
25th and 75th percentiles are frequently referred to as the lower and upper quartiles,
4-3
-------
Data Management and Analysis
respectively. The difference or spread between the lower and upper quartiles is known
as ihe interquartile range, and it encompasses the middle 50 percent of the data values.
Thus, in nonparametric statistics, the median and interquartile range are analogous to
the mean and one standard deviation above and below the mean in parametric statistics.
They also provide much more resistant measures of central tendency and variability.
For example, all the values below the lower quartile including the minimum can change
in magnitude, even drastically, without affecting the values of the median and
interquartile range.
The five-number summary gives five pieces of information about a data set instead of
just one or two. It also allows very useful but simple statements to be made, such as
"75 percent of the chloral hydrate data for the utilities sampled fell below 1.1 /jR/L."
Lastly, medians and other percentiles can still be determined for data sets containing
censored data. For example, for the following data set (n = 9):
<0.l. <0.l. <0.l. <0.1. 0.12, 0.43, 1.1, 2.0, 3.5
the lower quartile is
-------
GUIDE TO NOTCHED BOX-AND-WHISKER PLOTS
+
n
EXTREME OUTLIER
any value outside 3 interquartile ranges
measured from the 25 and 75 percentiles
OUTLIER
any value outside 1.5 interquartile ranges
measured from the 25 and 75 percentiles
MAXIMUM VALUE
this is the largest value (excluding outliers)
75 PERCENTILE
75% of the data are less than or
equal to this value (including outliers)
INTERQUARTILE RANGE
contains the data between
the 25 and 75 percentiles
95% CONFIDENCE INTERVAL
for the median (including outliers)
MEDIAN
50% of the data are above or
below this value (including outliers)
NOTE: Horizontal width of box is
proportional to square root of sample size (N)
-^- 25 PERCENTILE
25% of the data are less than or
equal to this value (including outliers)
MINIMUM VALUE
this is the lowest value (excluding outliers)
FIGURE 4-2
-------
Data Management and Analysis
outlier, and is excluded when determining the minimum and maximum for setting the
endpoints of the whiskers. For a perfectly symmetric normal distribution, the choice of
the 1.5 interquartile range cutoffs correspond to a probability of only 0.7 percent (p =
0.007. 0.0035 in each tail of the distribution) for a data point to lie outside those limits
(Hoaglin, Mosteller, and Tukey, 1983). The outlier criterion is only a guide as to
which data points should be more carefully scrutinized as potential outliers, not an
absolute rule.
Notched box-and-whisker plots portray the 95 percent confidence interval about the
median as a notch-width in the box. This confidence interval can be used to test the
pairwise significance of differences in medians between two data sets, which is a
nonparametric analog of the classical t-test for the significance of differences in means
of normally distributed data sets. An example is shown in Figure 4-3. In comparing
the box-and-whisker plots for three data sets (A, B and C in Figure 4-3), the 95 percent
confidence interval (notch-width) for one data set (A) would be compared to that of the
second data set (B), and then to that of the third data set (C). If the 95 percent
confidence intervals of any two medians do not overlap (medians for A and B, and A
and C). then the difference between those medians is statistically significant at a 95
percent confidence level. If the 95 percent confidence intervals of the medians of any
two data sets overlap (B and C), then the difference between those two medians is not
statistically significant at a 95 percent confidence level. It should be noted that this
type of comparison is pairwise only. In other words, even if box-and-whisker plots
from three different data sets are being compared, the 95 percent confidence intervals
for only two at a time (A and B, A and C, and B and C) can be compared for
determining whether or not there are statistically significant differences between the two
at a 95 percent confidence level. A more detailed discussion of the 95 percent
confidence interval about the median has been prepared by McGill, et al. (1978).
Bar Charts
A derivative form of the box-and-whisker plot was also created and used for this study.
As shown in Figure 4-4. standard bar graphs were made with the height of the bar
equal to the median of the data set. The interquartile range was then superimposed on
the bar. being drawn as an I-shaped "error bar" (more properly, "variability bar") with
the horizontal cross-members of the "I" at the lower and upper quartiles. Thus, this
form of bar chart represents a "three-number summary", omitting the minimum and
maximum values. These graphs were used in cases where multiple compounds or
chemical class totals were portrayed on the same graph, for which differences in
medians are not relevant. Rather, they provide a relative indication of location (central
tendency) and spread (variability). For those DBFs that had relatively high MRL values
and the 25th percentile. median and 75th percentile were less than the MRL. the bar
chart might be interpreted as giving a misleading representation of the concentration.
In such circumstances, the bar was drawn solid black rather than cross-hatched. For
purposes of comparison, the MRL is indicated on the bar chart in some figures where
the median was greater than the MRL, as shown in Figure 4-4.
Star Symbol Plots
Another type of graphical presentation that was used for this study was the star symbol
plot (STSC Inc.. 1987). The star symbol plot is not part of the standard repertoire of
4-5
-------
Example of Notched Box-and-Whisker Plot
for Three Data Sets
B
FIGURE 4-3
-------
GUIDE TO BAR CHARTS USING MEDIAN,
PERCENTILES AND MINIMUM REPORTING LEVEL
INTERQUARTILE RANGE
contains the data between
the 25 and 75 percentiles
75 PERCENTILE
75% of the data are less than
or equal to this value
MEDIAN
50% of the data are above
or below this value
25 PERCENTILE
25% of the data are less than
or equal to this value
MINIMUM REPORTING LEVEL
lower limit of quantitation
NOTE - ASTERISKS ON CHARTS INDICTE THE FOLLOWING:
* 25 percentile < Minimum Reporting Level
** 25 percentile and median < Minimum Reporting Level
*** 25 percentile, median, and 75 percentile < Minimum Reporting Level
FIGURE 4-4
-------
Data Management and Analysis
exploratory data analysis and nonparametric statistics, as are box-and-whisker plots. It
is. however, a unique way of visually portraying differences in multivariate (multiple
variable) "profiles" between entities.
A star symbol plot is used to graphically display the relative magnitude of the values of
multiple variables for two or more entities. As an example, consider the four
compounds which make up the regulated quantity, total trihalomethanes as the four
variables to be portrayed. This example is illustrated in Figure 4-5. On a star symbol
plot, the magnitude of each variable is plotted as a "ray" of a "star" emanating from a
common center point, starting with the first variable at a "0 degrees" position, and with
the remaining variables being spaced evenly around the full 360 degrees of a circle as
in trigonometry. For the four trihalomethanes, a four-ray star results, with each ray of
the star at 0. 90. 180, and 270 degrees (right angles), respectively. For the general
case of "n" variables, n rays of the star will be plotted, each separated by an angle of
360/n degrees.
Chapter 5 of this report presents the actual application of star symbol plots to this
study, at which point this concept will be made more discernible.
4-6
-------
EXAMPLE GUIDE TO STAR PLOTS
CHBr3
CHC13
CHBrC12
CHBr2Cl
FIGURE 4-5
-------
Section 5
Baseline Sampling Results
and Discussion
-------
SECTION 5
BASELINE SAMPLING RESULTS AND DISCUSSION
In this section, results of the quarterly baseline sampling are presented and discussed.
Methods of data analysis and presentation were described in Section 4. Table 5-1 lists
the abbreviations used in the figures in this section.
OVERVIEW OF BASELINE DATA
Figures 5-1 through 5-10 present an overview of the results of the baseline sampling
program for all four quarters combined. Table 5-2 shows the baseline median DBF
values for each quarter as well as for all four sampling quarters combined. It should be
noted that a class total median value is not the sum of the medians of individual
compounds, but rather the median value of the sum of all the compounds within that
class. To illustrate, the median for a compound (chloroform, for example) during a
sampling quarter was the median of the 35 measured values of chloroform for tnat
quarter. For each utility, the sum of the individual compounds within each class
(THMs. for example) was computed, and then the median of these 35 sums was
determined. Finally, the sum of the individual halogenated DBF compounds measured
in this study (XDBPsum) was computed, and then the median of these 35 sums was
determined.
In examining these data, it should be noted that they represent clearwell effluent
samples. Distribution system sampling, which was performed for some of the process
modification studies described in Section 6 of this report, indicated that some DBFs
(such as THMs) increased in the chlorinated distribution systems of some utilities, while
the same compounds remained stable in chloraminated distribution systems. In
addition, it is important to note that the disinfection practices of some of the 35
participating utilities, such as the use of chloramines, were employed to meet the
current TTHM MCL of 0.10 mg/L, and not to meet the requirements of the proposed
Surface Water Treatment Rule (SWTR). Thus, DBF levels at some utilities would most
likely be different if their current disinfection practices required modification in order to
meet proposed concentration-time (CT) requirements of the SWTR.
Figure 5-1 is a summary of the four-quarter median concentrations of each DBF class
(THMs, HANs, HKs. HAAs and ALDs) and the miscellaneous compounds. The
median value of THMs was 36 //g/L, and the median value of HAAs was 17 j/g/L. On
a weight basis, ALDs were the next most prevalent compound, with a total median
concentration of 5.7 //g/L.
Figure 5-2 shows the contribution of each of the median class totals and miscellaneous
compounds to the total concentration of DBFs measured in this study. On a weight
basis. THMs were the largest class of DBFs detected in this study, comprising 54.5
percent of the total measured DBFs. The second largest fraction was HAAs,
comprising 25.4 percent of the total. The data indicate that the median level of THMs
was approximately twice that of HAAs. Figure 5-2 also shows that the third largest
fraction detected was the ALDs (formaldehyde and acetaldehyde), which comprised 8.5
percent of the measured DBFs. These two low molecular weight aldehydes were
5-1
-------
TABLE 5-1
LIST OF ABBREVIATIONS USED IN DBF STUDY
ABBREVIATION1
DEFINITION
Disinfection By-Products
DBF
XDBPMItn
Haloacetic acids
HAA
MCAA
DCAA
TCAA
MBAA
DBAA
Haloketones
HK
l.t-DCP
U.I-TCP
Haloacetonitriles
HAN
DCAN
TCAN
BCAN
DBAN
Aldehydes
ALD
FRM
ACETAL
Disinfection by-product
Sum of measured halogenated disinfection by-products
Haloacetic Acid
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Haloketone
1,1 -Dichloropropanone
1,1,1-Trichloropropanone
Haloacetonitrile
Dichloroacetonitrile
Trichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
Aldehydes
Formaldeyde
Acetaldehyde
-------
TABLE 5-1
List of Abbreviations, Continued
ABBREVIATION1
DEFINITION
Chlorophenols
CP
DCP
TCP
PCP
Chlorophenol
2,4-DichIorophenol
2,4,6-Trichlorophenol
Pentachlorophenol
Other Disinfection By-Products
CH
CHP
CNC1
Others
PI
FI
FE
CE
MRL
NA
THM
TOC
TOX
UV-254
Chloral Hydrate
Chloropicrin
Cyanogen Chloride
Plant Influent
Filter Influent
Filter Effluent
Clearwell Effluent
Minimum Reporting Level
Not analyzed
Trihalomethane
Total Organic Carbon
Total Organic Halogen
Ultraviolet light absorbance at 254 nanometers
'Disinfection by-products not listed are represented by their chemical formulas.
-------
TABLE 5-2
DISINFECTION BY-PRODUCTS IN DRINKING WATER
SUMMARY OF BASELINE SAMPLING MEDIAN VALUES
1st 2nd 3rd 4th All
Disinfection By-Products Quarter Quarter Quarter Quarter Quarters
(//g/L) (Spring) (Summer) (Fall) (Winter) Combined
Trihalomethanes
Chloroform
Bromodichloromethane
Dihromochloromethanc
Bromoform
Total Trihalomethanes 34 44 40 30 36
Haloacetonitriles
Trichloroacetonitrile < 0.012 <0.012
Dichloroacetonitrilc 1.2 1.1
Bromochloroacetonitrile 0.50 0.58
Dibromoacetonitrile 0.54 0.48
Total Haloacetonitriles 2.8 2.5 3.5 4.0 3.3
Haloketones
I.l-Dichloropropanone 0.52 0.46 0.52 0.55 0.52
1,1,1-Trichloropropanone 0.80 0.35 Q.60 0.66 Q.60
Total Haloketones 1.4 0.94 1.0 1.8 1.2
Haloacetic acids
Monochloroacetic acid < 1.0
Dichloroacetic acid 7.3
Trichloroacetic acid 5.8
Monobromoacetic acid <0.5
Dibromoacetic acid 0.9
Total Haloacetic acids 18 20 21 13 17
-------
Table 5-2
Disinfection By-Products In Drinking Water
Summary of Median Values, Continued
Aldehydes
Formaldehyde
Acetaldehyde
1st
Quarter
(Spring)
NA
NA
2nd
Quarter
(Summer)
5.1
2.7
3rd
Quarter
(Fall)
3.5
2.6
4th
Quarter
(Winter)
2.0
1.8
All
Quarters
Combined
3.6
2.2
Total Aldehydes
Miscellaneous
NA
6.9
5.5
4.2
5.7
Chloropicrin
Chloral hydrate
Cyanogen chloride
2,4,6-Trichlorophenol
Halogenated DBPsum
Total Organic Halide
Plant Influent Characteristics
Total Organic Carbon, mg/L
Ultraviolet absorbance, cm"1
Chloride, mg/L
Bromide, mg/L
0.16
1.8
0.45
<0.3
64
150
NA
NA
NA
NA
0.12
3.0
0.60
<0.4
82
180
2.9
0.11
28
0.07
0.10
2.2
0.65
<0.4
72
170
2.9
0.11
32
0.10
0.10
1.7
0.80
<0.4
58
175
3.2
0.13
23
0.07
0.12
2.1
0.60
<0.4
70
170
3.0
0.11
29
0.08
NA = Not Analyzed
Note (I): Total class median values are not the sum of the medians of the individual compounds, but
rather the medians of the sums of the compounds within that class.
Note (2): The halogenated DBPsum median values are not the sum of the class medians for all
utilities, but rather the medians of the halogenated DBPsum values for all utilities. This
value is only the sum of halogenated DBFs measured in this study.
-------
I«
5
»
[TJ
B3
Disinfection By-Product Concentration
by DBP Class Four Quarters
7O
60
50
4O
30
20
10
o
THM TnhalomBthnnes
HAN Hotoocetonitriies
HK Hatohetonw
-
HAA Holoocetlc Acids
ALO AldoMyOa*
CHP Ctitoroptcrtn
CH Chloral Hydrate
CNCl Cyanogen CNortd»
TCP TrlcNoroofwool
>^/r,
w
Y/yfiffl/fc
mm.
7W. ,«-«. «— — ^
" '
PP
^^^^W
^Ui
THM HAN HK HAA ALD CHP CH CNCl TCP
DBP Class
FIGURE 3-1
Percent of Sum of DBP Class Medians
By DBP Class Four Quarters
HAA 25.4%
ALD 8.5%
THM 54.5%
MISC 5,O%
HAN 4.9%
HK 1.8%
THVI TrifiolometnanM
HAN Haloacatontril**
HK Haloketonea
HAA Haloacetic Acid*
ALD Aldafiydas
MISC:
CHP CrKoropic/in
CH CMoral HyOrata
CNCl Cyanogan Cttorida
TCP
FIGURE 5-2
-------
Baseline Sampling Results and Discussion
initially discovered as by-products of ozonation (Glaze et al., 1989a; Yamada and
Somiya, 1989), and they also appear to be by-products of chlorination.
A breakdown of the class medians of the non-THM DBFs measured in this study is
shown in Figure 5-3 on a weight basis. HAAs are by far the largest contributor to non-
THM DBFs, comprising 56.8 percent of the total measured compounds. The two next
largest fractions are ALDs and HANs, comprising 19.0 and 11.0 percent of the total,
respectively. Chloral hydrate was the next largest fraction, comprising 6.9 percent of
the total measured non-THM compounds on a weight basis.
Figure 5-4 shows the four-quarter median concentrations of the four THM compounds.
Additionally, Figure 5-5 shows the same data as a percentage of the total level of
THMs. These figures indicate that chloroform was detected at the highest levels for the
THM compounds, comprising 56.2 percent of the total. The four quarter median level
of chloroform was 14 />g/L. Bromodichloromethane and dibromochloromethane had
median levels of 6.6 and 3.6 /vg/L, respectively. The median level of bromoform was
0.57 fjg/L. representing 2.3 percent of the total THMs.
Figure 5-6 presents the median HAA concentrations by compound for the four sampling
quarters. Trichloroacetic acid (TCAA) and dichloroacetic acid (DCAA) were the two
HAAs detected in the highest concentrations, with median levels of 5.5 and 6.4 /i/g/L,
respectively. These results are consistent with those of other researchers, which have
shown that the aqueous chlorination of humic and fulvic acids yielded TCAA and
DCAA as the major chlorinated by-products (Quimby et al., 1980; Christman et al.,
1983; Miller and Uden, 1983; DeLeer et al., 1985). Dibromoacetic acid (DBA A) was
detected at a four-quarter median concentration of I.I jjg/L. The median
concentrations of both monochloroacetic acid (MCAA) and monobromoacetic acid
(MBAA) were below their minimum reporting levels of 1.0 and 0.5 >ug/L, respectively.
Four-quarter median concentrations of the HKs are presented in Figure 5-7. Both of
these compounds had median levels less than 1 /wg/L.
Median HAN concentrations for the baseline sampling are shown in Figure 5-8.
Dichloroacetonitrile was detected at the highest concentration among the HANs, with a
median level of 1.2 //g/L. Bromochloroacetonitrile (BCAN) and dibromoacetonitrile
(DBAN) had median levels of 0.57 and 0.50 /ug/L, respectively. The median
concentration of trichloroacetonitrile (TCAN) was below its minimum reporting level of
0.029
Figure 5-9 presents the median levels of cyanogen chloride, chloral hydrate,
chloropicrin and 2,4,6-trichlorophenol (TCP). Chloral hydrate had the highest
concentration of these miscellaneous compounds (2.1 /yg/L), followed by cyanogen
chloride (0.60 //g/L). The median concentration of chloropicrin was 0.12 yug/L.
During the first sampling quarter, TCP was detected at low levels at four of the 22
utilities sampled for this compound, yet it was not detected in any samples collected in
subsequent sampling quarters. Pentachlorophenol analyses were conducted for 22
utilities during the first sampling quarter, but was not detected in any of the samples.
Analyses were also conducted for 2,4-dichlorophenol at 12 utilities during the first
sampling quarter. This compound was only detected at one utility, at a concentration
equal to the detection limit (2 //g/L).
5-2
-------
Percent of Sum of Non-THM DBP
Class Medians by Class Four Quarters
ALD 19.0%
HAN 1 1.O%
HAA 56.8%
CH 6.9%
HK 3.9%
CNCI 2.O%
CHP O.4%
THM
HAN
HK
HAA H»IO*C»tlC Acids
ALD AI
-------
Percent of Gum of THM Compound Medians
by Compound Four Quarters
CHBrCl2 26.9%
CHBr2Cl 14.6%
CHBr3 2.3%
CHCI3
RGUR65-5
Haloacetic Acid Concentrations
by HAA Compound Four Quarters
01
0)
0
I i
•1
I
a
n
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
MCA A
DCAA
TCAA
MBAA
DBAA
25th parcentile value below
25th parcantil* and nr»Oan
values below
MCAA DCAA TCAA
HAA Compound
FIGURE 5-6
DBAA
-------
1
§
*-*
to
I
n
•6
O
Haloketone Concentrations
by HK Compound - Four Quarters
1.1 -DCP 1.1 -Dichloropropanone
1.1.1 -TCP 1,1.1 -Trichloropropanone
1.1-DCP
1.1.1-TCP
HK Compound
FIGURE 5-7
Haloacetonitnle Concentrations
by HAN Compound - Four Quarters
s
*-
05
o
<
(0
O
•*•*••*•
FCAN
TCAN Tnchloroacetonitrile
DCXXN Dichloroacetonitnle
BCAN Bromochlofoacetonitrile
DBAN Oibromoacetonitnle
*** 25th pefcentile. median, and 75th
percentile values below
DCAN
BCAN
HAN Compound
FIGURE 5-8
DBAN
-------
-
5
o
g
U
a
c
ro
Miscellaneous DBP Concentrations
by Compound - Four Quarters
CNCI Cyanogen Chloride
CH Chloral Hydrate
CHP Chloropicrin
TCP Tnchlorophenol
*** 25th parcentue, median, and 75th
percentile values below
1 -
0
CNCI
CHP
MlSC Compound
FIGURE 5-9
TCP
0)
U
Q
c
5
TD
Aldehyde Concentrations
by Compound - Four Quarters
FRM =; Formaldehyde
ACETAL = Acetaldehyde
•x- 25th percentile value
below MRL
Formaldehyde
Acetaldehyde
ALD Compound
FIGURE 5-10
-------
Baseline Sampling Results and Discussion
Median concentrations of the ALDs are shown in Figure 5-10. The four-quarter
median level of formaldehyde was 3.6 /t/g/L and the median level of acetaldehyde was
2.2/;g/L.
STAR PLOT ANALYSES
One of the stated objectives of this study was to "determine the seasonal nature of
DBFs as a function of temperature, total organic carbon, pH, and other water quality
parameters; and to determine the effects of changes in treatment processes and/or
disinfectants on the production of DBFs". Under ideal conditions, these objectives
could be approached and properly treated as a problem in formal statistical
experimental design. The ideal conditions would include a very large population of
United States utilities staffed and equipped for DBF sampling, with the investigator free
to perform any adjustments necessary at each plant to implement strict experimental
controls in a classical statistical level-response context. Furthermore, approximately
equal numbers of utilities would be using the same raw water source types, influent
water quality levels, and oxidation/disinfection schemes. However, ideal conditions are
not possible in field studies such as this one. Rather, the group of 35 utilities
represented an unbalanced mixture of differing source waters, raw water qualities,
treatments and disinfection schemes. The following paragraphs illustrate these points
with the aid of star symbol plots.
Figure 5-11 shows the guide to the star plots used in this section. Using data from the
summer sampling quarter for each plant in the study, the raw water characteristics of
TOC and UV-254 absorbance were plotted at 0 and 180 degrees to form what can be
termed the "organics axis", while raw water bromide and chloride levels were plotted at
90 and 270 degrees to form the "inorganics axis". It should be noted that the
magnitude of each ray of the star plots is normalized to the utility with the highest level
of the parameter under consideration. Utility 10 had the highest chloride and bromide
levels (640 mg/L and 3.00 mg/L, respectively) and Utility 29 had the highest TOC and
UV-254 levels (19.04 mg/L and 0.697 cnv1, respectively). However, because the levels
of these parameters were so high at these two utilities, relative differences between
levels of these parameters for the other utilities tended to be minimized. Thus, these
two utilities were excluded from all star plots and the plots were normalized to the
utilities with the next highest levels of these influent parameters, Utility 24 (115 mg/L
of chloride, 0.54 mg/L of bromide) and Utility 21 (10.57 mg/L of TOC, and UV-254
equal to 0.358 cnr1).
Figure 5-12 shows the star plots for the 33 utilities (excluding Utilities 10 and 29),
labeled by utility number. From the relative sizes and various shapes of the "stars", it
is immediately apparent that the raw water quality characteristics varied considerably
among the participating utilities.
Figure 5-13 presents the same star plots, but with the utilities sorted by source water
type, each source water type on a separate page of the figure. This plot clearly begins
to illustrate the problem of trying to ascribe differences in DBF production to discrete
and unrelated causal factors, in this case source water origin. The star plots for the
groundwater sources are a good illustration of this point. The groundwater "stars" vary
widely in both size and shape; and thus, in the relative influences of bromide and
organic carbon, both of which have a major impact on the level and speciation of DBFs
5-3
-------
GUIDE TO STAR SYMBOL PLOTS
Bromide
UV Absorbance
TOC
Chloride
Each "ray" of the star emanating from the center point
represents the relative magnitude of that parameter
FIGURE 5-11
-------
Plant Influent By Utility ID Number
Summer Quarter
6
9
11
13
FIGURE 5-12
-------
Plant Influent By Utility ID Number
Summer Quarter
14
16
16
17
18
19
29
21
22
23
24
25
FIGURE 5-12 (Continued)
-------
Plant Influent By Utility ID Number
Summer Quarter
26
27
28
31
33
34
36
FIGURE 5-12 (Continued)
-------
Plant Influent by Source Water Type
Lake/Reservoir Sources
Summer Quarter
6
15
16
17
18
26
FIGURE 5-13
-------
Plant Influent by Source Water Type
Lake/Reservoir Sources
Summer Quarter
33
34
36
FIGURE 5-13 (Continued)
-------
Plant Influent by Source Water Type
Flowing Stream Sources
Summer Quarter
12
13
14
19
28
31
FIGURE 5-13 (Continued)
-------
Plant Influent by Source Water Type
Groundwater Sources
Summer Quarter
22
23
24
32
FIGURE 5-13 (Continued)
-------
Baseline Sampling Results and Discussion
produced upon oxidation/disinfection of that water. For example. Utility 24 has a high
level of bromide, while the level of TOC at this utility is relatively low. Utility 21
treated a colored groundwater with a high concentration of TOC. Finally, Utility I's
source water was a groundwater low in both inorganic and organic constituents. The
plots from the flowing stream and lake/reservoir illustrate similar conditions.
The next set of star plots (Figure 5-14) depicts each utility again, but sorted by
treatment type. Some overall trends are apparent in these plots. For instance, the
direct filtration and disinfection only utilities generally have relatively small "stars",
indicating the presence of relatively low levels of organics and inorganics, and hence
the ability to employ less-extensive treatment systems for these source waters. The fact
that the characteristics of the source water dictate to a large degree the type of
treatment required for that water serves to confound an attempt to ascribe differences in
clearwell effluent DBF levels simply to treatment type. It may be misleading to
consider concentrations of TTHMs, for instance, produced by plants utilizing direct
filtration or disinfection only and conclude that THM concentrations at these plants
were caused solely by the treatment practices of the plants.
Figure 5-15 presents utilities labeled by disinfection scheme. In theory, this would
seem to be the most promising classification for identifying causal differences in DBF
production. However, examination of these star plots reveals several confounding
factors. For example, all other parameters being equal, chlorine-only utilities would be
expected to have higher TTHMs than the chloraminating utilities. However, the four-
quarter median TTHM value was actually higher for the
prechlorinating/postammoniating utilities (approximately 57 //g/L) than for the
chlorine-only utilities (approximately 34 ^g/L). (These findings will be discussed in
more detail later in this section.) This may be understood, in part, by considering the
widely varying raw water qualities of the utilities employing these two disinfection
scenarios, as shown in Figure 5-15. While it appears that, in general, the
prechlorinating/postammoniatmg utilities may have a higher percentage of relatively
large "stars" compared to the chlorinating utilities, the differences in these influent
parameters between the two disinfection schemes were not found to be statistically
significant (as will be discussed later in this section). However, it may be the
combination of high organics and bromide levels that required four of the 10
prechlorinating/postammoniating utilities to use postammoniation for THM control.
No overall trend in raw water quality characteristics is observed in the star plots of any
of the disinfection schemes. Thus, characteristics such as reactivity of precursor
material, chlorine dose and contact time, pH, water temperature, and precursor removal
within the treatment processes must also play a role in the formation and speciation of
DBFs. The star plots presented here illustrate some of the difficulties arising from
attempts to find discrete, unrelated causal factors for DBF production.
SEASONAL VARIATION
Figures 5-16 through 5-34 show the variation in the water quality parameters and DBFs
measured in this study as a function of sampling season. The data are presented as
notched box-and-whisker plots, showing the median, interquartile range, minimum and
maximum values, outliers, and the 95 percent confidence intervals of the medians. The
number of samples (n) is also shown on each plot. A more detailed discussion of this
type of data presentation was included in Section 4 of this report.
5-4
-------
Plant Influent By Treatment Type
Conventional Utilities, Summer Quarter
11
12
13
FIGURE 5-14
-------
Plant Influent By Treatment Type
Conventional Utilities, Summer Quarter
14
FIGURE 5-14 (Continued)
-------
Plant Influent By Treatment Type
Softening Utilities, Summer Quarter
21
22
23
24
26
27
28
31
FIGURE 5-14 (Continued)
-------
Plant Influent By Treatment Type
Direct Filtration Utilities, Summer Quarter
16
17
18
19
FIGURE 5-14 (Continued)
-------
Plant Influent By Treatment Type
Disinfection-Only Utilities, Summer Quarter
32
33
34
35
FIGURE 5-14 (Continued)
-------
Plant Influent By Disinfection Scheme
Chlorine Only, Summer Quarter
6
11
13
15
16
17
22
23
FIGURE 5-15
-------
Plant Influent By Disinfection Scheme
Chlorine Only, Summer Quarter
26
31
33
34
36
FIGURE 5-15 (Continued)
-------
Plant Influent By Disinfection Scheme
Chlorine /Ammonia, Summer Quarter
12
14
27
28
30
FIGURE 5-15 (Continued)
-------
Plant Influent By Disinfection Scheme
Summer Quarter
Chloramines Only
18
Ozone/Chlorine
19
32
Ozone/Chloramines
FIGURE 5-15 (Continued)
-------
Baseline Sampling Results and Discussion
Influent Water Quality
The seasonal effects on influent water quality parameters measured in this study (TOC,
UV-254. chloride and bromide) are shown in Figures 5-16 through 5-19. Data from
only the summer, fall and winter quarters are presented since these influent analyses
were not instituted until the second sampling quarter.
As shown in Figure 5-16. median levels of TOC did not change appreciably by season,
ranging from 2,9 to 3.2 mg/L. The three-quarter median influent TOC level was 3.0
mg/L. There was not a significant difference in the medians of any two sets of
quarterly influent TOC data. In other words, the 95 percent confidence interval for the
spring quarter median was compared to that of the summer quarter, then to the fall
quarter. The overlapping 95 percent confidence intervals indicate that the difference in
the median TOC levels of any two quarters is not statistically significant. The "folded
over" box for the fall quarter data shown in Figure 5-16 results from the 95 percent
confidence interval of the median extending lower than the 25th percentile value. The
notch width corresponds to the 95 percent confidence interval of the median, and the
horizontal line across the box corresponds to the 25th percentile value.
Figure 5-17, a plot of UV-254 absorbance as a function of sampling quarter, indicates a
trend similar to the influent TOC levels. Although the winter median (0.13 cm1)
appears slightly higher than the summer or fall medians (0.11 cnr1 for both quarters),
the medians of the summer and fall seasons are within the 95 percent confidence
interval for the median of the winter quarter, indicating no statistical difference in the
medians of any two seasons.
A plot of influent chloride levels by season is shown in Figure 5-18. Again, there is
little seasonal variation indicated in this figure. The highest median influent chloride
level was 32 mg/L, occurring in the fall season. The three-quarter median chloride
level was 29 mg/L. Figure 5-19 shows the seasonal variation in influent bromide levels
on a seasonal basis. Quarterly medians ranged from 0.07 to 0.10 mg/L of bromide.
with a three-quarter median of 0.08 mg/L. Note the presence of one outlying data
point in each quarter that occurs at substantially higher concentrations than the 75th
percentiles and other outliers. These outlying data points represent Utility 10, which
had extremely high influent bromide levels. Bromide and its impact on DBF formation
and speciation will be discussed in detail later in this section.
Classes of DBF Compounds
Figure 5-20 is a plot of the seasonal variation of the sum of halogenated DBFs
(XDBP ) measured in this study. The median of the summer quarter (82 //g/L) is
higher than the median of the winter quarter (58 */g/L); however, this difference is not
statistically significant since the 95 percent confidence intervals about the medians of
both quarters overlap. The median values of XDBPS11I11 for the spring and fall quarters
were 64 and 72 //g/L, respectively.
Figure 5-21 is a plot of TTHMs by sampling quarter. As would be expected based on
seasonal temperature differences, the highest median TTHM level occurred in the
summer season and the next highest in the fall. For many utilities in California and the
South, the fall season can be almost as warm as the summer. The lowest median
5-5
-------
Influent Total Organic Carbon
Bg Quartar
Influent UU Ab>orbancB
By Quartar
am
J >•
a
6
C
0
SI *•
L
U
u
•H
c
a
o
— i
•
0 «
H
-1 1 1 1 1-
-
-
•
-
B
_
.
:
.
-
LJ LJ y^
> — < > — C / N
i i
...
9
u
\
N^ t.B
E
C
U)
01
-u
« «-*
fll
u
u
0
XI *-2
-------
Influent Chloride By Quarter
Influent Bromide By Quarter
7BI
67*
J
a
0 38*
TJ
•H
L
Q
L
0
19*
-i 1 1 ~ 1 n
•
•
-
i i i
4
3
J
a
81 *
•Q
•H
E
0
L
m
.
-r i i
:
-
•
•
-
"
iji i
i i — i 1 L— '
L_l J — i ' ' — •
SUV1ER FBLL WINTER
SUMMER POLL UIHTER 5UmEK
n= 35 35 35
n= 35 35 35
Quarter Quarter
RGURE5-18
FIGURE 5-19
-------
XDBPsum by Quarter
Trihalomethanei
Bg Quarter
_J
n
D
E
N
a.
DO
a
x
_L
_L
_L
-I
a
•
•
r
0
-t
i
•H
L
T
T
_L
SPBIMO SUMMER POLL
35 35 35
Quarter
FIGURE 5-20
UINTER
35
SPRINO
n- 35
35 35
Quarter
FIGURE 5-21
UINTEB
35
-------
Baseline Sampling Results and Discussion
TTHM level occurred in the winter. The impact of water temperature on the formation
of THMs and other DBFs will be discussed in detail later in this section. Figure 5-22
is a plot of median chloroform levels by sampling quarter. There are no significant
differences between the medians of any two quarters and only three outliers appear on
the plot. However, plots of bromodichloromethane, dibromochloromethane and
bromoform (Figures 5-23, 5-24 and 5-25, respectively) indicate progressively greater
numbers of outliers, although there are no statistically significant differences between
the medians of any two quarters for these compounds. The increasing number of
outliers is most likely due to the presence of high levels of brominated THMs in
utilities with high influent bromide concentrations. Note the extremely high levels of
bromoform (over 50 /ug/L) of several of the outliers in Figure 5-25. The impact of
influent bromide levels on the production of brominated DBFs is discussed in detail
later in this section.
Following the same trend as median TTHM levels, median total HAA levels were
higher in summer (20 pg/L) and fall (21 /ug/L) than in winter and spring (13 and 18
/t/g/L. respectively), as shown in Figure 5-26. However, there was no statistical
difference between the medians of any two quarters. Figures 5-27 and 5-28 show
DCAA and DBAA concentrations on a quarterly basis. Neither of these plots indicate a
significant difference between the medians of any two quarters. However, following the
same trend observed in the plots of chloroform and bromoform, the plot of DBAA
shows many more outliers than the plot of DCAA.
In contrast to the trend of increasing median concentrations with increasing seasonal
temperatures seen in the TTHM and HAA plots. Figure 5-29 shows that the median
level of total HANs was highest in winter and lowest in summer, although the
differences were not statistically significant. This same trend is observed in the plot of
quarterly median HK concentrations. Figure 5-30. The role of HANs and HKs as
reactive intermediates rather than stable endproducts of chlorination reactions is
discussed later in this section.
Figure 5-31 illustrates the seasonal variation in aldehyde concentrations. Only three
quarters of data are presented since the ALD analysis was not instituted until the second
sampling quarter. The plot indicates that the level of ALDs in the summer quarter (6.9
/ug/L) was higher than winter median (4.2 yug/L), but the difference was not statistically
significant at a 95 percent confidence level.
