ENVIRONMENTAL ACTION TEAM
FOR DRINKING WATER
MISSION REPORT
REPUBLIC OF KOREA
SEPTEMBER 2 -13,1997
US-ASIA Environmental Partnership

-------
TRIP REPORT
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
DRINKING WATER ENVIRONMENTAL ACTION TEAM
REPUBLIC OF KOREA
September 2-13, 1997
Respectfully Submitted bv:
David Parker, Team Leader
USEPA Region 4, Atlanta, GA
X 2 9 1/&0
Chris Thomas
USEPA Region 4, Atlanta, GA
¦r 9^ ^9
Eric Bissonette
USEPA OGWDW/TSC
Cincinnati, OH
Hiba Shukairy
USEPA OGWDW/TSC
Cincinnati, OH
Mike Leonard
Atlanta-Fulton Co. Water Resources Commission
Atlanta, GA

-------
TABLE OF CONTENTS
I. EXECUTIVE SUMMARY
II, BACKGROUND
III. ITINERARY
IV. TECHNICAL ISSUES
A. Disinfectants and Disinfection By-Products
1.	Disinfection By-Products
2.	Application of Ozone
3.	References
B. Source Water Issues
1.	Source Water Overview
a.	General Observations
b.	Surface Water
c.	Ground Water
2.	Surface Water Quality Modeling Information
3.	Control of Nitrogen and Phosphorus Through Source Water Protection
4.	Algae Control for Taste & Odor (esp. Microcystis aeruginosa)
5.	Public Participation and Availability of Information to the Public
C. Conventional Filtration Issues
1.	Declining Rate Filter Operational Theory
2.	Filtration Processes - Operation, Maintenance and Backwashing
3.	American Water Works Association (AWWA) Filter Media Guide
D. GAC/BAC Issues
E.	Drinking Water Sludge Issues
1. Ulsan Sludge Removal From Sedimentation Basins
F.	Ammonia Concerns
1.	Ammonia Control (Breakpoint Chlorination vs. Ozonation)
2.	High pH Air Stripping to Remove Ammonia
G.	Miscellaneous
1.	Planning for Future Growth in Plaint Design
2.	Distribution System Management
a.	Microbial Monitoring and Control
b.	Corrosion Control
c.	Distribution System Preventative Maintenance
3.	Public Perception of Drinking Water Quality
4.	Fluoride - Guidelines on Selection and Methods of Addition
5.	Scum Formation Issues
V. CONCLUSIONS
VI. ACKNOWLEDGMENTS

-------
LIST OF APPENDICES
A.	G-7 Project & Summary
B.	Summers, R.S., et. al. "Assessing DBP yield: uniform formation conditions" - Journal
A WW A, 88(6), 1996.
C.	Introduction from "The Chlorine Dioxide Handbook" AWWA, Denver, CO (1998)
D.	Oxidation of Bromide; Parameters Used for Ozonation Design Purposes; and Bibliography
E.	Declining Rate Filtration
F.	Direct Filtration
G.	AWWA Standard for Filtering Material
H.	Wang, J.Z., et. al. "Biofiltration performance: part 1, relationship to biomass" - Journal
AWWA, December 1995
I.	Urfer, D., et. al. "Biological filtration for BOM and particle removal: a critical review" -
Journal AWWA, December 1997
J. Water Fluoridation

-------
I. EXECUTIVE SUMMARY
Since July 1995, US-Asia Environmental Partnership (US-AEP) sponsored EPA
Environmental Action Teams (coordinated by EPA Region 4) have worked with the Korean
Ministry of Environment on an "Advanced Drinking Water Treatment Project". Action Teams in
1995, 1996, and 1997 interacted with Korean Government Officials and traveled to water utilities
throughout the nation to identify issues of concern, provide on-site technical assistance, and to
identify potential sales opportunities for the U.S. drinking water industry in South Korea.
In early September, 1997, an Environmental Action Team (EAT) returned to South Korea
to address previously identified issues and to provide expert on-site technical assistance
concerning the following drinking water subjects: selection, application, operation, and
optimization of advanced drinking water treatment technologies (ozone, Granular Activated
Carbon (GAC) filtration, and Bacterial Activated Carbon (BAC) filtration); technology upgrades
and process optimization for existing conventional drinking water treatment plants; control of
disinfection by-products from ozone and chlorine disinfection; source water protection; and
consultation in regard to regulatory development and implementation issues.
The 1997 EAT consisted of the following members:
David Parker [team leader] - EPA Region 4, Atlanta, Georgia
Chris Thomas - EPA Region 4, Atlanta, Georgia
Dr. Hiba Shukairy - EPA Cincinnati's OGWDW Technical Support Center
• Eric Bissonette - EPA Cincinnati's OGWDW Technical Support Center
Mike Leonard - General Manager of the Atlanta-Fulton County Water Resources
Commission
The EAT visited Korea from September 2-13, 1997, and was hosted by US-AEP, the
South Korean Ministry of Environment (MOE), and the Korea Institute of Construction
Technology (KICT). The EAT traveled to drinking water treatment plants at the following
locations: Taejon, Pusan, Ulsan, Taegu, and Kimhae. While traveling, the EAT met with local,
provincial, and national government officials, along with academic researchers, water treatment
plant operators and management, and consulting engineers.
The Action Team conducted an International Symposium on Drinking Water Treatment
Issues on September 12, 1997. This was held at the National Institute for Environmental
Research in Seoul. The Symposium was attended by over 100 South Korean Government
Officials; Environmental Regulators; Drinking Water Researchers from academia and the
government; Municipal Officials; and Drinking Water Utility Managers & Operators. Symposium
Topics included forthcoming changes to the US Drinking Water Standards; current EPA Drinking
Water Research Initiatives; Source Water Protection in the US; Application Issues and Current
Research on Ozone and GAC/BAC filtration; Conventional Drinking Water Treatment
Optimization using EPA's Composite Correction Program; and State-of-the-art Technology
Applications at Drinking Water Treatment Facilities in the USA.
1