The chloropicrin levels shown in Figure 5-32 indicate that there was little seasonal
variation, with all four quarterly medians falling within the range 0.10 to 0.16 //g/L.
Figure 5-33 indicates the same lack of significant variation in quarterly cyanogen
chloride medians. The median concentrations of cyanogen chloride were within the
range 0.45 to 0.80 /yg/L for the four sampling quarters. The plot of chloral hydrate
concentrations shows no significant difference between the medians of any two quarters
(Figure 5-34).
Although Figures 5-20 through 5-34 do not show statistically significant differences
between the median DBF concentrations from season to season, seasonal variation in
DBF levels is demonstrated in Table 5-3. In this table, DBF values for the 35 utilities
are presented for the 25th percentile, median and 75th percentile for the four sampling
quarters. For TTHMs, the 25th percentiles for the summer and winter quarters were
5-6
-------
Chloroform By Quarter
Bramodichloramethani
By Quarter
a
E
a
o
a
u
\
a
^
c
r
a
o
-H
U
•H
TJ
0
0
(D
I
SPftlNO
35
35 35
Quarter
UIMTER
35
SPRINO
n= 35
35 35
Quarter
WINTER
35
FIGURE 5-22
FIGURE 5-23
-------
Dibromochlorom«thani
BU Quarter
Bromoform BU Quarter
••
J
\
n ••
11
•
*J
II
E «•
0
L
0
.-•
•^
0
0
E
0
L a.
a
•H
a
•
-T— 1 1 1 ' ^
-
_
m
,
_
-
" • * *
*
- .
- w 2 2 3
c 1
"
• •
S\
_J
0
^^
1 **
0
Q
E
Q
m
ai
i
n -T 1 1 1 r-
-
•
h— • "
"
•
.
. .
1
.
=1= =±2 L2 = :
"l 1 1 _l L
„_ SPRINQ «unrcn fMJ. UIHTER
SPHINQ ***** MU. UIKTEB
35 , 35
Quarter Quarter
FIGURE 5-24
FIGURE 5-25
-------
Haloacetie Acid*
BU Quarter
D
w
u
•H
u
-------
Dibromoacetic Acid
Bg Quarter
Haloacetonitrilee
Bg Quarter
T
3
V
T)
U
u
•H
4J
II
U
a
o
_L
_L
D
•H
L
+J
•H
0
4J
•
U
0
H
II
z
T
T
SPRING
35
35 35
Quarter
UINTER
35
SPRINO
n= 35
35 35
Quarter
UINTER
35
FIGURE 5-28
FIGURE 5-29
-------
Haloketonea
Bg Quarter
Aldehydes by Quarter
O
0
+J
n
i
_L
j
\
n
in
o
•o
3
r
u
TJ
_L
3PBINQ
n= 35
r«U-
35 35
Quarter
FIGURE 5-30
WINTER
35
SUMMER
n= 33
35
Quarter
FIGURE 5-31
UINTEB
33
-------
Chloropicrin By Quarter
Cyanogen Chloride By Quarter
a..
a
^
_J
\
a
3
^ l.B
C
•H
I.
U
•H
a.
0 »
i.
o
u
*.*
*
II iii
-
_
-
"
•
•
~
: zj M H M
i
"
13
^
J
\
a "
3
^
§|
.•0 .
L
0
^^
u .
c
ll
01
0
c «
o
2
*
-
_
-
-
—
_
:
_
•
-
•
.
-
•
; s i B 5
_J 1 1 1 1 L-
SPRING SUMMER PALI- UINTER
SPRING SUMMER P«U- UINTEB
n= 35 35 34 35 n= 20 35 34 35
Quarter Quarter
FIGURE 5-32 FIGURE 5-33
-------
Chloral Hydrate By Quarter
Influent Total Organic Carbon
By Tr«atm»nt
3*
a*
J
a
3 «
^
•
•
t
•D "
I
•H
9
i. 11
0
r
a
.
•
1 1 1 1 I
-
-
•
•~~ —
•
~
-
~
.
: :
™ •
_
; 0 H U i :
iii i
3B
J »•
\
n
c
0
J3 IB
•
0
U
.^
c
« It
a
L
O
•jj
4J
0 6
H
•
I I 1 I 1
•
•
•
.
'
~
.
1
— _
"
•
2' '_
_
*T S ^ :
i 11 i
SPRIHD SUMWR PMJ. WINTER CNU OFL 019 SPT
n- Ii8 15 12 30
n= 23 35 35 35 u- 16 5 A 10
Quarter Treatment
FIGURE 5-34 FIGURE 5-35
-------
TABLE 5-3
COMPARISON OF SEASONAL DBF LEVELS
Quarter
25th
Percentile
Median
75th
Percentile
Spring
Summer
Fall
Winter
Total Trihalomethanes (/ig/L)
22
28
20
15
34
44
40
30
57
76
54
48
Sum of Haloacetic Acids Gt/g/L)
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Summer
Fall
Winter
10
11
9
10
17
20
21
13
Sum of Haloacetonitriles G/g/L)
1.4
1.6
1.3
1.3
2.8
2.5
3.4
4.1
Sum of Aldehydes G/g/L)
4
2
2.3
7
5.5
4
42
35
3t
26
6.0
7.0
7.8
6.0
12
9
7
-------
Baseline Sampling Results and Discussion
28 and 15 ^g/L. respectively, indicating a decrease of almost 50 percent from summer
to winter. The 75th percentile for TTHMs was 76 //g/L in the summer and 48 //g/L in
the winter, indicating a decrease of almost 40 percent from summer to winter.
Although the seasonal differences for HAAs, HANs and ALDs were not as substantial
as those observed for THMs, seasonal differences for these DBFs are apparent in the
table. For instance, the 75th percentile for the sum of HAAs was 42 //g/L in the
spring, and decreased almost 40 percent to 26 /j g/L in the winter. Furthermore, the
medians for HAAs were highest in the summer and fall, but the 75th percentile was
highest in the spring. From the data presented in Table 5-3, it is clear that seasonal
variations influenced levels of DBFs measured over the sampling period.
Table 5-4 is a summary of clearwell effluent TTHMs and influent temperature, TOC
and bromide levels for all 35 utilities, comparing data from the summer and winter
quarters. This table illustrates that seasonal variations in the reported parameters were
different for individual utilities. For instance, Utility 1 had a summer water
temperature of I8°C and a winter temperature of I4°C. Bromide and TOC levels at
Utility I varied only slightly as well, and this lack of variability is reflected in the
TTHM levels (7.9 and 9.0 /ug/L in the summer and winter quarters, respectively). At
Utility 20. there was essentially no seasonal variation in TOC and bromide
concentrations, but water temperature varied from 26 to 8°C from summer to winter,
and TTHM levels in the summer were significantly higher (76 and 24 //g/L in the
summer and winter quarters, respectively). At Utility 23, the water temperature was
similar both quarters (10°C in the summer and 8°C in the winter); however, TOC and
bromide levels were higher in the winter (4.49 mg/L and 0.58 mg/L, respectively) than
in the summer (3.27 mg/L and 0.44 mg/L, respectively). These differences are
reflected in the higher TTHMs in the winter versus the summer quarter (24 versus 3.8
/yg/L. respectively). From the data presented in this table, it is apparent that seasonal
variations in influent water quality and effluent DBFs do not exist uniformly for utilities
around the nation, but rather for individual utilities.
VARIATION BY TREATMENT TYPE
Figure 5-35 shows the three-quarter influent TOC levels as a function of treatment type
(conventional, direct filtration, disinfection only, and softening). The plot includes the
number of data points in each category (n) as well as the number of utilities within each
category (u). There is no statistically significant difference between the median values
for conventional, direct filtration and disinfection only utilities. However, the softening
utilities participating in this study had a significantly higher median influent TOC than
either conventional or direct filtration utilities. This is perhaps due to the inclusion of
several utilities treating highly-colored ground and surface waters by softening.
Figure 5-35 is the only plot of water quality parameters as a function of treatment type.
As discussed above regarding the star plot analyses, levels of DBFs measured in this
limited study of 35 utilities cannot be ascribed simply to discrete causal factors such as
treatment type and source water type. Thus, plots of DBF levels as a function of
treatment type are relatively meaningless and could be misleading if taken out of
context. However, some special issues relating to DBF levels in disinfection-only
utilities are discussed later in this section.
5-7
-------
TABLE 5-4
COMPARISON OF SEASONAL TTHMs AND INFLUENT WATER QUALITY
Summer Quarter
Winter Quarter
Utility
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Effluent
TTHMs
0/g/L)
7.9
90
6.1
9.1
42
98
63
27
95
43
67
71
24
114
27
28
94
7.6
15
76
56
60
3.8
40
34
164
57
36
108
73
77
3.1
44
46
38
Influent Values
Temp.
18
30
28
23
30
24
21
20
31
28
27
23
19
22
19
15
30
17
20
26
25
6.6
10
22
NA
27
30
28
29
27
28
26
18
15
18
TOC
(mg/L)
0.75
2.4
3.2
2.8
5.4
4.5
2.8
1.6
3.6
4.6
2.1
2.6
1.1
2.6
1.3
2.7
3.0
1.8
1.8
7.4
11
5.5
3.3
0.60
7.1
5.2
2.9
2.4
19
3.6
3.0
2.6
3.2
2.6
3.1
Br
(mg/1)
0.07
0.12
0.02
0.27
0.12
0.06
0.15
0.02
0.06
3.0
0.07
0.41
1.0
3.5
0.01
0.05
0.01
0.02
0.10
0.04
0.18
0.02
0.44
0.54
0.22
0.19
0.33
0.09
0.16
0.10
0.05
0.07
0.01
0.01
0.01
Effluent
TTHMs
0/g/L)
9.0
60
6.1
12
32
33
30
12
40
47
28
85
26
76
22
36
29
0.7
5.9
24
48
38
24
43
9.1
98
36
16
259
66
57
0.8
25
53
20
Influent Values
Temp.
14
27
4.5
9.6
15
4.0
13
9.5
5.0
6.0
8.4
9.0
11
11
13
12
9.5
4.5
5.0
8.1
23
10
8.1
20
18
3.0
10
4.5
23
2.0
1.1
24
3.1
6.2
2.8
TOC
(mg/L)
0.96
2.5
3.2
2.7
5.5
4.9
2.7
1.7
4.5
5.3
3.0
2.8
1.5
3.9
1.4
2.7
3.2
1.9
1.5
6.9
11
4.0
4.5
0.74
6.9
3.7
3.8
2.6
17
20
3.2
2.5
3.5
3.3
3.5
Br
(mg/L)
0.08
0.12
0.02
0.47
0.13
0.07
0.07
0.02
0.10
2.8
0.06
0.79
0.01
0.32
0.01
0.07
0.01
0.03
0.04
0.03
0.17
0.02
0.58
0.44
0.14
0.68
0.35
0.04
0.21
0.11
0.05
0.06
0.01
0.03
0.01
NA = Not analyzed
-------
Baseline Sampling Results and Discussion
VARIATION BY SOURCE WATER TYPE
Figures 5-36 through 5-39 show the influent water quality parameters measured in this
study as a function of source water type (flowing stream, groundwater, and
lake/reservoir). The 105 TOC measurements performed for this study had a minimum
of 0.6 mg/L and a maximum of 19.9 mg/L. As seen in Figure 5-36, there is very little
difference between the medians of any two source water types. The highest TOC levels
occurring in this study were measured at Utility 29, treating a highly-colored surface
water, and one anomalous reading at Utility 30 in the winter quarter. The other two
influent TOC measurements at Utility 30 were less than 5 mg/L. The 25th percentile
TOC level for the groundwater sources was the lowest; however, the presence of
colored groundwaters resulted in a median TOC value statistically comparable to that of
the other two source water types.
Influent UV-254 absorbance by source water type is illustrated in Figure 5-37.
Although the median UV-254 absorbance level is higher for the groundwater utilities
than the flowing stream or lake/reservoir utilities, the difference is not statistically
significant. The outlier points shown for the flowing stream utilities are the three
quarterly readings from Utility 29, reflecting the high organic content of this utility's
source water.
The influent chloride levels plotted by source water type in Figure 5-38 show very little
variability in chloride concentrations from source to source. The influent bromide data
plotted in Figure 5-39 indicate the same lack of variation between the source water
types. The highest outlier points on both plots represent Utility 10, treating a
lake/reservoir source with extremely high chloride and bromide levels. No plots of
DBF concentrations by source water type are included for the reasons discussed
previously with respect to the star plot analyses and the variation of DBF data by
treatment type.
VARIATION BY DISINFECTION SCHEME
Figures 5-40 through 5-54 illustrate influent parameters and DBF levels measured in
this study as a function of disinfection scheme (chlorine only, "C12"; prechlorination
and postammoniation. "CI2NH3"; and chloramination, "NH2C1"). Results for the
preozonation/postchlorination and preozonation/postchloramination utilities are not
presented here because the very small number of data points precluded meaningful
observations about those data. However, aldehyde data from the ozonating utilities will
be discussed later in Section 5, and further discussion will be included with the results
of the ozonation treatment studies in Section 6.
Influent Parameters
Figure 5-40 shows the overall influent TOC levels as a function of disinfection scheme.
The medians show very little variation among utilities using the various disinfection
practices. Although Figure 5-41. a plot of influent UV-254 absorbance as a function of
disinfection scheme, does not show statistically significant differences between the three
categories, this figure does indicate that prechlorinating/postammoniating utilities had
higher influent UV-254 levels than the chlorinating or chloraminating utilities. One
5-8
-------
Influ«nt Total Organic Carbon
By SourcB
Influ«nt UU-264 Abmorb.nci
By Source
SN ••
_l
a
E
0
JJ
o
u
•H
c
a
L
O
+J
0
u
E
ID
01
II
U
a
o
a
-------
Influent Chloride by Saurci
Influent Bromide by Source
a
•a
•H
L
a
•H
o
\
a
E
w
•H
E
O
DO
/IN
PS
33
11
21
7
Sourci
LR
51
17
33
ll
ou
21
7
Source
LR
51
17
FIGURE 5-38
FIGURE 5-39
-------
Influent Total Organic Carbon
By Disinfection Schema
Influent UU Abaorbanci
By Disinfection Schemi
M
s-\
—1 >•
\
a
E
W
C
0
.Q IB
i.
•
0
U
•H
C
« '•
a
L.
O
i
•H
•
.u
0 B
^~
•
1 III
-
-
•
• —
"
.
.
- ' _
'
_
*
•
i
— _
~
-
i ' 1 I 1 1 1
^_j rj n
/ — \ \ / v/
r i ^^ m
1
i i i i i
•.•
• -T
^%
E
U
\
v/ •.•
E
C
Tj I.B
ID
(VI
4J
• *.4
•
U
c
4 *.3
o
L
0
D i.a
-------
Baseline Sampling Results and Discussion
possible explanation for these results is that utilities with high UV-254 absorhance most
likely have high color levels and require free chlorine contact time for color removal.
However, the high UV-254 levels also indicate the presence of dissolved organic matter
which could function as THM precursors. Thus, subsequent ammonia addition is
required in order to control THMs.
Figures 5-42 and 5-43 illustrate overall influent chloride levels plotted with and without
the inclusion of Utility 10, respectively. Because of the relatively small number of data
points in the chloraminating category, the inclusion of Utility 10's extremely high
influent chloride levels produces an oddly-shaped box-and-whisker plot. Whether or
not Utility 10 is included, however, the overlapping 95 percent confidence intervals
indicate there is no statistical difference between the medians of any two quarters. In
Figures 5-44 and 5-45. influent bromide levels are plotted by disinfection scheme both
with and without the inclusion of Utility 10. In both plots,
prechlorinating/postammoniating utilities have a higher median influent bromide level
than chlorinating utilities, and the difference is statistically significant at a 95 percent
confidence level.
Classes of DBF Compounds
Figure 5-46 illustrates levels of XDBP1lim by disinfection scheme. The plot indicates
that the prechlorinating/postammoniating utilities have a higher median value of
XDBPslim (approximately 94 //g/L) than either the chlorinating or chloraminating
utilities (approximately 62 and 23 //g/L, respectively), and that the median XDBPS(im
level of the chlorinating utilities is higher than that of the chloraminating utilities.
These differences are statistically significant at a 95 percent confidence level.
Figure 5-47, a plot of overall median TTHM levels by disinfection scheme, reflects the
same trend observed in Figure 5-46. The median TTHM level for the
prechlorinating/postammoniating utilities (approximately 57 //g/L) was significantly
higher than that of the chlorinating and chloraminating utilities (approximately 34 and
12 //g/L. respectively), and the median TTHM level of the chlorinating utilities was
substantially higher than that of the chloraminating utilities, although the difference was
not statistically significant.
The fact that the prechlorinating/postammoniating utilities produced higher levels of
XDBPSI1I)) than the chlorinating utilities is contrary to an intuitive expectation. While
the plots of influent water quality parameters did not indicate significant differences in
TOC or UV-254 levels between the three disinfection schemes (Figures 5-40 and 5-41),
the median UV-254 level was higher for the prechlprinating/postammoniating utilities
than for the other two disinfection schemes. The higher levels of UV-254 absorbance
for the prechlorinating/postammoniating utilities may indicate higher levels of DBP
precursors (i.e.. higher humic substances content of the dissolved organic matter) than
in the influents of utilities using the other disinfection schemes. Thus, the higher
median level of XDBPslim may reflect the higher concentrations of DBP precursors in
the prechlorinating/postammoniating utilities.
Another factor contributing to the unexpectedly high levels of DBPs occurring at the
prechlorinating/postammoniating utilities may be the higher bromide levels measured at
these utilities. The data plotted in Figure 5-45 indicated that the
5-9
-------
D
M
•n
•H
r
o
Influent Chloride
By Disinfection Scheme
Influent Chloride
Bg Disinfection Scheme (w/o
T
_
\
a
L
0
-I
o
c
II
1
H
T
0-3
NH2CI.
CL2
CL2NH3
NMSCL
58 29 9
20 1Q .3
Disinfection Scheme
FIGURE 5-42
. 58 29 6
.20 10 2
Disinfection Scheme
FIGURE 5-43
-------
Influent Bromide
By Disinfection Schemi
Influent Bromide
By Disinfection Scheme
-------
XDBPsum By Disinfection Schemi
Trihalomethenes
By Disinfection Schemi
X
Q 3»»
0.
s -
•x.
-I
a
c
r
-P
o
T
HH2CL.
11
78 liO
21 11
-------
Baseline Sampling Results and Discussion
prechlorinating/postammpniation utilities had a higher median influent bromide level
than the chlorinating utilities, and this difference was statistically significant at a 95
percent confidence level. Research by Aizawa, et al. (1989) found "...the
concentration of total THM increased with the augmentation of bromide ions with the
same amount of chlorine dosage. The increase in THMs is up to two times higher than
in the absence of bromide ions." As discussed later in this section, utilities treating
waters high in bromide had relatively high concentrations of bromoform and other
brominated DBFs. Since bromoform is the heaviest of the THM compounds, its
presence can result in elevated TTHM levels.
In interpreting the DBF data presented as a function of disinfection scheme, it should
be noted that baseline data samples for this study were collected at the clearwell
effluents of the participating utilities. Figure 2-2 illustrated the process trains of the
plants at which samples were collected. Of the 11 prechlorinating/postammoniating
utilities participating in this study, eight did not add ammonia until the filter effluent
and filtration was the process immediately upstream of the clearwell. Thus, these
utilities maintained a free chlorine residual throughout most of the in-plant detention
time, and levels of DBFs measured at these utilities reflect the free chlorine contact
time. The differences in DBF concentrations between the
prechlorinating/postammoniating utilities and the chlorinating utilities would be more
apparent had the sampling taken place within the distribution systems of the
participating utilities.
As discussed previously in the section entitled "Star Plot Analyses", all other factors
being equal, it would be expected that the chlorinating utilities would have the highest
median TTHM level compared to utilities employing the other two disinfection
schemes. Figures 5-46 and 5-47 highlight the danger of oversimplifying the influences
of causal factors on DBF production. Taken out of context, these two figures could
give the impression that based on median levels, prechlorinating/postammoniating
utilities could not meet a revised THM standard if it were lowered to 50 //g/L, while
utilities utilizing only free chlorine would be able to meet this lower standard. Such a
perception would not consider other potential causal factors found in this and other
DBF studies, such as influent TOC levels, influent UV-254 absorbance, influent
bromide concentration, free chlorine contact time, pH, and temperature, among others.
A correct interpretation of the results presented in Figures 5-46 and 5-47 is that, of the
35 utilities participating in this study, the utilities practicing
prechlorination/postammoniation had higher median levels of halogenated DBFs and
TTHMs than the utilities employing chlorination or chloramination. The UV-254 data
presented previously in Figure 5-41 may suggest that these results are due to the
presence of THM and DBF precursors in higher levels in the source waters of the
prechlorinating/postammoniating utilities, and that prechlorination/postammoniation
may be the most effective oxidation/disinfection method for such source waters.
Figure 5-48 shows overall HAA concentrations by disinfection scheme. Both
chlorinating and prechlorinating/postammoniating utilities had significantly higher HAA
levels than chloraminating utilities. The chlorinating and
prechlorinating/postammoniating utilities had median HAA levels of approximately 22
and 19 /yg/L. respectively, while the chloraminating utilities had a median HAA level of
only 8 /yg/L.
5-10
-------
H«lo«c«tic Acids
By Disinfection Schemi
Heloecetonitriles
By Disinfection Schemi
IB!
/^
_J
\
a
j !»•
N-'
•
D
•H
U
-------
Baseline Sampling Results and Discussion
Overall HAN levels are shown in Figure 5-49 by disinfection scheme. This plot reflects
the same trend seen in the plots of XDBP,.im) and TTHM levels as a function of
disinfection scheme; that is, a significantly higher HAN concentration for the
prechlorinating/postammoniating utilities compared to either the chlorinating or
chloraminating utilities. However, in contrast to the plots of XDBPsum and TTHM
medians, Figure 5-49 shows a number of outliers at high levels for the chlorinating
utilities, indicating that some chlorinating utilities produced relatively high levels of
HANs (in the approximate range of 12 to 21
Figure 5-50 is a plot of overall HK concentrations by disinfection scheme. This plot
indicates that the median level of HKs was significantly higher for chlorinating and
prechlorinating/postammoniating utilities compared to chloraminating utilities.
Figure 5-5 1 shows ALD levels as a function of disinfection scheme. In this plot,
prechlorinating/postammoniating utilities produced the highest median ALD
concentration, significantly higher than either the chlorinating or chloraminating
utilities. Influent ALD levels and ALD production as a function of the
oxidation/disinfection schemes utilized by water treatment facilities will be discussed
later in this section and in Section 6.
Overall chloropicrin and chloral hydrate levels are shown in Figure 5-52 and 5-53,
respectively, as a function of disinfection scheme. Both plots suggest the same trend of
the highest median for chlorinating utilities, lower for prechlorinating/postammoniating
utilities, and the lowest median for chloraminating utilities. For chloropicrin levels, the
differences between medians are significant for the chlorinating and chloraminating
utilities. For chloral hydrate, the median level is significantly higher for chlorinating
and prechlorinating/postammoniating utilities compared to chloraminating utilities.
Figure 5-54 is a plot of cyanogen chloride data for the three disinfection scenarios.
The plot indicates that the prechlorinating/postammoniating utilities had a significantly
higher cyanogen chloride level (approximately 2 /ug/L) compared to both the
chlorinating and chloraminating utilities (less than 0.5 /i/g/L). These results will be
evaluated in detail later in this section in the "Special Issues" discussion.
DBF AND INFLUENT PARAMETER CORRELATIONS
The objective of this section is to examine correlations of various DBP classes,
individual DBPs and various water quality parameters. Correlations were determined
between:
o DBP classes
o Individual DBPs
o Individual DBPs and influent water quality parameters
o Influent water quality parameters
In total, over 300 correlations, as shown in Appendix D. were determined. Figures
5-55 to 5-68 present a selected number of these. Note that in some figures, the
correlations are re-evaluated without including certain outliers. In some cases, the
outlier is an anomalous data point not observed in other samplings at the same utility.
In other cases, utilities with very high concentrations of the parameter under
5-11
-------
Aldahydsa Bg Di»inf«ctian Schams
By DiHinfaction Schvmi
\
a
o
ii
o
-4
3
a
TJ
£
H
-------
Chloropicrin Bg Disinfection Scheme
Chloral Hgdrate
By Disinfection Schema
...
1
^
1
\ ..a
a
3
c
•H
L • »
U
a.
0
L
0
H ••*
r
o
t.2
«
i
•
-
j
1
n °
u =
1
CL2
78
21
w
"
CC2HH3
II
1 r~
-
-
-
"
*
"
•
1 i
HH3CL
«
28
G
\
a
D »•
4J
i
L
TJ 18
31
X
^
fj
i. i.
0
H
£.
U
5
•
-i r i i i_
_
•
— ~
-
"
.
.
_
•
•
•
_ —
-
-
l l
: r\ 2 ;
"l 1 1 L.
CLl Cl-2>*O NH2CL
11 n - 72 3* U
4 u • 2] 10 "i
Disinfection Schem
FIGURE 5-52
Disinfection Scheme
FIGURE 5-53
-------
Cyanogen Chloride
By Disinfection Scheme
T
T
-I
a
3
•a
•H
L
0
H
a
n
a
O
CL2
KM30.
71 32 n
21 10 *
Disinfection Scheme
FIGURE 5-54
-------
Baseline Sampling Results and Discussion
consideration can create what may be an artificially high correlation coefficient; thus a
more realistic value is obtained by treating those utilities as outliers.
Correlations with THMs
In this study, analyses were performed through four analytical fractions for a total of 19
individual halogenated DBFs (XDBPs). Concerns have been expressed as to the
practicality of performing several different DBF analyses in a utility's routine
monitoring program, and the question of employing a surrogate has been raised.
Figure 5-55 presents various correlations with THMs. There was a strong correlation
between TTHMs and the sum of XDBPs measured in this project (r=0.96). As THMs
represent the largest DBF fraction detected in this study, the data were reevaluated by
comparing TTHMs to the sum of non-THM XDBPs. In this instance, r decreased to
0.76. However, the latter comparison does not mean that THMs cannot be used as a
surrogate or predictor of the sum of all XDBPs. It should be noted, though, that
correlations between classes of compounds were often low (e.g., comparing TTHMs to
haloketones yields an r of only 0.06). Additionally, some of the other individual DBFs
may require separate monitoring based on their health effects and their formation and
control relative to THMs. For example, THM production can be minimized with the
use of chloramines, whereas cyanogen chloride formation can be increased, as will be
discussed in detail later in this section.
Figure 5-55 also presents the correlation of HANs with THMs (r=0.78). The median
ratio of the sum of the HANs to THMs for all four quarters was 0. II (the 25th
percentile was 0.073 and the 75th percentile was 0.15). The linear regression equation
for this relationship is:
IHANs) = 0.78 + 0.087[THMs]
In general, it appears that the concentrations of HANs were approximately one-tenth of
the concentrations of THMs. Another study found that the concentration of DCAN
averaged about 10 percent of the THM concentration (Oliver, 1983). The variations
from these generalized relationships are probably due, at least in part, to the different
effect of pH on these two DBF classes; this is discussed in greater detail later in this
section.
Correlation with HAAs
Figure 5-56 presents various correlations with HAAs. The correlation of HAAs with
XDBPSUI)I was 0.87. If THMs are not included in the correlation, r increases to 0.98.
However. HAAs comprise approximately 57 percent of the non-THM XDBPs (see
Figure 5-3). If both HAAs and THMs are subtracted from the XDBP and then
correlated with HAAs, r equals 0.77. Furthermore, a correlation of 0.74 is found if
only HAAs are subtracted from XDBP The latter two correlations may be useful in
helping to predict the sums of non-THM, non-HAA XDBPs.
Correlations with Influent Water Quality Parameters
The correlations of TOC with THMs and UV-254 absorbance, and UV-254 absorbance
with THMs. are shown in Figure 5-57. Several researchers have investigated the
5-12
-------
Correlations with Trihalom»thanes
-I
E
Q.
S
X
308
299
188
XDBPsum us THM«
t
r=0.96
sa lea isa see asa 3ea
n-1*»0
THMS (ug/L)
XDBPs-THMs us THMs
H
§
s
X
ise
eo
48
• '-I
O • B>
r=0.76
se iaa isa aee sse 3aa
0=1^0
THMS
J
20
16
12
HANS us THMs
r»0.78
sa iaa isa sea 2sa 3aa
n-1^0
THMa (UQ/L)
FIGURE 5-55
-------
Correlation* with Haloacetic Acids
XDBPsum us HAA
XDBPs-THMs us HAAs
aea
^
j
a
3
i 29e
a.
s
X
180
A
a
a
O
a
B
O
a
a
°.i >*.' '
5? & . '"f^ °
i»'f*'°
•&' r«0.87
Jjfca
fC~ .
169
129
J
S
a
2
•
| 89
i
§
a
CO
o
X 48
9
a
0
o
a
a
!
%v°°
a o <* a
Bag ° OD
a«T» °
0 &** *
f^K
f r=0.98
9
L. — '
28 48 68 89 199 128
HAAS (ug/L)
28 49 68 88 188 128
n=1 kQ
HAAS
-------
Correlations with Influent Parameters
J
a
c
300
250
see
15B
198
58
THMs us TOO
4 8 12 IB
n-105
Influent TOC
28
THMs us UU Abi
380
569
209
158
188
69
8.2
8.4
8.S
8.8
n=102
Influent UW-254 (/cm)
8.8
§ a.e
\
U)
ru
§
9'4
8.2
UU Abs us TOC
r-0.79, 0.85*
4 8 12 IB
n-102, 101*
Influent TOC (mg/L)
28
Excludes indicated outlier
HGURE 5-57
-------
Baseline Sampling Results and Discussion
relationships among TOC, UV-254 absorbance, and THM formation potential
(THMFP) (AWWA Organic Contaminants Committee, 1985; Amy et al., 1987;
Chowdhury et al.. 1988). For this study, the correlation between the UV-254 and TOC
of plant influents was high (i.e., r=0.85, without outlier). Determining this correlation
by water source did not improve the correlation (Figure 5-58). It should be noted that
neither TOC or UV-254 correlated well with the TTHMs of the plant effluents.
Specifically, during the last three sampling quarters, the correlation of TTHMs with
influent TOC was 0.48: and for TTHMs with UV-254, r=0.41. Because utilities apply
chlorine doses based on numerous considerations (e.g., the chlorine demand of the
water, disinfection requirements, taste and odor control), in actuality the TTHMs
detected do not necessarily reflect the THMFP of that water. Also, since a large
percentage of the utilities in this study used chloramines and several utilized ozone as a
preoxidant. it is not surprising to find a poor correlation between the influent TOC and
the effluent DBFs. A mathematical model accounting for chlorine dose, TOC
concentration, bromide level, temperature, pH and other parameters is currently being
explored by other researchers to determine whether THM levels are predictable (Amy et
al.. 1987).
Correlations with Bromide and Brominated DBFs
Figure 5-59 presents correlations of plant influent chloride with plant influent bromide,
and influent bromide with bromoform. Influent chloride had a very high correlation
with influent bromide (r=0.97). The three outliers noted were from the same utility; if
these are not included, the correlation coefficient decreases to 0.86. Because of the
high correlations, chloride may be used as a predictor for bromide. Using all the data,
linear regression analysis yields the equations:
|Br ] = -0.0500 + 0.0044|Cl-j (with outliers)
|Br | = -0.0071 -I- 0.0034|CI] (without outliers)
The correlation of influent bromide with bromoform is also presented in Figure 5-59.
The correlation for this relationship was 0.57 unless the noted outliers are excluded
(r=0.69).
Relationships of influent bromide with chloropicrin, 1,1,1-TCP, TCAA and chloroform
are presented in Figure 5-60. In each case, an exclusion relationship is demonstrated,
i.e.. the presence of bromide appeared to exclude the presence of the particular DBP,
and the inverse was also observed. Mutual exclusion of bromide with these compounds
is also demonstrated by their negative correlation coefficients.
The formation of chloropicrin due to chlorination of: 1) nitrogenous organic
compounds, 2) non-nitrogenous organic substances in the presence of nitrites, and 3)
humic substances, has been well documented in the literature (Coleman et al., 1976;
Sayato et al.. 1982; Duguet et al.. 1984; Thibaud et al., 1986). Additionally, bromide
has been shown by many researchers to play an important role in the formation of
trihalomethanes (Rook, 1976). This study has shown, however, that bromide may also
play a significant role in precluding the formation of certain other compounds. The
exclusion relationship of bromide with chloropicrin is consistent with that observed by
Thibaud, et al. (1988). These researchers showed that increasing bromide
5-13
-------
Correlations with Influent Parameters
U«t»r Source i« F9
Uater Source ie QU
a.a
E 0.6
u
10
nj
C
5
*C 8.2
00
O • •
r - 0.82, 0.90
4 8 12 16 28
n -35, 3*» *
Influent TOC (mg/L)
8.5
0.4
U
ni
3
.1
0.8A
3 6 9 12 15
n = 21
Influent TOC (mg/L)
Ueter Source im LR
ID
ru
8.3
8. 26
8.2
e. IB
8. 1
0.8E
f
I
r = 0.66
n = k6
Influent TOC (mg/L)
* Excludes Indicated Outlier
FIGURE 5-58
-------
Correlations with Influent
Chloride and Bromide
2.5
J
V.
a
•H
0
L.
CD
1.5
e.s
Br— us CL-
CHBr3 us Br—
r=0.97, 0.86*
89
60
a
a
a
o
n
£8
. r=0.57, 0.69---
208 486 688
n=105, 102^
Znfluant Chloride (tng/L)
688
2. 5
0.5 1 1.5 2
n-105, 102*
Influwrt Bromid* (mgxL)
Excludes indicated outliers
FIGURE 5-59
-------
Correlations with Influent Bromide
J
1.6
1.2
9'4
CHP us Br-
r«-0.22, -0.25*
1.6
2.6
0.6 1
n-10^*, 101*
Influent Bromide (mg/L>
1,1, 1-TCP us Si—
a 4
a
o
-H
u
L
&.°
r>-0.26, -0.35^
0.6 1 l.S 2 2.E
n-105, 102*
Influant Bromide (mg/L>
TCAA us Br—
CHC13 us Br-
40
S
? 30
w
TJ
• H
€
•H 20
a
B
.
a
i •
i |
i •
5 1 .
• L «
o E-.
* r.
£ 10 h
" Is' .
L O .
H F . r-0.27, -0.35*
^^•^0«« • ««
1SB
120
i
S
3
w "•
_
t
0
g 60
0
«^
5
a
1 a
o
*u
t*
i
ii1'
Ul .
30 »:
if
B\.' r=-0.25, -0.29'1-'
• Li '^tmAt • > i*-i
0.6 1 1.6 2 2.6 3
n-105, 102*
Influent Bromide (mg/L>
8.6 1 1.6 2 2.5
n-105, 102*
Influent Bromide (mg/L>
Excludes indicated outliers
FIGURE 5-60
-------
Baseline Sampling Results and Discussion
concentrations led to a decrease in the production of chloropicrin
(trichtoronitromethane) and to the formation of brominated halopicrins (i.e.,
bromodichloronitromethane, dibromochloronitromethane and tribromonitromethane).
These halopicrins are analogous to the THMs and appear to be influenced by bromide
in a similar manner.
The exclusion relationship of bromide with 1,1,1-trichloropropanone is supported by
work performed by Croue and Reckhow (1988). These researchers showed that in the
presence of 0.4 mg of bromide per mg of carbon, the chlorinatiqn of a particular fulvic
acid yielded decreased concentrations of 1,1,1 -TCP. Additionally, several new
brominated compounds were detected including chloro-, bromopropanones. As
1,1.1 -TCP has been demonstrated to be an intermediate DBF which can hydrolyze to
chloroform, the chloro-, bromopropanones may be precursors of the brominated
THMs. Since bromide appears to play an important role in DBF formation, it is not
unusual that other exclusion relationships such as with TCAA and chloroform were
observed, as shown in Figure 5-60. A more detailed discussion of the effect of bromide
levels on DBF production is presented later in this section.
The pattern of exclusion relationships with bromide and chorinated DBFs is similar to
that shown with bromoform and chlorinated DBFs. Figure 5-61 presents correlations of
chloroform, 1.1.1-TCP, chloral hydrate and TCAA with bromoform. In each case a
negative correlation was observed. Koch and Krasner (1989) observed that chloral
hydrate concentrations were higher in two water sources low in bromide when
compared to a third source water high in bromide. Additionally, in this study, when
bromoform was correlated with a brominated compound such as dibromoacetonitrile
(Figure 5-62). a fairly high correlation was achieved (r=0.88). A similarly high
correlation was observed when dibromoacetic acid was correlated with
dibromoacetonitrile (r=0.85).
Correlations with Chloroform
In an effort to evaluate whether chloroform could be employed as a surrogate
parameter for other selected DBFs, various correlations with this compound were
determined. Figure 5-63 shows the correlations of chloroform with DCAA and TCAA.
Correlations with all utilities participating in the study and with non-ozonating utilities
are presented because it has been shown that ozonation can reduce TCAA precursors
(Reckhow and Singer. 1984; Bruchet et al., 1985; Lykins et al.. 1986; Dore et al.,
1988). Additionally, it has been shown conceptually that ozonation can shift the DBF
speciation from chloroform and TCAA to DCAA (Reckhow and Singer, 1985). There
was very little difference observed whether or not ozonating utilities were included in
the correlation for either compound; however, only three utilities utilized ozone in the
baseline studies and. thus, may not have impacted the overall correlations for the 35
utilities. Correlations for DCAA and TCAA with chloroform when all utilities were
included were 0.85 and 0.80, respectively.
Correlations of chloroform with chloral hydrate, 1,1,1-TCP and chloropicrin are
presented in Figure 5-64. The best correlation among the three compounds was with
chloral hydrate (r = 0.85). The poor correlation with 1,1.1-TCP (r=0.52) may be
attributed to the instability of the compound. Reckhow and Singer (1985) showed that
1,1.1-TCP was an intermediate of chloroform. They reported that approximately 7.5
5-14
-------
Correlations with Bromoform
CHC13 us CHBr3
1,1,1-TCP us CHBr3
150
120
J
Q
90
8
L
0
w
g 60
O
£
U
38
A
O
r=-0.30
a
B
" 0 0 *
«. « « 4 IB - n B-S "
8
n
3
6
I
n
I
a. A
0
i
i
.
* 1
u f
t L"
H 2 f
3 | • „ r=-0.29
L" '•
F a A o ° ° «
e fc. 2 * tf m * • 9 1
28 48 68
n=1'*0
Bromoform (ug/L)
88
28 48 68
n=1 ^0
Bromarorm
-------
Correlations with Dibromoac»tanitrile
CHBr3 us DBAN
DBAA us DBAN
80
60
L
n
29
r=0.88
28
"
TJ
I
0
r-0.85
0 3 6 9 12 IS
Oibronoacvtonitril*
e 3 6 9 12 15
Dibromo«c«tonitril« (ug/L)
FIGURE 5-62
-------
Correlations with Chloroform
DCAA us CHC13
All Utilities
DCAA us CHC13
Utilities wXo Ozone
50
I *'
TJ
0 30
U
•H
4J
U 20
0
a
-4
r
u 10
a
a
* B
° . '•
a
a
B
<•• B
* . "
B* " °
. •%°.| "
if*'* B
"•':"" •
Ter*
£ " »." r-0.86
ee
\ 49
•D
'u 39
U
+J
1
u se
2
a
^ 19
0 0
°. '.
a
a
a
(0° a
ff -
B
.* •-
. -^'.B '
S» • *' B
//."" '
'° kir*
B*-Js- r=0.85
60
BO
40
30
U
I -
•-4
I. 10
39 69 99 129 160
n=1^0
Chloroform (ugXL)
TCAA us CHC13
All Utilities
o.so
"
•a 49
•H 39
k 2e
10
39 69 90 129
n-UO
Chlorororm (ug/U)
189
39 69 99 128 158
n=129
Chloroform
-------
Correlations With Chloroform
24
20
J
\
a
2 16
4J
L
CH us CHC13
r=0.85
30 68 90 120 150
n=128
Chlorafarn (ugXL)
1,1,1-TCP us CHC13
a.
a
a 4
u
L
h
' -0.52
30 60 90 128 150
n=1i*0
Chloroform (ug/L)
J
a
L
a
-H
r
u
1.6
1.2
0.4
CHP us CHC13
> >
a* •
•• *«
30 60 90
n-139
Chlorafarm (ug/L)
120 150
HGURE 5-64
-------
Baseline Sampling Results and Discussion
percent of the chloroform produced following 72 hours of reaction time between
chlorine and a particular fulvic acid at pH 7 passed through the 1,1,1-TCP
intermediate. This issue will be expanded upon later in Section 5.
Chloroform correlated poorly with chloropicrin (r=0.49). This may be due to several
factors, including the concentration of nitrogenous organic compounds. That
chloropicrin can be formed by nitro-compounds and amino acids has been reported by
several researchers (Coleman et at., 1976; Sayato et al., 1982; Thibaud et al., 1986).
Additionally, chloropicrin precursors can be produced during the oxidation of various
non-nitrogenous compounds, such as phenols, if nitrite is present (Duguet et al., 1985).
Correlations with DCAA and TCAA
Correlations of the four trihalomethanes and DCAA are shown in Figure 5-65. The
best correlation was found with chloroform (r=0.86). However, as the THMs shift to
the more brominated species, the correlations decrease until an exclusion relationship is
observed between DCAA and bromoform (r=-0.33). This progression is consistent
with the bromide correlations discussed above, which showed that bromide and various
chlorinated DBFs were related by exclusion.
Figure 5-66 presents correlations with TCAA. The correlation of DCAA with TCAA
was 0.85, while that with DCAN was 0.76. The figure also shows that the correlation
with 1,1,1-TCP was 0.77. The latter correlation was better than that discussed above
between 1.1.1-TCP and chloroform (r=0.52). This observation is consistent with the
fact that at high pH. chloroform is stable while the other two compounds may undergo
hydrolysis.
Correlations with Formaldehyde
The correlation of acetaldehyde and formaldehyde is shown in Figure 5-67. A
correlation of 0.64 was observed between the two. Because it has been shown in this
study and in others (Lykins et al., 1986; Glaze et al., I989a) that ozonation produces
aldehydes during water treatment, correlation of acetaldehyde with formaldehyde were
determined for only non-ozonating utilities. Excluding the outlier, the correlation
improved to 0.78.
Correlations with Total Organic Halide
Figure 5-68 presents the correlations of XDBPsum with TOX on a molar basis. With all
utilities included in the analysis, the correlation was 0.70. However, if disinfection-
only utilities are not used (since TOX increased in concentration between time of
sampling to time of receipt at Metropolitan's laboratory; see Section 3 of this report for
details), the correlation improved to 0.86, or 0.77 without the outliers. When only
utilities which employ chloramines as a final disinfectant are included in the correlation
(since TOX concentrations in chloraminated effluent appeared to be least affected by
transit time to Metropolitan's laboratory for preservation; see Section 3 for details), r is
equal to 0.78. The correlations found in this study are not as high as those found by
Singer (1988) who correlated the THM formation potential with the TOX formation
potential of various raw waters (r=0.96, n = 60, reported on /ug/L basis). However, the
correlation found by this researcher between TTHMs and TOX on finished waters
5-15
-------
Correlations with Dichloroacetic Acid
CHCl2Br us DCAA
169
128
J
\
68
39
CHC13 us DCAA
r=0.86
a
•H
Q
68
48
28
o
On o
18 28 38 48 58
Oichloro«c«tic Acid (ug/t)
CHBr2Cl us DCAA
18 28 38 48 58
.
Dichloro«c«tic Acid (ugxl_)
CHBrG us DCAA
88
3
X
3 68
w
g
£
I 48
0
L
0
u
o
0 28
L.
a
,•4
Q
a
a
o
a a
a
a
e a
a
a* a.8
r-0.03
o •
Jo a g
ft *2^ ° * B °
^^ ^^ ° o* °8 f ° *
88
68
*»
^
J
L 40
a
a
!
o
20
a
a
a
"> 0
D.
1
a
J
°.
a
e
D°°' - r=-0.33
a B
a° o o
^r-»f?J^ -r.- - -
18 28 38 48 58
18 28 30 48 58
Oichloro»c«tic Acid (ug/L>
Oichlaroacatic Acid Cug/L)
FIGURE 5-65
-------
Correlations with Trichloroacetic Acid
DCAA us TCAA
1,l,1-TCP us TCAA
68
J
3
TJ
U 38
-------
Correlation with Formaldehyde
Non-0xon«ting Utilitii
IB
12
j
a
.:-
-%•/
la
j
a
IB 20 38
n=95
Formaldehyd"
-------
I '
t«d
og»n«
M
L
Molar XDBP» uer«u« Molar TOX
All Utilities
-••» *'.
r-0.70, 0.60*
a 12
2« 24
n-138. 136*
Molar Vox
Utilities using Chlaraninae
As Final Disinfectant
r=0.78
• s 4 a a !• 12
Molar TOX
* Excludes Indicated Outliers
Net Diainfsetion-Oniu Utilities
i
TJ
41
r-0.86,0.77'1;
24
n- 122. 120*
Molar TOX
FIGURE 5-68
-------
Baseline Sampling Results and Discussion
(r=0.89. n=!66) was closer to the correlation found in this study. The difference in
correlations between the formation potentials of raw water and instantaneous
measurements on finished water is probably due to the variety of conditions with which
each plant treats its water.
SPECIAL ISSUES
In evaluating the results of the baseline data collection for this study, a number of
issues were identified which could not be classified under the previous headings of this
section. These special issues are discussed below.
Effect of pH
The HAN and HK data presented in Figures 5-29 and 5-30 tend to support
relationships reported by other researchers in which HANs and HKs were found to be
reactive intermediates rather than stable endproducts such as THMs. For instance,
Reckhow and Singer (1985) reported that when a fulvic acid solution was chlorinated,
concentrations of DCAN and 1,1,1-trichloropropanone (1,1,1-TCP) declined over time
after quickly reaching an initial peak while chloroform, TCAA and DCAA increased in
concentration with increasing contact time. These researchers also found the same
trend with increasing pH. demonstrating that DCAN and 1,1,1-TCP hydrolyzed at high
pH; while chloroform concentrations increased with increasing pH, indicating that this
compound is a product of the hydrolysis reactions of DCAN and 1,1,1-TCP.
Additionally. Gurol. et al. (1983) found that the presence of free chlorine greatly
increased the degradation rate of 1.1,1-TCP and the formation of chloroform.
In order to investigate the impact of pH on the DBP concentrations measured in this
study, several DBPs were plotted as a function of clearwell effluent pH. In actuality, it
is the pH (which varies) throughout the treatment process that impacts DBP formation,
but it is the final pH which determines the stability of the DBPs and their fate in the
distribution system (i.e.. that received by consumers). In these plots, clearwell effluent
pH values from all four quarters of baseline data collection were divided into four
intervals of varying magnitude, but with an equal number of measurements in each
interval. (Usually, plots with equal interval widths would be a preferable way to show
the data: however, because there were very few data points in the lowest pH interval
and most of the data points were in the highest pH interval, the plot was prepared with
varying interval widths and equal numbers of data points in each interval.) Figure 5-69
illustrates the levels of XDBPfum plotted in this manner. No significant difference
between XDBPsum values are indicated by the data in Figure 5-69. The problems
inherent in any attempts to ascribe differences in DBP levels to single causal factors
such as pH have been discussed previously in this section; and Figure 5-69 may indicate
the influence of many confounding factors. However, it may also be the case that pH
impacts the levels of individual DBP compounds which make up the parameter
XDBPsunr rather than influencing the level of XDBPsum itself.
To further explore this issue. TTHMs were plotted as a function of clearwell effluent
pH in Figure 5-70. As observed in the plot of XDBP , there are no statistically
significant differences between the median levels of TTHMs at a 95 percent confidence
level. HANs are plotted as a function of pH in Figure 5-71. This plot, in general,
indicates decreasing levels of HANs with increasing pH. When clearwell effluent pH
5-16
-------
XDBPsum By Effluent pH
J
a
E
3
«
a.
m
Q
X
13
T.BS-8.23
33
111
O.24-8.7
8.11-9.78
33
13
pH Range
FIGURE 5-69
J
a
3 «•
£
+J
II
0
•H
L
Trihalomethanas
By Effluent pH
"T
4.6-7.BE
13
7.E6-8.23
33
pH Range
FIGURE 5-70
\
J 1.
a.71-9.78
33
13
-------
Haloacetonitriles
BU Effluent pH
Dichloroacet.anit.rile
Bg Effluent pH
T
_t
a
•H
L
•H
I "
II
U
0
IB
J
a
•H
L
+J
•H
0
4J
II
U
•
0
L
0
H
U
•H
Q
T
T
T
J_
T.M-e.83
•.24-V.7
33 ^
ft f*
pH Range
FIGURE 5-71
B.-71-S.7B
33
13
4.6-7.ft
3<«
13
T.ea-a.33
33 3*»
Hi l"t
pH Range
FIGURE 5-72
9.71-9.78
33
13
-------
Baseline Sampling Results and Discussion
values occurred within the range 8.24 to 9.78, the median level of HANs was
approximately 2.5 to 3.5 /ug/L, whereas the HAN medians were 4.5 to 5.5 /ug/L in the
4.6 to 8.23 pH ranges. The difference between the median HAN concentrations for the
7.56 to 8.23 pH range and the 8.71 to 9.78 pH range is statistically significant at a 95
percent confidence level. Thus, Figure 5-71 tends to support the finding that HANs
hydrolyze at high pH. When DCAN was plotted as a function of clearwell effluent pH,
no statistically significant difference in median concentrations occurred over the four pH
ranges (Figure 5-72). However, the 75th percentile value for the 4.6 to 7.55 pH range
(3.7 /yg/L) is approximately twice the 75th percentile value for the 7.56 to 9.78 pH
ranges (1.6 to 2.0/ug/L).
A plot of 1.1,1-TCP as a function of pH is shown in Figure 5-73. This figure
illustrates that statistically significant differences occurred between 1,1,1-TCP
concentrations at low pH (4.6 to 7.55) and at high pH (7.56 to 9.78). Figure 5-73 is
thus an indication that hydrolysis of this haloketone occurred at basic pH. This same
trend was observed in a plot of TCAA as a function of pH. The median TCAA
concentration was higher within the pH range 4.6 to 7.55 than within the pH ranges
greater than 7.55; however, the differences were not significant at a 95 percent
confidence level. Miller and Uden (1983) reported that TCAA concentrations declined
substantially as pH increased from 6 to 10 in the chlorination of a fulvic acid solution.
Miller and Uden (1983) also found that chloral hydrate increased in concentration from
pH 4 to pH 7, and then decreased as pH increased to 10. These researchers reported
that the product of chloral hydrate decomposition at elevated pH was chloroform.
However, chloral hydrate results from the baseline data of this study indicated that
there were no statistically significant differences between the median concentrations of
this compound occurring within the four clearwell effluent pH ranges.
Effect of Temperature
Seasonal effects on DBP levels were discussed previously in this section. Plots of DBP
concentrations showed some indications of temperature-related effects, although some
seasonal impacts may be due to changes in the nature of naturally-occurring organics as
well. The impact of temperature on THM formation has been documented by a
number of researchers over the past decade, including Stevens, et al. (1976), who
found that chloroform concentrations more than doubled (from approximately 100 to
over 225 jug/L) when incubation temperature was increased from 25 to 40°C in raw
Ohio River water receiving 10 mg/L of chlorine.
In order to evaluate temperature effects on DBP levels measured in this study,
XDBP,lim, TTHMs and HANs were plotted as a function of influent water temperature.
Data from all four quarters of baseline data were divided into four temperature ranges
of equal magnitude. Figure 5-74 illustrates that levels of XDBP,um were strongly
influenced by temperature. Although there is very little difference in the three lower
temperature ranges (from 1.1 to 23.4°C), in the range 23.5 to 3I.O°C, XDBPS was
significantly higher at a 95 percent confidence level. The same trend is observed in
Figure 5-75. a plot of TTHMs by temperature range. In the three lower temperature
ranges, median TTHMs varied between approximately 25 and 35 /ug/L, while in the
highest temperature range, the median TTHM concentration was over 65 /ug/L. The
same trend was observed for HANs as a function of temperature, as shown in Figure
5-76, although the differences were not statistically significant.
5-17
-------
J
\
a
o
a.
o
a.
o
o
H
U
•H
•.
H
*
1»1-Trichloropropanone
By Effluent pH
_L
D
a.
a
x
4.e-7.ee
34
13
7.68-8.23
33
14
pH Range
FIGURE 5-73
e. 7i-9. 70
"
3
Sum of Halogenated DBPi
By Influent Temperaturi
e.6-ie,« ie.i-33.4 aa.e-3i.8
1,7 44 27
17 27 26 17
Temperature Range (°C)
FIGURE 5-74
-------
\
a
3 "•
•
+J
E
a
•H
t
K
Trihalomethanes
By Influent. Temperaturi
ai
J
a
•H
L
+J
•H
c
•
u
0
«
i.i-«.•
n- 2)
u- 17
a.*-ie.i
1C. 1-23. 4
23.S-31.
27
17
" L° n ''
Temperature Range (UC)
FIGURE 5-75
Haloacetonitrilee
By Influent Tamperaturi
i.i-a.B ».e-i«.« ie.i-23.4
21 47 44
17 27 26
27
17
Temperature Range (°C)
FIGURE 5-76
-------
Baseline Sampling Results and. Discussion
Brominated DBFs
During the first quarter of baseline sampling, a high correlation was found between
dibromoacetic acid (DBAA) and dibromochloromethane (correlation coefficient r =
0.91). In addition, relatively high levels of the measured brominated DBFs were
detected at some utilities. These findings suggested that the influence of bromide
present in the raw water should be evaluated; therefore, bromide and chloride analyses
of the plant influents were added beginning with the second quarter of baseline
sampling.
Among the 35 utilities in this project, bromide levels in the plant influents ranged from
<0.01 to 3.00 mg/L. High bromide levels were found in each of the three types of
source waters. Figure 5-19 presented the raw water bromide levels at the 35 utilities
for the latter three sampling quarters.
Table 5-5 shows DBF data for a utility with high bromide levels (Utility 12). Not only
was there a shift in THMs to the more brominated species at high bromide levels, but
the same situation existed for the HANs and HAAs. Dichloro- and trichloroacetic acid
(DCAA and TCAA). which are commonly found in other DBF studies (Uden and
Miller, 1983; Norwood et al., 1986) as well as this one, were detected at low levels at
Utility 12 when bromide levels were high; instead. DBAA was the major HAA
detected.
Additionally, there were seasonal shifts in the raw water bromide concentration at this
utility. Such shifts were observed in some utilities as a result of drought conditions and
saltwater intrusion problems. During the summer of 1988, 0.41 mg/L bromide was
detected at Utility 12. In the fall of 1988 and winter of 1989, higher bromide levels
were detected (0.78 to 0.79 mg/L). As Table 5-5 shows, the change in distribution of
brominated DBFs was consistent with the change in bromide level. For example,
bromoform and dibromoacetonitrile (DBAN) represented (on a weight basis) 36 and 67
percent of the sum of their respective class fractions in the summer, when the bromide
level was lowest. These percentages increased to approximately 66 and 85,
respectively, in the fall and winter, when the bromide levels were higher. Similar
results have been documented by other researchers, such as Lange and Kawczynski
(1978). when investigating the impact of bromide concentration on THM speciation.
When bromide levels at Utility 12 increased from 0.41 to 0.78 mg/L from summer to
fall, 1988. there was a shift to a higher percentage of brominated DBFs, as discussed
above. This shift is illustrated in Figure 5-77. However, this figure indicates that the
shift for the HANs and HAAs was less dramatic than that for the THMs, as the
majority of HANs and HAAs were already brominated during the summer, 1988
sampling. However, bromoform only represented 36 percent of the THMs in summer
sampling, while it increased significantly (to 68 percent) by the fall. One possible
explanation is that the kinetics for the formation of DBAN and DBAA may be faster
than that for bromoform. If this is the case, this could explain why the correlation
coefficients for DBAN and DBAA versus bromoform (i.e., 0.88 and 0.82. respectively)
are not higher, as the formation of the fully brominated species for each DBF class may
proceed to completion at different kinetic rates.
5-18
-------
TABLE 5-5
DBF CONCENTRATIONS AT UTILITY
WITH SEASONAL CHANGE IN BROMIDE LEVELS
:
Component
PLANT
Total Organic Carbon
Chloride
Bromide
Summer
1988
INFLUENT
2.6
111
0.41
Utility 12*
Fall Winter
1988 1989
VALUES, mg/L
2.2 2.8
215 202
0.78 0.79
Summer*
1989
2.9
44
0.14
CLEARWELL EFFLUENT VALUES, /ig/L
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Total Trihalomethanes
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dihromoacetonitrile
Total Haloacetonitriles
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Total Haloacetic Acids
1 . 1 -Dichloropropanone
1,1,1 -Trichloropropanone
Total Haloketones
Chloropicrin
Chloral Hydrate
Cyanogen Chloride
4.7
13
28
26
72
<0.012
0.74
1.6
4.6
6.9
<1.0
2.9
1.6
1.0
14
20
0.36
0.24
0.60
0.018
0.53
2.3
1.4 0.86
7.5 6.5
25 24
72 53
106 84
< 0.029 < 0.029
0.24 0.19
0.96 1.4
7.0 11
8.2 13
<1.0
-------
100
90 -
80
I 7°
CO
§ 60
CL
CQ —
Q 50
i 40
LU
O 30
LU
D_
20
10
SUMMER FALL SUMMER
1988 1988 1989
UTILITY 12 DISINFECTION BYPRODUCTS SPECIATION
FIGURE 5-77
-------
Baseline Sampling Results and Discussion
When the bromide level at Utility 12 decreased to 0.14 mg/L in the summer of 1989,
chlorinated DBFs began to predominate. As would be expected from the
bromide/chlorinated DBF exclusion relationships, the highest levels of chlorinated
ketones. chloropicrin and chloral hydrate were detected during the low bromide
sampling period.
The high bromide concentrations and associated impacts on DBF speciation observed at
Utility 12 were the result of saltwater intrusion. However, high levels of chloride and
bromide were measured in waters not limited to coastal origins with modern saltwater
intrusion problems. A mid-South utility (Utility 10), which is located inland, derives it
waters from two lakes, one of which is high in mineral content. In dry years,
evaporation results in an increase in salinity, and in particular, chloride levels. During
the three quarters that chloride and bromide were measured, levels ranged from 561 to
680 mg/L and 2.8 to 3.0 mg/L, respectively. In the Midwest, high bromide levels
were detected in waters from two utilities (Utilities 23 and 26). Utility 23 had 47 to
152 mg/L chloride and 0.44 to 1.19 mg/L bromide. Utility 26 had 69 to 251 mg/L
chloride and 0.19 to 0.68 mg/L bromide. As Table 5-6 indicates, these utilities have a
higher level of brominated DBFs than chlorinated ones, due to the presence of high
bromide levels. Thus, the presence of brominated DBFs is not restricted to coastal
areas, such as in California or Florida, experiencing saltwater intrusion problems in
their source waters. According to Standard Methods (1989), the bromide content of
groundwaters and stream base-flows can also be affected by connate water (ancient
seawater that was trapped in sedimentary deposits at the time of geological formation).
In addition, industrial and oil-field brine discharges can contribute to the bromide in
source waters.
In the 35-utility study, there was a very good correlation between bromide and chloride
levels (correlation coefficient r = 0.97), as discussed previously in this section. Since
the levels of these ions were atypically high at Utility 10. the data were re-examined by
excluding the levels detected at that utility. The correlation was still high (r = 0,86).
As discussed previously, an equation for predicting bromide levels from chloride levels
was determined, excluding the outlier points from Utility 10:
[Br ] = -0.0071 + 0.00341C1)
This equation was then applied to the data from Utility 12, where the source of
chloride and bromide is due to saltwater intrusion problems, and the results are
reported in Table 5-7. The measured and predicted bromide levels agreed to 10
percent, on the average. Furthermore, the concentrations of chloride and bromide in
seawater are 18.980 and 65 mg/L. respectively (Sverdrup et al.. 1942). From these
data, if chloride and bromide were only from sea water diluted with unsalty freshwater.
then:
|Br | = 0.0034[C1|
Despite the similarity of the equations, there will be more variability in the prediction
when more than modern sea water intrusion is involved. However, since the correlation
holds in general for the 35 utilities, one should be able to predict a relative level of
bromide in waters where only chloride measurements were made.
5-19
-------
TABLE 5-6
DBF CONCENTRATIONS AT INLAND UTILITIES
WITH HIGH BROMIDE LEVELS*
Mid-South
Component Util. #10
Midwest
Util. #23
Midwest
Util. #26
PLANT INFLUENT VALUES, mg/L
Total Organic Carbon
Chloride
Bromide
CLEARWELL
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Total Trihalomethanes
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
Total Haloacetonitriles
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Total Haloacetic Acids
4.9
561
2.9
EFFLUENT
0.59
2.9
9.2
40
53
< 0.029
0.24
I.I
6.7
8.0
-------
TABLE 5-7
UTILITY 12: INFLUENCE OF SALTWATER INTRUSION
Bromide Level
Sampling Chloride
Period mg/L
Summer 1988
Fall 1988
Winter 1989
Summer 1989
1 II
215
202
44
Measured
mg/L
0.41
0.78
0.79
0.14
Predicted*
mg/L
0.37
0.72
0.68
0.14
Percent
Difference*
9.8
7.7
14
0.0
•Percent difference = 100 x (measured - predicted)/measured.
'Prediction based on relationship derived from this study:
[Br-| = -0.0071 + 0.0034|C1]
-------
Baseline Sampling Results and Discussion
As indicated above, the production of THMs in those utilities with high bromide levels
can shift to the more brominated species. The same phenomenon was observed for
HAAs and HANs. It is fortunate that many brominated DBFs were included in the
study: otherwise, the DBF levels of some utilities would be misrepresented. Yet,
research has shown that other brominated species exist which were not included in the
analytical methods for this DBF study. GC/MS analysis revealed the presence of HAAs
containing both bromine and chlorine atoms (Slocum et al., 1987) as is observed with
THMs; however, analytical standards for these compounds did not exist commercially.
Research also indicated the formation of brominated trihalonitromethanes in a manner
similar to the production of chloropicrin (trichloronitromethane) (Thibaud et al., 1988).
These findings emphasize the fact that brominated DBFs, not just chlorinated DBFs,
are important in chlorinated drinking water.
Aldehydes
Results of aldehyde analyses of cleanvell effluents were discussed previously in this
section. When ALDs were added to the list of analytes after the second quarter, and it
became clear that ALDs were occurring in the effluents of many of the 35 utilities (not
exclusively ozonating utilities), the issue of whether these compounds were originating
in the plant influents or within the plant still needed to be resolved. To determine if
formaldehyde and acetaldehyde were produced by the disinfectants/oxidants used at the
plants or if these compounds originated from the source water (e.g., from a biogenic
process), all 35 utilities were sampled at the plant influents (preserved with mercuric
chloride) during the fourth quarter of baseline sampling.
Table 5-8 shows the aldehyde levels in the plant influents and effluents. Formaldehyde
was found in 16 of the 34 influents analyzed, at levels of 1.2 to 13 /ug/L. The median
level of formaldehyde in plant influents for all 34 utilities was <1.0 /ug/L, and the
median level for only the 16 utilities where formaldehyde was detected in the influents
was 2.8 /ug/L. Acetaldehyde was found in 12 of the 33 influents analyzed, at levels of
1.1 to 16 /ug/L. The median level of acetaldehyde in plant influents for all 33 utilities
was < 1.0 /ug/L. and the median level for only the 12 utilities where acetaldehyde was
detected in the influents was 2.0 /ug/L.
At the three ozone plants, it is clear that formaldehyde was a product of the
oxidation/disinfection process. Acetaldehyde was found in the effluents of all three
ozone plants; however, it was detected at a higher level in the plant influent of Utility
25. The other plants shown in Table 5-8 employed either free chlorine only or
chloramination (primarily with pre-chlorination). At some plants, no formaldehyde or
acetaldehyde was detected in the plant influent, but these aldehydes were detected in
the effluent. Where these aldehydes were detected in the influent, they were either at a
very low level compared to that detected in the effluent (e.g.. Utility 29, a chlorinating
utility, had 2.0 versus 8.0 /ug/L formaldehyde in the plant influent and effluent,
respectively) or at a level comparable to that detected in the effluent (e.g., Utility 14
had 6.4 versus 4.1 /ug/L formaldehyde in the plant influent and effluent, respectively).
From these limited data, formaldehyde and acetaldehyde appear to be present because
of a combination of the effects of plant disinfection processes and influent water
quality, the combination varying from one utility to another.
5-20
-------
TABLE 5-8
ALDEHYDE LEVELS IN PLANT INFLUENTS AND EFFLUENTS*
Utility
Number
4
10
18
2
5
7
9
12
14
21
28
30
1
3
6
8
11
13
15
16
17
20
22
23
24
26
29
31
33
34
35
27
25
19
32
Disinfection
Scheme
NH2CI
NH2C1
NH2C1
CI2, NH3
CI2, NH3
CI2, NH3
CI2, NH3
CI2, NH3
CI2, NH3
CI2. NH3
CI2, NH3
CI2, NH3
CI2
CI2
CI2
CI2
CI2
c!2
Cl,
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2
CI2. CI02
O3. NH2CI
0,. CI2
03. CI2
Formaldehyde, A/g/L
Plant Clearwell
Influent Effluent
ND
1.7
1.2
NA
4.9
ND
3.9
ND
6.4
3.0
2.7
ND
ND
ND
13
ND
3.4
ND
ND
ND
ND
ND
1.2
ND
1.4
ND
2.0
ND
5.8
ND
ND
1.4
3.2
ND
1.4
ND
ND
ND
NA
ND
ND
8.7
2.0
4.1
6.9
3.5
4.3
ND
ND
7.6
NA
2.0
1.1
ND
1.6
ND
ND
4.3
ND
ND
2.1
8.0
3.3
3.6
1.8
2.1
4.6
19
7.5
21
Acetaldehyde, //g/L
Plant Clearwell
Influent Effluent
ND
ND
1.9
NA
4.5
ND
4.4
ND
2.4
10
1.7
1.3
ND
ND
NA
ND
ND
2.1
ND
ND
1.6
1.4
ND
ND
ND
ND
ND
ND
ND
ND
ND
1.1
16
ND
ND
1.5
1.2
1.5
NA
3.4
1.8
6.1
1.6
4.5
8.5
3.0
2.2
ND
ND
NA
NA
1.8
1.3
ND
2.0
ND
ND
2.2
1.0
1.4
2.1
4.6
2.6
1.3
2.4
1.5
1.9
5.5
3.9
2.1
* Fourth sampling quarter.
ND = Not detected (< 1.0 //g/L).
NA = Not analyzed; analytical problem with sample.
-------
Baseline Sampling Results and Discussion
Table 5-9 lists the aldehyde levels at three ozone plants during baseline sampling and at
four plants that utilized only chlorination or prechlorination/postammoniation but had
high formaldehyde levels (j>IO /t/g/L during at least one sampling). The latter four
utilities had 10 //g/L or more of formaldehyde at a time when they had approximately
100 /vg/L (or more) TTHMs. These limited data suggest that when a chlorinating
utility's water quality and treatment practices produce a high level of THMs, they can
potentially produce a high level of formaldehyde as well. As will be discussed in
Section 6 regarding treatment studies where ozonation was implemented, there can be a
"tradeoff" of halogenated DBFs for aldehydes in implementing ozonation. Levels of
halogenated and non-halogenated DBFs in the clearwell effluents and distribution
systems of ozonation plants will be discussed in Section 6. In addition, the issue of the
formation and removal of aldehydes in ozonation plants will also be discussed in
Section 6.
Cyanogen Chloride Results
Typically, chloramines are used as a means of lowering THM levels in treated waters.
Of the 142 cyanogen chloride analyses conducted for this study (35 utilities for three to
four quarters), 32 samples represented prechlorination/postammoniation and another 11
represented chloramines only. As seen in Figure 5-54 (presented previously), the
prechlorinating/postammoniating utilities had a significantly higher cyanogen chloride
median concentration (approximately 2 //g/L) compared to either the chlorinating or
chloraminating utilities (both of which had medians of less than 0.5 /t/g/L).
Additionally (not shown in Figure 5-54), one utility with
preozonation/postchloramination had a median cyanogen chloride level of
approximately 7 fJg/L. Since some of these disinfection schemes represent small data
sets, making it difficult to interpret the data, Figure 5-78, a plot of cyanogen chloride
by final disinfectant (regardless of whether prechlorination/postammoniation or
chloramination were used) was generated. As seen in the figure, the utilities that
deliver chloraminated water were demonstrated to have a statistically higher level of
cyanogen chloride.
Research in Japan has shown that cyanogen chloride was sometimes formed in the
presence of certain ami no acids and hypochlorous acid, but was always formed in the
presence of the amino acids tested when both hypochlorous acid and the ammonium ion
(i.e.. chloramines) were present (Hirose et al., 1988). Other research (Ohya and
Kanno. 1985) found that cyanogen chloride was formed by the reaction of humic acid
with hypochlorous acid in the presence of the ammonium ion. It was found that the
amount of cyanogen chloride was at a maximum when the reaction mixture contained a
ratio of 8 to 9 ppm of chlorine to 1 ppm of ammonia (as nitrogen), and that the
maximum yield of cyanogen chloride increased as increasing amounts of hypochlorous
acid were added. Furthermore, these formation patterns were reproduced with three
raw waters from Japan. These data imply that cyanogen chloride may be more readily
formed in chloraminated systems.
In separate research endeavors, N-chloroglycine was formed as a result of the reaction
of monochloramine with glycine under conditions typical for drinking water (Margerum
and Gray. 1978). and it has been suggested that the formation of cyanogen chloride is
caused by the reaction of glycine with chlorine (Kopfler et al., 1975). Studies have
demonstrated the formation of organochloramines by the use of inorganic chloramines
5-21
-------
TABLE 5-9
LEVELS OF ALDEHYDES AND THMS IN SELECTED CLEARWELL EFFLUENTS
Utility
Formaldehyde, pgIL Acetaldehyde, fig/L
Summer Fall Wlnler Summer Foil Winter
1988 1988 1989 1988 1988 1989
TTHMs, //g/L
Summer Fall
1988 1988
Winter
J989
Chlorine or Chloramine Plants:
2
9
26
29
Ozone
19
25
32
10 8.2 NR* 4.2 4.4 NR
12 6.2 8.7 4.1 2.8 6.1
17 8.6 2.1 6.0 7.1 2.1
NR 13 8.0 4.3 5.2 4.6
Plants:
5.8 10 7.5 4.8 5.3 3.9
31 22 19 15 9.9 5.5
30 24 21 3.5 2.8 2.1
90
95
164
109
15
34
3.1
82
54
100
180
20
16
1.4
60
40
98
259
5.9
9.0
0.72
*NR = not reported; analytical problem wilh sample.