-------
This report addresses specific technical issues which were raised or observed during the
EAT's visit to Korea.
H. BACKGROUND
The Republic of Korea has embarked upon the "G-7 Project" (see Appendix A) which is
designed to increase the levels of drinking water treatment technology to that of the G-7 member
nations by the year 2000. As part of this project, the South Korean Government is evaluating
Advanced Drinking Water Treatment Technologies such as Ozone and GAC Adsorption and
Biological Filtration.
Concern over deteriorating source water quality has led the South Korean drinking water
industry to evaluate different types of treatment for biological contaminants, volatile organic
chemicals (VOCs), synthetic organic chemicals (SOCs), inorganic chemicals (IOC;), total organic
carbon (TOC), and disinfection by-products (DBPs). Surface water in South Korea, which is the
major source of drinking water, is typically characterized by significant levels of untreated sewage
(both domestic and industrial wastes), storm water and agricultural runoff. Surface water
contamination is monitored by Biological Oxygen Demand measurements and ammonia
concentration.
While in Korea, the EAT was asked to evaluate ongoing Advanced Water Treatment
(AWT) research projects, pilot plants, and construction of full-scale AWT facilities. The team
traveled to many sites in South Korea where they addressed existing operational problems and
provided input for enhancements to current/planned research and operational procedures.
m. ITINERARY
The EAT traveled to the following sites within the Republic of Korea and met with the
following people:
Day 1 [September 4] - Seoul - Met with US-AEP, KICT, and MOE personnel
Contacts: Mr. Chi-Sun Lee, US-AEP Director
Mr. Je-Ha Yang, US-AEP Deputy Director
Ms. Eun-Hee Rim, US-AEP
Dr. Seongho Hong, Soongsil University
Dr. Sang-Eun Lee, KICT Executive Director
Mr. Hyun-Je Oh, KICT Senior Researcher
Mr. Kuk Hyun Chung, MOE Director General (Water Supply and Sewage
Bureau)
Mr. Chong Chun Kim, MOE Director (Water Supply & Sewage Policy Division)
Mr. Sung Soo Kim, MOE Deputy Director (Water Supply & Sewage Policy
Division)
2

-------
Day 2 [September 5] - Taejon - Met with US-AEP, MOE, KICT, and Korea Water Resources
Corporation (KOWACO) personnel
Contacts: Mr. Je-Ha Yang, US-AEP Deputy Director
Mr. Sung Soo Kim, MOE Deputy Director (Water Supply & Sewage Policy
Division)
Mr. Bong-Woo Shin, MOE Assistant Chief (Cheong Joo Envir. Branch Office)
Mr. Won Jae Kim, KICT Senior Researcher
Mr. Young Song Choi, KOWACO Director General
Mr. Hwan Ki Lee, KOWACO Head of Research Lab
Mr. Chang Jin Ahn, KOWACO Water Supply & Sewerage Res. Team Manager
Dr. Jeong-Hyun Kim, KOWACO Senior Researcher
Dr. Jae-Heung Yoon, KOWACO Senior Researcher
Day 3 [September 6] - Seoul - Toured the Drinkiiij Water Research facilities at Soongsil
University
Contacts: Dr. Seongho Hong, Soongsil University
Day 4 [September 7] - Seoul - OFF DAY
Day 5 [September 8] - Travel to Pusan - Site visit to the City of Pusan (Deok San) Water
Treatment Plant
Contacts: Mr. Je-Ha Yang, US-AEP Deputy Director
Mr. Sung Soo Kim, MOE Deputy Director (Water Supply & Sewage Policy
Division)
Mr. Won Jae Kim, KICT Senior Researcher
Mr. Jong Gee Lee, Pusan Waterworks Headquarters, Deputy Director General
Mr. Kwang Jae Lee, Facility Section Director, Pusan Waterworks Headquarters
Mr. Seong Geun Lee, Water Supply Dept. Mgr., Pusan Waterworks Headquarters
Mr. Shi-Joong Kim, Director, Deok San Water Treatment Plant, Pusan
Mr. Fred Nichols, Senior Field Service Engineer, PCI-WEDECO (USA)
Day 6 [September 9] - Travel to Ulsan - Site visit to the City of Ulsan Water Treatment Plant
Contacts: Mr. Je-Ha Yang, US-AEP Deputy Director
Mr. Sung Soo Kim, MOE Deputy Director (Water Supply & Sewage Policy
Division)
Mr. Won Jae Kim, KICT Senior Researcher
Mr. Sang-Goo Lee, Director General, Ulsan Water Supply Headquarters
Dr. Byoung-Ho Lee, University of Ulsan
Day 7 [September 10] - Travel to Taegu - Site visit to the City of Taegu (Du-Ryu) Water
Purification Plant
Contacts: Mr. Je-Ha Yang, US-AEP Deputy Director
Mr. Sung Soo Kim, MOE Deputy Director (Water Supply & Sewage Policy Div.)
3

-------
Mr. Won Jae Kim, KICT Senior Researcher
Mr. Sung Gun Kim, Vice Chief, Taegu City Water Supply Headquarters
Mr. Jum Chul Baek, Chief, Water Distribution Division, Taegu City Water Supply
Headquarters
Mr. Ki Hyun Cho, Chief, Taegu City Water Supply Headquarters
Mr. Sung Keun Kim, Director, Taegu City Water Supply Headquarters
Mr. Won Chung, Chief, Du-Ryu Water Purification Plant
Mr. Mu Su Park, Chief, Maegok Water Purification Plant
Mr. Young Hwa Chung, Water Resources Section, Distribution Division, Taegu
City Water Supply Headquarters
Mr. Mu Ryong Kim, Advanced Water Treatment Section, Du-Ryu Water
Purification Plant
Dr. Jae-Hyoung Kang, Director, Institute of Water Quality Water Works
Day 8 [September 11] - Travel to Kimhae - Site visit to the City of Kimhae, return to Seoul
Contacts: Mr. Je-Ha Yang, US-AEP Deputy Director
Mr. Sung Soo Kim, MOE Deputy Director (Water Supply & Sewage Policy
Division)
Mr. Won Jae Kim, KICT Senior Researcher
Mr. Chae Ock Lim, Ulsan City Water Supply
Mr. Eui Jae Lee, Ulsan City Water Supply
Day 9 [September 12] - International Symposium - National Institute of Environmental Research
Contacts: Mr. Chi-Sun Lee, US-AEP Director
Mr. Je-Ha Yang, US-AEP Deputy Director
Dr. Jong-Suk Kim, Director General, National Institute of Environmental Research
Mr. Chong Chun Kim, MOE Director (Water Supply & Sewage Policy Division)
Mr. Sung Soo Kim, MOE Deputy Director (Water Supply & Sewage Policy
Division)
Dr. Chaisung Gee, KICT Director (Environmental Engineering Division)
Mr. Won Jae Kim, KICT Senior Researcher
IV. TECHNICAL ISSUES
The following issues were either brought up by Korean water officials for the EAT to address, or
were observed while on-site by EAT members.
A. Disinfectants and Disinfection by-Products
1. Disinfection By-Products
Disinfection of drinking water is used to ensure that a microbially safe
water is provided to the consumers. Chlorine disinfection is relatively inexpensive,
4

-------
easy to use, and generally effective in inactivating bacteria and viruses commonly
found in drinking water sources, and it can provide residual disinfection for the
distribution system. Therefore, it is one of the most common disinfectants used in
the US. Disinfectants can, however, lead to the formation of disinfection by-
products (DBPs), some of which are of health concern.
Although chlorine is a potent disinfectant, it reacts with the natural organic
matter (NOM) and anthropogenic organic matter, found in water sources, to form
some potentially mutagenic and carcinogenic by-products. NOM can be composed
of humic and fulvic acids, chlorophyll, proteins, carbohydrates, polysaccharides
and other components of algae and bacteria.
In the 1970's, trihalomethanes (THMs) were the first identified DBPs found
in chlorinated drinking water and were determined to be of health concern. The
US Environmental Protection Agency (EPA) set drinking water regulations to limit
the concentration of this DBP class to less than the maximum contaminant level
(MCL) of 0.10 mg/L. Current regulations are being set in two stages: Stage 1 of
the Disinfectants/Disinfection By-Product (D/DBP) Rule will decrease the MCL
for total THMs (TTHMs) to 0.080 mg/L and set new standards for the sum of five
haloacetic acids (HAA5) at 0.060 mg/L. For Stage 2 of the D/DBP Rule, it is
proposed that the MCLs for TTHMs and HAA5 be lowered to 0.040 and 0.030
mg/L, respectively {Federal Register, 1994). Stage 1 of the D/DBP Rule will be
promulgated in November 1998 and Stage 2 will be proposed in November 2000
after regulatory negotiations between participants from the EPA, representatives of
the water industry, and other interested parties.
To date, only some DBPs have been identified in disinfected drinking
waters. The regulations are set for two DBP classes (TTHMs and HAA5). The
sum of these two classes, calculated as chlorine equivalence, make up 30 to 40
percent of the total organic halogen (TOX), a summary parameter that measures
the sum of halogenated DBPs.
Several approaches are used to assess the formation of DBPs in drinking
water. The formation potential (FP) test is an indirect measure of the precursor
compounds and usually determines the maximum amount of DBPs formed in that
water. The test is designed to determine DBPFP of a sample buffered at pH 7,
chlorinated with excess free chlorine to yield a residual of 1 to 5 mg/L after
holding the sample for 7 days at 25 °C. Another method to evaluate DBP
formation is the simulated distribution system (SDS) test, where waters are
chlorinated under site-specific conditions that reflect the chlorine dose, holding
time, pH and temperature found in distribution system.-; This test is site-specific
and varies seasonally for the same treatment plant. Both test methodologies are
described in Standard Methods (1995).
5

-------
The FP test is a good measure of the maximum potential of NOM in a
water source to form DBPs. However, chlorination conditions are not
representative of actual conditions. The seasonal variability in distribution
conditions makes it difficult to compare DBP formation for a particular water.
Differences in distribution system conditions make it difficult to compare DBP
formation between utilities. A Uniform Formation Conditions (UFC) test was
developed to allow for comparisons between DBP formation data generated from
different sources (Summers et al., 1996). The test is described in detail in
Appendix B. Chlorination of a water sample is done under the following
conditions:
pH: 8.0 ±0.2
Temperature: 20.0 ± 1.0 °C
Incubation time: 24 ± 1 hr
Chlorine residual: 1.0 ± 0.4 mg/L as free chlorii.; after 24 hr.
Under these conditions that represent average conditions in US distribution
systems, DBP formation comparisons can be made between various waters.
Precursor removal prior to disinfectant addition is practiced to decrease the
formation of DBPs. This results in a decrease in NOM as measured by total
organic carbon (TOC) concentration and a decrease in the disinfectant demand,
thus, requiring less disinfectant addition to provide a residual for the distribution
system. Drinking water treatment (DWT) processes, such as conventional
coagulation and biological filtration, generally result in a decrease in TOC
concentration. Stage 1 of the D/DBP Rule requires utilities to remove a
percentage of their TOC concentration based on their initial TOC concentration in
the raw water and on their alkalinity (NODA, 1997).
Another effective method to decrease DBPs is the use of alternative
disinfectants such as chloramines, chlorine dioxide and ozone. These disinfectants
vary in their oxidative power and disinfection effectiveness, but generally result in
lower halogenated DBP formation than chlorine.
Chloramines
Chloramines are formed from the reaction of chlorine with ammonia.
Generally, they are used for disinfection, for taste and odor control, to limit the
formation of chlorinated DBPs and to provide a residual for the distribution system
to control biofilm growth. Monochloramine is the principal species formed in
drinking waters at neutral pH. Chloramination is becoming more common in the
US, given the more stringent THM regulations. DBP formation is dependent on
the method of addition of chlorine-ammonia in the process. THM formation is
higher if the plant adds chlorine and follows it by ammonia addition in a latter
6