-------
Cyanogen Chloride
By Final Disinfectant
IB
14 —
13
r
01
T3
•H
r
o
Q)
01
D
ID
3
O
"
!•
CHLOHXNB
78
CHLOAAMXNeS
Final Disinfectant
FIGURE 5-78
-------
Baseline Sampling Results and Discussion
in the treatment of water (Scully and Bempong, 1982). The cyanogen chloride results
presented above, as well as the presence of moderate concentrations of TOX in
chloraminated waters with low levels of TTHMs. indicate a need to further identify
chloramine by-products.
DBF Levels of Disinfection-Only Utilities
Figure 5-35. presented previously, illustrated influent TOC levels as a function of
treatment type. DBF levels as a function of treatment type have not been discussed
further in this section for the reasons presented previously in Section 5, under the
heading "Star Plot Analyses". However, some findings of this study with respect to
DBF levels occurring in disinfection-only utilities warrant further evaluation.
The overall median concentration of TTHMs for direct filtration and disinfection-only
utilities were the lowest of the four treatment types, although the differences were not
statistically significant. However, the median level of HAAs in disinfection-only
utilities (approximately 35 //g/L) was significantly higher than that of either
conventional or direct filtration utilities (approximately 17 and 9 /ug/L, respectively);
and this difference was significant at a 95 percent confidence level. Additionally, the
median concentration of HAAs for softening utilities was only approximately 23 //g/L,
although this was not significantly lower than the HAA median for disinfection-only
utilities. This trend was even more pronounced when comparing the median HK
concentrations of disinfection-only utilities (approximately 3.8 //g/L) with those of
conventional, direct filtration and softening utilities (all of which occurred within the
approximate range of 0.8 to 1.4 //g/L), and the differences were statistically significant
at a 95 percent confidence level.
Samples from the disinfection-only utilities were collected very shortly after the
addition of the disinfectant (in all cases, free chlorine, although one disinfection-only
utility also employed preozonation). There was very little free chlorine contact time
before sample collection since none of these utilities had detention time in a clearwell
after the addition of chlorine.
The role of HKs as reactive intermediates rather than stable endproducts was discussed
previously. The HK data from the disinfection-only utilities seem to suggest that this
class of compounds are formed very rapidly upon chlorination, and there was
insufficient contact time before sampling for the reaction to proceed further toward the
formation of chloroform. A factor that may influence HAA formation in disinfection-
only utilities is that the disinfectant is applied directly to the organic material present in
the raw water, before the precursor material has been lowered in concentration, or
altered in molecular weight distribution or other characteristics, by the addition of
treatment chemicals, filtration or other treatment processes.
Removal of TOC During Treatment
Figure 5-79 illustrates the removal of TOC within the filtering plants included in this
study. The percent removal of TOC from the plant influent to the filter influent of
each plant was calculated, and the mean value of percent removal was plotted in the
figure. In a similar manner, the removal from the filter influent to the clearwell
effluent, and from the plant influent to the clearwell effluent were also calculated and
5-22
-------
O
0)
DC
U
O
c
a
O
^
0)
CL
Mean Tota Organic Carbon Hernova
Through Filtering Utilities' Processes
40
3O
20
10
o
PI Plant Influent
Fl Filter Influent
EFF Clearwell Effluent
Mean PI TOG = 4.O mg/L
Mean EFF TOG = 2.8 mg/L
2O.9
PI TO Fl
Fl TO EFF
FIGURE 5-79
23.9
PI TO EFF
-------
Baseline Sampling Results and Discussion
are shown in the figure. The plot indicates that most of the TOC removal
(approximately 21 percent) occurred prior to filtration, that is, by sedimentation, and
only 5 percent of the remaining TOC was removed by filtration. Overall, TOC removal
within the filtering plants averaged approximately 24 percent.
The data shown in Figure 5-79 reflect the TOC removal achieved by the participating
utilities, where much of the process operation is most likely focused on turbidity
control. In Section 6. results are presented for two utilities that were able to achieve
higher TOC removals than the mean values noted in Figure 5-79. (It should be noted
that for one of the two utilities which were capable of increasing TOC removal by
increasing alum doses, the enhanced precursor removal was only an incremental
increase over that achieved in the baseline sampling.) From the data collected for this
study, it is not possible to draw conclusions as to whether or not the levels of TOC
removal illustrated in Figure 5-79 can be improved under all or most circumstances by
optimizing the coagulation process. Further research in this area is required, especially
since the TOC in some surface waters does not appear amenable to removal by
conventional treatment (Chadik and Amy, 1983).
TOC removal by coagulation may be further optimized by controlling coagulation pH,
although many utilities do not have the capability to control pH at the rapid mix basin.
Of course, the economics of increased chemical consumption and associated sludge
production, as well as increases in total dissolved solids due to acid and base addition,
must be weighed against the advantages of enhancing precursor removal by increasing
coagulant doses and adjusting the coagulation pH. The coagulation studies will be
discussed in detail in Section 6 of this report.
Comparison of THM Levels from USEPA DBF Study and AWWARF THM Survey
The median TTHM levels previously reported in Table 5-2 were 34, 44, 40 and 30
//g/L for the spring, summer and fall quarters of 1988 and the winter quarter of 1989.
These data were compared to the TTHM values obtained in a survey of 727 utilities
around the United States conducted for the American Water Works Association
Research Foundation (AWWARF) in 1987 (McGuire and Meadow, 1988). The median
TTHM concentrations in the AWWARF survey for the spring, summer, fall and winter
seasons were 40, 44. 36 and 30 //g/L, respectively. (The AWWARF survey reflected
more than 67 percent of the population represented by water utilities serving more than
10.000 customers.)
Because of the similarity of TTHM levels for the DBF study and the AWWARF survey,
the data were further evaluated. Compliance with the THM regulation is based on a
running annual average for each utility (USEPA, 1979); therefore, mean values were
computed for each of the 35 utilities for the four sampling quarters. The AWWARF
survey utilized the means of three years of quarterly data. The means for both projects
are plotted on a frequency distribution curve in Figure 5-80. A log scale was used for
the ordinate axis in order to compress the displayed range, not to imply a log-normal
distribution Visually, the 35-utility DBF study appears to represent a THM frequency
distribution very similar to that of the 727-utility AWWARF survey. The major
difference is in the data for low TTHM levels (less than or equal to 25 //g/L). As
discussed in Section 2 of this report, utility selection for this study attempted to achieve
a balance among utilities which had reported low, medium and high TTHM levels in
5-23
-------
FREQUENCY DISTRIBUTIONS OF
THM SURVEY DATA
O)
CO
til
O
300
200
100
50
20
10
5
USEPA/CDHS
DBP SURVEY
MEANS OF 4 QTRS
(n = 35)
AWWARF
THM SURVEY
(n = 727)
_L
I I I I I 1 |
I I
10 30 50 70 90 95 99 99.9 99.99
PERCENT LESS THAN OR EQUAL TO GIVEN CONCENTRATION
FIGURE 5-80
-------
Baseline Sampling Results and Discussion
the AWWARF survey. Thus, fewer utilities with very low TTHM levels were included
in the DBF study than are found nationwide, and this may be the cause of the
differences in the data below 25 /ug/L of TTHMs between the two surveys. A statistical
comparison of the two distributions, by means of a Kolmogorov-Smirnov test (Hoel,
1971). indicates that the hypothesis that these samples are from the same distribution is
not rejected at a significance level of 0.01.
In addition, a notched box-and-whisker plot of the TTHM levels from the two surveys
is shown in Figure 5-81. Both sets of survey results have median TTHM values of 39
//g/L. and the 95 percent confidence intervals for the two medians overlap, indicating
that the two groups are statistically similar in terms of central tendency. Furthermore,
their minima and maxima are comparable, indicating a similarity in variability.
It should be noted that the AWWARF survey reported distribution system THM data,
whereas results of this study represent clearwell effluent THM levels. THM levels from
the clearwell effluents of utilities employing chloramines as a final disinfectant may
reflect more closely the THM levels found in their distribution systems.
Comparison of USEPA Study and CDHS DBF Study Results
Figure 5-82 illustrates the levels of XDBP for the 25 utilities participating in the
USEPA study and the 10 utilities in the CDHS study, as well as for the combined 35
utilities. The four-quarter median level of XDBPsum was substantially lower for the
CDHS study, approximately 48 //g/L. compared to approximately 75 /vg/L for the
USEPA study, although the difference was not statistically significant at a 95 percent
confidence level. Additionally, the 75th percentile values from each study are
approximately equal, as are the ranges.
5-24
-------
J
?
c
j
j
I
•H
L
AUUARF 12-Quart«r u». USEPA
4-Quart«r Maans
Total Halaganated DBPi
By Study
a
E
M
Q
X
1
_L
_1_
727
,,-
"
n=
u=
EPA
100
25
40
Study
FIGURE 5-81
10
Study
FIGURE 5-82
ALL
140
35
-------
Section 6
Treatment Modification
Studies • Results and
Discussion
-------
SECTION 6
TREATMENT MODIFICATION STUDIES - RESULTS AND DISCUSSION
In this section, results of the treatment modification studies will be presented and
discussed. A total of ten studies were performed for the combined USEPA and CDHS
DBF projects. Five of these studies focused on me use of ozonation as a method of
DBF control, two investigated coagulation for DBF precursor removal, two studies
evaluated chlorine dioxide for DBF control, and one involved the use of GAC to lower
concentrations of DBF precursors.
OZONATION STUDIES
Because of the increasing use of ozone in the United States for disinfection and control
of DBFs, five treatment modification studies focused on the use of this oxidant at
various water treatment plants. The utilities which participated in this project were
selected because ozonation processes were installed at their treatment plant or because
they were able to provide a pilot plant which could evaluate ozonation. Table 6-1
shows the type of source water and various disinfection schemes employed at each
utility in order to evaluate the effect of ozone on the formation of DBFs.
Utility 6
Utility 6 operates a plant in the eastern United States which had a capacity of 52 mgd
at the time of the study. Sampling at this plant was conducted before and after ozone
was incorporated into the treatment process, which is presented schematically in Figure
6-1. As a conventional treatment plant, chlorine was added in four places: to the raw
water, before flocculation, and before and after filtration. Ammonia was also added
after filtration to convert the free chlorine to chloramines for residual disinfection. As
the schematic shows, there was a change to direct filtration when ozonation was
installed, the flocculation and sedimentation basins having been converted to
flotation/skimming tanks. As a result of changing the plant configuration and
implementing ozone, the utility was able to employ lower doses of chlorine for shorter
contact times. Under this configuration, chlorine was applied before and after
filtration, ammonia being added before the second chlorination point to produce
chloramines for residual disinfection. Therefore, free chlorine contact time was limited
to that required to prevent biological growth in the filters.
Sampling locations included distribution system Location 3, which corresponds to a
residence time of 7 hours. Because of delays in the switch to ozone and changes in
water quality, the "before" and "after" samples (before and after the implementation of
preozonation) were taken approximately four to six months apart. The first "after"
sample was collected on March 13, 1989. However, as indicated in Table 6-2, the
temperature at Location 3 was 7°C lower as compared to the November 21, 1988
conventional treatment sampling at that location. The raw water TOC concentration
was also lower in March compared to November, Consequently, a second ozonated
"after" sample was collected on May 15, 1989 when the distribution system water
temperature was comparable to that of the November sampling. Additionally, Table
6-2 shows that the samples collected on this sampling date and on November 21, 1988,
6-1
-------
TABLE 6-1
DISINFECTION SCENARIOS, PLANT SCALES AND SOURCES
FOR UTILITIES PARTICIPATING IN STUDY
Source
Utility Scale Water
Disinfection Scenarios
Abbreviations
19
25
Full Reservoir
Full/
Pilot Reservoir
Full Flowing
Stream
Full
Lake
1) Prechlorination. postammoniation C12. NH3
2) Preozonation. postchlorination
postammoniation O3. CI2, N
I) Pre- and postchlorination (full scale) CI2
2) Prechlorination, postammoniation C12, NH3
(full scale)
3) Preozonation. postchloramination O3, NH2C1
(pilot scale)
1) Pre- and post chlorination CI2
2) Preozonation, postchlorination O3, Cl,
1) Prechloramination only NH2C1
2) Pre- and ppstozonation, post
chloramination O3, NH:CI
3b Pilot Lake
1)
2)
3)
4)
5)
Prechlorination only
Prechloramination only
Preozonation. prechlorination
Preozonation. prechloramination
Preozonation with hydrogen peroxide
addition, prechloramination
CI2
NH2CI
03, CI2
03. NH,CI
03. H2d2.
NH:CI
Note: The prefix "pre-" in disinfection scenarios denotes addition prior to the rapid mix. The
prefix "post-" denotes addition directly before or after filtration.
-------
OZONE
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
PRECHLORINATION / POSTAMMONIATION
Chlorine
t
Chlorine
Alum
Polymer
Chlorine
Sodium Hydroxide
OZONE/CHLORINE / CHLORAMINES
Chlorine
Alum
Pol \rner
000000
0000
0000000
Ammonia
Chlorine ^
Sodium Hydroxide
SCHEMATIC OF UTILITY 6 TREATMENT PROCESSES
FIGURE 6-1
-------
TABLE 6-2
UTILITY 6 TREATMENT STUDY
Water Quality Parameters
TOC
(mg/L)
CONVENTIONAL (11/21/88
Raw Water
Location 3
4.25
2.84
UV
Absorbance
at 254 nm Chloride
(cm1)
DBF Sampling)
0.178
0.102
(mg/L)
43
NA
Bromide
(mg/L)
0.05
NA
TOX
0>g/L]
NA
240
pH
1
7
7
.8
.6
Free
Chlorine
Residual
(mg/L)
NA
NA
Total
Chlorine
Residual
(mg/L)
NA
>2.0
Temp.
(°C)
9
13
OZONE (5/15/89 DBF Sampling)
Raw
Location 3
4.28
3.60
OZONE (3/13/89 Aldehyde
Raw Water
Ozone Contactor
Effluent (OCE)
Filler Influent (FI)
Filter Effluent (FE)
Plant Effluent (PE)
Location 1 (LI)
Location 2 (L2)
Location 3 (L3)
NA = Not Analyzed
ND = Not Detected
Note: I. All compa
3.75
3.72
3.64
3.18
3.20
3.18
3.23
3.20
0.144
0.059
Profile)
0.137
0.093
0.092
0.070
0.069
0.071
0.072
0.070
risons of conventional
66
NA
66
NA
NA
NA
NA
NA
NA
NA
0.06
NA
0.07
NA
NA
NA
NA
NA
NA
NA
NA
100
NA
NA
60*
76*
80*
82
76
89
versus ozone treatment are
7
7
7
8
7
8
8
8
8
8
.5
.8
.7
.0
.6
.0
.0
.1
.1
.2
made
NA
ND
NA
NA
0.5
0.1
NA
NA
NA
NA
with
NA
2.0
NA
NA
1.0
0.5
1.0
0.7
0.8
0.4
14
15
6
5
5
5
5
7
6
6
samples collected on 11/21/88 and 5/15/89. repectively. The data on
these two sampling dates represent the water quality parameters associated
with Figures 6-2 to 6-9.
2. Locations represent distribution system sampling points in order of
increasing residence time.
*These TOX samples were dechlorinated and preserved in the field.
-------
Treatment Modification Results and Discussion
had similar raw water TOC and pH. Consequently, all comparison of "before" and
"after" oronation samples were made using data from these two sampling dates. Table
6-2 shows that the total chlorine residual at Location 3 with preozonation (May 15,
1989 sampling) was 2.0 mg/L. Before ozone was implemented, the measured
distribution system residual was greater than 2.0 mg/L; historical records showed that
the residual at this point was usually 2.8 mg/L. The table also indicates that TOX
decreased by 58 percent with the preozonation disinfection scheme, despite the higher
concentration of TOC measured in the distribution system. However, it should be
noted that less free chlorine and a shorter contact time were employed when
preozonation was utilized.
The effects of ozonation on DBFs measured at Location 3 is presented in Figures 6-2 to
6-9. As shown in Figures 6-2 through 6-4, a reduction of 56 to 66 percent was
observed for all XDBP classes except HKs, which increased slightly from 3.6 to 4.4
j/g/L. Chloral hydrate, chloropicrin and cyanogen chloride decreased slightly. Figures
6-5. 6-6 and 6-7 show that most of the individual THMs, HAAs. and HANs decreased
in concentration. Because of the low bromide levels in the raw water (0.05 to 0.06
mg/L), there were no observed increases in concentrations of brominated DBFs. The
only individual DBF compound to increase in concentration was l.l-DCP (1.4 to 2.8
*/g/L). as shown in Figure 6-8. Concentrations of the miscellaneous compounds are
illustrated in Figure 6-9.
The effect of water temperature on formation of DBFs after ozonation was implemented
is shown in Figure 6-10. Samples were collected when the temperature was 6°C and
approximately 2 months later when the temperature had increased to I5°C. The
residence time at the distribution sampling point was 7 hours. The data show that DBF
class totals increased only slightly with the 9 degree increase in water temperature.
From this study, it is not possible to directly attribute the reduction in most XDBPs to
ozone since lower doses of chlorine and shorter contact times were employed. It
cannot be determined if ozone caused a decrease or modification of DBF precursor
material or if lower DBF levels can be attributed solely to a decreased use of chlorine.
Examination of Figure 6-6 shows that the use of preozonation affected individual HAAs
differently (50 percent reduction in DCAA versus a 78 percent decrease in TCAA). For
this study, the following order of increasing effect of preozonation on controlling the
formation of these DBFs was:
I, I, I-TCP < DCAA < TOXj< CHC13 < DCAN < TCAA
According to bench-scale research on preozonation/postchlorination treatment of fulvic
acids by Reckhow and Singer (1985), the following order of increasing precursor
destruction for nearly all of the investigated doses were:
1.1,1-TCP < DCAA
-------
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY—PRODUCT FORMATION AT UTILITY 6
FIGURE 6-2
OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY —PRODUCT FORMATION AT UTILITY 6
/•g/L
FIGURE 6-3
-------
c
0)
U
a
m
a
Effect Of CI2.NH2CI and O3.CI2.NH2CI
on DBP Formation (Utility 6)
15O
100
50
o
XDBPsum THM
CI2.MH2CI
(1 V/21/88J
O3.Q2MH2CI
(5/15/69)
Dist. Sys. Residence Time (t)
t = 7 nrs
* TCAN Mow
»* IwdAA twlow t
NA
HAN
HK
HAA
40
ALD
DBP Class
FIGURE 6-4
Effect Of CI2.NH2CI and O3.CI2.NH2C
on THM Formation (Utility 6)
0
U
30
20
10
CI2.NH2CI
(1 1/21/88)
O3.CI2MH2CI
(5/ 1 5/89)
Dist. Sys. Residence T.me (t)
t = 7 nrs
'"HC.I j
CHBfQ2
CHBr2Cl
CHBr3
1 HM Compound
FIGURE 6-5
-------
O)
o
c
-------
oi
ro
i.
c
O
u
4.O
3.O
2.0
1.0
O.O
Effect Of CI2.NH2CI and O3.CI2.NH2CI
on HK Formation (Utility 6)
1.1-DCP
CI2JSH2CI
(1 1/21/88)
03.02^20
(5/15/89)
Dist. Sys. Residence Time (tl
t= 7 trs
1.1.1-TCP
HK Compound
FIGURE 6-8
S1
O
U
20
15
10
Effect of CI2.NH2CI and O3.CI2.NH2CI
On Misc. DBP Formation (Utility 6)
CUP
» B«low MRL
CI2^H2CI
11 1/21/88)
O3.CI2AH2CI
(5/15/89)
Disl System Residence Tim* (t)
t= 7 frs
CH
CNCI
TCP
C'BP Compound
FIGURE 6-9
-------
g
•*-*
rtJ
(J
Q.
Effect Of Seasonal Temperatures
on DBP Formation (Utility 6)
150
100
50
o
03.CI2MH2CI
3/13/89 6 C
O3.CI2M-I2CI
5/15/89 IS C
Oist. Sys.
t = 7 hra
MA - Not Analyzed
Tinr»
-------
Treatment Modification Results and Discussion
Stability of Aldehydes Through the Plant and Distribution System. The effect of
ozone on the formation of aldehydes is presented in Figure 6-11. No formaldehyde or
acetaldehyde was detected in the plant influent. Immediately after ozonation, though,
concentrations of formaldehyde and acetaldehyde increased to approximately 15 and 5
//g/L. respectively. The plant and distribution system profile shows that these
compounds were stable after formation. As noted above, chlorine is added in the
treatment process before filtration; a 0.5 mg/L free residual (1.0 mg/L total residual)
was detected in the filter influent. As will be discussed in more detail below, the
maintenance of a disinfectant residual through the filter most likely precluded the
aldehyde-removal capabilities of the filter.
Utility 7
Three different treatment scenarios were studied at Utility 7 (Figure 6-12), which is a
400-mgd conventional treatment plant. Chlorine-only and prechlorination/
postammoniation options were studied at full scale. A chlorine dose of 2.3 mg/L was
applied at the plant influent. Additional chlorine was dosed at the filter influent: I.I
mg/L during the chlorine-only test and 0.6 mg/L during the postammoniation option.
Chlorine-only DBF samples were collected from the filter effluent just prior to
postammoniation; the prechlorination/postammoniation samples were taken just after
ammonia (0.49 mg/L as nitrogen) was added to the filter effluent. There were
approximately 4 hours of free chlorine contact time under this scenario. In order to
study preozonation at this utility, a 6-gpm pilot plant was employed. The pilot plant
followed similar treatment processes as the full-scale plant but included ozone, which
was applied at a dose of 2.0 mg/L. The pilot filter effluent was chloraminated (0.5
mg/L ammonia as nitrogen. 1.5 mg/L chlorine) with no free chlorine contact time.
Samples for each scenario were collected after 2 and 24 hours using the simulated
distribution system (SDS) protocol. The SDS test was described in detail in Section 3
of this report. In these tests, samples were dosed with disinfectants and held under
conditions representative of Utility 7's distribution system to give an estimate of the
levels of DBFs that can be formed under realistic environmental conditions (Koch et al.,
1989). For the SDS tests performed at Utility 7, samples were incubated at 25°C for 2
and 24 hours to simulate samples from the clearwell effluent and the distribution
system, respectively. The preozonated pilot-plant sample was buffered to pH 8.2 to
8.3. The chloraminated filter effluent sample had received final pH adjustment in the
treatment plant. However, the chlorinated filter effluent sample was collected at a
point in the plant prior to final pH adjustment, so the SDS sample was raised to a pH
of 8.2 by the addition of a 1-percent sodium hydroxide solution. The latter two
samples had sufficient chlorine or chloramine levels for the SDS testing. The
preozonated pilot plant effluent required a 1.5 mg/L chloramine dose, with ammonia
added prior to chlorine dosing.
The chlorine residuals at the sampling locations, as well as other water quality
parameters, are shown in Table 6-3. The data indicate that the pH and temperature at
the sampling points for all the disinfection schemes were similar. The pilot plant
received the same raw water as the full-scale plant, so the observed difference in
treated-water TOC levels was probably due to a slightly greater TOC removal during
the ozone-treatment trial.
6-3
-------
OZONE RAPID MIX FLOCCULATION SEDIMENTATION
FILTRATION
CHLORINE ONLY
Polymer
Chlorine
T
Chlorine
Alum
Polymer
PRECHLORINATION/ POSTAMMONATION ONLY
Polymer
Chlorine
T
Chlorine
Alum
Polymer
Ammonia —
Sodium Hydroxide
OZONE / CHLORAMINES
Alum
Polymer
Polymer
000000
0000
0000000
oomocnoo
.
o
o
s
V
k
•s,
Ammonia
Chlorine
SCHEMATIC OF UTILITY 7 TREATMENT PROCESSES
FIGURE 6-12
-------
TABLE 6-3
UTILITY 7 TREATMENT STUDY
Water Quality Parameters
TOC
-------
Treatment Modification Results and Discussion
Effects of ozonation on DBFs at Utility 7 after the 24-hour SDS tests are shown in
Figures 6-13 and 6-14. Preozonation followed by concurrent addition of ammonia and
chlorine after filtration decreased the levels of THMs, HAAs, HANs, and chloral
hydrate as compared to chlorination-only or prechlorination/postammoniation. Very
little difference in concentrations of HKs, cyanogen chloride or chloropicrin was
observed between any of the treatments.
Aldehydes were not analyzed at the time of the study. However, subsequent pilot
testing at Utility 7 indicated formation of aldehydes in the ozone contactors, the levels
of which (1) remained the same through the filtration step when chlpramines were
applied upstream of the filters, and (2) decreased through the filtration step when
chloramines were not added until after filtration (Montgomery, 1989), as shown in
Figures 6-15 and 6-16 for formaldehyde. Since the ozone residual is short-lived, when
no chlorine or chloramines were added prior to filtration, it is probable that biological
activity had devejoped on the filter media; the data suggest that this activity was
capable of removing biodegradable material produced by ozonation, such as the two
aldehydes studied here. This is consistent with other studies which have shown that
biologically operated filters can decrease materials such as assimilable organic carbon
(Van der Kooij et al., 1982; Montgomery, 1989).
The use of filtration with biological activity is typical of many European ozone plants;
however, it is atypical in water treatment plants in the United States. As the plant
schematics for the other ozonating utilities in this study indicate (see Figures 6-1, 6-23,
6-32 and 6-46), chlorine or chloramines were added after the ozone contactor and
before the filters, thus probably minimizing microbiological growths in the filters. An
important finding of the ozonation studies conducted for this project is the large
increases observed in aldehyde formation whenever ozonation is employed followed by
secondary disinfection prior to filtration (see Figures 6-33 and 6-47). The placement of
secondary disinfection with respect to filtration requires further study in order to
minimize aldehyde levels in the finished waters of ozonating plants.
Table 6-3 shows that 24-hour SDS TOX concentrations decreased as free chlorine
contact time decreased in each successive treatment train, the lowest (38 /wg/L) being
observed when ozonation/chloramination (with no free chlorine contact time) was
tested.
Figures 6-17 to 6-22 present the DBF data for the treatment modification study as a
function of contact time. Results from SDS tests at 2 and 24 hours are shown. Under
the chlorine-only scenario, all DBFs increased with increasing residence time except for
the haloketones and chloropicrin. Under the prechlorination/postammoniation scheme,
all DBFs were relatively stable between the 2-hour and 24-hour tests, except for
1.1,1-TCP which decreased and cyanogen chloride which increased. For the
preozonation/postchloramination treatment scenario, most DBFs remained at or below I
to 2 //g/L each, except for cyanogen chloride, which was detected at 2.3 to 3.4 vg/L.
Utility 19
This utility, which was studied at full scale (Figure 6-23), is a 600-mgd
preozonation/direct filtration facility. During the trial when only chlorine was
employed, ozone was taken offline and pre- and post-chlorine doses were applied to
6-4
-------
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY-PRODUCT FORMATION AT UTILITY
FIGURE 6-13
•Includes -1 .'lours of frw? i-hloi ii:.
contact t Line l>etorcj ijostJJinnm it
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY —PRODUCT FORMATION AT UTILITY 7
*g/L
*Fncludes 4 hours of free chlorine
contact tune before [jostonrnonuit ion
FIGURE 6-14
-------
3
9
LU
Q
QC
O
u.
25
20
15
1O
CHANGE IN FORMALDEHYDE CONCENTRATION
THRU PROCESS TRAIN. STATE PROJECT WATER
Reference: Montgomery. 1989
O
GUIDE TO FIGURES:
PIA = plant influent
OE3 = ozone contactor effluent
. PI = Mlsr influent
FE - filter effluent
Disinfectants
Upstream of Filters
+ 1- 2 mg/L O3.
no NH2C.I
A A 2 mg/L O3.
1.Smart- N-I2CI
O O2 moA. 03.
PIA
OE3
PL AM I LOCATION
FIGURE 6-15
LU
Q
a
O
LL
CHANGE IN FORMALDEHYDE CONCENTRATION
THRU PROCESS TRAM COLORADO RIVER WATER
Reference: Montgomery, 1989
GUIDE TO FIGURES:
PIA = plant influent
O63 • ozone contactor effluent
Fl = filter riflueni
FE = tiller effluent
Disinfectants
Upstream of Filters
+ 1- 2 man. O3.
no NH2CI
10 -
5 -
O
o-
- A 2 mo/L O3.
1.5moA_ N-CCI
-O2 mr>L O3.
1.5mp/l- N-GCI
PIA
OE3
PLANT LOCATION
FIGURE 6-16
-------
Effect Of CI2. CI2/NH3 & O3/NH3/CL2
on DBP Formation (Utility 7)
"bi
o
HM Compound
1 2
CH&3
RGURE6-18
-------
Effect Of CI2. CI2/NH3 & O3/NH3/CI2
on HAN Formation (Uti ity 7)
0)
o
o
U
O
CI2 only
rTTI O3/N-I3/CI2
Dist. Sys Residence Time (t)
1: t = 2 hrs
2: t = 24 hrs
» = Below MRL
1 2
TCAN
1 2
DC AN
SCAN
HAN Compound
RGURE6-19
Ol
o
o
U
1.5
1.0
Effect Of CI2. CI2/NH3 & O3/NH3/CI2
on HK Formation (Utility 7)
O.5
o.c
Dist. Sys. Residence Time (T)
1: t = 2 hrs
21 t = 24 hrs
I.I DCP
HK Compound
FIGURE 6-20
-------
Effect Ot CI2. CI2/NH3 & O3/NH3/CI2
on HAA Formation (Utility 7)
1^
1O
3
i 8
*-*
(D
t 6
O
S
U 4
<
<
2
,%
_
~
|!
g |/ fte
g|/ ^J
;
j
^
J3
x
r
^
•'
1 »
12 12
MCAA DCAA
B??3 CI2 only
^ C.2/^3
EZ] 03^3/02
Oist. Sys. Residence Tune (0
I
'•
;
1
'/
\
\
^
'
;
f
j
\
/i ^
3_l
1: t = 2 nrs
2: t = 24 hrs p
%
^ ra^ |3,
d_ ^^1 ^fe
|
|
!
,
3 *
-T|
3
2 12 12
TCAA MBAA DBAA
HAA Compound
FIGURE 6-21
u
5
u
Effect Of CI2. CI2/NH3 & O3/NH3/CI2
on Misc. DBP Formation (Utility 7)
12
10
7
2 -
EZZ3 Q2 any
IB CI2/M-O
DISI. Sy&. Rosidenca
1: t » 2 Hr»
2: t » 24 IYS
* - Below IvAL
1 2
CHP
FIGURE 6-22
-------
OZONE
RAPID MEX FLOCCULATION
FILTRATION
CHLORINE ONLY
T
Ferric Chloride
Polymer
Chlorine
Chlorine
1
OZONE / CHLORINE
Ferric Chloride
Polymer
000000
0000
0000000
000000000
.
o
o
Chlorine -
Intermittent'
Chlorine
SCHEMATIC OF UTILITY 19 TREATMENT PROCESSES
FIGURE 6-23
* Chlorine was not added at this point on day of sampling.
-------
Treatment Modification Results and Discussion
100 percent Los Angeles Aqueduct (LAA) water for five days with doses on the day of
sampling of 1.8 and 0.3 mg/L, respectively. On the fifth day, samples were collected
at predetermined points: at the clearwell effluent and in the distribution system at
residence times of 4.3 and 11 hours. When preozonation was reinstated,
prechlorination was stopped and an ozone dose of 1.3 mg/L was applied; a post-
chlorine dose of 1.0 mg/L was used. However, during this ozone/chlorine treatment,
the treatment plant was using 78 percent LAA water/22 percent State Project water
(SPW). Therefore the ozone/chlorine treatment was repeated when the plant was
utilizing 100 percent LAA water. At that time, an ozone dose of 1.7 mg/L was applied
and a 1.5 mg/L post-chlorine dose was used. The chlorine residuals at the sampling
locations, as well as other water quality parameters, are shown in Table 6-4. Free and
total chlorine residuals and TOC concentrations were similar in the "before" and
"after" samplings when 100 percent LAA water was treated. The temperature for the
chlorine-only samples was slightly lower. The blended water had a higher TOC level,
since SPW has a higher TOC than LAA water.
Figures 6-24 and 6-25 present concentrations of DBP classes and the miscellaneous
DBPs. All data are presented after 11 hours of residence time in the distribution
system and represent only the 100 percent LAA water samplings. As shown in Figure
6-24. decreases of 13 yug/L and 8.7 /yg/L were observed for TTHMs and HAAs,
respectively, after ozone implementation with subsequent chlorination. A 2.3 j/g/L
increase in chloral hydrate was observed. The effect of ozonation on HANs, HKs and
chloropicrin formation is shown in Figure 6-25. After ozone addition, the
concentration of HANs was reduced by 1.2 /ug/L, while small increases were observed
for HKs and chloropicrin. Cyanogen chloride analysis was not performed during the
ozonation trial; it was only detected at levels equal to or slightly above the MDL during
the chlorine-only experiment. Changes in TOX concentrations before and after
ozonation are presented in Table 6-4. For the 100 percent LAA water studies at the
11-hour residence time samplings, a decrease in TOX of 62 //g/L was observed after
ozonation was applied.
Figures 6-26 to 6-31 present the data for the treatment modification study in relation to
residence time in the distribution system. Levels of XDBP classes and individual DBPs
are shown at the three different distribution system residence times: 0 hours (clearwell
effluent), and 4.3 and 11 hours. In addition, ozone/chlorine data are presented when
the treatment plant was using 100 percent LAA water versus 78 percent LAA water/22
percent SPW. In general, the data show that all DBPs increased with increasing
residence time except for 1,1 -DCP and cyanogen chloride.
Figure 6-26 shows that when only LAA water was used, decreases were observed in the
DBP class totals for THMs, HAAs and HANs with ozone implementation. However, as
shown in Figures 6-27 to 6-29, the di- and tri-brominated species of these DBP classes
increased with ozonation. Although the level of total HAAs decreased with ozone
implementation by approximately 50 percent at the 11-hour residence time sample
point, the level of DBAA increased by more than 100 percent. At Utility 19, bromide
was not assayed at the time of the study; however, 0.04 mg/L was detected in a
subsequent sampling of the same raw water source.
The change in source water from 100 percent LAA water to a blend with SPW further
shows the effect of bromide on concentrations of DBPs produced by ozonation followed
6-5
-------
TABLE 6-4
UTILITY 19 TREATMENT STUDY
Water Quality Parameters
Pre-
Ozone
TOC Dose
(mg/L) (mg/L)
Cleanvel] Eflluenl
O3/CI,M 1.50 1.3
oycu" 1.22 1.7
Cl;h 1.20 0.0
Distr. PI. 1
( 4.25 hours )
O3/CI:H 1.48 1.3
O,/CI,h 1.20 1.7
Cl,'1 1 .25 0.0
Distr. Pt. 2
( 1 1 hours )
O3/Cl:a 1.56 1.3
O3/Cl,h 1.21 1.7
Cl,h 1.28 0.0
Chlorine Free Total
Dose Chlorine Chlorine pH
Pre, Post TOX Residual Residual Temp.