-------
stage in the process. Chloramines however, are weaker disinfectants than chlorine
and form cyanogen chloride, a DBP that may be of health concern. Typical
concentrations of chloramines used as disinfectant in drinking water range from 1.5
to 2.7 mg/L.
Chlorine Dioxide
Chlorine dioxide, a strong oxidant is mostly used for the control of tastes
and odors. It is a strong disinfectant that does not react with NOM to form
THMs, unless a significant amount of chlorine is present from the manufacturing
process. Chlorine dioxide will react with various constituents in raw water and will
dissociate into chlorite (C102") and chloride (CI") as the main degradation species.
Chlorate (C103) may also be formed because of inefficient generation of chlorine
dioxide. The MCL for c'llorite is set at 1.0 mg/L and required compliance
monitoring is monthly because of the acute toxic effect of this by-product. To
date, no MCL is set for chlorate. The regulatory perspective, health effect
concerns and summary of advantages and disadvantages of the use of chlorine
dioxide are presented in Appendix C from Gates (1998).
One of the most significant problems with the generation of chlorine
dioxide is its purity. C102 is generally formed by reacting sodium chlorite
(NaClOj) solutions with an oxidizing agent in mechanical generators. The
oxidizing agent can be 1) gaseous or aqueous chlorine; 2) acid in combination with
a hypochlorite solution; and 3) a mineral acid by itself or combined with chlorine.
It is difficult to generate pure solutions of chlorine dioxide and the purity is driven
by cost and recovery. The regulations require 95 percent yield C102 generation
with the feed stream containing less than 5 percent chlorine by weight, to allow the
utility to obtain CT (disinfectant residual concentration X contact time) credit for
using chlorine dioxide. Generators can be designed to meet the criteria but not
without operator training and experience to optimize the process.
The chemistry of chlorine dioxide generation is very complex but can be
simplified as the reaction of sodium hypochlorite with:
1)	Chlorine gas
2 NaC102 + Cl2 (g)	2 C102 (g) + 2 NaCl	(1)
2)	Hypochlorite
2 NaC102 + HOC1	2 C102 (g) + NaCl + NaOH	(2)
3)	Hydrochloric acid
5 NaC102 + 4 HC1	4 C102 (g) + 5 NaCl + 2 H20	(3)
While the main objective is to minimize chlorine contamination during
application in drinking water treatment, it is necessary to have excess chlorine to
7

-------
drive the generation chemistry Optimizing the generation process is a very
important objective. Details on the use of chlorine dioxide in water treatment
including chemical properties and reactions, regulatory concerns, drinking water
treatment process design and operation and health effects can be found in Gates
(1998).
Ozone
Ozonation is gaining in use in the United States, for disinfection purposes,
DBP control, and for improvement in aesthetic quality including taste and odor,
color removal and iron and manganese oxidation. Ozone is typically installed at
drinking water treatment plants for multiple purposes. One of the most common
objective is to enhance disinfection while decreasing the amount of chlorine or
chloramine that is needed to maintain a residual for the distribution system.
Ozone's oxidation of the chlorine-reactive sites in NOM results in lower chlorine
demand and in decreased DBP formation. This is important in light of the more
stringent drinking water regulations.
In aqueous media, water generally participates in the reactions of ozone
with organic compounds. Under ozonation conditions generally used for microbial
inactivation in DWT, the ozone doses are not high enough to cause mineralization
of the organic matter to carbon dioxide, but some degradation of the organic
matter occurs. Because of the nature of these types of oxidation reactions,
ozonation has been used to control taste and odor compounds, color, and some
synthetic organic compounds. The reaction of ozone with organic matter also
results in an increase in the biodegradability of the organic matter. Smaller, more
hydrophilic and oxygen rich compounds are generated by the ozonation process.
Such compounds, if not controlled by subsequent biological filtration, will serve as
substrate and promote regrowth in the distribution system.
Ozone reacts with NOM and generates oxidation by-products such as
aldehydes and keto acids. Most of the identified organic ozonation by-products
can generally be well controlled in biologically active filters. A detailed description
of the reactions of ozone with NOM is presented in Appendix D (Shukairy, 1998).
Another by-product of the ozonation process is bromate. It is an inorganic
by-product formed by the oxidation of bromide (Br") by ozone. Bromide ion in
source waters may originate from natural sources, from anthropogenic processes,
and/or as an impurity in chlorinating solutions. The most common natural
contribution is from saltwater intrusion in aquifers.
The interest in bromate formation stems from the fact that it is carcinogenic
and one of the contaminants in drinking water that will be regulated by Stage 1 D-
8

-------
DBP Rule. The MCL for bromate is set at 0.010 mg/L. Potassium bromate,
which is also used as an oxidizing agent in making bread, is nephrotoxic in animals
(Kurokawa et al., 1990) and in humans (Gradus et al., 1984). It is a genotoxic
carcinogen and possesses both initiating and promoting potential for renal
carcinogenesis (Kurokawa et al., 1990). Based on a multistage model used by the
USEPA to extrapolate the results of carcinogenesis experiments with high doses in
animals to low doses in humans, the expected MCL at the 10"6, 10"5, and 104
cancer risk levels are 0.08, 0.8 and 8 ng/L, respectively. The MCL of 0.010 mg/L
that is proposed in the D-DBP Rule is based on analytical limitations and not on
the risk level (Federal Register, 1994). The minimum detection level (MDL)
range reported by analytical laboratories is in the range of 3 to 10 ng/L.
The oxidation of bromide by ozone is of particular importance in drinking
water treatment. Br" reacts with ozone with an appreciable rate constant of * 60
M"1 s"1 (14). A thorough description of the reaction of bromide with ozone,
specifically the kinetics of formation of hypobromous acid (HOBr) and bromate
(Br03") is presented in Haag and Hoigne (1983). Ozone oxidizes bromide to form
HOBr which reacts further with ozone, only when it exists in its ionized form, as
hypobromite ion (OBr~). Hypobromite is oxidized to bromate and to an
unidentified species that regenerates bromide. The following scheme of reactions
describe the formation of bromate (Haag and Hoigne, 1983).
Equations 4 and 5 represent the catalytic decomposition of ozone. HOBr/OBr"
concentrations are pH dependent, and the equilibrium shifts towards the
protonated form at lower pH. The dissociation constant (pKa) is 8.66 at 25 °C.
The reaction of ozone with HOBr is negligible, however, in the presence of
organic matter, HOBr reacts to form bromine-substituted DBPs.
Krasner and co-workers surveyed bromate formation in the US. They
estimated that the 20th, median, and 80th percentile for bromate occurrence in
surface waters using ozone for predisinfection is in the range of 0.5-0.8, 1-2 and 3-
5 ng/L, respectively. The 90th and 95th percentile are in the range of 5-20 ng/L
(Krasner et al., 1993). Therefore, because of the proposed regulations, some
utilities would not meet the MCL for bromate. These utilities would either choose
not to employ ozone in their DWT or operate the ozonation system such that
bromate formation is decreased. Bromate formation is affected by the reaction
conditions.
03 +Br" 	 02 + OBr*
03 + OBr" 	 202 + Br"
203 + OBr 	 202 + Br03
(4)
(5)
(6)
9

-------
Impact of pH: Because of the distribution of the species of free bromine
(HOBr/ OBr"), the pH effect on the formation of ozonation by-products is
significant. At low pH, HOBr dominates and the reactions with the organic matter
to form bromine-substituted by-products become the most important reactions. At
higher pH, the equilibrium shifts to OBr and this ion can be further oxidized to
bromate. Therefore, decreasing the pH of ozonation will decrease bromate
formation, but may enhance the formation of bromine-substituted DBPs. The
impact of pH on bromate formation and control has been investigated by many
researchers (AJlgeier et al., 1995; Siddiqui and Amy, 1993). Refer to Appendix D
for a detailed description of the impact of operational parameters on bromate
formation from Shukairy (1998).
Effect of Temperature: An increase in temperature will increase the rate
of ozone decomposition as the activation enc. gy increases. At the same time, an
increase in temperature will increase the reaction rates. (See Appendix D).
Effect of Ammonia Addition: TOBr and bromate formation decrease upon
the addition of ammonia. This occurs because free bromine, a result of the
oxidation of bromide by ozone, reacts very rapidly with ammonia, faster than the
reaction of bromide with ozone. The following reactions explain the reaction of
HOBr and OBr" with ammonia.
HOBr + NH3	 NH2Br + H20	(7)
OBr" + NH3	 NH2Br + OH"	(8)
The formation of dibromamine and tribromamine is also possible. Combined
bromine reacts with ozone to form nitrate and bromide (Haag et al., 1984).
NH2Br + 303 + 2H20	 N03" + Br" + 2H30+ + 302	(9)
Equation 9 indicates that ammonia appears to compete with other reactions that
involve free bromine, resulting in ozone consumption and an enhancement of its
decomposition. Consequently, a decrease in bromate formation occurs because of
enhanced ozone decomposition.
Although ammonia addition can decrease bromate and TOBr formation, it
does enhance ozone decomposition and may form other unknown by-products. In
Drinking Water Treatment plants that employ ozonation for disinfection, the
presence of ammonia should be accounted for in determining the required ozone
concentration for the purpose of disinfection.
Effect of Contact Time: In the presence of bromide and ozone residual,
ozone contact time has a significant effect on bromate formation. For a given
10

-------
ozone residual, increasing the contact time results in an increase in bromate
formation. (See Appendix D)
Effect of Ozone Dose and Residual: In the presence of bromide, increasing
ozone dose results in an increase in bromate formation. For the reaction to take
place, the ozone demand of the water should be satisfied and a residual ozone
should be available. For examples of this effect, see Appendix D.
To minimize residual concentrations, while providing adequate C*Ts,
staging ozone contacting is practiced. Krasner and co-workers (1993) have shown
that staging ozone application into two or three contactors may be beneficial in
providing adequate inactivation concentrations and CT credit while controlling
bromate formation because of actual residual decrease in the contactors.
2. Application of Ozone
Ozone may be applied for several reasons such as:
~	Disinfection and algae control
~	Oxidation of taste and odors
Iron and manganese oxidation
*¦ Decreasing DBP formation potential
~	Oxidation of organic pollutants
The location for ozone application depends on the treatment objectives and
the water quality characteristics. The most common locations include the raw
water (preozonation), following sedimentation (intermediate), and both raw water
and post sedimentation (two-stage application). In most DWT plants, ozone is
applied for multiple reasons. When iron and manganese oxidation is required, or
when ozone is used for aiding in coagulation, it is used at the head of the treatment
plant. If it is used to enhance biological filtration, then it is used as an intermediate
application prior to the filters (provided enough time is allowed for the ozone
residual to dissipate prior to the filters). Taste and odor control are best achieved
with intermediate ozonation (Langlais et al., 1991). For other uses, ozone may be
applied anywhere in the treatment train, depending on the objective.
A plant needs to determine the objective for applying ozonation and the
water quality characteristics. A determination of the ozone demand of the water is
essential. Therefore, in designing optimal ozone application the following have to
be considered:
~	Location of ozone application
*¦ Ozone dose requirements (Ozone demand has to be determined)
~	Contact time
11

-------
It is important to ensure that ozone application (selected location and dose)
are not detrimental to downstream processes.
The important parameters used for design purposes, treatability studies and
analytical determination of ozone is presented in Appendix D from Shukairy
(1998). More details can be found in Langlais et al. (1991).
3. References
AJlgeier et al., Proceedings, AWWA Annual Conference, Anaheim, CA (1995).
Gates, D., The Chlorine Dioxide Handbook, AWWA, Denver CO (1998).
Gradus, B. et al., American Journal of Nephrology, 4:188 (1984).
Haag, W.R. and Hoigne, J. Environmental Science & Technology, 17:5:261
(1983).
Haag , W.R. et al., Water Research, 18:9:1125 (1984).
Krasner, S.W. et alJournal AWWA, 85:1:73 (1993).
Kurokawa, Y. Et al., Environmental Health Perspectives, 87:309 (1990).
Langlais, B. et al. (Editors), Ozone in Water Treatment Application and
Engineering, Lewis Publishers, Chelsea, MI (1991).
Shukairy, H.M., Ozonation in Drinking Water Treatment, Chapter in
Encyclopedia of Environmental Analysis and Remediation, in press, (1998).
Siddiqui, M.S. and Amy, G.L., Journal AWWA, 85:1:63 (1993).
Standard Methods for the Examination of Water and Wastewater, APHA,
AWWA and WEF, Washington, DC. (19th Edition, 1995).
Summers, R.S. et al., Journal AWWA, 88:6:80 (1996).
USEPA, National Primary Drinking Water Regulations, Disinfectants and
Disinfection By-Products, Proposed Rule, Federal Register, 40, Parts 141 and
142, July 1994, EPA 81 l-Z-94-004.
USEPA, National Primary Drinking Water Regulations, Disinfectants and
Disinfection By-Products, Notice of Data Availability, Federal Register, (1997).
12