(mg/L) 0/g/L) (mg/L) (mg/L) TO
0.0. 1.0 73 0.7 0.9 8.1 17.5
0.0. 1.5 68 1.1 1.2 8.1 18
1.8.0.3 140 1.0 .15 8.1 15.2
0.0. 1.0 100 0.68 0.71 7.7 16.8
0.0. 1.5 79 1.0 1.0 8.1 19.0
1.8.0.3 140 0.92 1.00 7.9 15.0
0.0, 1.0 92 0.55 0.68 7.8 16.8
0.0, 1.5 88 0.52 0.54 8.2 19.5
1.8,0.3 150 0.63 0.68 7.9 15.5
"Source water was 78 percent LA Aqueduct water and 22 percent State Project Water.
''Source water was 100 percent LA Aqueduct water.
-------
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY —PRODUCT FORMATION AT UTILITY 1 9
«g/L
FIGURE 6-24
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY—PRODUCT FORMATION AT UTILITY 19
-fe
ug/L
NA: Not Analyzed
FIGURE 6-25
-------
7O
s
4—•
w.
•*-•
§
u
Q
60 -
5O
4O
30
20
10
0
Effect of O3/CI2 & CI2 Only on
DBP Formation (Utility 19)
CI2. 1OO% LAA
O3/CI2.1OOV. LAA
P773 03/CL2. 78% LAA. 22* SPW
Dist. Sys. Residence Time (t)
1:1 = 0. Cleafwell Eff.
2: t = 4.3 ITS
3: t = 11 hrs
L nrv
1 2 3
XDBPsum
1 2 3
THM
1 2 3
HAN
DBP Class
FIGURE 6-26
1 2 3
HK
1 2 3
HAA
_J
3
g
«-<
10
I
U
h-
20
15
10
O
Effect of O3/CI2 & CI2 Only on
THM Formation (Utility 19)
1 2
CHO3
CI2 . 1OO% LAA
O3/CI2. 1OO% LAA
fTTTI O3/CI2 . 78% LAA. 22% SPW
Dist- Sys. Ftesidence Time (t)
1: t = 0. C1««rw«ll Bit.
2: t = 4.3 tra
3: t = 1 1 rrs
1 2 3
CHClBr2
1 2 3
CHBri',12
11 1M Compound
FIGURE 6-27
2 3
CHBr3
-------
9
g
4-*
ID
O
<
<
10
a
o
Effect of O3/CI2 & CI2 Only on
HAA Formation (Utility 19)
CI2. 1OO% LAA
O3/CI2. 1OO% LAA
O3/C.I2 . 78% LAA 22% SPW
Dist Sys. Residence Time ttJ
1 I = O. Ciearwell Eff.
2: t = 4.3 hrs
3: I = 1 1 hrs
1 2 3
MCAA
HAA Compound
FIGURE 6-28
s
*-^
w
4-»
§
<
2.5
2.0
1.5
1.O
0.5
O.O
Effect of O3/CI2 & CI2 Only on
HAN Formation (Utility 19)
CI2 . 100% LAA
O3/CI2 . lOOH LAA
O3/CI2. 76% LAA 22% SPW
Dist. Sys. Residence Tune (t)
1: t =0. Claarwaii Erf
2: t = 4.3 hrs
3: t = 1 1 hrs
* = Below K/Ftt.
123 123 123
TCAN DCAN BCAN
HAN Compound
FIGURE 6-29
1 2 3
DBAN
-------
(J
O.5 -
0.0
Effect of O3/CI2 & CL2 Only on
HK Formation (Utility 19)
CI2. 10OK LAA
O3/CI2. 10OS LAA
O3/CI2. 78% LAA 22% SPW
Dial. Sys. RMKtonca Tim* (t)
1: t = O. Cl«w«»ali Eft
2: t = 4.3 hrs
3: t = 1 1
1.1.1 ~ TCP
HK Compound
FIGURE 6-30
I
ro
u
10
8
0
Effect of O3/CI2 & CI2 Only on
Misc. DBP Formation (Utility 19)
(%%%! CI2. 10CM LAA
^B O3/CI2. IOO% LAA
(7771 Q3/C12. 76% LAA 22S SPW
Oi»t Sy*. RMKfenca Tim* (1)
1: t « 0. Cl««rw4ll Eft.
2: t * A3 fr»
3: t • 11 rvt
* ~ BttlOW iLifil
o - Not Analyzed
-
* « «A..
1
^Bv
^
!
>
^
^
/
\
'/
y,
/.
I
I
V
Ea°ra im0r7i «~°^,
1 2
CHP
1 2 3
CH
DBP Compound
FIGURE 6-31
1 2
CMC I
-------
Treatment Modification Results and Discussion
by chlorination. A comparison of the two shows that when SPW was employed, a shift
to increasing concentration of the brominated species occurred. Although bromide was
not measured during this treatment scenario, a subsequent sample of a blend of 74
percent LAA water and 26 percent SPW showed that the bromide level had increased to
0.1 mg/L. These data are consistent with those discussed in Section 5 regarding Utility
12. where seasonal shifts in bromide concentrations promoted increases in some
brominated DBFs. In addition, ozone can react with bromide ions in the raw water,
causing the formation of hypobromous acid (HOBr) (Dore et al., 1988). Reactions of
HOBr with natural organic material can produce bromoform and other brominated
DBFs, as evidenced at Utility 19.
Utility 25
Sampling at Utility 25 was conducted at full scale at a conventional treatment plant
with a capacity of 90 mgd. A schematic of the plant before and after the
implementation of ozonation is presented in Figure 6-32. Samples were collected
before the plant went online with two-stage ozonation. During this period, concurrent
addition of ammonia (1.6 mg/L) and chlorine (8.0 mg/L) was practiced at the rapid
mix; there were no other points of disinfectant addition. Sampling of ozonated water
was conducted seven days after the plant switched to routine operation of ozonation to
insure that the distribution system was well-flushed with ozonated water. The oxidant
was applied to both raw and settled water at 4 mg/L per stage, and chlorine (5.0 mg/L)
and ammonia (1.0 mg/L) were added concurrently prior to filtration. Both before and
after ozone implementation, samples were collected at various points in the plant and
distribution system. Water quality parameters of the raw water and at the various
sampling points are shown in Table 6-5.
The effects of the treatment modification on concentrations of DBFs for this utility are
shown in Figures 6-33 and 6-34. In these plots, data from Location 4 in the
distribution system (residence time of 18 to 20 hours) are employed. The DBF classes
with the highest concentrations when chloramines-only were employed were TTHMs
and HAAs. It should be noted, though, that the total chlorine dose and residual
concentrations were higher during this sampling than during the ozone study. Despite
concurrent addition of chlorine and ammonia, Figure 6-33 shows that under the
chloramines-only scheme, the TTHM and HAA concentrations were relatively high,
both being 44 //g/L. This is in contrast to the study at Utility 36 which showed that
the chloramines-only scenario was very effective in limiting the formation of these
compounds. The relatively high levels of these DBF classes at Utility 25 were probably
a result of some free chlorine contact time due to inadequate mixing and/or the high
raw water TOC concentration (7.7 mg/L).
Concentrations of TTHMs and HAAs were reduced to 6.7 //g/L and 15 ,t/g/L,
respectively, after the treatment modification was implemented. Concentrations of the
measured aldehydes increased from 10 to 57 /yg/L with the switch to ozone. Figure
6-34 shows that the ozone/chloramines treatment slightly decreased the concentration of
chloral hydrate and HANs as compared to chloramines only; however, small increases
were observed for HKs, cyanogen chloride and chloropicrin. Finally, Table 6-5 shows
that TOX decreased by 89 /wg/L when ozone treatment was implemented.
6-6
-------
OZONE
RAPID MIX FLOCCULATION SEDIMENTATION OZONATION FILTRATION
CHLORAMINES ONLY
T
Polymer
Alum
Chlorine
Ammonia
Lime
1
OZONE / CHLORAMINES
Polymer
Alum
Chlorine
Ammonia
1
000000
0000
ooooooo
000000000
Lime
o
o
x
\
x
s,
^>V&£
•^:*\^."fr'\
]
i
000000
0000
ooooooo
000000000
I
1
SCHEMATIC OF UTILITY 25 TREATMENT PROCESSES
FIGURE 6-32
-------
TABLE 6-5
UTILITY 25 TREATMENT STUDY
Water Quality Parameters
uv
Absorbance
TOC at 254 nm
(mg/L) (cm1)
CONVENTIONAL
Raw
Filter Influent1
Filter Effluent1
Clearwell Effluent
Location 1
Location 2
Location 3
Location 4
OZONE
Raw
2nd O, Contactor Inf.
Filter Influent
Filler Effluent
Clearwetl Effluent
Location 1
Location 2
Location 3
Location 4
7.72
4.63
4.63
4.85
4.67
4.72
4.69
4.76
8.01
5.47
5.25
5.06
5.23
5.65
5.48
5.44
5.51
0.255
0.096
0.081
0.095
0.087
0.091
0.086
0.088
0.233
0.054
0.067
0.041
0.046
0.050
0.049
0.053
0.047
Chloride
(mg/L)
34
NA
NA
NA
NA
NA
NA
NA
34
NA
NA
NA
NA
NA
NA
NA
NA
Bromide
(mg/L)
0.22
NA
NA
NA
NA
NA
NA
NA
0.23
ND
ND
ND
ND
ND
ND
ND
ND
TOX
U/g/Lt
20
180
200
170
150
140
150
150
NA
29
57
65
67
64
56
64
61
Free
Chlorine
pH Residual
(mg/L)
8.3
7.3
8.1
9.0
9.0
8.9
8.9
9.0
NA
9.8
9.2
9.2
9.4
9.2
9.3
9.4
9.3
NA
NA
NA
NA
O.I
0.1
O.I
O.I
NA
NA
NA
NA
NA
NA
NA
NA
NA
Total
Chlorine
Residual
(mg/L)
NA
4.0
4.0
3.5
2.9
3.2
3.5
3.5
NA
ND
2.7
2.7
2.2
1.8
1.2
1.3
I.I
'No lime added
NA - Not Analyzed
ND - Not Detected
Note: Temperature for all studies ranged from 26.0 to 30.0°C.
Locations represent distribution system sampling points in order of increasing residence
time.
-------
OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY—PRODUCT FORMATION AT UTILITY
FIGURE 6-33
EFFECT OF VARIOUS DISINFECTION SCHEMES Or J
DISINFECTION BY—PRODUCT FORMATION AT UTILITY 2
FIGURE 6-34
-------
Treatment Modification Results and Discussion
Stability of DBFs Through the Plant and Distribution System. Figures 6-35 to 6-45
show a profile of DBFs through the treatment plant and at four locations in the
distribution system. Explanations of abbreviations used in the figures and distribution
system residence times at the point of sampling are presented in Table 6-6. During the
sampling for conventional treatment, the plant temporarily went offline and lime was
not applied immediately after startup. The filter influent and filter effluent samples
were collected after the plant startup, so these samples had abnormally low pH values
(7.3 and 8.1. respectively, as compared to 8.9 to 9.0 at the distribution system
sampling points which represents water produced when lime addition was underway) as
shown in Table 6-5.
Figures 6-35 to 6-39 present the DBF profiles for XDBPsum, THMs, HAAs, DCAA and
chloropicrin. The data show that for both treatment scenarios, the DBFs were formed
immediately after chlorine and ammonia addition and remained stable through the plant
and into the distribution system. During the ozonation study, some low levels of DBFs
were detected in the second-stage ozone influent. It is believed that some minimal
level of chlorine or chloramines was added in the treatment train prior to this sampling
point.
DBF profiles of HANs. HKs, and chloral hydrate are depicted in Figures 6-40 to 6-42.
In each case, during the chloramines-only treatment scenario, higher concentrations
were found in the filter influent and filter effluent than in the clearwell effluent or
distribution system. As noted above, the pH values were 0.8 to 1.7 units lower for
these two sampling points during the chloramine-only study because of a temporary
discontinuation of lime addition. The data suggest that at the higher pH values these
compounds are unstable and break down to other final products. As discussed in
Section 5, other researchers have reported that HANs and HKs were reactive
intermediates rather than stable endproducts such as THMs and HAAs (Reckhow and
Singer, 1985). They also found that with increasing pH, DCAN and 1,1.1-TCP
underwent hydrolysis. Likewise, Miller and Uden (1983) found that chloral hydrate
decomposed at elevated pH. When ozonation was applied, smaller concentrations of
these compounds were detected, but similar trends in relation to stability were
observed. In each case, concentrations in high pH waters decreased with increasing
residence time.
The cyanogen chloride profile (Figure 6-43) indicates that this compound was unstable
in Utility 25's distribution system. In contrast, cyanogen chloride increased in Utility
6's distribution system (e.g., from 4.7 /ug/L at the clearwell effluent to 9.9 /ug/L at
location 3 during the chlorine/chloramines study). Both utilities have chloramines as
the final disinfectant; however, the pH of Utility 6's distribution system was 7.5 to 8.0.
These limited data imply that cyanogen chloride may be unstable at a pH of 9, as
evidenced at Utility 25.
Figures 6-44 and 6-45 show the profiles for the aldehyde sum and for formaldehyde
only, respectively. In the chloramines-only scenario, these compounds were formed
after disinfectant addition and remained relatively stable through the plant and
distribution system. In the ozone/chloramines treatment scenario, higher concentrations
of aldehydes were detected and a similar stability profile was observed. It should also
be noted that these data are consistent with those described above at Utility 6.
6-7
-------
TABLE 6-6
EXPLANATION OF ABRREVIATIONS USED IN UTILITY 25
DBF PROFILES AND RESIDENCE TIME OF DISTRIBUTION SYSTEM
SAMPLING POINTS
Abbreviation
Explanation
RAW
2OI
FI
FE
CE
LI
L2
L3
L4
Plant Influent raw water
Second stage ozonation influent
Filter Influent
Filter Effluent
Clearwell Effluent
Location 1 in Distribution System (4-5 hours residence time)
Location 2 in Distribution System (8-9 hours residence time)
Location 3 in Distribution System (9-10 hours residence time)
Location 4 in Distribution System (18-20 hours residence time)
-------
O)
g
ro
c
?
o
U
13
OJ
ro
I
0
to
htfects ol NH2CI and O3/NH2CI Treatment
c.in Halogenated DBP Formation, Utility 25
120
100 -
NA - Not analyzed
*•* = NO hme addition
RAW 2OI Fl FE CE L1 l_2 L3
Treatment & Residence Time
FIGURE 6-35
L4
o
U
m
o
Effects of NH2CI and O3/NH2CI Treatment
on Total THM Formation at Utility 25
70
60
50
4O
30
2O -
10 -
NH2CI
O3/NH2CI
NA = riot analyzed
*•* = No lime addition
•**•
RAW ' 2Ol
Treatment & Residence Time
FIGURE 6-36
-------
O)
D
g
ro
u
o
U
70
60
50
40
3O
20
10
O
Effects of NH2CI and O3/NH2CI Treatment
on HAA Formation at Utility 25
V777A NH2CI
O3/NH2CI
NA = Not analyzed
*-*- = No lime addition
* - Below MRL
NAN,
RAW ' 2OI
L2
Treatment & Residence Time
FIGURE 6-37
L3
L4
0)
U
-------
O
8
0
c
U
Effects of NH2CI and O3/NH2CI Treatment
on Chloropicnn Formation at Utility 25
0.6
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
0.0
NA = Not Analyzed
* = Below MRL
Treatment &• Residence Time
FIGURE 6-39
Oi
O
O
5
U
<
I
Effects of NH2CI and O3/NH2CI Treatment
on HAN Formation at Utility 25
NA = Not analyzed
*••* = No lime addition
CE L1 L2 L3
0.0
RAW 2OI Fl
Treatment &• Residence Time
FIGURE 6-40
-------
-------
(TJ
O
0
u
8.0
6.0
4.0
2.0
0.0
Effects of NH2CI and O3/NH2CI Treatment
on CNCI Formation at Utility 25
W77X NH2CI
BBB O3/NH2CI
NA = Not analyzed
» = Below N/I3L
** = No lime addition
-X--X-
RAW 2OI Fl
FE CE
L1
L2 L3
L4
Treatment & Residence Time
FIGURE 6-43
o
ffl
k.
*^
0)
o
o
u
Q
Effects of NH2CI and O3/NH2CI Treatment
on ALD Formation at Utility 25
100
90
80
7O
6O
50
4O
30
20
10
0
NH2CI
O3/MH2CI
NA = Not analyzed
* = Below MRL
•X--X- = No lime addition
= Formaldehyde not analyzed
RAW 2OI
Treatment & Residence Time
FIGURE 6-44
-------
O)
g
^->
ra
O
O
U
-------
Treatment Modification Results and Discussion
Moreover, since disinfectant was applied before filtration, there was probably no
opportunity for biological degradation of these compounds in the filters.
Utility 36
At Utility 36, a pilot plant with a flow rate of 5 gallons per minute was employed to
evaluate five different treatment scenarios on the formation of DBFs. Conventional
treatment was employed with options to add preozonation (2 mg/L), with and without
hydrogen peroxide addition (0.67 mg/L), and free chlorine (6.5 mg/L) or chloramines
(2,1 mg/L chlorine and 0.5 mg/L ammonia) as disinfectants in the rapid mix. The
various treatment scenarios are shown in Figure 6-46. Samples were collected from the
filter effluent and held for 24 hours at ambient temperature in order to simulate the
retention time in a distribution system. After the 24-hour incubation time, samples
were transferred to individual sample bottles containing the appropriate dechlorination
agents and preservatives. Various water quality parameters are shown in Table 6-7.
Table 6-8 shows the effect of ozonation on the individual DBF compounds produced
with chlorination in 24-hour samples. With ozonation, chloroform decreased from 42
to 37 //g/L. However, increases were observed for the brominated THMs, resulting in
an increase in TTHMs. The level of bromide measured in the raw water at Utility 36
was 320 fjg/L. Dichloroacetic acid was essentially the same level, with and without
preozonation, while trichloroacetic acid was reduced by 50 percent with preozonation.
The varying effect of ozone on HAA precursors is consistent with bench-scale
experiments performed by Reckhow and Singer (1985). Other notable changes were
observed for formaldehyde and acetaldehyde, which increased from 13 to 36 >ug/L and
II to 16 fjg/L, respectively, with ozonation. Additionally, chloral hydrate increased
from 19 to 28 //g/L. Figure 6-47 shows the effect of the five different disinfection
schemes on various DBF classes and chloral hydrate. The highest levels of TTHMs,
HAAs. HANs and chloral hydrate were observed with those schemes which employed
chlorine as a final disinfectant. By comparison, large decreases were observed under
any disinfection scheme which employed chloramines. For example, as compared to
the chlorine-only disinfection scenario, chloramines-only, ozone/chloramines. and
ozone/hydrogen peroxide/chloramines reduced TTHM levels by 96, 97 and 98 percent,
respectively.
When chloramines-only were applied, only 6.1 fjg/L of TTHMs were formed. This
result is in contrast to that found during the the study at Utility 25, where fairly high
levels of TTHMs (44 fjg/L) were formed when chloramines-only were employed.
Thus. Utility 36 data further support the probability that there was some free chlorine
contact time at Utility 25 although concurrent addition of ammonia and chlorine was
practiced.
Figure 6-47 shows that at Utility 36, chloramines only were as effective in reducing the
levels of chlorinated DBPs as ozone/chloramines or ozone/hydrogen
peroxide/chloramines. Similar findings are shown in Table 6-7 for TOX
concentrations. Unlike the study at Utility 25, where the chloramine dose was reduced
after preozonation. the chloramine doses and residuals at Utility 36 were similar
whether or not ozone was employed. Aldehydes were produced in greatest
concentrations under the disinfectant schemes which employed ozone. The level of the
6-8
-------
OZONE
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
CHLORINE ONLY
t
Chlorine
Ferric Sulfate
Polymer
I
O
O
— ^-
\
V
k,
x
^.
CHLORAMINES ONLY
*
Chlorine
Ammonia
Ferric Sulfate
Polymer
I
O
O
s.
X
\
\
^^
r — -
OZONE/CHLORINE
Chlorine
t
000000
0000
0000000
000000000
Ferric Sulfate
Polymer
J .
O
O
\
X
\
s
— ^-
SCHEMATIC OF UTILITY 36 TREATMENT PROCESSES
FIGURE 6-46
-------
OZONE
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
OZONE/CHLORAMINES
"1
000000
0000
0000000
000000000
*
Chlorine
O
O
s
\
\
s,
Ferric Sulfate
Polymer
OZONE AND PEROXIDE / CHLORAMINES
Hydrogen
Peroxide
000000
0000
0000000
000000000
t
Chlorine
O
O
Ferric Sulfate
Polymer
SCHEMATIC OF UTILITY 36 TREATMENT PROCESSES
FIGURE 6-46 (Continued)
-------
TABLE 6-7
UTILITY 36 TREATMENT STUDY
Water Quality Parameters
uv
Absorbance
TOC at 254 nm Chloride Bromide TOX
(mg/L) (cm1) (mg/L) (mg/L) *g/L)
Free Total
Chlorine Chlorine
pH Residual Residual
(mg/L) (mg/L)
INFLUENT (at 9 a.m.) 4.28
FILTER EFFLULENT
(Immediate)
C12
NH2CI
O3 and CU
O3,NH,CI
O3/H2O2.NH2CI
4.01
3.87
4.10
4.03
4.00
0.091
0.045
0.060
0.036
0.044
0.044
46
NA
NA
NA
NA
NA
0.32
NA
NA
NA
NA
NA
NA
300
67
280
59
81
8.24
7.63
7.55
7.72
NA
7.71
NA
2.9
NA
1.6
NA
NA
NA
3.4
1.4
2.1
1.3
1.3
FILTER EFFLUENT
:4 hours)
cu
NH,CU
O, and'CI,
OVNH,CI
O3/H,(52.NH,.CI
3.87
3.83
4.05
3.91
3.88
0.031
0.049
0.034
0.034
0.035
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
380
75
330
69
73
7.56
7.59
7.61
7.76
7.69
0.8
NA
0.2
NA
NA
1.4
1.0
0.3
0.8
0.8
NA - Not Analyzed
ND - Not Detected
Note: Temperature for all studies ranged from 24.8°C to 26.9°C.
-------
TABLE 6-8
DISINFECTION BY-PRODUCT CONCENTRATIONS
BEFORE AND AFTER OZONE ADDITION AT UTILITY 36
Disinfection By-Products
Chlorine
Only
0/g/L)
Ozone/
Chlorine
Trihalomethanes
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Total Trihalomethanes
Haloacetic Acids
Monochloroacetic Acid
Dichloroacetic Acid
Trichloroacetic Acid
Monobromoacetic Acid
Dibromoacetic Acid
Total Haloacetic Acids
Haloketones
I. I -Dichloropropanone
I.I.I -Trichloropropanone
Total Haloketones
Haloacetonitriles
Trichloroacetonitrile
Dichloroacetonitrile
Bromochloroacetonitrile
Dibromoacetonitrile
Total Haloacetonitriles
Aldehydes
Formaldehyde
Acetaldehyde
Total Aldehydes
Miscellaneous
Chloropicrin
Chloral Hydrate
Cyanogen Chloride
42
50
50
9.5
152
4.0
23
22
3.8
II
64
0.24
1.8
2.0
-------
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY —PRODUCT FORMATION AT UTILITY
EFFECT OF VARIOUS DISINFECTION SCHEMES ON
DISINFECTION BY-PRODUCT FORMATION AT UTILITY
FIGURE 6-48
-------
Treatment Modification Results and Discussion
measured aldehydes for ozone/chlorine was 52 //g/L as contrasted to 24 /ug/L and 9.1
A/g/L for chlorine-only and chloramines-only schemes, respectively.
Figure 6-48 shows the effect of the five different treatments on HKs, cyanogen chloride
and chloropicrin. The data indicate that formation of cyanogen chloride was most
influenced by the final disinfectant. For those disinfection schemes which employed
chloramines, cyanogen chloride levels were approximately 8 to 15 times greater than
those which employed chlorine as a final disinfectant. Additionally, preozonation
resulted in a higher cyanogen chloride level than that produced by chloramines alone.
Chloropicrin concentrations were approximately 2 to 4 times greater when the
disinfection scheme employed preozonation. Although these results were consistent
with those found by others (Becke et al., 1984; Hoigne and Bader, 1988), the levels of
chloropicrin detected were low, ranging from 0.22 to 1.07 //g/L. The cited studies
reported increases in chloropicrin in systems where preozonation was followed by
chlorine addition. At Utility 36, the highest increase in chloropicrin concentration was
observed in the trial which employed ozone followed by chloramination.
The use of hydrogen peroxide in conjunction with ozone has been shown to enhance
the efficiency of ozone in oxidizing some organic materials (Glaze et al., 1987; Aieta et
al.. 1988: Wallace et al., 1988). In this study, peroxide was added at a ratio of 0.3 to
1 (hydrogen peroxide to ozone) on a weight basis; chloramines were employed as the
final disinfectant. When compared to ozone/chloramine treatment, only small changes
were observed in DBF formation using the ozone/hydrogen peroxide combination.
Effects of Holding Time on DBFs. Figures 6-49 to 6-58 show the effect of holding
time on levels of DBFs which were measured in the pilot plant effluent (t=0; the
residual disinfectant contact time through the plant before filtration was approximately
one hour) and after 24 hours. The greatest increases in XDBPS(im or XDBP class total
concentrations over 24 hours were observed in the treatment scenarios which employed
free chlorine as a final disinfectant (Figures 6-49 to 6-53). Similar results were
observed for chloral hydrate (Figure 6-54). For chloropicrin, however, the
ozone/chloramines treatment scenario produced the greatest increase over a 24 hour
period (Figure 6-55). It should be noted, though, that the concentration even after the
24-hour holding time was low (1.07 jug/L).
As noted above, cyanogen chloride levels after 24 hours were 8 to 15 times greater
under treatment scenarios which employed chloramines as a final disinfectant. Figure
6-56 shows that levels of this compound increased over the holding time under the
ozone/chloramines scenario, while it decreased under those which employed chlorine as
a final disinfectant.
Figure 6-57 and 6-58 show that once aldehydes were formed, they remained stable over
24 hours. This is consistent with the data found at Utility 6. It should also be noted
that since a disinfectant residual was carried through the filter, there was probably no
significant biological activity to lower the aldehyde levels formed from the various
disinfectants.
6-9
-------
(0
t-
«-•
8
u
TJ
0)
Effects of Various Treatments on
Halogenated DBP Formation at Utility 36
400
3OO -
2OO -
G3S3 CL2
MH3.CL2
03.CL2
17773 O3JSH3.CL2
100 -
t=0 hrs t=24 hrs
Conventional Filter Media
Treatment & Residence Time
RGURE 5-49
5
«-•
(0
i
Effects of Various Treatments on
THM Formation at Utility 36
200
150 -
100 -
t-O hrs
t=24 hrs
Filter Media
Treatment. & FHesidence Time
FIGURE 6-50
-------
o
ro
k_
•f-'
-------
ID
t_
4—'
0)
u
^)
8
*-^
to
L.
*->
-------
5
4—'
(0
k_
**
-------
00
Q
<
o»
U
*
>
I
m
b
u.
Effects of Various Treatments on
ALD Formation at Utility 36
t=O hrs t=24 hrs
Conventional Filter Media
Treatment & Residence Time
FIGURE 6-57
Effects of Various Treatments on
Formaldehyde Formation at Utility 36
t=0 hrs t=24 hrs
Conventional Filter Media
Treatment & Residence Time
FIGURE 6-58
-------
Treatment Modification Results and Discussion
Discussion
As shown in the ozonation studies discussed above, the formation of DBFs is strongly
influenced by the particular disinfection scheme employed at the treatment plant. In
order to meet the anticipated Surface Water Treatment Rule disinfection requirements
as well as DBF regulations, many utilities are faced with selecting alternative
treatments. Three general disinfection treatment change scenarios for utilities
considering ozonation include:
1) Those which currently employ only free chlorine for disinfection in their
treatment process and will switch to ozone and chlorine for primary and
residual disinfection, respectively;
2) Those which currently employ only chloramines for disinfection, but will
switch to ozonation followed by chloramination;
3) Those which currently employ only free chlorine in their treatment and will
switch to ozonation followed by chloramination.
Figures 6-59 to 6-64 summarize results from this study based on the above DBF control
options. All the data represented in these figures correspond to the terminal
distribution system points studied. Changes in concentrations of TOX, XDBPSU11) and
other DBF classes except HKs are presented as percent change. Individual compounds
and HKs are shown as actual changes in concentration. Changes were calculated by
comparing concentrations of DBFs before and after ozonation was added to the
treatment process.
Modification from Chlorine to Ozone/Chlorine. Figures 6-59 and 6-60 present data
for utilities where treatment plants were modified from employing only chlorine for
disinfection to ozone/chlorine treatment. At Utility 19, preozonation decreased the
concentration of TOX and most classes of the measured DBFs. In contrast, a 7 percent
increase in XDBP was observed at Utility 36; this was primarily due to the observed
increase in TTHM concentration. That ozonation can increase TTHMs has been
observed in other studies (Riley et al., 1978; Lawrence, 1977; Umphres et ah, 1979).
Increased levels of THMs after preozonation may be attributed to an increase in
precursor materials by formation of: I) m-hydroxy aromatic compounds; or 2)
secondary precursors, i.e., aliphatic carbonyl compounds (Glaze et al., 1982); or to an
increase in brominated THMs due to bromide ion in the raw water (Dore et al., 1988).
Figures 6-24 and 6-25. and Table 6-8. show the individual DBF concentrations which
were measured during the modification studies at Utilities 19 and 36. As mentioned
previously, the data show decreases in chloroform for the two utilities and a general
shift to the brominated THMs after ozonation. This was particularly evident at Utility
36 where the increase in TTHMs after ozone addition was due primarily to increases in
dibromochloromethane and bromoform. These changes were greater than the upper
control limits of the respective compounds, i.e.. the differences were not due to
analytical variability. As discussed in Section 5. levels of brominated THMs in treated
water have been shown to be associated with bromide concentrations in the raw water.
The same shift to the brominated species is seen to varying extents upon examination of
6-10
-------
Percent Change in DBP Concentrations c*je to Switch from
CHLORINE ONLY to OZONE/CHLORINE Treatment
UTILITY 19
UTILITY 36
TJ
O
a
CO
o
u
0)
"c
If}
a
TOX
XDBPsum
TTHM
HAA
HAN
ALD
NA
- 10O
-50
0
50
1OO
ISO
% Change in DBP and TOX Concentrations
FIGURE 6-59
Change in DBP Concentrations due to Switch from
CHLORINE ONLY to OZONE/CHLORINE Treatment
UTILITY 19
V777A UTILITY 36
y
•p
o
a
i
CO
c
c
c
in
LJ
HK
CHP
CH
'-"NCI
r-JA
(O)
O
1O
Change in DBP Concentration (ug/L)
FIGURE 6-60
-------
Treatment Modification Results and Discussion
the concentrations of dibromoacetic acid. Although the levels of total HA As decreased
by approximately 50 and 14 percent at Utilities 19 and 36, respectively, the level of
dibromoacetic acid increased by more than 100 percent at Utility 19, and by 18 percent
at Utility 36. Ozone can react with bromide ions in the raw water causing the
formation of hypobromous acid (HOBr) (Dore et al., 1988). Reactions of HOBr with
natural organic matter can produce bromoform and other brominated DBFs. When
preozonation and postchlorination is practiced, competition exists between
hypochlorous acid and HOBr for organic matter, leading to varying concentrations of
chlorinated and brominated DBFs (Dore et al., 1988).
In this study, HAAs represented the second largest fraction of halogenated DBFs on a
weight basis. The data presented in Figure 6-28 and Table 6-8 confirm work by other
researchers who found DCAA and TCAA to be major by-products of chlorination of
humic and fulvic acids (Quimby et al., 1980; Christman et al., 1983; Miller and Uden,
1983: DeLeer et al.. 1985). At Utility 19, chloroform and TCAA were decreased by
68 and 82 percent after preozonation and postchlorination. DCAA was decreased by
50 percent, but remained the largest contributor to the HAA fraction. At Utility 36,
the TCAA concentration was decreased by 41 percent while DCAA decreased only
slightly. Dore, et al. (1988) showed that ozonation followed by chlorination of fulvic
acids reduced TCAA levels but did not greatly affect DCAA. Reckhow and Singer
(1985) demonstrated decreases in chloroform and TCAA precursors after similar
treatment while DCAA precursors remained largely unchanged. Moreover, the latter
researchers predicted the destruction of dichloroacetonitrile precursors and an
enhancement of 1,1,1-TCP. In this study, ozone addition resulted in a decrease in
HANs and a small increase in HKs at Utilities 19 and 36, respectively.
At Utility 36. 19 //g/L of chloral hydrate were detected after the chlorination-only
treatment, and at Utility 19, 6.3 //g/L. After ozonation and chlorination, those
concentrations were increased to 28 and 8.6 //g/L, respectively.
Modification from Chloramines to Ozone/Chloramines. Four utilities modified
treatment from chloramines-only or prechlorination/postammoniation to ozone/
chloramines. Figures 6-61 and 6-62 show that, in each study, the modification was
effective in reducing levels of all classes of halogenated DBF compounds except for
HKs. However, increases in this DBF class were less than 0.7 //g/L in all cases. Small
increases were observed for chloropicrin and cyanogen chloride in two of the studies,
while chloral hydrate decreased in three of them. The most significant increase was
observed in the concentrations of aldehydes. Preozonation increased the sum of
formaldehyde and acetaldehyde over 300 percent in the utilities where these compounds
were measured.
Modification from Chlorine to Ozone/Chloramines. Figures 6-63 and 6-64 present a
summary of the results of the two plants which modified treatment from chlorine-only
as a disinfectant to ozone/chloramines. Decreases of at least 80 percent in
concentrations of TOX. XDBPslim and sums of all halogenated classes of DBF
compounds except for HKs were observed. Aldehydes increased by 67 percent and
cyanogen chloride by 2.9 //g/L for the treatment modification study at Utility 36. The
greatest decrease among the other compounds was detected in concentrations of chloral
hydrate, which was reduced by approximately 9 and 15 //g/L at Utilities 7 and 36,
respectively.
6-11
-------
o
CL
I
>
CD
O
4->
O
0)
M—
c
in
Q
Percent Change in DBP Concentrations due to Switch from
CHLORAMINES ONLY to OZC>E/CHLORAMINES Treatment
UTILITY
TOX
XDBPsum
THM
HA A
HAN
ALD
-1OO
-50
100
% Change in DBP and TOX Concentrations
•» Includes 4 hours of free chlorine contact time before postammoniation
HGURE 6-61
Change in DBP Concentrations due to Switch from
CHLORAMINES ONLY to OZOE/OLORAMINES Treatment
o
3
o
a.
i
CD
g
o
0)
Q
UTILITY
HK
CHP
UTILITY 36 77777A UTILITY 6
V/////////////////7//,
CNCI
Change in DBP Concentration (ug/L)
* Includes 4 hours of free chlorine contact time before postammoniation
FIGURE 6-62
-------
Percent Change in DBP Concentrations due to Switch from
CHLORINE ONLY to OZONE/CHLORAMINES treatment
UTILITY 7
UTILITY 36
0
a
i
m
u
Q
TOX
XDBPsum
TTHM
HAA
HAN
ALD
-100
-5O
0
50
1OO
% Change in DBP and TOX Concentrations
FIGURE ^63
Change in DBP Concentrations due to Switch from
CHLORINE ONLY to OZONE/CHLORAMINES Treatment
UTILITY 7
UTILITY 36
y
E
a
i
m
g
•*-•
u
"c
Q
HK
CHP
CH
CNCI
(0)
-20
-15
-1O
-5
0
Change in DBP Concentration (ug/L)
FIGURE 6-64
10
-------
Treatment Modification Results and Discussion
CHLORINE DIOXIDE STUDIES
Chlorine dioxide studies were conducted at Utility 16 and Utility 37. Both utilities
employed chlorine dioxide as a preoxidant, and used chlorine as a final disinfectant.
Utility 16
Utility 16 operates a large (400 mgd) direct filtration treatment system for a low-
organics reservoir water in the western United States. The plant utilizes free chlorine
for both preoxidation and final disinfection for most of the year, but periodically
switches to chlorine dioxide preoxidation to control THMs and taste and odor. The
plant utilized chlorine-only treatment for all quarterly baseline samples collected for
this study. The plant was sampled on November 21. 1988 for the chlorine-only
treatment, and on March 21, 1989 for the chlorine dioxide/chlorine treatment. Figure
6-65 illustrates the treatment trains employed on the two sampling days.
Samples were collected at the plant influent, filter influent, clearwell effluent, and at
two distribution system locations (approximate residence times of 45 minutes and 7
days). DBF analyses were conducted on the clearwell effluent and distribution system
samples only. On the chlorine-only sample date, the plant was operating with a pre-
chlorine dose of 2 mg/L (applied to the plant influent) and a post-chlorine dose of I
mg/L (applied to the filter effluent). On the chlorine dioxide/chlorine sample date, the
plant was operating with a chlorine dioxide dose of 0.5 mg/L (plant influent) and a
chlorine dose of 1.9 mg/L (filter effluent).
An emergency situation occurred on the March sampling date. A chlorine storage tank
developed a leaky valve, requiring that the tank be emptied. The contents of the tank
were discharged into the water being treated in the plant, resulting in a significant
amount of free chlorine in addition to the chlorine dioxide. Thus, the in-p!ant samples
were unusable for purposes of this study. However, the two distribution system
samples were collected prior to the chlorine leak problem and the results from these
samples are reported in Table 6-9 and illustrated in Figure 6-66.
Figure 6-66 indicates very little difference in the levels of DBFs produced by the two
different oxidation/disinfection schemes. In fact, at distribution system Location 2, the
chlorine dioxide/chlorine treatment produced slightly higher levels of XDBP THMs,
HANs and HKs than the chlorine-only treatment. With both chlorine dioxide/chlorine
and chlorine-only treatment, levels of all DBFs increased with increasing residence time
in the distribution system.
Chlorine dioxide residual, and chlorite and chlorate levels were measured at the
laboratory of Dr. Gilbert Gordon at Miami University in Oxford, Ohio. Preserved
samples were shipped overnight to Dr. Gordon on both the November and March
sampling dates. Dr. Gordon's laboratory utilized a flow injection analysis method to
analyze for chlorine dioxide, chlorite and chlorate (Gordon et al., 1989; Themelis et
al.. 1989). Chlorine dioxide residual levels are reported in Table 6-9. Neither chlorite
nor chlorate were detected in either of the chlorine-only samples. However, for the
chlorine dioxide/chlorine samples, 0.30 mg/L of chlorite and 0.14 mg/L of chlorate
were detected at Location I. Also, 0.09 mg/L of chlorite and 0.28 mg/L of chlorate
were detected at Location 2.
6-12
-------
RAPID MIX FLOCCULATION FILTRATION
CHLORINE ONLY
Chlorine
Ferric Chloride
Zinc Orthophosphate
CHLORINE DIOXIDE / CHLORINE
Chlorine Dioxide
Ferric Chloride
Zinc Orthophosphate
SCHEMATIC OF UTILITY 16 TREATMENT PROCESSES
FIGURE 6-65
-------
TABLE 6-9
UTILITY 16 TREATMENT STUDY
WaCer Quality Parameters
uv
TOC Absorbance
(mg/L)
(cm1)
Chlorine Chlorine
Residunl Dioxide
pH Temp. (mg/L) Residual
(°C) Free
Total
(mg/L)
Chlorite
(mg/L)
Chlorate
(mg/L)
TOX
(PHIL)
CHLORINE ONLY
C
I
Rau Water
Location 1
Location 2
2.69
2.64
2.62
0.098
0.083
0.082
7.72 14.8 NA
7.59 15.1 0.6
7.65 16.5 0.9
NA
0.9
1.5
ND
ND
ND
ND
ND
ND
ND
ND
ND
-------
o>
ro
c
o
5
u
a
CD
Q
Effect Of CI2 and OO2/CI2
on DBP Formation (Utility 16)
/UL>
60.0
50.O
40.O
30.O
20.O
10.O
O.O
-
-
-
-
/
/
/
's
'',
T
^
/
^
^
^
^
7]
^
^
^
7
\
'/
'',
\
\
/
m^
%% CIO2/CI2
Dist. Sys. Residence Time (t)
1: t = 0.75 hrs
2: t = 16O hrs
NA - Not Analyzed
|-|
mWi *^**A SHI ilL
12 12 12 12 12 12
xCGP'jtm THM HftN hK HAA ALD
UBP Class
RGURE6-66
-------
Treatment Modification Results and Discussion
Utility 37
Utility 37 did not participate in the quarterly baseline sampling program for this study.
hut was selected for a treatment study because the plant's configuration and operation
offered a unique opportunity for a side-by-side comparison of DBF levels produced by
chlorine dioxide and chlorine-only preoxidation. This Southeastern utility treats a low-
organics river water with a conventional treatment process. The plant typically treats
approximately 30 mgd.
The configuration of the plant provides for two separate treatment trains. In 1981. the
plant began seasonal use of chlorine dioxide preoxidation (added following flocculation)
in one of the trains for THM control. The other train has chlorine preoxidation. Free
chlorine is used for residua! disinfection in both trains; however, samples were collected
before the addition of chlorine for residual disinfection. Schematic diagrams of the two
treatment trains are shown in Figure 6-67.
The chlorine dioxide dose was approximately 0.9 mg/L and the chlorine dose was
approximately 2.25 mg/L for preoxidation. However, the chlorine dioxide generator
product could not be analyzed directly; it could only be analyzed after addition of the
generator product to the flocculation basin effluent. Because chlorine dioxide reacts
very rapidly, the actual dose was most likely higher than 0.9 mg/L, although the
sample was taken immediately after addition 01 the generator product to the flocculated
water. To further illustrate this point, plant operations personnel reported that the
applied chlorine dose was 2.25 mg/L; however, the total chlorine residual measured
immediately after application of the chlorine to the flocculated water was only 1.15
mg/L.
All samples for this study were collected on May 19, 1989. Water quality parameters
for this utility are reported in Table 6-10. On the sampling date, it was determined
that there was some degree of mixing between the two treatment trains, since low levels
of chlorine dioxide were measured in the chlorine-only treatment train and vice versa.
Additionally, although the chlorine dioxide generator was operated to produce the
lowest practical levels of free chlorine in the generator product, detectable levels of free
chlorine were measured immediately after addition of the generator product to the
flocculated water (0.2 mg/L). Thus, the preoxidant in one treatment train was
predominantly chlorine dioxide (referred to as the "chlorine dioxide" treatment train).
and the preoxidant in the other train was predominantly free chlorine (referred to as the
"chlorine" treatment train).
Samples were collected at the plant influent and sedimentation basin effluent. DBF
analyses were performed on the sedimentation basin effluent samples. The
sedimentation basins had a detention time of approximately 2.5 hours. Following the
sedimentation basins, there was mixing of the two trains in the filter influent flume.
Chlorine dioxide, chlorite and chlorate measurements were performed on-site by flow
injection analysis. Chlorite levels were 1.16 mg/L in the chlorine dioxide treatment
train and 0.07 in the chlorine train (measured at the sedimentation basin effluent).
Chlorate concentrations were 0.53 and 0.22 mg/L in the chlorine dioxide and chlorine
treatment trains, respectively (sedimentation basin effluent).
6-13
-------
RAPID MIX/FLOCCULATION SEDIMENTATION
FILTRATION
CHLORINE
Chlorine
Fluoride
Chlorine
Solids
Contact
Clarifier
Lime
(Not Operational)
CHLORINE DIOXIDE / CHLORINE
Solids
Contact
Clarifier
Lime
SCHEMATIC OF UTILITY 37 TREATMENT PROCESSES
FIGURE 6-67
-------
TABLE 6-10
UTILITY 37 TREATMENT STUDY
Water Quality Parameters
UV Chlorine
Absorbance Residual
TOC at 254 nm pH Temp (mg/L)
(mg/L) (cm ') (°C) Free Total
Chlorine
Dioxide
Residual Chlorite Chlorate TOX
(mg/L) (mg/L) (mg/L) (//g/L)
Raw Water 1.01 0.050 7.0 15 ND ND NA NA NA NA
CHLORINE
Scd Basin
Effluent 1.03 0,023 6.7 15 0.2 0.35 0.19 0.07 0.22 140
CHLORINE DIOXIDE
Sed Basin
Effluent 1.11 0.028 6.8 15 0.05 0.06 0.21 1.16 0.53 69
NA - Not Analyzed
ND - Not Detected
-------
Treatment Modification Results and Discussion
Results of the DBF sampling are presented in Figures 6-68 through 6-73. Figure 6-68
shows the effect of the two different preoxidants on XDBPsum and the DBF classes.
From this figure, it is apparent that even in the relatively short detention time in the
sedimentation basins, the use of chlorine dioxide preoxidation resulted in lower levels
of all measured DBFs compared to chlorine treatment. XDBP was almost 50%
lower with chlorine dioxide treatment. However, the figure indicates that chlorine
dioxide treatment produced detectable levels of DBFs, most likely due, at least in part,
to the presence of free chlorine in the chlorine dioxide generator product.
Figure 6-69 illustrates THM levels resulting from the two preoxidation schemes. The
advantages of chlorine dioxide as a THM control method are apparent, since levels of
chloroform, bromodichloromethane and dibromochloromethane were 52, 75 and 74
percent lower, respectively, with the chlorine dioxide treatment compared to the
chlorine-only treatment.
HAN formation was also substantially lower with the chlorine dioxide treatment, as
illustrated in Figure 6-70. The concentration of DCAN was over 60 percent lower with
chlorine dioxide preoxidation (1.0 /ug/L) compared to free chlorine preoxidation (2.9
//g/L). Figure 6-71 shows HK levels produced by the two preoxidation schemes.
Again, chlorine dioxide treatment produced significantly lower levels of the two HK
compounds. The concentration of I, I, I-TCP was significantly lower for the chlorine
dioxide treatment (0.2 /vg/L) as compared to the chlorine treatment (1.6 //g/L). As this
ketone is a chloroform precursor, the previously discussed lower TTHM concentration
with the use of chlorine dioxide treatment is consistent.
HAA levels are plotted in Figure 6-72. MCAA concentrations were approximately the
same for the two treatments and DC A A levels were only 28 percent lower for the
chlorine dioxide treatment. However, the concentration of TCAA was over 70 percent
lower with chlorine dioxide treatment. Figure 6-73 shows the levels of miscellaneous
compounds for the two treatments. Very little difference in the concentrations of
chloropicrin and cyanogen chloride was observed. Chloral hydrate was significantly
lower with chlorine dioxide treatment (1.8 //g/L) compared to chlorine treatment (4.2
//g/L).
Discussion
Use of chlorine dioxide has been identified as a treatment technology capable of
enabling utilities to lower THM levels. Chlorine dioxide preoxidation allows chlorine
addition to be delayed until after coagulation, sedimentation and filtration processes
have removed precursors to some extent. Chlorine dioxide does not form THMs,
although a small amount of TOX may be formed (Chow and Roberts, 1981;
Fleischacker and Randtke. 1983; Werdehoff and Singer, 1986; Lykins and Griese,
1986). Additionally, some researchers have found that chlorine dioxide is capable of
lowering the concentration of THM and TOX precursors by oxidation (Rav-Acha et al.,
1985: Werdehoff and Singer, 1986; Lykins and Griese, 1986). Thus, chlorine dioxide
preoxidation in combination with free chlorine residual disinfection would be expected
to produce lower levels of DBFs.
Although the results of the study at Utility 16 were inconclusive with respect to
reductions in levels of DBFs produced by chlorine dioxide preoxJdation, Utility 37
6-14
-------
_!
O
o
U
a
m
Q
Effect of C'2 AND CIO2
«;>n L'BP Formation (Utility 3i')
XDBP
THM
DbP Class
FIGURE 6-68
HA A
ALD
5
*-•
(0
k_
4-*
0)
u
i
40
30
20
10
Effect of CI2 and CIO2
on THM Formation (Utility 37)
CHCI3
CI2
CIO2
Below MRL
CHBrCll
CHBr2Cl
CHBr3
THM Compound
FIGURE 6-69
-------
o
-t-*
ro
V.
(J
Z
4.O
3.0 -
1.O -
o.o
Effect of CI2 and CIO2
on HAN Formation (Utility 37)
TCAN
DCAN
BCAN
I-IAN Compound
FIGURE 6-70
8
*-*
(0
*-•
1
o
(J
Effect of CI2 and CIO2 .
on Haloketone Formation (Utility 37)
0.5 -
0.0
1.1-DCP
1.1,1-TCP
HK Compound
FIGURE 6-71
-------
Oi
s
o
5
u
30
20
10
O
Effect of CI2 and CIO2 on
Haloacetic Acid Formation (Utility 37)
CI2
CI02
* Below MRL
MCAA
DCAA
TCAA
HA A Compound
FIGURE 6-72
MB A A
fl
I
O
Effect of CI2 and CIO2
on Misc. DBP Formation (Utility 37)
CHP
CNCl
UBP Compound
FIGURE 6-73
TCP
-------
Treatment Modification Results and Discussion
results indicated that chlorine dioxide preoxidation produced substantially lower levels
of many DBFs, even with the presence of free chlorine in the generator product and
with some degree of mixing between the two process trains.
The chlorine dioxide generator product was optimized for this study to contain the
lowest practical levels of free chlorine. It is interesting to note that routine operation
of the generator at Utility 37 resulted in approximately equal amounts of chlorine
dioxide and free chlorine in the generator product (before generator optimization, 1.60
mg/L of chlorine dioxide and 1.56 mg/L of total chlorine were measured upon addition
of the generator product to the flocculated water). The plant superintendent reported
that the generator is operated in this way to avoid "kerosene" and "bleach" odor
complaints from consumers that occurred when the level of free chlorine in the
generator product was minimized. Household odors associated with chlorine dioxide
treatment, including "kerosene" and "chlorinous", have been investigated by Hoehn, et
al. (1989). This chlorine dioxide study reports on DBF levels produced when the
operation of the chlorine dioxide generator was optimized and does not necessarily
reflect the levels experienced under routine operations at Utility 37.
Chlorite and chlorate are inorganic by-products of chlorine dioxide. Currently, the
USEPA recommends that the combined residuals of chlorine dioxide, chlorite and
chlorate not exceed 1.0 mg/L in distribution systems (USEPA, 1983). Additionally,
chlorine dioxide, chlorite and chlorate are among the disinfectants and DBFs targeted
by the USEPA's Drinking Water Priority List (Federal Register, 1988).
Levels of chlorite and chlorate in the distribution system samples of Utility 16 were low
and the distributed water easily met the recommended level of 1.0 mg/L combined
chlorine dioxide, chlorite and chlorate residual. Levels of chlorine dioxide, chlorite and
chlorate were not measured in the distribution system of Utility 37, but based on the
levels measured in the sedimentation basin effluent, this utility may not have been able
to meet the 1.0 mg/L recommended limit under the conditions employed for this study.
COAGULATION STUDIES
Coagulation studies were conducted in order to study the effect of coagulant dose on
DBF precursor removal. Utilities 3 and 12 were selected because they were able to
adjust their alum dose at full scale without severely compromising the quality of the
finished water. Each applied low, medium and high doses of aluminum sulfate, after
which samples were collected and analyzed for DBFs and various water quality
parameters.
Utility 3
Utility 3 is a conventional treatment plant in the eastern United States with a 10.5-mgd
capacity. Figure 6-74 presents a process schematic of the plant and shows the three
doses of alum which were applied during the study. Each alum dose was applied on a
different day; a period of at least 48 hours passed between initiating each alum dose to
assure adequate flushing of the plant before collecting samples. Samples of the plant
influent raw water, filter influent and filter effluent were collected and analyzed for
TOC and UV-254 absorbance. Turbidity measurements were made on-site at these
three locations. Bromide and chloride were also measured in the plant influent.
6-15
-------
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
ALUM COAGULATION
Caustic Soda. Alum
Potassium Permanganate
Chlorine
Polyphosphate, Fluoride
Caustic Soda, Chlorine
To
Distribution
Alum Dosages (ppm)
Low = 10
Medium = 19
High = 40
SCHEMATIC OF UTILITY 3 TREATMENT PROCESS
FIGURE 6-74
-------
Treatment Modification Results and Discussion
Chlorine is normally applied at two places in the plant, at the entrances to the clearwell
and to the distribution system. However, during this study, samples from the filter
effluent were collected and 24-hour SDS tests were conducted. Thus, no chlorine was
applied in the plant to the water that was sampled. A 3.5 mg/L dose of chlorine was
applied for the SDS testing. A description of the protocol for the SDS test was
provided in Section 3 of this report.
Chlorine residuals, as well as other water quality parameters, are presented in Table
6-11. The table shows that the pH values for each sampling point in the plant were
similar for all three alum doses (5.5 to 5.6 at the filter influents and effluents).
Additionally, chloride and bromide levels were low, being detected at 5 mg/L and 0.01
to 0.02 mg/L, respectively. Influent turbidity ranged from 1.1 to 1.6 NTU, with filter
effluent values of 0.21 NTU for the low alum test and 0.04 to 0.06 NTU for the high
and medium alum tests.
Table 6-11 also shows that the influent TOC concentrations ranged between 2.97 and
3.11 mg/L and the UV-254 absorbance between 0.120 and 0.133 cnr1. The percent
removals of these water quality parameters at each alum dose are presented in Table
6-12. Percent removals of TOC increased from 25 to 50 percent in the filter influent
water as the alum dose increased from 10 to 40 mg/L. At the low alum dose, a greater
percent of TOC was removed through filtration (20 percent) than at the medium and
high alum doses (15 and 9 percent, respectively); this was probably due to the better
settling characteristics of the floe at these doses.
A pattern similar to TOC removal was observed for UV-254 absorbance. Percent
removal ranged from 39 to 69 percent through flocculation/sedimentation at the various
alum doses. An additional 9 to 26 percent removal was obtained through filtration.
Figure 6-75 presents the effect of increasing alum dose on XDBPSIII11 and DBP classes.
In general, DBPs decreased with increasing alum dose. XDBPsum was lowered from
150 to 94 yug/L, and THMs from 86 to 55 //g/L as alum doses increased from 10 to 40
mg/L. Figure 6-76 shows that chloroform was the predominant THM; this is consistent
with the low bromide levels found in the plant influent. Chloroform concentrations
were reduced from 74 to 45 x/g/L as alum doses increased. Changes in coagulant dose
had little effect upon the other chlorinated THMs, although these were detected only at
levels less than 11 /sg/L.
The effect of alum dose on other individual DBPs is presented in Figures 6-77 to 6-80.
DCAA and TCAA followed patterns similar to chloroform (Figure 6-77) as did
1.1.1-TCP (Figure 6-78), DCAN (Figure 6-79) and chloral hydrate (Figure 6-80). It
should be noted that some DBPs may have undergone hydrolysis since the SDS tests
were conducted at pH 8.2; however, this effect would have been consistent for all alum
doses tested.
As part of the two coagulation studies, residual aluminum was measured in the plant
influent and in the actual distribution system at a residence time of approximately 24
hours. Table 6-11 shows that levels in the plant influent were similar on the three
sampling dates (46 to 50 /ug/L) and that the levels in the distribution system were
higher (75 to 124 //g/L), although residual aluminum was the lowest in the distribution
system sample with the highest alum dose.
6-16
-------
TABLE 6-11
UTILITY 3 TREATMENT STUDY
Water Quality Parameters
Sampling TOC
Location (mg/L)
Plant Influent
Low Alum
Mecl Alum
Hi Alum
Filter Influent
Low Alum
Meil Alum
Hi Alum
Filler Effluent
Low Alum
Med Alum
Hi Alum
24 Hr. Simulated
Low Alum
Med Alum
Hi Alum
3.11
3.08
2.97
2.32
1.87
1.48
1.71
1.41
1.22
UV
absorbince Chloride
(cm"1) (mg/L)
0 122
0.133
0.120
0.074
0.047
0.037
0.043
0.030
0.026
5
5
5
NA
NA
NA
NA
NA
NA
Bromide
(mg/L)
0.02
0.01
0.02
NA
NA
NA
NA
NA
NA
Turbidity
(NTU)
1.32
1,62
1.09
1.10
0.68
0.44
0.21
0.06
0.04
Aluminum
tf/g/L)
47
30
46
NA
NA
NA
NA
NA
NA
TOX
U/g/L)
NA
NA
NA
NA
NA
NA
NA
NA
NA
Free
Chlorine
Residual
(mg/L)
ND
NA
ND
ND
ND
ND
ND
NA
ND
pH
6.4
fi.3
6.4
5.5
5.5
5.5
5.5
5.6
5.5
Temp.
<°C)
14.5
16.0
14.5
14.0
16.0
13.5
14.5
16.0
13.5
Distribution System Test
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
12<
871
75*
240
190
150
1.2
1.3
1.9
8.2
8.2
8.2
25
25
25
NA = Not Analyzed
ND = Not Delected
* = Aluminum data from actual plant distribution samples.
-------
TABLE 6-12
PERCENT REMOVAL OF TOTAL ORGANIC CARBON
AND ULTRAVIOLET ABSORBANCE AT 254 nm BY
VARIOUS ALUM DOSES AT UTILITY 3
PLANT INFLUENT PLANT INFLUENT
TO FILTER INFLUENT TO FILTER EFFLUENT
(Percent Removal) (Percent Removal)
Total Organic Carbon
Alum Dose
Low 25 45
Medium 39 54
High 50 59
UV-254 Absorbance
Alum Dose
Low 39 65
Medium 65 77
High 69 78
-------
8
u
200
150 -
1OO -
50 -
Effect of Alum Dose
on DBP Formation (Utility 3, SDS)
-------
_J
3
U
1O -
0
Effect of Alum Dose
on HAA Formation (Utility 3. SDS)
Low Alum Dose
Medium Alum Dose
High Alum Dose
MCAA
DCAA
HAA Compound
RGURE6-77
o»
ra
I
o
U
1.0O
0.75
0.50
0.25
0.00
Effect of Alum Dose
on HK Formation (Utility 3, SDS)
Low Alum Dose
Medium Alurn Dose
High Alum Dose
1.1-DCP
1.1,1-TCP
HK Compound
FIGURE 6-78
-------
o>
(0
U
3 -
O
Effect of Alum Dose
on HAN Formation (Utility 3. SDS)
-**•*•
TCAN
Low Alum Dose
Medium Alum Dose
High Alum Dose
* Below MRL
DCAN
BCAN
HAN Compound
FIGURE 6-79
DBAN
o>
-------
Treatment Modification Results and Discussion
Utility 12
Utility 12 is a conventional treatment plant in the western United States with a 72-mgd
capacity. Figure 6-81 presents a process schematic of the plant and shows the three
doses of alum applied during the study. Chlorine was added at two locations in the
plant: before the rapid mix (1.8 mg/L dose) and before filtration (1.3 mg/L dose).
The total chlorine contact time was approximately 100 minutes. Ammonia was added
approximately 4 minutes after filtration, prior to the clearwell, in order to convert the
free chlorine residual to chloramines. Chlorine residuals and other water quality
parameters are shown in Table 6-13.
Samples of the raw water, sedimentation basin effluent, filter effluent and clearwell
effluent were collected and analyzed for all DBFs. Sampling was conducted a
minimum of three plant detention times after alum doses were changed.
Table 6-13 presents various water quality parameters for each alum dose applied. Free
and total chlorine residuals and temperatures were similar throughout the study. Since
the plant had no capability to control pH before or during the sedimentation process,
the pH values decreased as alum dose increased. It is important to note that the
bromide levels were 0.14 to 0.15 mg/L during the treatment modification study. These
bromide levels are in contrast to the quarterly sampling, in which 0.41 mg/L and 0.79
mg/L were detected during the summer, 1988 and winter, 1989 quarters, respectively.
The effect of elevated levels of bromide on DBFs as compared to the concentrations
found during the treatment study was discussed in more detail in Section 5 of this
report.
Residual aluminum was measured in the plant influent and clearwell effluent. Table
6-13 shows that increasing the alum dose did not increase the residual aluminum
leaving the plant. In fact, removals of 71 to 76 percent of influent aluminum were
observed. These results are in contrast to that observed at Utility 3, indicating a need
to better understand aluminum chemistry in relation to the coagulation process.
Influent turbidity ranged from II to 17 NTU, with clearwell effluent values of 0.14
NTU for the low alum test and 0.06 to 0.07 NTU for the medium and high alum tests.
Table 6-14 shows that the low alum treatment removed 33 percent of the influent TOC,
while the medium and high alum doses removed 47 and 46 percent, respectively, as
measured in the filter effluent. It is important to note, however, that most of the TOC
removal occurred in the sedimentation basins, with little or no additional removal
having occurred through filtration. Removal of UV-254 absorbance ranged from 76 to
86 percent from plant influent to filter effluent.
Figures 6-82 to 6-88 present the DBF data from the clearwell effluent in relation to
alum dose. Figure 6-82 shows the effect of alum dose on DBF concentrations by class.
For XDBP , concentrations decreased from 87 to 69 //g/L as the alum dose increased
from 25 to 75 mg/L; for THMs, the levels decreased from 53 to 39 /ug/L. Little or no
change was observed for the other DBF classes. Figures 6-83 to 6-88 present
individual DBF concentrations by DBF class and the miscellaneous compounds. In
general, individual THMs and HAAs decreased slightly with increasing alum dose; little
or no change was observed for individual HKs, HANs and ALDs, or for chloropicrin,
chloral hydrate, and cyanogen chloride.
6-17
-------
RAPID MIX
FLOCCULATION
SEDIMENTATION
FILTRATION
ALUMJCOAGULATION
Alum
Chlorine
Fluoride, Polymer
Lime, Chlorine
Sodium Hydroxide
Ammonia
Alum Dosages ( ppm)
Low = 24.6
Medium = 45.7
High = 73
SCHEMATIC OF UTILITY 12 TREATMENT PROCESS
FIGURE 6-81
-------
TABLE 6-13
UTILITY 12 TREATMENT STUDY
Water Quality Parameters
Sampling
Location
UV
Absorbftncc
TOC at 234 nm
(mg/L)
(cm-1)
Chloride
(mg/L)
Free Toul
Chlorine Chlorine
Bromide Turbidity Aluminum Residual Residual
(mg/L) (NTU) OWg/L) (mg/L) (mg/L)
pH Temp.
Plant Influent
Inw Alum
Mod. Alum
High Alum
Scdimcntnlion
Basin Effluent
Ijow Alum
Mcd. Alum
High Alum
Filter Effluent
Low Alum
Mod. Alum
High Alum
2
2
2
2
1
1
1
1
1
.XX
.92
.76
.07
.55
.42
.92
54
.48
0.123
0.123
0.121
0.051
0.032
0.027
0.030
0.021
0.017
50
44
49
NA
NA
NA
NA
NA
NA
0.15
0.14
0.15
NA
NA
NA
NA
NA
NA
11.0
17.0
13.5
2.2
1.2
1.7
O.J3
0.04
0,04
450
350
370
NA
NA
NA
NA
NA
NA
ND
ND
ND
0.1
0.1
0.1
1.0
1.0
0.8
ND
ND
ND
0.2
0.2
0.5
1.2
1.1
1.0
8.6
8.5
8.6
7.4
7.0
6.7
7.4
7.4
6.9
25
23
23
23
22
22
23
22
22
Clcarwcll Effluent
Low Alum
Mcd. Alum
High Alum
2.
1.
1.
00
57
42
0.041
0.025
0.022
NA
NA
NA
NA
NA
NA
0.14
0.06
0.07
110
94
107
0.1
0.1
0.1
1.0
1.0
1.0
8.6
8.7
8.6
23
22
22
NA = Not An«ly7fil
ND = None Detected
-------
TABLE 6-14
PERCENT REMOVAL OF TOTAL ORGANIC CARBON
AND ULTRAVIOLET ABSORBANCE AT 254 nm BY
VARIOUS ALUM DOSES AT UTILITY 12
PLANT INFLUENT PLANT INFLUENT
TO FILTER INFLUENT TO FILTER EFFLUENT
(Percent Removal) (Percent Removal)
Total Organic Carbon
Alum Dose
Low 28 33
Medium 47 47
High 49 46
UV-254 Absorbance
Alum Dose
Low 59 76
Medium 74 83
High 78 86
-------
-------
o»
0)
o
b
C
20
10 -
'Effect of Alum Dose
on HAA Formation (Utility 12)
Low Alum Dose
Medium Alum Dose
High Alum Dose
I iAA Compoui ul
FIGURE 6-84
DBA A
Ol
o
o
2.00
Effect of Alum Dose
on HK Formation (Utility 12)
Medium Alum Dose
O.25 h
o.oo
1.1-DCP
1.1,1-TCP
HK Compound
FIGURE 6-85
-------
(0
o
U
Effect of Alum Dose
on HAN Formation (Utility 12)
Medium Alum Dose
» Balow MinMTun Reporting Level
TCAN
DCAN
BCAN
HAN Compound
FIGURE 6-86
DBAN
8
u
3
2
0
Effect of Alum Dose
on Misc. DBP Formation (Utility 12)
CNCI
Low Alum Dose
Medium Alum Dose
High Alum Dose
CH
DBP Compound
FIGURE 6-87
CHP
-------
o»
5
•«-^
a
-------
Treatment Modification Results and Discussion
That greater DBF removal was not observed with increasing alum dose was
undoubtedly due to the utility's prechlorination practices. Approximately 1.8 mg/L of
free chlorine was added to the raw water, with 75 minutes of contact time from the
point of addition to the sedimentation basin effluent. Free chlorine residuals measured
in the basin effluent for all three trials were 0.1 mg/L. Consequently, DBF formation
occurred before and during TOC removal processes, as demonstrated by the similar
concentrations of THMs detected in the sedimentation basin effluent (Table 6-15).
However, from the sedimentation basin effluent to the clearwell effluent, an additional
18 /yg/L of THMs formed at the low alum dose while only 9 //g/L were produced at
high alum dose. A similar trend, although to a lesser extent, was observed for HAAs.
Table 6-15 also shows that THMs increased from the plant influent to the clearwell
effluent. However, other classes of DBF compounds such as HKs, and to a lesser
extent. HANs. decreased from the filter effluent to the clearwell effluent; similar results
were observed for TOX. Since Utility 12 adds sodium hydroxide to increase the pH for
corrosion control, the observed decreases were probably due to hydrolysis of these
compounds. These results are consistent with those observed by Reckhow and Singer
(1984) who showed that TOX and chloroform exhibited opposite pH dependencies.
However, as noted in Section 3, problems with pn-site dechlorination of TOX samples
may have accounted for additional TOX formation during shipping of the chlorinated
in-plant samples (this should not have been a problem for the chloraminated clearwell
effluent sample). Additionally. Reckhow and Singer (1985) showed the instability of
1.1.1-TCP at high pH, as did Oliver (1983) for DCAN.
Discussion
Several studies have focused upon the use of alum coagulation for removal of THMs
and other organic precursor materials from water (Kavanaugh. 1978; Babcock and
Singer, 1979; Semmens and Field, 1980; Dempsey et al., 1984; Hubel and Edzwald.
1987). The optimum pH for coagulation of humic substances with alum has been
reported to be between 5 and 6 (Kavanaugh, 1978; Babcock and Singer 1979). As
noted above, studies at Utility 3 were conducted at pH 5.5 to 5.6. At Utility 12, the
pH ranged from 6.7 to 7.4 in the sedimentation basin.
Using a 40 mg/L dose of alum. Utility 3 was able to achieve a TOC removal of 50
percent in the sedimentation basin effluent (filter influent). At this sampling point at
Utility 12 for a 73 mg/L alum dose, TOC was reduced by 49 percent. These percent
removals are consistent with those reported by Singer (1988), which ranged from 32 to
66 percent. In general, increasing alum dose decreased the levels of DBFs that were
produced. At Utility 3. the order of increasing percent removal of DBF and TOX
precursors. TOC. and UV-254 absorbance at the high alum dose versus the low alum
dose (reference dose) was as follows:
TCAA > UV-254. DCAA > DCAN, TOX, THMs > HKs, TOC
These removals were similar to those reported by Reckhow and Singer (1984) on Black
Lake. North Carolina treated water for UV-254, and TOX and DBF precursors:
UV-254 > DCAN, TCAA > TOX, THM, DCAA, > 1,1,1-TCP
6-18
-------
TABLE 6-15
CONCENTRATIONS 0>g/L) OF VARIOUS DISINFECTION BY-PRODUCTS AND TOTAL
ORGANIC HALIDE AT VARYING ALUM DOSES AT UTILITY 12
DBF
Trihalo-
meihancs
Haloacctic
Acids
'Haloaccto-
nitriles
Halokciones
Chloropicrin
Chloral
Hydrate
Cyanogen
Chloride
Total Organic
Halide
Alum
Dose
low
medium
high
low
medium
high
low
medium
high
low
medium
high
low
medium
high
low
medium
high
low
medium
high
low
medium
high
Plant
Influent
0.33
0.13
0.16
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
18
18
16
Sedimentation
Basin
Influent
35
34
30
16
18
17
4.9
5.0
4.9
2.5
2.7
3.0
0.059
0.068
0.070
2.6
2.8
2.6
2.0
2.1
2.2
140
110
120
Filter
Effluent
45
43
36
27
26
23
6.1
6.1
6.0
3.0
3.0
3.2
0.088
0.094
0.10
3.6
3.6
3.1
2.7
2.9
3.3
210
170
170
Cleat-well
Effluent
53
48
39
25
25
17
5.1
5.0
4.8
1.4
1.5
1.6
0.088
0.10
0.094
3.4
3.3
3.0
3.5
3.7
3.8
140
120
110
ND = Not Detected
-------
Treatment Modification Results and Discussion
It was not possible to assess removals of these compounds for Utility 12 since the plant
influent was chlorinated. It is important to note that Utility 12 practices
prechlorination for disinfection and taste-and-odor control. Thus, the advantages of
optimized coagulation can be lost if a suitable preoxidant/disinfectant is not available.
Utilities 3 and 12 showed greater TOC removal (39 to 50 percent, from plant influent
to filter influent) at the medium and high alum doses as compared to the average TOC
removal by the filtering utilities participating in this study (21 percent). It should be
noted, however, that the treatment practices by the various utilities surveyed focused on
turbidity, rather than on TOC removal. It is not possible to determine whether high
TOC removals, as observed at Utilities 3 and 12, would result at other utilities by
improving the coagulation process. Such an assessment would need to be conducted on
a case by case basis, since some surface waters may not be amenable to greater TOC
removals by conventional treatment.