-------
B. Source Water Issues
1. Source Water Overview
a.	General Observations' Source water issues at all drinking water
treatment facilities that were visited were very similar. The operators were
concerned with high ammonia and total organic carbon (TOC) concentrations in
their raw water, algae control in their sedimentation basins, seasonal taste and odor
problems, concerns about organic and inorganic contaminants associated with
industrial and municipal wastewater discharges upstream of their intakes,
disinfectant by-products, and large seasonal variations in their raw water quality.
A point of interest is that no local water treatment plant operator voiced concern
over microbiological contaminants in the raw water.
These issues clearly point to national questions regarding the effectiveness
of the wastewater point source control program and the lack of emphasis on
nonpoint source pollution control. The level of uncertainty of what is going on
upstream of drinking water intakes may also be an indication of minimal public
access to environmental information, little emphasis on public environmental
education, or possibly an overall absence of upstream information. Regardless of
the cause of this uncertainty, the EAT strongly recommends that source water
assessments be conducted with the results being made available to the public and
water supply owner/operators. Source water assessments conducted in
conjunction with an ambient water quality monitoring program for the watersheds
serving as sources for drinking water intakes would be even better.
During the 1997 trip to Korea and in previous trips, actual monitoring data
of raw water, treatment plant unit operations, and of finished water were rarely
shared with the EAT. This makes it very difficult for EAT members to provide
quality judgements and recommendations regarding MOE program effectiveness
and treatment plant operation optimization. It is not clear how much and what
kind of actual monitoring information is even available.
b.	Surface Water: The EAT was told that 50-60% of municipal sanitary
wastewater and all industrial wastewater is treated and that point source
discharges are regulated through a permitting system similar to the U.S.'s NPDES
program. However it is unclear how the Korean point source discharge program
works. Many questions and issues remain regarding the effectiveness of the point
source control program, including: permit limit development; permit compliance
monitoring; who is responsible for enforcement (local or central government?);
does meaningful enforcement take place; which cities/wastewater collection
systems get treatment plants; which treatment plants get permits and which ones
do not; are all outfalls from the wastewater collection system known and are they
13

-------
all permitted; wastewater collection system coverage; industrial pretreatment
program effectiveness, and ambient surface water quality monitoring. Many
straight pipe discharges were observed coming from homes and going to creeks
and ditches. Given the population density of Korea and the low percentage of
domestic wastewater collection and treatment, the surface water resources of
Korea are certainly impacted by point source discharges.
The high percentage of untreated domestic wastewater undoubtedly is a
significant contributor to the nutrient and ammonia loadings which enter Korean
surface water resources. An adequate wastewater collection infrastructure must
be put in place. Also, the point source pollution control program must effectively
hold wastewater treatment plants accountable for the quality of their effluent
discharged into surface waters. The EAT recommends that Korea aggressively
implement their point source conu A program and make it a priority to expand the
wastewater collection infrastructure and treatment capabilities. Until adequate
infrastructure and until regulatory accountability exists, industrial and municipal'
point source pollution will continue to be a significant environmental and human
health issue for the Republic of Korea.
Agricultural activity on the Korean landscape was extensive. Rice is the
primary crop and requires a large amount of water to grow. As water moves
down through a Korean watershed, it is diverted many times through series of rice
fields. Non-point source pollution mitigation practices including fertilizer
application optimization and proper pesticide/herbicide application, storage, and
management are not likely to be occurring. Also, since the availability of
agricultural land is so important, there is no incentive for Korean farmers to utilize
buffer strips or riparian zones between agricultural land and surface water bodies.
Based on EAT observations, the non-point source pollution impacts to Korean
surface waters, especially nutrient loadings, are expected to be significant.
Expansion of the urban landscape into undeveloped and traditionally
agricultural lands is extensive. Row after row of high density, high rise apartments
are being built all over Korea. Roads and other associated infrastructure systems
needed to support high population densities are also being constructed. Available
space is a precious commodity in Korea and as a result, no storm water
detention/retention systems were associated with the construction activities, and
no observable sediment and erosion controls were in place at any of the
construction sites. Sediment loadings to surface waters as a result of the scale and
extent of the development activities are expected to be extensive.
c. Ground Water: The status and utilization of ground water resources of
Korea is very difficult to ascertain. The large size of the populations served by
most of Korean public water supplies, in conjunction with the geology and
14

-------
hydrogeology of most of the country generally prohibits the use of ground water
as a source of supply. Some smaller cities are using ground water as a source of
supply and it is certain that wells exist in rural areas which supply water to
individual homes or clusters of homes. Conversations with Korean government
officials indicated that ground water resources were a very low priority and that
little was known about the status of the Korean ground water resources.
Conversations with local officials revealed that there is a high level of
concern within the Korean population about the quality of ground water.
Historically, there have been many agricultural wells used to supply water to the
rice fields. Over the long history of the country many wells have been created with
a relative few abandoned in a way that would help maintain aquifer integrity.
Other ground water contaminant concerns seems to be those associated with the
Ku.ean War and the large military presence in the country since the 1950's. The
potential for aquifer contamination, especially surficial aquifers, appears to be high.
With ground water resources largely ignored and having a very low government
priority, it is difficult to make any definitive statement regarding its status.
An additional ground water/surface water relationship that should not be
overlooked is that of nutrient loading. United States research on the Chesapeake
Bay watershed has indicated that ground water discharge accounts for roughly half
of the nitrate flowing into Chesapeake Bay from non-tidal streams. Given the
shallow, surficial aquifers in many of the agricultural areas of Korea it is not
beyond possibility that significant nitrate loads from ground water to surface water
could exist in Korea as well.
Note: Information about the Chesapeake Bay study can be found on the U.S.
Geological Survey's home page at:
www.usgs.gov/public/press/public_affairs/press_releases/index.html or by calling
the Chesapeake Bay Science Program at 410/238-4252.
2. Surface Water Quality Modeling Information
At the International Drinking Water Symposium in Seoul, several questions
were asked regarding the water quality modeling tools that are currently available.
Copies of the EPA document "Compendium of Tools for Watershed Assessment
and TMDL Development" have been provided to USAEP as a reference. These
documents throughly document all of the modeling software tools that were
available as of 1997. Additional watershed assessment and total maximum daily
load (TMDL) development resources may be found on EPA's homepage at
www.epa.gov.
15