Assessing costs associated with increased alum doses was not part of the scope of this
project. However, applying higher alum doses for greater DBF precursor removal will.
in many cases, increase the concentrations of total dissolved solids due to pH
adjustment for coagulation. This is particularly the case for well buffered waters.
Moreover, the amount of chemical sludge produced due to elevated alum doses will be
increased. Therefore, costs associated with additional sludge handling, treatment and
disposal should be considered.
GRANULAR ACTIVATED CARBON STUDY
The GAC study conducted at Utility 11 was unique among the treatment studies
conducted for this project, in that samples were collected during a GAC contactor run.
rather than collecting one or two sets of samples at different treatment conditions. In
all. six sets of samples were collected over a period of almost four months.
Process Description
Utility 11 operates a large (235 mgd) conventional treatment facility for a low-organics
river water. This Midwestern utility also operates demonstration facilities for GAC
adsorption and regeneration at the site of the full-scale plant. Numerous pilot and
demonstration-scale studies have been conducted at this utility in recent years to
evaluate the performance and economic feasibility of GAC for removing trace organic
compounds from the plant's source water, a river water receiving industrial discharges
and susceptible to spill events. Studies at this utility have also evaluated the feasibility
of GAC treatment for removing THM and TOX precursors (McGuire et al.. 1989:
DeMarcoet al., 1981: Clark, 1987; Miller and Hartman, 1982).
A schematic diagram of the full-scale and demonstration facilities is presented in Figure
6-89. The full-scale plant's process train includes rapid mix with alum or alum and
polymer addition: flocculation; lamella separation; reservoir storage for 3 days; addition
of lime, fluoride and intermittent addition of ferric sulfate to a hydraulic jump;
sedimentation and filtration. Free chlorine is added at the hydraulic jump, and at the
filter influent and effluent. For the GAC demonstration facilities, a sidestream was
diverted after presedimentation to a rapid sand filter (identical to the full-scale plant's
filters) and then to two separate GAC contactors, each with a hydraulic capacity of I
6-19
-------
RAPID MIX FLOCCULATION SEDIMENTATION FILTRATION
PRETREATMENT:
Alum
Polymer
1
T
V
\
\
\
^
Settler
\ Reservoir /
^\ Storaee /
\ (3da>si /
CONVENTIONAL PLANT:
-
Chlorine
Lime
Fluoride
Feme Sulfate'
DEMONSTRATION PLANT:
T
Fluoride
Ferric Sulfate'
* Ferric Sulfate added intermittently (in use on 5/8/89,6/12/89 and 7/17/89)
SCHEMATIC OF UTILITY 11 TREATMENT PROCESS
FIGURE 6-89
-------
Treatment Modification Results and Discussion
mgd. The GAC facilities received unchlorinated water and no chlorine was added
during treatment in the demonstration plant until after GAC treatment (GAC effluent
samples for this study were collected before the addition of chlorine).
The GAC contactors have a diameter of 11 feet. The contactor sampled for this study
had a 15-foot carbon depth and was operated at an empty bed contact time (EBCT) of
15 minutes. The carbon was Calgon Filtrasorb 400 (12x40 mesh).
GAC Contactor Operation
Previous studies at this utility have found that dioxins were formed upon regeneration
of GAC if chlorinated water had been applied to the carbon, due to the combustion of
adsorbed chlorinated compounds (DeMarco et al.. 1988). Thus, recent operation of the
GAC contactors at this utility involves somewhat intermittent operation to avoid
application of chlorinated water to the GAC.
At Utility II, the rapid sand filter upstream of the GAC contactors is backwashed with
chlorinated (finished plant) water. Filter run lengths varied during the GAC contactor
run from approximately 12 to 50 hours. When backwash was initiated, the GAC
contactor was taken off-line. The sand filter was backwashed and then flushed with
unchlorinated water until such time as residual chlorine was undetectable in the filter
effluent. For this reason, each time the dual media filter was backwashed, the GAC
contactor was off-line for periods of 1.5 to as much as 4 hours. Thus, the GAC
contactor was operated somewhat intermittently, and the "run time" of the contactor
did not coincide with "calendar time". Although the sampling period for this study
was April 10. 1989 to July 31. 1989, a period of 113 calendar days, the actual run
time of the GAC contactor was 95 days. One "run day" of the GAC contactor was
defined by the utility as the time required to pass I million gallons of water through the
contactor at the I -mgd nominal hydraulic loading rate of the contactor.
Treatment Study Description
The objectives of the treatment study at Utility 11 were twofold:
o To evaluate the effectiveness of GAC over time in removing DBF precursors;
and
o To determine DBF levels produced by conventional treatment as practiced in
the full-scale plant.
The first objective was achieved by sampling the GAC column influent and effluent and
performing SDS tests on those samples. Immediate DBF samples were also collected at
the GAC influent and effluent as a control (no chlorine added), to determine if the
source water contributed to measured levels of DBFs. The second objective was
achieved by collection of DBF samples in the conventional plant and at one distribution
system location.
Table 6-16 summarizes the sampling plan for the treatment study. The sampling
locations are defined in Table 6-17. Six sample sets were collected during the
treatment study. Samples were collected at the conventional and demonstration plants
on Mondays. Bottles were filled for immediate DBF analyses, as well as for SDS
6-20
-------
TABLE 6-16
UTILITY 11 TREATMENT STUDY
SAMPLING PLAN
Sample
Location
1.
2.
3.
4.
5.
6.
7.
8.
9.
10
Raw
Al
A2
Bl
B2
B21
B22
B3
B31
.B33
Sample
Type
Immediate
Immediate
Immediate
Immediate
Immediate
SDS 4.5
SDS 4.5 tab.
Immediate
SDS 4.5
SDS 2.0
pH
X
X
X
X
X
X
X
X
X
X
Temp.
X
X
X
X
X
X
X
X
X
X
Analyses
C12 TOC/UV
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Cl-/Br~ DBPs ALDs Comments
(incl. Only
ALDs)
X X
X
X
X
X
X
X
X
11.A3
12.Blank
Immediate
SDS 4.5
Collected 3 days after
collection of
Samples 1-10
Buffered to pH 8.2,
addition of Br~ to
level measured in
raw water
Note: DBF analyses include THMs, HANs, HKs, HAAs, CH, CHP, CNCl and ALDs.
1. SDS protocol: Cl2 dose = 4.5 mg/L, pH = 8.2, Temp. = 25°C, Time = 3 days.
2. SDS protocol: Clz dose = 4.5 mg/L, pH = 8.2, Temp. = Temp, of clearwell effluent at time of sampling,
Time = 3 days.
}. SDS protocol: Cl 2 dose = 2.0 mg/L, pH . ambient, Temp. = 25°C, Time = 3 days.
-------
TABLE 6-17
UTILITY 11 SAMPLING LOCATIONS
ABBREVIATION LOCATION
Raw Plant Influent
AI Full-Scale Plant. Filter Influent (After primary and secondary
sedimentation, and prechlorination)
A2 Full-Scale Plant. Clearwell Effluent
A3 Full-Scale Plant, Distribution System (Detention time = 3 days)
Bl Demonstration Plant, Filter Influent (After primary
sedimentation, no chlorination)
B2 Demonstration Plant, Filter Effluent/GAC Contactor Influent
B3 Demonstration Plant, GAC Contactor Effluent
-------
Treatment Modification Results and Discussion
testing of the unchlorinated GAC influent and effluent, and were shipped overnight to
Metropolitan for analysis and set-up of the 3-day SDS tests. On Thursdays following
the Monday sampling dates, the distribution system was sampled. A distribution
system sampling point was selected which had an approximate residence time of 3 days,
determined by fluoride tracer studies (conducted by Utility 11) and checked monthly
(by Utility 11) by comparing simulated distribution system THM and actual distribution
system THM sample results. Thus, the water sampled at this point on a given
Thursday was the same water leaving the plant's clearwell effluent the previous
Monday. In this way. formation of DBFs in the distribution system could be evaluated,
although not on a strictly controlled basis.
Simulated Distribution System Testing
The SDS test protocol was described in Section 3. For this study, all SDS testing was
performed at Metropolitan according to Metropolitan's SDS protocol, which had not
been developed specifically to simulate Utility II 's distribution system. The protocol
called for buffering to pH 8.2, and incubating at 25°C for three days. The applied
chlorine dose of 4.5 mg/L was determined adequate to insure a chlorine residual of at
least 0.5 mg/L after 3 days of incubation. Two separate chlorine demand tests were
conducted on Utility 11 's unchlorinated GAC column influent water, one in mid-
February and the other in early April. 1989. before start-up of the GAC column.
Although the GAC column was expected to remove a substantial amount of the chlorine
demand, especially at the beginning of the column run. it was determined that the dose
of 4.5 mg/L would be applied to both the GAC column influent and effluent in order to
provide a basis of comparison for the SDS DBFs in the column influent and effluent.
In practice, the chlorine dose required by the GAC column effluent is significantly
lower than that required by the column influent.
An additional SDS sample was run on the GAC column effluent at identical conditions
as the other contactor effluent SDS samples, except the chlorine dose was 2.0 mg/L
and the samples were run at ambient pH. These samples were run to approximate
operating conditions anticipated by Utility 11 when future full-scale GAC facilities are
on-line (i.e.. lower chlorine dose and no lime addition) (see Section 3 of this report).
On two sampling dates, a third SDS protocol was used, in which a GAC influent
sample was incubated at the temperature of Utility 11's clearwell effluent at the time of
sample collection. Results of this test provided a basis of comparison between actual
and SDS data (see Section 3 of this report).
Simulated Distribution System Testing Results and Discussion
Water quality parameters for the immediate samples collected on all six sampling dates
are reported in Table 6-18. As reported in the table. TOC levels in the raw water
ranged from 1.75 to 2.83 mg/L, and TOC levels in the GAC column influent (sampling
point "B2") fell within the range 1.37 to 2.31 mg/L. The TOC removal performance
of the GAC column is plotted in Figure 6-90. TOC breakthrough, defined by Utility
11 as a TOC concentration of 1.0 mg/L in the combined effluent of on-line GAC
contactors (only one of which was sampled for this study), did not occur within the
95-day sampling period. The GAC was very effective for TOC removal. For the first
25 run days, the contactor effluent TOC remained the same as that measured on Run
Day 0.2 (0.1 mg/L), which represents the non-adsorbable fraction of the TOC. After
6-21
-------
TABLE 6-18
UTILITY 11 TREATMENT STUDY
Water Quality Parameters
Sampling TOC
Location (mg/L)
Absorbance
at 254 nm
(cm1)
Free
Chlorine
Residual
(mg/L)
Total
Chlorine
Residual
(mg/L)
pH
Temp.
(°C)
Run Day 0.2 (4/10/89)
Raw
Al
A2
A3 (4/13/89)
Bl
B2
B3
2.83
.88
.77
.78
.88
.80
0.10
0.175
0.072
0.063
0.062
0.074
0.070
0.032
NA
2.08
1.60
0.9
NA
NA
NA
NA
2.30
1.78
NA
NA
NA
NA
7.6
8.2
8.6
8.6
7.0
7.1
10.0
10
16
9
11
9
12
9.4
Run Day 13 (4/24/89)
Raw
Al
A2
A3 (4/27/89)
Bl
B2
B3
Run Day 25 (5/8/89)
Raw
Al
A2
A3 (5/11/89)
Bl
B2
B3
2.10
.54
.51
.52
.87
.52
0.10
.75
.59
.56
.49
.60
.37
0.10
0.148
0.050
0.048
0.021
0.058
0,058
0.029
0.161
0.094
0.024
0.022
0.087
0.042
0.003
NA
2.23
1.66
0.7
NA
NA
NA
ND
2.10
1.64
0.7
ND
NA
ND
NA
2.41
1.75
0.8
NA
NA
NA
ND
2.30
1.88
0.8
ND
NA
ND
7.6
8.5
8.5
8.7
7.3
7.3
7.3
7.6
9.0
8.9
8.8
7.2
NA
6.8
14
12
12
15
12
12
13
15
16
14
16
15
16
15
Run Day 54 (6/12/89)
Raw
Al
A2
A3 (6/15/89)
Bl
B2
B3
2.47
2.24
2.15
2.00
2.14
2.24
0.20
0.116
0.066
0.034
0.042
0.058
0.054
0.003
ND
1.94
1.53
0.3
ND
ND
ND
ND
2.23
1.68
0.4
ND
ND
ND
7.7
8.6
8.4
8.6
7.4
7.5
7.2
23
23
23
22
23
23
23
-------
TABLE 6-18 (Continued)
UTILITY 11 TREATMENT STUDY
Water Quality Parameters
Sampling
Location
Run Day 82
Raw
Al
A2
TOC
(mg/L)
(7/17/89)
2.42
2.22
2.08
A3 (7/20/89) 1 .95
B1
B2
B3
Run Day 95
Raw
Al
A2
A3 (8/
Bl
B2
B3
2.15
2.16
0.60
(7/31/89)
2.59
2.39
2.13
3/89) 2.17
2.30
2.31
0.85
Absorbance
at 254 nm
(cm1)
0.090
0.066
0.032
0.036
0.056
0.056
0.009
0.098
0.115
0.035
0.038
0.064
0.060
0.015
Free
Chlorine
Residual
(mg/L)
ND
1.58
2.08
0.4
ND
ND
ND
NA
1.71
2.05
0.3
NA
NA
NA
Total
Chlorine
Residual
(mg/L)
ND
1.82
2.48
0.5
ND
ND
ND
NA
1.96
2.18
0.4
NA
NA
NA
pH
7.6
8.7
8.4
8.4
7.5
7.4
7.2
7.7
8.7
8.4
8.5
7.6
7.4
7.2
Temp.
("0
27
27
27
26
27
27
27
27
27
27
NA
27
27
27
NA = Not Analyzed
ND = Not Detected
-------
U
TOG vs Run Time for GAG
Column Influent and Effluent. Utility 1 1
+—+GAC mf A AGAC Eff
Immediate Immediate
-A"
4G 6O
Run Days
RGURE 6-90
.-A''
8O
100
1.0
TOG Breakthrough Profile
At utility 1 1
u
u
u
o
O.6
O.4
O.2
0.0
0
2O
4O 60
Run Days
FIGURE 6-91
80
100
-------
Treatment Modification Results and Discussion
another 29 run days, contactor effluent TOC had only increased to 0.2 mg/L. It was
not until the sampling on the 82nd run day that a substantial increase in contactor
effluent TOC was measured (0.6 mg/L). A further 13 run days had increased the
contactor effluent TOC to 0.85 mg/L, showing evidence of imminent TOC
breakthrough.
These TOC removal results are consistent with published data from previous GAC runs
at Utility 11's demonstration facilities. DeMarco, et al. (1981) reported a non-
adsorbable TOC level of approximately 0.2 mg/L in an example plot of a GAC run at
Utility 11 (16-minute EBCT). For the same run, the column effluent TOC was within
the range of 0.8 to 1.2 mg/L after 95 days, while column influent TOC values ranged
from approximately 1.3 to 2.8 mg/L during the first 95 days of column operation.
Additionally. Lykins. et al. (1988) reported non-adsorbable TOC levels of 0.2 from a
GAC run at Manchester. New Hampshire (23-minute EBCT), and 0.5 mg/L from
Jefferson Parish. Louisiana (20-minute EBCT).
Figure 6-91 is the TOC breakthrough profile for Utility II, plotted in the form C/C0 as
a function of run time. Additionally, Figure 6-92 shows breakthrough profiles from
three previous GAC column runs at Utility II (15-minute EBCT) published by Miller
and Hartman (1982). In comparing Figures 6-91 and 6-92, even though the shapes of
the curves are different, the values of C/C0 at approximately 90 to 100 days range from
approximately 0.4 to 0.6 in the breakthrough profiles shown in Figure 6-92, and the
value of C/C0 at 95 days is approximately 0.4 in Figure 6-91. It is interesting to note
that the operation of the GAC columns for the data in Figure 6-92 included
prechlorination upstream of the columns, which prevented or at least reduced the
potential for microbiological activity within the GAC bed. The curve shown in Figure
6-91 was produced from data without chlorination upstream of the GAC column.
The TOC removal results from Utility 11 are favorable compared to published results
from other utilities. For example, in an extensive pilot study conducted by
Metropolitan (McGuire et al., 1989), influent TOC values for Colorado River water
ranged from approximately 2.0 to 3.1 mg/L (only slightly higher than the influent TOC
range for Utility II). However, using a pilot GAC column with an EBCT of 15
minutes and 12x40 mesh carbon, a column effluent TOC of 0.85 mg/L occurred after
only 30 run days, and after 95 run days, the column effluent TOC was 1.7 mg/L.
TOC removal results from GAC studies at other utilities occur between the favorable
results achieved at Utility 11 on the one hand, and the relatively unfavorable results of
Metropolitan on the other. A pilot study was conducted at Jefferson Parish, Louisiana
using Mississippi River water and l-mgd GAC contactors with a 20-minute EBCT.
Results of this study, reported by Lykins, et al. (1988) and McGuire, et al. (1989).
indicated that TOC was steadily removed by the GAC columns for up to 160 days, with
C/C0 continuing to increase during that time. Influent TOC concentrations generally
ranged from approximately 2.5 to 3.5 mg/L. After 120 to 160 days, C/C0 was
reported in the range of 0.6 to 0.7 in three separate column runs at Jefferson Parish.
At Manchester, New Hampshire, pilot study results indicted that after approximately 35
days. C/C0 had increased to approximately 0.7 and fluctuated around that value for the
remainder of the 130-day column run (Lykins et al., 1988). Influent TOC for the lake
water used in this study generally ranged from 2.2 to 2.8 m|/L. Results of a pilot
study conducted at Philadelphia, Pennsylvania, on Delaware River water was reported
6-22
-------
1.0
0.9
0.8
0.7
1O
CONTACTOR A, 15 MM. EBCT
SEPT. is, ieao TO DEC. so, ioeo
CONTACTOR C, 15 MR EBCT
AUG. 11, 1080 TO NOV. 12, 198O
CONTACTOR O, 15 MUt EBCT
SEPT. 1, 1080 TO DEC. 16, 1080
2O 3O 4O SO 6O
DAYS OF COLUMN OPERATION
70
80
1OO
TOC Breakthrough Profiles for Three
Demonstration-Scale Contactor RUM
•I Utility II (McGulre et •!., 1989>.
FIGURE 6-92
SDS TTHMs vs Run Time for GAG
Column Influent and Effluent, Utility 1 1
+—+ GAC Inf A A GAG Eff
3-day SDS 3-day SDS
i-
2OO
15O
1OO
5O
CIS = 4,5
pH = 3.2
Terrp. = 25 C
OO3
All
.-A
.-.••.• r:',". ".",".• ri-.Qr -.--.-1-1-1 -r TV-.-•.•-."•:
0
20
40
60
80
100
Run Days
nr.l IRE 6-93
-------
Treatment Modification Results and Discussion
by McGuire. et al (1989). In this study, influent TOC concentrations ranged from 1.86
to 2.94 mg/L and the columns were operated at an EBCT of 15 minutes. C/C0
increased steadily over the first 65 days of the column run to approximately 0.4, then
increased rapidly after that time until C/C0 reached 0.6 after 85 days. Thus, it is
apparent that the effectiveness of GAC for DBF precursor removal, and hence the
feasibility of GAC as a DBF control method, is site-specific, with Utility 11's results
being favorable compared to those of some other utilities.
It was beyond the scope of the treatment study at Utility 11 to completely characterize
TOC and SDS DBF removal through the entire duration of the GAC contactor run.
Rather, the objectives of this study focused on TOC and SDS DBF removal in the early
breakthrough stage. Thus, as seen in Figures 6-90 and 6-91, sampling did not extend
into the steady state region of the TOC breakthrough curve.
GAC treatment was also effective in removing UV-254, as is apparent in examining the
data presented in Table 6-18. The initial UV-254 data appear anomalous, in that GAC
effluents had 0.03 cnr1 UV-254 values during the first two samplings and 0.003 cm'
for the next two samples, even though TOC in the GAC effluent remained low (0.1 to
0.2 mg/L) during this period. However, as TOC concentrations in the GAC effluent
increased on Run Days 54, 82 and 95 (GAC effluent TOC of 0.20, 0.60 and 0.85
mg/L, respectively), the UV-254 steadily increased from 0.003 to 0.009 and 0.015 cm
•', respectively.
The impact of GAC treatment on levels of SDS DBFs is illustrated in Figures 6-93
through 6-102. The GAC contactor influent and effluent SDS values plotted in these
figures were measured in the samples dosed at 4.5 mg/L of chlorine, buffered to pH
8.2. and incubated for 3 days at 25°C. As shown in Figure 6-93, SDS TTHMs were
extremely low (approximately 6 /ug/L or less) in the contactor effluent for the first 54
run days, indicating very effective removal of THM precursors. After 82 run days.
SDS TTHMs had increased to 34 /ug/L and after 95 run days, SDS TTHMs increased
to 47 ug/L. SDS TTHMs in the contactor influent were 104. 80, 74, 136, 116 and
134 in the 0.2. 13. 25. 54. 82 and 95 run day samples, respectively. Thus, GAC
treatment led to reductions in SDS TTHM levels of >97, >92, >97 and >96 percent
on the first four sampling days, respectively; and 71 and 65 percent on the 82nd and
95th run days, respectively.
DeMarco, et al. (1981) reported similar simulated distribution system THM results for
a previous GAC contactor run at Utility 11 (16-minute EBCT). The SDS procedure
used by these researchers included adding sufficient chlorine to produce an initial free
chlorine residual of 2.5 mg/L, buffering to pH 8.2 and incubating for three days at the
temperature of the full-scale plant's finished water. After 95 days of column operation,
column effluent SDS TTHMs were less than 30 //g/L, while column influent SDS
TTHMs ranged from approximately 60 to 160 //g/L during the first 95 days of column
operation.
As was the case for TOC removal, the effectiveness of THM precursor removal by GAC
is also site specific. For example, SDS THM data published by McGuire, et al. (1989)
from GAC pilot studies conducted by Metropolitan on Colorado River water indicate
much less effective THM control than was achieved at Utility 11. Metropolitan's SDS
6-23
-------
Treatment Modification Results and Discussion
conditions for the latter testing included adding 1 mg/L of chlorine, buffering to pH 8.2
and incubating for five days at 25°C. For a I5-minute EBCT (12x40 mesh carbon),
SDS TTHMs in the column effluent were approximately equal to the SDS TTHM levels
in the column influent (40 to 52 jjg/L) after only 30 days of column operation.
Figures 6-94 and 6-95 illustrate the levels of chloroform and bromoform at Utility 11 in
the 3-day SDS. 5 (4.5 mg/L of chlorine, buffered to pH 8.2, 25°C) GAC contactor
influent and effluent samples. SDS chloroform concentrations in the GAC influent
occurred within the range 52 to 100 //g/L. AH of the SDS chloroform levels in the
GAC effluent samples were less than 2 */g/L through the first 54 run days, then
increased to 8.9 and 14 */g/L on the 82nd and 95th run days, respectively. Thus, GAC
treatment was very effective for lowering the levels of chloroform in the SDS samples.
Figure 6-95 illustrates SDS4 5 bromoform levels in the GAC influent and effluent
samples. Bromoform was not detected in any of the contactor influent SDS samples.
and was not detected in the contactor effluent samples for the first 54 run days.
However, after 82 run days, bromoform was detected in the contactor effluent SDS
samples, at 2.0 and 2.8 //g/L on the 82nd and 95th run days, respectively. These data
indicate that GAC caused a shift from the chlorinated species to the brominated species
in the column effluent samples. This is reported by Utility II to be a typical, though
unexplained, phenomenon at their GAC facilities. To further illustrate this occurrence.
on the 82nd and 95th run days, chloroform comprised 70 and 72 percent, respectively.
of the SDS TTHMs in the contactor influent; while chloroform only comprised 30 and
26 percent of the SDS TTHMs in the contactor effluent on those sampling days. On
those same sampling days, bromoform comprised less than 1 percent of the SDS
TTHMs in the contactor influent, but 6 percent of the contactor effluent SDS TTHMs
were contributed by bromoform.
Similar trends to those observed for SDS THMs were found for SDS HAAs. Figure
6-96 shows the SDS4, results for HAAs in the GAC column influent and effluent. For
the first 25 run days, no HAAs were detected in the contactor effluent SDS samples.
although the contactor influent SDS samples contained HAA levels ranging from
approximately 25 to 40 yug/L during that time. SDS HAA levels gradually increased in
the contactor effluent samples over the subsequent sampling dates, although the GAC
remained very effective at lowering SDS HAA concentrations over the 95 days of
sampling. The shift from chlorinated to brominated species in the GAC contactor
effluent SDS samples is seen in Figures 6-97 and 6-98. DCAA in the contactor effluent
SDS samples remains significantly lower than in the contactor influent SDS samples.
However. DBAA concentrations in the contactor influent SDS samples remained low
throughout the sampling period (less than or equal to 1.8 /ug/L), while SDS DBAA
levels in the effluent samples (3.0 and 4.6 /ug/L on the 82nd and 95th run days.
respectively) exceeded those in the influent SDS samples after 82 run days. In the SDS
samples from the 82nd and 95th run days, DBAA accounted for approximately 3
percent of the total measured HAAs in the GAC influent, but accounted for 26 to 29
percent of the total measured HAAs in the contactor effluent.
Raw water bromide levels were 0.03, 0.05,
-------
Ol
rn
U
U
SDS CHCI3 vs Run Time For GAG
Column Influent and Effluent, Utility 1 1
+—+GAC mf A AGAC Eff
3-day SDS 3-day SDS
15O
100
so -
CI2 = 4.5 mo/L
pH = 8.2
Temp. = 25 C
* Below MRL
0
4O
6O
80
1OO
Run Days
FIGURE 6-94
SDS CHBr3 vs Run Time For GAG
Column Influent and Effluent. Utility 1 1
+—+GAC Inf A AGAC Eff
3-day SDS 3-day SDS
^
3
--K
.3
n 2
U
1
0,
C
CI2 - 4.5 man. » influent below K/WL
f*"1 = 8-2 »* influont »na «ffkj«ot t»iow M=«.
Temp. = 25 C
.A
,-'
x
,-'
-
•** -X--X- -*--* -X-J(-/' -)<- -K-
) 20 4O 6O 80 10O
Run Days
FIGURE 6-95
-------
SDS HAAs vs Run Time for GAG
Column Influent and Effluent. Utility 1 1
+ +GAC Inf A A GAG Eff
3-day SDS 3-day SDS
1
?
o
rti
c
o
0
O
<
I
«u
70
60
50
40
30
20
10
0
t
c
CI2 = 4.5 mg/L * AM ^j^ IMOW ^^g
pH = 8.2 *•* MCAA, TCAA. and I*BAA below MRLs
Tenx>. = 25 C **•» KfiAA balow k/«_ +
+/^^
S~~~~~
,/
* X^
- +-^-X
-
•jr^r-*r _ , - £i
) 2O 40 60 8O 1C
Run Days
FIGURE 6-96
DO
5
ra
o
o
O
<
<
O
Q
SDS DCAA vs Run Time for GAG
Column Influent and Effluent, Utility 1 1
+—+GAC Inf A A GAG Eff
3-day SDS 3-day SDS
30
20
10
0
CI2 = 4.5 mg/L
pH = 8.2
Temp. = 25 C
* Below MR.
* * * A
j . j. j . I -L .ijd^. .i. J. J. i. i^y—*.-y l'1' i ' i It- ' • • ' 1 ' ' ' '-*—*--
2O 40 60
Run Days
FIGURE 6-97
SO
...A
100
-------
o>
0)
0
SDS DBAA vs Run Time for GAG
Column Influent and Effluent, Utility 1 1
+—+GAC inf
3-day SDS
6 r
0
CI2 = 45 mg/L
pH = 8.2
Temp. = 25 C
.^-..J...! .1 .I£L ... J.J.,.1^1
0 2O
A AGAC Eff
3-day SDS
* Below
4O
6O
Run Days
FIGURE 6-98
8O
1OO
o»
-------
Treatment Modification Results and Discussion
influent bromide. This low level of bromide did not lead to bromoform production and
caused only low levels of DBAA formation in the column influent SDS samples. The
shift in speciation from chlorinated to brominated compounds in the GAC contactor
effluent may be due to the following:
o Because bromide is not adsorbed by GAC, it is possible that the increased
percentage of brominated THMs and HAAs in the column effluent is due to
the increased ratio of bromide to precursor material after significant levels of
TOC have been removed in the GAC contactor. Bench-scale TOC dilution
experiments conducted at Metropolitan (McGuire et al., 1989) indicated that
as the TOC of a water was diluted while the bromide concentration was held
constant, the speciation of THMs shifted toward the brominated compounds.
o Some researchers have found that the selectivity of precursors for either
bromination or chlorination reactions may be a function of precursor
molecular weight (Schnoor et al.. 1979), and the average of which is most
likely altered significantly by adsorption within the column. However, other
researchers have found no relation between THM speciation and molecular
weight (Glaze et al.. 1980).
Removal of SDS HANs by GAC treatment is plotted in Figure 6-99. Very low levels
of HANs were observed in the GAC contactor effluent SDS samples, with HAN
concentrations showing evidence of slightly increasing toward the end of the sampling
period. SDS HK removal is illustrated in Figure 6-100. Overall, very low
concentrations of HKs were found in the GAC contactor effluent SDS samples.
SDS chloral hydrate levels are plotted in Figure 6-101 for the GAC column influent and
effluent. Contactor influent levels are relatively high, in the range of 9.5 to 22 //g/L.
The GAC was very effective in removing SDS chloral hydrate for the first 54 days of
column operation, with concentrations at or near detection limits for the first four
samples. However, concentrations gradually increased in the next two contactor
effluent SDS samples, with a level of 4.0 //g/L measured on the 95th run day. As will
be discussed below, chloral hydrate levels in Utility I I's distribution system were also
relatively high.
SDS chloropicrin concentrations in the contactor influent and effluent are plotted in
Figure 6-102. Levels of this compound in the GAC column effluent SDS samples were
at or near detection limits throughout the sampling period. Cyanogen chloride and
2.4.6-trichlorophenol were not detected in any GAC column influent or effluent SDS4 ?
samples.
Table 6-19 summarizes TOX results for the six sampling dates at all sampling
locations. Because the GAC facilities received unchlorinated water, data from the GAC
column influent (B2) indicate that some TOX was present in the raw water (12 to 26
//g/L). which would be expected due to the industrial nature of the raw water source.
On Run Days 0.2 through 82, the 12 to 16 /ug/L of TOX in the GAC column influent
was removed (or lowered in concentration) by GAC treatment, since no TOX was
detected (detection limit equals 10 //g/L) in the GAC effluent on those days. However,
on the 95th run day, 26 //g/L of TOX was measured in the unchlorinated GAC column
influent and 21 //g/L of TOX was measured in the unchlorinated GAC column effluent,
6-25
-------
u
0)
ro
ro
p
£1
u
SDS HKs vs Run Time for GAG
Column Influent and Effluent. Utility 1 1
+—+GAC inf
3-day SDS
A AGAC Eft
3-day SDS
1.0
0.8
0.6
0.4
0.2
O.O
CI2 = 4.5 mo/L
pH - 6.2
Tennp. - 25 C
* 1.1.1-TCP below HFL.
*•» Botn i-K» batow ivH_s
«»» 1.1 -OCP Mow I*RL
.-A.
•A*..
...-A
•A"
) 10 2O 3O 4O 50 6O 7O SO 9O 10O
Run Days
FIGURE 5-100
SDS CH vs Run Time for GAG
Column Influent and Effluent. Utility 1 1
- GAG Inf ---A--- GAC Eff
3-day SDS 3-day SDS
30
CI2 = 4.5 mo/l_
PH = 8.2
Tenrp. = 25 C
* Below N/R_
20
10
A'f"
U
4O
6O
ao
10O
Run Days
FIGURE 6-101
-------
y
a
o
o
6
SDS CHP vs Run Time for GAG
Column Influent and Effluent, Utility 1 1
+ +GAC Inf A AGAC Eff
3-day SDS 3-day SDS
CI2 = 45 mg/L
pH = 8.2
Tetrp. = 25 C
» Beio*
. I . I .1. .'. -t. J- 1 . I ^ ' J_J.J.*.t.L.«-
4 O 6O
Run Days
FIGURE 6-102
-'-'•tfr
SO
A
1OO
Ol
20O
Formation of TTHMs
In Distribution System (Utility 1 1)
Clearwell
Effluent
Dist System
T = 3 days
4/10/89 4/24/89 5/8/89 6/12/89 7/17/89 7/31/89
Sampling Date (Week of)
FIGURE 6- 103
-------
TABLE 6-19
UTILITY 11 TREATMENT STUDY
TOX RESULTS (in //g/L)
Run Day 0.2 13 25 54 82 95
Sampling Date 4/10/89 4/24/89 5/08/89 6/12/89 7/17/89 7/31/89
Location/
Condition
A2/lmmed.1 140 110 150 250 260 290
A3/Immed.2 180 150 170 240 270 341
B2/Immed. 16 12 13 16 14 26
B2/SDS4.53 220 170 180 270 270 300
B3/Immed. < 10 < 10
-------
Treatment Modification Results and Discussion
indicating that the halogenated organic compounds present in the raw water were not
removed by GAC treatment.
As indicated in Table 6-19, the SDS45 TOX concentrations in the GAC column influent
samples ranged from 170 to 300 jt/g/L, while concentrations in the GAC effluent
samples ranged from only 10 to 86 //g/L. SDS^5 TOX levels in the GAC effluent
remained below 20 x/g/L in samples collected during the first 54 run days, but more
than doubled on each successive sampling date after 54 run days.
A comparison of SDS data with actual in-plant and distribution system data is made in
Table 6-20 for three of the six sampling days for the levels of chloroform, TTHMs,
DCAA. HAASU1I1 and TOX. The demonstration plant filter effluent (B2) SDS, 5 was run
to provide a controlled basis of comparison with the SDS45 GAC column effluent (B3)
samples as discussed in the previous paragraphs. The distribution system sampling
point (A3) had an approximate residence time of 3 days, but actual in-plant chlorine
doses varied between 2.7 and 4.8 mg/L, temperatures in the distribution system varied
between 11 and 27°C. and distribution system pH ranged from 8.4 to 8.8 over the
95-day sampling period.
In comparing the DBFs measured in the B2 SDS and A3 samples, varying results are
seen. For the samples collected the week of 4/10/89, there are substantial differences
in the levels of chloroform and TTHMs between the B2 SDS4 5 and A3 samples. These
differences were probably due, in part, to the difference in temperatures (A3 was I I°C
and B2 was incubated at 25°C). However, for the samples collected the week of
7/31/89, the SDS and actual conditions were very similar, and the DBF data shown in
Table 6-20 for that week agree to 22 percent or better. The data presented in Table
6-20 illustrate the importance of utilizing a standardized SDS protocol for the
evaluation of DBF precursor removal by GAC at Utility 11. By conducting SDS tests.
variables such as pH. temperature, chlorine dose and holding time were held constant
over the sampling period so that the only independent variable was TOC.
DBF Levels Produced by Conventional Treatment: Results and Discussion
Although the primary objective of the treatment study at Utility 11 was to evaluate
removal of DBF precursors by GAC adsorption, the sampling program at this utility
offered an opportunity to investigate levels of DBFs produced in conventional treatment
and in a chlorinated distribution system, and thus expand upon the DBF data collected
under the baseline sampling program.