-------
3 Control of Nitrogen and Phosphorus Through Source Water Protection
The EAT recommends that Korea first minimize nutrient loadings to
surface waters through an aggressive implementation of its point source regulatory
program. Nonpoint source nutrient and sediment loadings should be addressed
through regulation, education, and/or through incentives to encourage
environmentally sound farming practices. The EAT encourages Korea to
investigate much of the agriculture research and literature available on nutrient
balances, pesticide application and management practices, and agriculture nonpoint
source mitigation practices. TMDLs should be developed and allocated to
dischargers in each watershed.
In addition, the EAT encourages the Korean government to investigate
other possible alternatives for reducing nutrient loads. One practice that has been
implemented in some municipal areas of the United States, including metropolitan
Atlanta, GA, is that of eliminating or reducing the percentage of phosphates in
laundry detergents. This practice is particularly effective in reducing algal
concentrations in phosphorus limited surface waters. An additional resource to
review for similar types of pollution prevention practices can be found in pollution
prevention documents available through the EPA homepage.
4. Algae Control for Taste & Odor (esp. Microcystis aeruginosa)
Blue-green algae, such as Microcystis aeruginosa have been recognized for
decades as one cause of taste and odor problems in water supplies. Recent
research has indicated that certain blue-green algae, including Microcystis
aeruginosa, Anabaena, Oscillatoria, Aphanizomenon, and Nodularia can produce
toxins. These include Hepatotoxic Tumor-Promoting Peptides and Alkaloids,
Neurotoxic Alkaloids, and Endotoxins. Although risks posed to humans from
these toxins tend to be chronic and are usually associated with toxin concentrations
greater than 1 ug/L, a recent case of toxic algae poisoning was responsible for 26
deaths due to liver failure at a dialysis center in Caruaru, Brazil, in February 1996.
Drinking water regulators around the world have taken notice of the
potential health problems associated with blue-green algae toxins. Australia and
Canada have established 0.5 - 1.0 ug/L drinking water standards for Microcystes.
Japan has proposed a limit of 0.11 ug/L for Microcystin in drinking water. The
U.S. EPA has added Cyanobacteria (blue-green algae), other freshwater algae, and
their toxins to the "Drinking Water Contaminant Candidate List". In the past,
EPA has stated that control of these organisms is best handled through good
watershed management practices. The reason that EPA has decided to add the
algae and their toxins to the final "Drinking Water Contaminant Candidate List"
because (1) pathogenic algae and their toxins are not necessarily associated with
16

-------
fecal contamination and thus may not be effectively controlled by the existing
Surface Water Treatment Rule (SWTR) or the proposed Enhanced Surface Water
Treatment Rule (ESWTR), and (2) some data suggest that current treatment
techniques may be particularly inadequate in controlling algal toxins. EPA's
decision to add the algae and their toxins to the final "Drinking Water Contaminant
Candidate List" will make them a priority for research to determine what triggers
toxic algal growth in source water and the effectiveness of water treatment
practices.
Traditionally, water systems in the United States faced with a blue-green
algae bloom will treat their source water copper sulfate to kill the algae.
Unfortunately, this practice can liberate high concentration of toxins if the blue-
green algae which are causing the bloom are also producing toxins. Blue-green
algae can be physically removed in the drinking water t. eatment process without
liberating the toxins by using either microscreens or dissolved air flotation. The
toxic compounds such as Microcystin can be removed in the drinking water
treatment process by any of the following methods: adsorption with granular
activated carbon (GAC), adsorption with powder activated carbon (PAC), ozone
oxidation (note that high TOC levels can be an interference), and oxidation with
potassium permanganate. Other types of treatment which have been successful in
the control of problematic blue-green algae blooms include in-reservoir treatments
(aeration and/or artificial circulation to prevent phosphorus release from the
sediments) and biological methods such as harvesting of the algae.
In order to prevent these algal problems, water systems must prevent the
growth of algae in the source water, which makes it very important to control the
input of nutrients such as nitrogen and phosphorous into source water. Watershed
protection programs can be a real asset to regulatory agencies as they work to
limit nitrogen and phosphorus inputs.
5. Public Participation and Availability of Information to the Public
One of the central foundations of United States environmental policy is that
of public involvement and public availability of environmental information. The
EAT encourages Korea to engage stakeholders and the general public in the
development and implementation of environmental statutes, regulations, and
policy. An educated and involved public is key to balancing all of the competing
interests in a nation's environment. In the United States, source water assessments
are required for all public water supply systems and these assessments are required
to be made available to the public. Watershed information is also made available
on the national scale through the "Surf your Watershed" section of the EPA
homepage. Similar systems are being established at the state and local levels.
17

-------
C. Conventional Filtration Issues
1.	Declining Rate Filter Operational Theory
The Atlanta-Fulton County Water System utilizes declining rate filters at
their plant. Appendix E is a discourse on operational theory of declining rate
filters. The information is from the Atlanta-Fulton County Water System's
operational manual.
2.	Filtration Processes - operation, maintenance and backwashing
Information concerning filtration can be found in Appendix F. This information is
from the American Water Wo;!.s Association's book entitled Water Quality and
Treatment and from .
3.	American Water Works Association (AWWA) filter media guide
Please refer to Appendix G for this information.
D. GAC/BAC Issues
There are many advantages to using biological filtration in DWT such as
decreasing the regrowth potential in the water (biostabilization) and decreasing the
subsequent DBPs formed from the addition of chlorine or chloramine. Treatment plants
that use ozonation are encouraged to follow it by biological filtration. This is because
biological filtration can control some of the organic ozonation by-products (OBPs) that
are formed from the oxidation of NOM by ozone. These by-products are generally small
molecular size compounds, oxygen rich, and represent the more hydrophilic components
of NOM. OBPs are generally biodegradable and can serve as substrate for
microorganisms. If not controlled by biological filtration, they may promote microbial
regrowth in the distribution system.
Biodegradable organic matter (BOM) can be utilized by a wide range of
microorganisms which include pathogens. Therefore, BOM can be utilized in a controlled
manner by biofiltration, or in an uncontrolled manner in the distribution system. In
biological filters, the heterotrophic bacteria attached to the media as biofilm will oxidize
BOM.
The main objectives of biological filtration are to control microbial regrowth,
decrease disinfectant demand, oxidize DBP precursors, and improve tastes and odors.
NOM is made up of a biodegradable organic fraction [biodegradable dissolved organic
carbon (BDOC)] and a recalcitrant fraction. The BDOC, which is the amount of DOC
18