Figure 6-103 illustrates TTHM levels measured in Utility I I's clearwell effluent and
distribution system. As noted previously, the clearwell effluent (A2) was sampled on
Monday of each sampling week, and the distribution system sampling point (A3),
having an approximate residence time of 3 days in the system, was sampled 3 days
later. This figure illustrates two important points. First, the seasonal change in THM
production from spring to summer conditions is apparent as the water temperature at
A2 increased from 9 to 27°C. Additionally, the chlorine demand of the water increased
over the sampling period, since applied chlorine doses increased from 2.7 to 4.8 mg/L
in the plant, while free chlorine residuals at the distribution system sampling point
decreased from 0.9 to 0.3 mg/L from the 4/13/89 to the 8/3/89 samples (see Table
6-18). TTHMs measured at A2 increased from 32 to 84 //g/L from the early April to
6-26
-------
TABLE 6-20
UTILITY 11 TREATMENT STUDY
COMPARISON OF SDS AND DISTRIBUTION SYSTEM RESULTS
Location
4/10/89
A3 (4/13/89)
B2/SDS 4.5
5/8/89
A3 (5/11/89)
B2/SDS 4.5
7/31/89
A3 (8/3/89)
B2/SDS 4.5
Free C12
Temp. pH C12 Dose Residual TOC CHC13 TTHMs DCAA HAAsu. TOX
°C (mg/L)
-------
Treatment Modification Results and Discussion
the late July samples. Increased formation of THMs in the warm weather months was
observed in the baseline data collected for this study and DBF production was found to
be strongly influenced by water temperature (see Section 5). It is apparent that the
increase at Utility 11 is caused by factors such as the higher water temperatures and
higher TOC levels. TOC levels at A2 were 1.77, 1.51, 1.56. 2.15, 2.08 and 2.13
mg/L on the 6 sampling dates (in chronological order), indicating an increase from
April to July. It is interesting to note, however, that UV-254 values at A2 exhibit a
different trend than TOC, with Uy-254 equal to 0.063, 0.048, 0.024, 0.034, 0.032
and 0.038 cm ' on successive sampling days from early April to late July.
The second important issue illustrated in Figure 6-103 is the production of DBFs in the
distribution system of Utility II. TTHMs measured at A3 were substantially higher
than levels measured at A2 on all six sampling dates. For the first two sampling dates,
TTHM concentrations were approximately double after 3 days of retention in the
distribution system compared to the clearwell effluent TTHMs. For the six samples
collected for this study, increases of 29, 36, 33, 57, 42 and 60 /ug/L of TTHMs
occurred in the distribution system.
Figure 6-104 shows the fate of DCAN in Utility I I's distribution system. In contrast
to the increased levels of TTHMs observed in the previous figure, DCAN levels
decreased with residence time in the distribution system, with the decrease increasing in
magnitude over the six sampling dates. This figure highlights an issue discussed in
Section 5. where results of this study confirmed the results of other researchers who
found that DCAN was a reactive intermediate rather than a stable endproduct of the
chlorinalion of natural organic matter. DCAN is known to undergo a base-catalyzed
decomposition (Trehy and Bieber, 1981), and the hydrolysis rate is significantly higher
at pH 8.5 than at neutral pH (Croue and Reckhow, 1988) (the pH at A3 ranged from
8.4 to 8.8.).
Chloral hydrate levels at Utility 11 measured during the baseline data collection were
relatively high compared to other utilities, exceeding the 75th percentile values in the
summer and fall samples. The chloral hydrate concentrations measured in Utility i I's
clearwell effluent samples in April through July, 1989 are consistent with those
recorded in the summer quarter baseline sample in 1988. Figure 6-105 illustrates the
concentrations of chloral hydrate at A2 and A3 for the six sampling dates. In the first
four samples, chloral hydrate levels approximately doubled from A2 to A3, however the
increase measured at A3 was significantly lower in the last two samples. In the
6/12/89 sample, the highest level of chloral hydrate was measured (20 /ug/L), and this
compound was the fourth most prevalent of the measured DBF compounds on a weight
basis, exceeded only by chloroform (94 /ug/L), bromodichloromethane (26 /ug/L), and
DCAA (26 //g/L).
6-27
-------
en
O
Q
Formation of DCAN
In Distribution System (Utility 11)
Clearwell HH Dist System
Effluent T = 3 days
4/10/89 4/24/89 5/8/89 6/12/89 7/17/89 7/31/89
Sampling Date (Week of)
FIGURE 6-104
0)
4—
ro
ti
ro
b
n
U
Formation of CH
In Distribution System (Utility 11)
Clearwell
Effluent
Dist System
T = 3 days
4/10/89 4/24/89 5/8/89 6/12/89 7/17/89 7/31/89
Sampling Date (Week of)
FIGURE 6-105
-------
Section 7
Summary and Conclusions
-------
SECTION 7
SUMMARY AND CONCLUSIONS
In this section, major study results are summarized and conclusions based on study
results are drawn. Additionally, research needs arising from study results are discussed.
BASELINE RESULTS
The median TTHM value for (he four quarters of baseline sampling at clearwell
effluents was 39 ^g/L (computed on a running annual average basis for each individual
utility). The THMs represented the largest class of DBFs measured in this study (on a
weight basis), comprising approximately 54 percent of the measured compounds. The
second largest DBF fraction measured in this study was HAAs, with a median
concentration of 17 /vg/L. which represents approximately 25 percent of the measured
compounds. Thus, the median level of TTHMs was approximately twice that of HAAs.
The third largest fraction detected was the aldehydes, comprising approximately 9
percent of the measured compounds, with a median concentration of 5.7 //g/L.
Little difference was observed in the overall median concentrations of influent water
quality parameters and concentrations of DBF compounds in clearwell effluents from
quarter to quarter in the baseline samples. Because of the nationwide distribution of
participating utilities and the wide variations in seasonal weather conditions and water
temperatures, "season" or "sampling quarter" proved to be a somewhat arbitrary
method of categorizing the data. Water temperature, however, was found to have a
significant impact on concentrations of DBF compounds. Levels of XDBPsnni and
THMs were found to be higher in the highest temperature range when the data were
sorted into four equal temperature ranges. The differences in DBF levels between the
highest temperature range and any of the three lower ranges were statistically
significant at a 95 percent confidence level. In addition, plant effluent pH was found
to strongly impact a number of DBFs which are unstable and hydrolyze at high pH.
Raw water quality characteristics varied considerably among the 35 participating
utilities. Relative concentrations of TOC, UV-254 absorbance. chloride and bromide
indicated that classification of the utilities and their DBF levels by source water type.
treatment type and disinfection scheme was too simplistic to account for other factors
having considerable impact on DBF production.
Classification of the utilities by disinfection scheme indicated that influent parameters
varied considerably among the utilities within each category. Prechlorinating/
poslammoniating utilities were found to have a higher median influent bromide
concentration than chlorine-only utilities, with the difference being statistically
significant at a 95 percent confidence level. Significant differences were also observed
between median concentrations of some DBFs as a function of disinfection scheme.
Median levels of XDBPMm, and TTHMs for the prechlorinating/postammoniating
utilities (approximately 94 and 57 //g/L. respectively) were significantly higher than
those of either chlorinating (62 and 34 j/g/L. respectively) or chloraminating utilities
7-1
-------
Summary and Conclusions
(2.1 and 12 //g/L. respectively). The same trend was observed in median concentrations
of HANs. aldehydes and cyanogen chloride, and the differences were also statistically
significant at a 95 percent confidence level. However, the prechlorinating/
postammoniating utilities treated water with the highest median level of UV-254 (0.15
cm1 as compared to 0.10 cm1 for chlorine-only and chloraminating utilities). Since
UV-254 may be an indicator of THM formation potential, the former utilities may
require postammoniation to minimize further THM formation in their systems.
A strong correlation was found between TTHMs and the sum of measured halogenated
DBFs (r=0.96). However, since TTHMs represented over 50 percent of XDBPsum. the
correlation coefficient for TTHMs and the sum of non-THM DBFs was determined and
found to be 0.76. Some of the other DBF classes correlated well with TTHMs (e.g..
HANs). while others correlated poorly with TTHMs (e.g., HKs).
HAAs correlated strongly (r = 0.98) with the sum of non-THM XDBPs. When both
HA As and TTHMs were subtracted from XDBP and were then correlated with
HAAs. r equaled 0.77. Fn addition, a correlation of 0.74 resulted between HAAs and
XDBPslim minus HAAs. The latter two correlations may be useful in helping to predict
the sums of non-THM. non-HAA XDBPs.
The correlation between the UV-254 absorbance and TOC of plant influent samples was
strong (r=0.85). although neither UV-254 nor TOC correlated well with the TTHMs of
plant effluents. Influent chloride correlated strongly with influent bromide (r=0.97);
thus, chloride may be used as a predictor for bromide. Exclusion relationships were
found for bromide with chloropicrin, 1.1,1-TCP, TCAA, chloroform, and other
chlorinated DBFs: that is. the presence of bromide appeared to exclude the presence of
the particular DBF. and the inverse was also observed.
Bromide present in the source water was found to shift the distribution of THMs,
HANs and HAAs to the more brominated species. High bromide levels were observed
not only in utilities susceptible to tidal influences and saltwater intrusion, but at inland
utilities as well.
Of the 35 utilities included in this study, only three employed ozone, yet almost all 35
had detectable levels of formaldehyde and acetaldehyde. These aldehydes were found
in some plant influent samples, and chlorination alone was found to produce these
compounds.
Chloramines are recognized as an effective control strategy for THMs and other DBFs.
However, for most waters studied in this project, cyanogen chloride was found to be
preferentially produced in chloraminated water. The distribution of cyanogen chloride
could statistically be divided into two groups, depending on whether the final
disinfectant was chlorine or chloramines.
The TOC removal within filtering plants included in the baseline sampling program
averaged approximately 24 percent. It should be noted that the treatment practices of
the utilities participating in the baseline sampling most likely focused on turbidity
control and were not optimized with respect to TOC removal.
7-2
-------
Summary and Conclusions
The TTHM data from this study of 35 utilities was compared to that of the THM
survey conducted by the American Water Works Association Research Foundation in
1987. which involved 727 utilities around the nation. Based on running annual
averages, frequency distribution curves for the two sets of data were similar. The
hypothesis that the data from the two studies were from the same distribution was not
rejected at a significance level of 0.01, using a Kolmogorov-Smirnov test.
PROCESS MODIFICATION RESULTS
Ozone in conjunction with chlorine or chloramines as final disinfectants was generally
effective in lowering concentrations of classes of halogenated DBFs, except lor HKs.
However, observed increases in HKswere equal to or less than 1.0 //g/L. The extent to
which DBF levels were decreased or increased after implementation of ozonation
depended primarily on the final disinfectant which was employed. Where comparable,
ozonation followed by chloramination was more effective in reducing levels of
halogenated DBF classes and chloral hydrate than ozone followed by chlorination. For
all ozone studies, ozone addition resulted in decreased TOX concentrations.
Although the use of ozone and chlorine or chloramines appeared to be effective in
minimizing the formation of compounds in the major halogenated DBF classes, shifts to
greater concentrations of the brominated species were observed for THMs and HAAs
when free chlorine was used as a final disinfectant. Increases in chloropicrin were
observed in some cases where ozone was used first, regardless of the final disinfectant;
however, these increases were always less than 1.0 //g/L. When chloramines were
employed after ozonation. increases in cyanogen chloride concentrations were observed
at some utilities.
Aldehyde concentrations increased substantially whenever ozonation was employed.
These increases ranged from 67 to 459 percent, depending on the treatment
modification implemented. However, at one utility, the use of filtration to which a
disinfectant was not applied, indicated that these aldehydes could be removed. It is
important to note that free chlorine also produced aldehydes, although to a lesser extent
than ozone.
At one utility studied, chlorine dioxide preoxidation with free chlorine for residual
disinfection was not found to lower concentrations of DBFs when compared to chlorine-
only oxidation/disinfection. However, at another utility, chlorine dioxide preoxidation
was found to be an effective control method for DBFs compared to chlorine
preoxidation. even though free chlorine was detected in the chlorine dioxide generator
product. Use of chlorine dioxide preoxidation at this utility led to decreases of
approximately 50 percent in levels of XDBPsum, TTHMS, HANs, HKs and HAAs
compared to free chlorine preoxidation.
At two utilities studied, increasing alum dose increased removal of DBF precursors and
resulted in lower concentrations of DBFs. However, at one utility, chlorine was added
before the removal of TOC in the coagulation, flocculation, sedimentation and filtration
processes, resulting in less effective DBF control.
7-3
-------
Summary and Conclusions
At one utility. GAC was very effective in removing TOC during the initial 54 clays of
column operation. After the 54th run clay. TOC in the column effluent increased
steadily. Levels of DBFs in column influent and effluent samples which were
chlorinated and held under strictly controlled conditions indicated that DBFs followed
the same trend as TOC: that is. very low levels for the first 54 days of column
operation, followed by a steady increase. Concentrations of individual THMs and
HAAs in the GAC filter effluent indicated that GAC treatment caused a shift from
chlorinated to brominated species.
Although this study focused on DBF concentrations and control of DBFs by treatment
modifications, it should be noted that in full-scale applications, DBF control strategies
must be evaluated not only on their effectiveness for limiting concentrations of DBFs,
but on their economic and operational feasibility as well. Evaluation of these aspects of
DBF control at full scale were beyond the scope of this study.
RESEARCH NEEDS
Several areas warranting further research are apparent from the study results.
Additional research on the effectiveness of currently accepted DBF control methods,
such as chloramination and ozonation, is justified due to the detection of
chloramination by-products, such as cyanogen chloride, and ozonation by-products,
such as formaldehyde, in this and other studies. Additionally, this study focused on the
occurrence and control of DBFs, rather than the economic and operational feasibility of
the DBF control strategies investigated, subjects of critical importance for utilities faced
with meeting future DBF regulations. Consequently, operation and maintenance
considerations and associated costs are subjects in need of further investigation.
Another area of needed effort is the development of water quality data for utilities
around the nation for modeling purposes, especially TOC and bromide data. In order
to develop a background of information on the present and projected future treatment
capabilities of the nation's utilities with respect to DBFs and DBF precursors, it is
essential that TOC be more routinely measured. The feasibility of monitoring plant
performance based on TOC, instead of a parameter such as turbidity, in order to
control DBF production also warrants investigation.
Treatment modification studies at two utilities also demonstrated that incremental
increases in TOC removal could be accomplished by increasing alum doses. More
research needs to be focused on enhancing TOC removal by increased coagulant doses
and on overall optimization of the coagulation process. Moreover, data from such
research needs to be developed on a nationwide basis.
This study highlighted the role of bromide in the formation of DBFs and the
importance of brominated analogs, such as brominated HAAs and possibly brominated
picrins. Further investigation 01 brominated DBF compounds is required for continued
progress in the characterization of compounds contributing to TOX.
A great deal of effort in this project was devoted to development of analytical methods
tor the compounds of interest, and effective sampling and preservation techniques.
7-4
-------
Summary and Conclusions
Further research is needed to expand the list of analytes investigated in future studies
and protect valuable data by improving sample preservation methods.
This study focused on levels of DBFs produced in full-scale drinking water treatment
facilities. In evaluating DBP production within plants and distribution systems, bench-
scale (simulated distribution system) research was also conducted for this study to
provide a controlled environment from which comparisons between various treatments
could be made. In this report, results of a great deal of laboratory research from other
studies were compare with the results of this study. However, further research is
required to more fully apply the results of strictly controlled laboratory research to
"real world" applications in full-scale water treatment facilities. Operation of full-scale
facilities is not easily compared to laboratory experiments, in that raw water quality and
plant operation may vary from day to day or season to season, and oxidants and
disinfectants are added at multiple points with various contact times and degrees of
mixing, among other differences. Further research in these areas will increase
understanding of the presence and control of DBFs in drinking water.
7-5
-------
References
-------
REFERENCES
Aieta. E.M.; Reagan, K.M.; Lang, J.S.; McReynplds, L.; Kang, J.W.; & Glaze,
W.H. Advanced Oxidation Processes for Treating Groundwater Contaminated
With TCE and PCE: Pilot-Scale Evaluations. Jour. AWWA, 5:64 (1988).
Aizawa. T.; Magara, Y.; & Musashi, M. Effect of Bromide Ions on
Trihalomethane (THM) Formation in Water. Aqua. 38:165 (1989).
Amy. G.L.; Chadik, P.A.; & Chowdhury, Z.K. Developing Models for Predicting
Trihalomethane Formation Potential and Kinetics. Jour. AWWA. 79:7:89 (1987).
Babcock. D.B.: & Singer. P.C. Chlorination and Coagulation of Humic and
Fulvic Acids. Jour. AWWA, 71:3 (1979).
Becke. C.: Maier. D.; & Sontheimer, H. Herkunft von Trichlornitromethan in
Trinkwasser. Vom Wasser. 62:125 (1984).
Bruchet, A.; Tsutsumi, Y.; Duguet, J.P.; & Mallevialle, J. Characterization of
Total Halogenated Compounds During Water Treatment Processes. In: Water
Chlorination: Chemistry. Environmental Impact and Health Effects, Vol.5. R.L.
Jolley. et al.. eds. Lewis Publ.. Inc.. Chelsea, MI (1985).
Chadik. P.A.: & Amy, G.L. Removing Trihalomethane Precursors from Various
Natural Waters by Metal Coagulants. Jour. AWWA, 75:10 (1983).
Chow. B.M.; & Roberts. P.V. Halogenated Byproduct Formation by CIO-, and
CI2. Journal. Environ. Engrg. Div. - ASCE, 107:EE4 (1981).
Chowdhury. Z.K.; Cline, G.C.; Wiesner. M.R.; & Hiltebrand, D.J. Monitoring
of Flocculation Using UV-Absorbance. Proceedings, AWWA Water Quality
Technology Conference, St. Louis. MO (Nov. 13-17, 1988).
Christman. R.F.; Norwood, D.L.; Milington, D.S.; Johnson, J.D.; & Stevens.
A. A. Identity and Yields of Major Halogenated Products of Aquatic Fulvic Acid
Chlorination. Environ. Sci. & Technol., 17:10:625 (1983).
Clark. R.M. Evaluating the Cost and Performance of Field-Scale Granular
Activated Carbon Systems. Environ. Sci. Technol.. 21:6 (1987).
Coleman. W.E.: Lingg. R.D.: Melton R.G.: & Kopfler, F.C. The Occurrence of
Volatile Organics in Five Drinking Water Supplies Using GC/MS. In:
Identification and Analysis of Organic Pollutants in Water. L.H. Keith, ed. Ann
Arbor Sci. Publ. Inc.. Ann Arbor" Ml (1976).
Croue. J.P.: & Reckhow. D.A. The Destruction of Chlorination Byproducts with
Sulfite. Presented at the AWWA Annual Conference, Orlando, FL (June 19-23,
1988).
-------
DeLeer, E.W.B.; Damste, J.S.S.; Erkelens, C.; & De Galan, L. Identification of
Intermediates Leading to Chloroform and C-4 Diacids in the Chlorination of
Humic Acid. Environ. Sci. Technol., 19:6:512(1985).
DeMarco, J.; Miller, R.; & Hartman, D.J. Discovery and Elimination of Dioxins
From a Carbon Reactivation Process. Jour. AWWA, 80:3 (1988).
DeMarco. J.: Stevens. A.A.; & Hartman. D.J. Application of Organic Analysis
for Evaluation of Granular Activated Carbon Performance in Drinking Water
Treatment. In: Advances jn the Identification & Analysis of Organic Pollutants in
Water. Vol. 2. L. H. Keith, ed. Ann Arbor Sci. Publ. Inc.T Ann Arbor, MI
(T9TT).
Dempsey. B.A.; Ganho. R.M.; & O'Melia, C.R. The Coagulation of Humic
Substances by Means of Aluminum Salts. Jour. AWWA, 76:4 (1984).
Dore. M.; Merlet. N.: Legube. B.; & Croue, J.P. Interactions Between Ozone,
Halogens and Organic Compounds. Ozone Sci. & Engrg., 10:53 (1988).
Duguet, J.P.; Tsutsmui, Y; Buchet, A.; Mallevialle, J. Chloropicrin in Potable
Water: Conditions of Formation and Production During Treatment Processes. In
Water Chlorination: Environmental Impact and Health Effects Vol.5, R.L. Jolley,
et al.. eds. Ann Arbor Sci. Publ., Inc.. Ann Arbor, MI (1985).
Fair, P.S. (Chemist. Technical Support Division, U.S. Environmental Protection
Agency. Cincinnati. OH) Personal communication (May, 1988a).
Fair. P.S. Comparison of Two Haloacetic Acid Methods in Use at MWDSC. In-
house memorandum. Drinking Water Quality Assessment Branch, U.S.
Environmental Protection Agency. Cincinnati, OH (Nov. 28, 1988b).
Federal Register. National Primary Drinking Water Regulations; Substitution of
Contaminants and Drinking Water Priority List of Additional Substances Which
May Require Regulation Under the Safe Drinking Water Act. 53 FR 1892 (January
22/1988).
Fleischacker. S.J.; & Randtke, S.J. Formation of Organic Chlorine in Public
Water Supplies. Jour. AWWA, 75:3 (1983).
GJaze, W.H.. et al. Ozone as a Disinfectant and Oxidant in Water Treatment. In:
Disinfection By-Products: Current Perspectives, AWWA, Denver, CO (in press,
I989a).
Glaze. W.H.: Koga. M.; & Cancilla, D. Ozonation Byproducts. 2.
Improvement of an Aqueous-Phase Derivatization Method for the Detection of
Formaldehyde and Other Carbonyl Compounds Formed by the Ozonation of
Drinking Water. Environ. Sci. & Technol.. 23:7:838 (1989b).
Glaze. W.H.: Rang. J.W.; & Chapin, D.H. The Chemistry of Water Treatment
Processes Involving Ozone. Hydrogen Peroxide and Ultraviolet Radiation. Ozone
Sci. & Engrg.. 9:335 (1987).
-------
Glaze. W.H.; Peyton, G.R.; Lin, S.; Huang. R.Y.; & Burleson, J.L. Destruction
of Pollutants in Water with Ozone in Combination with Ultraviolet Radiation. 2.
Natural Trihalomethane Precursors. Environ. Sci. & Technol., 16:454 (1982).
Glaze, W.H.; Saleh, F.Y.; & Kinstley, W. Characterization of Nonvolatile
Halogenated Compounds Formed During Water Chlorination. In: Water
Chlorination: Environmental Impact and Health Effects, Vol. 3. R.L. Jolley, et al.
eds. Ann Arbor Sci. Publ., Inc.. Ann Arbor, MI (1980).
Gordon. G.; Yoshino. K.: Themelis. D.G.; Wood. D.W.; & Pacey, G.E.
Utilization of Kinetic Based Flow Injection Analysis Methods for the
Determination of Chlorine and Oxychlorine Species. Anal. Chem. Acta (in press,
1989).
Gurol. M.D.: Wowk. A.: Myers. S.: and Suffet. l.H. Kinetics and Mechanism of
Halofprm Formation: Chloroform Formation from Trichloroacetone. In: Water
Chlorination: Environmental Impact and Health Effects, Vol. 4, R.L. Jolley, et al.,
eds. Ann Arbor Sci. Publ., Inc.! Ann Arbor, MI (1983).
Helsel. D.R.: & Cohn, T.A. Estimation of Descriptive Statistics for Multiple
Censored Water Quality Data. Water Resources Research, 24:12:1997 (1988).
Hirose. Y.. et al. Formation of Cyanogen Chloride by the Reaction of Amino
Acids with Hypochlorous Acid in the Presence of the Ammonium Ion.
Chemosphere. 14:11/12:1717(1988).
Hoaglin. D.C.: Mosteller. F.; & Tukey. J.W. Understanding
Exploratory Data Analysis. John Wiley and Sons, Inc., New York. N
Robust and
Y (1983).
Hoehn. R.C., et al. Household Odors Associated with the Use of Chlorine
Dioxide During Drinking Water Treatment. Presented at the AWWA Annual
Conference Sunday Seminar "Identification and Treatment of Taste and Odor
Compounds". Los Angeles, CA (June 18. 1989).
Hoel. P.G. Introduction to Mathematical Statistics. John Wiley & Sons. Inc..
New York. NY (\97\).
Hoigne. J.: & Bader. H. The Formation of Trichloronitromethane (Chloropicrin)
and Chloroform in a Combined Ozonation/Chlorination Treatment of Drinking
Water. Wat. Res.. 22:3:313 (1988).
Hubel. R.E.: & Edzwald. J.K. Removing Trihalomethane Precursors by
Coagulation. Jour. AWWA. 79:7 (1987).
Kavanaugh. M.C. Modified Coagulation for Improved Removal of
Trihalomethane Precursors. Jour. AWWA, 70:11 (1978).
Koch. B.; & Krasner. S.W. The Occurrence of Disinfection By-Products in a
Distribution System. Proceedings. AWWA Annual Conference, Los Angeles, CA
(June 18-22, 1989).
-------
Koch. B.; Krasner, S.W.; & Sclimenti, M.J. A Simulated Distribution System
Disinfection By-product Method. Proceedings. AWWA Water Quality Technology
Conference, Philadelphia, PA (Nov. 12-16, T989).
Koch. B.. et al. Analysis of Halogenated Disinfection By-products by Capillary
Chromatography. Proceedings, AWWA Water Quality Technology Conference,
St. Louis. MO (Nov. 13 17, 1988).
Kopfler. F.C.. et al. Human Exposure to Water Pollutants. Presented at 169th
ACS Natl. Mtg.. Philadelphia, PA (April, 1975).
Krasner. S.W.: Sclimenti. M.J.; & Hwang, CJ. Experiences with Implementing a
Laboratory Program to Sample and Analyze for Disinfection By-products in a
National Study. In: Disinfection By-products: Current Perspectives, AWWA.
Denver. CO (in press.
Lange. A.L.: & Kawczynski. E. Controlling Organics: The Contra Costa County
Water District Experience. Journal. AWWA. 70:1 1 (1978).
Lawrence. J. The Oxidation of Some Haloform Precursors with Ozone. Presented
at the Third International Symposium on Ozone Technology, International Ozone
Institute. Paris (May, 1977).
Lykins. B.W.. Jr.: Clark, R.M.; & Adams, J.Q. Granular Activated Carbon for
Controlling THMs. Jour. AWWA, 80:5 (1988).
Lykins. B.W.. Jr.: Koffskey. W.; & Miller, R.G. Chemical Products and
Toxicologic Effects of Disinfection. Jour. AWWA. 78: 1 1 :66 (1986).
Lykins. B.W.. Jr.: & Griese. M.H. Using Chlorine Dioxide for Trihalomethane
Control. Jour. AWWA. 78:6 (1986).
McGill. R.: Tukey. J.W.: and W.A. Larsen. Variations of Box Plots. The
American Statistician. 32:1:12 (1978).
Margerum. D.W.: & Gray. E.T.. Jr. Chlorination and the Formation of N-Ch!oro
Compounds in Water Treatment. Presented at the 175th ACS NatT. Mtg.,
Anaheim. CA (March. 1978).
McGuire. M.J.. et al. Optimization and Economic Evaluation of Granular
Activated Carbon for Organic Removal. American Water Works Association
Research Foundation, Denver, CO (1989).
McGuire. M.J.: & Meadow. R.G. AWWARF Trihalomethane Survey. Jour.
AWWA. 80:1:61 (1988).
McGuire. M.J. Quality Assurance Project Plan: Disinfection By-Products in
Drinking Water. In-house report. Water Quality Division, Metropolitan Water
District of Southern California, Los Angeles, CA (June 15, 1988).
Metropolitan Water District of Southern California; & Montgomery, James M.,
Consulting Engineers. Inc. Disinfection By-Products in California Drinking Water,
Final Report to the California Department of Health Services (July, 1989).
-------
Miller, J.W.; & Uden, P.C. Characterization of Nonvolatile Aqueous Products of
Humic Substances. Environ. Sci. & Techno!.. 17:3:150(1983).
Miller, R.; & Hartman, D.J. Feasibility Study of Granular Activated Carbon
Adsorption and On-Site Regeneration. EPA-600/S2-82-087 (1982).
Montgomery. James M. Consulting Engineers, Inc. Evaluation of Ozone and
PEROXONE for Disinfection and Control of Disinfection By-Products and Taste
and Odor Compounds, Phase IV - Experiment 3, Project Status Report to the
Metropolitan Water District of Southern California (September. 1989).
Norwood. D. L.. et al. Using Isotope Dilution Mass Spectrometry to Determine
Aqueous Trichloroacetic Acid. Jour. AWWA. 78:4:175 (1986).
Ohya. T.; & Kanno, S. Formation of Cyanide Ion or Cyanogen Chloride
Through the Cleavage of Aromatic Rings by Nitrous Acid or Chloride. VIII. On
the Reaction of Humic Acid with Hypochlorous Acid in the Presence of
Ammonium Ion. Chemosphere. 14:11/12:1717(1985).
Oliver. B.C. Dihalocetonitriles in Drinking Water: Algae and Fulvic Acid as
Precursors. Environ. Sci. & Techonol.. 17:2:80(1983).
Organic Contaminants Committee, AWWA Research Division. Non-Specific
Organic Analysis for Water Treatment Process Control and Monitoring.
[Proceedings. AWWA Seminar on Non-Specific Organic Analysis for Water
Treatment Process Control and Monitoring (1985).
Quimby. B.D.: Delaney. M.F.: Uden. P.C.; & Barnes, R.M. Determination of
the Aqueous Chlorination Products of Humic Substances by Gas Chromatography
with Microwave Emission Detection. Anal. Chem.. 52:259 (1980).
Rav-Acha. C.: Serri. A.: Choshen, E.; & Limpni. B. Disinfection of Drinking
Water Rich in Bromide with Chlorine and Chlorine Dioxide, while Minimizing the
Formation of Undesirable By-products. Wat. Sci. Tech.. Vol. 17, pp. 611-621
(1985).
Reckhow. D.A.; & Singer. P.C. Mechanisms of Organic Halide Formation
During Fulvic Acid Chlorination and Implications with Respect to Preozonation.
In: Water Chlorination: Chemistry, Environmental Impact and Health Effects, Vol.
5. R.L. Jolley. et al.. eds. Lewis Publ.. Inc.. Chelsea, MI (1985).
Reckhow. D.A.: & Singer. P.C. The Removal of Organic Halide Precursors by
Pre-ozonation and Alum Coagulation. Jour. AWWA, 76:4:151 (1984).
Riley. T.L.; Mancy. K.H.; & Boettner, E.A. The Effect of Pre-ozonation on
Chloroform Production in the Chlorine Disinfection Process. In: Water
Chlorination; Environmental Impact and Health Effects, Vol. 2. R.L. Jolley. et al..
eds. Ann Arbor Sci. Publ., Inc.. Ann Arbor MI (1978).
-------
Rook. J. Haloforms in Drinking Water. Jour. AWWA, 68:168 (1976).
Sayato. Y.; Nakamora, K.; & Matsui, S. Mechanism of Formation of Chloroform
and Chloropicrin by Chlorination of Humic Acid. Suishitsu Qdaku Kenkyu 3,
127-134 (1982).
Schnoor, J.L., et al. Trihalomethane Yields as a Function of Precursor Molecular
Weight. Envir. Sci. & Technol., 13:1134(1979).
Scully. F.E.; & Bempong, M.A. Organic N-Chloramines in Drinking Water:
Chemistry and Toxicology. Environ. Health Persp.. 46:111 (1982).
Semmens. M.J.: & Field. T.K. Coagulation: Experiences in Organics Removal.
Jour. AWWA. 72:8 (1980).
Singer. P.C. Alternative Oxidant and Disinfectant Treatment Strategies for
Controlling Trihalomethane Formation. EPA/600/2-88/044 (1988).
Slocum, C. J.: Moore. L. A.; & Stevens, A. A. Analytical Criteria for Detection
of Disinfection Byproducts. Proceedings. AWWA Water Quality Technology
Conference. Baltimore. MD (November 15-20, 1987).
Sorrell. K.: & Brass. H. (U.S. Environmental Protection Agency, Cincinnati. OH)
Personal communication (1988).
STSC. Inc. Statgraphics User's Guide. STSC. Inc.. Rockville. MD (1987).
Standard Methods for the Examination of Water and Wastewater. 17th Edition.
APHA. AWWA. WPCF. Washington. D.C. (1989).
Standard Methods for the Examination of Water And Wastewater, 16th Edition.
APHA. AWWA. WPCF. Washington, D.C. (1985).
Stevens. A.A.; Slocum. C.J.; Seeger. D.R.; & Robeck, G.G. Chlorination of
Organics in Drinking Water. Jour. AWWA, 68:11 (1976).
Sverdmp. Johnson & Flemming. The Oceans, Prentice-Hall, Inc., New York. NY
(1942).
Themelis. D.G.: Wood. D.W.; & Gordon, G. Determination of Sub-mg L1
Levels of Chlorite Ion and Chlorate Ion by Using a Flow Injection System.
Anal.Chem. Acta. (in press. 1989).
Thibaud. T.; De Laat. J.; & Dore. M. Effects of Bromide Concentration on the
Production of Chloropicrin during Chlorination of Surface Waters. Formation of
Brominated Trihalonitromethanes. Wat. Res.. 22:3:381 (1988).
Thibaud. H.: Merlet. N.; & Dore. M. Formation de Chloropicrine lors de la
Chloration de Quelques Composes Organiques Nitres. Incidence d'une
Preozonation. Envir. Technol. Lett. 7, 163-176 (1986).
-------
Trehey. M.L.; & Bieber, T.I. Detection and Quantitative Analysis of
Dihaloacetonitriles in Chlorinated Natural Waters. In: Advances in the
Identification & Analysis of Organic Pollutants in Water, Vol. 2, L.H. Keith, ed.
Ann Arbor Sci. Publ. Inc.. Ann Arbor, MI (1981).
Tukey. J.W. Exploratory Data Analysis. Addison-Wesley, Reading, MA (1977).
Uden. P. C.; & Miller. J. W. Chlorinated Acids and Chloral in Drinking Water.
Jour. AWWA. 75:10:524 (1983).
Umphres. M.D.. et al. The Effects of Preozonation on the Formation of
Trihalomethanes. Ozonews. 6:3 Part 2 (1979).
United States Environmental Protection Agency. Trihalomethanes in Drinking
Water (Sampling. Analysis, Monitoring and Compliance). EPA 570/9-83-002.
USEPA (1983).
United States Environmental Protection Agency. National Interim Primary
Drinking Water Regulations; Control of Trihalomethanes in Drinking Water.
Federal Register. 44:231:68624-68707 (Nov. 29, 1979).
Van der Kooij. K.: Fisser, A.; & Hijnen W.A.M. Determining the Concentration
of Easily Assimilable Organic Carbon in Drinking Water. Jour. AWWA, 74:10
(1982).
Wallace. J.L.; Vahadi. B.: Fernandes, J.B.; & Boyden. B.H. The Combination of
Ozone/Hydrogen Peroxide and Ozone/UV Radiation for Reduction of
Trihalomethane Formation Potential in Surface Water. Ozone Sci. & Engrg.,
10:103 (1988).
Werdehoff, K.S.; & Singer, P.C. Effects of Chlorine Dioxide on Trihalomethane
and Total Organic Halide Formation Potentials and on the Formation of Inorganic
By-Products. Proceedings. AWWA Annual Conference, Denver, CO (June 22-26.
1986).
Yamada, H.; & Somiya, I. The Determination of Carbonyl Compounds In
Ozonated Water by the PFBOA Method. Ozone Sci.& Engrg. 11:2 (1989).
------- |