-------
consumed in a bioreactor, is made up of rapid and slow biodegradable fractions. BDOC is
also made up of assimilable organic carbon (AOC) which is the fraction of NOM that
supports the growth of particular strains of microorganism, Spirillum NOX and
Pseudomonas P-l 7. There are also specific biodegradable organic compounds that make
up part of the BDOC. These are carbohydrates, amino acids, aldehydes, ketoacids, and
carboxylic acids.
Conventional media filters develop a biomass and become biologically active if
disinfectant residuals are not maintained in the process water as it passes through the bed.
Both anthracite and sand have been used as attachment media, with reduced headloss
associated with the larger media. Biologically active granular activated carbon (GAC) as
applied in a filter adsorber or in a post-filter adsorber is similar to conventional media
filtration, except that an interaction between the adsorptive capacity of the GAC and
jiodegradation may be present. The macroporous nature of some of the GAC may
provide greater surface area for microorganism attachment compared to other media.
While contact times are typically 5 to 30 minutes, these processes can be effective in
controlling a significant fraction of BOM if sufficient biomass is maintained in the media.
Biomass concentrations typically increase over operational time and eventually reach a
plateaux. The filter then operates at steady state. The time for the filter to achieve steady
state removals is dependent on the parameter. Time to steady state for aldehydes is much
shorter than for DBP precursors.
There are many methods to assess attached biomass. These methods range from
direct microscopic methods, to growth on culture media and measurement of cellular
biochemicals. One method to assess viable biomass uses a phospholipid analysis.
Phospholipids are contained in cell membranes and are used very rapidly during
metabolism. With cell death, the cellular enzymes hydrolyze and release the phosphate
group. The total lipids can be extracted from the cells attached to the media and measured
quantitatively. The biomass measurement standard operating procedure is presented in
Appendix H.
A number of parameters have to be taken into account when designing biofilters.
Parameters such as configuration, empty bed contact time (EBCT), filtration rate, filter
media type, biomass concentration, the degree of preozonation, and backwashing. For
details, refer to Appendix I for a critical review of biological filtration by Urfer et al.
(1997).
The configuration of biofilters are generally either one-stage or two stage: the one
stage is less expensive but must be designed to optimize particle and BOM removal in one
filter; the two-stage approach is more flexible, effective, but more expensive. The two-
stage approach generally uses biologically active carbon (BAC).
19

-------
When GAC is first used as a filter media, it operates in an adsorptive mode. At the
same time, if the filter is operating under conditions amenable to biological growth, the
filter starts to develop biomass on the media. After the filter adsorptive capacity is
exhausted, the biomass builds up on the filter media and the filter will operate as a
biological filter - BAC. Ozonation prior to biological filtration enhances the
biodegradability of NOM.
The operational parameters such as filter loading, EBCT and temperature
significantly impact the performance with respect to the control of BOM. EBCT is a very
important design and operating parameter which significantly affects BOM removal.
EBCT and not the hydraulic loading is the key parameter for biological BOM removal.
However, benefits of using longer EBCTs are likely small except at colder temperatures.
In general, OBP and AOC removals occur at shorter contact times than the removal of
chlorination by-product precursors and chlorine demand. Under colder temperatuios, the
filter may require longer EBCTs for the same performance observed at higher
temperatures. It is expected that BOM removal at higher temperatures will be higher
because of more rapid microbial kinetics and mass transfer. This effect may also be
media-type dependent. The macroporous structure and irregular surface of GAC provides
a good surface for biomass development and some protection from shearing. The
adsorption of slower biodegradable NOM on the GAC can be biodegraded over a
continuos period by the attached bacteria. This makes GAC a good media for biological
filters.
Backwashing of filters is an important issue that needs to be addressed in the
design. The success of biological filtration is dependent on managing the amount of
biomass on the filters during the filtration cycle. It is better to backwash the filter with
water that has no residual disinfectants. However, filters can be backwashed with chlorine
or chloramine in the backwash water. Under these conditions, the performance of the
filters is slightly impaired for a time period after the backwash, but biodegradability is
restored after an operational time period. Biomass concentrations (top of filters) on filters
backwashed with waters that contain no disinfectants are generally one or two orders of
magnitude higher than those backwashed with chlorine in the water. Biomass is much
more affected, by chlorine in the backwash water, than BOM removal.
In general, biofiltration results in good removals of aldehydes (80 to 95 percent),
decrease in chlorine demand by 30 to 70 percent and a decrease in DBP precursors
(dependent on the particular DBP). Biofiltration results in biostabilized water and the
removal of some of the biodegradable fraction of NOM, in a controlled manner. Thus,
employing biological filtration, particularly following ozonation, decreases the risk of
microbial regrowth in the distribution system. A major issue that needs to be addressed in
designing biofilters is to ensure that biological filtration does not compromise pathogen
and particle removal.
20

-------
E. Drinking Water Sludge Issues
1. Ulsan Sludge Removal From Sedimentation Basins
While visiting the Water Treatment Plant in Ulsan, the EAT was presented
with a problem concerning sludge removal from the sedimentation basin.
The problem, as presented to the EAT, was that sludge collects in the
bottom of the basin but does not flow out of the collection basin due to
inadequate compression and riot enough head pressure to push the sludge
out through the collection system. Figure 1 illustrates the problem and
Figure 2 illustrates the EAT's proposed solution:
Ulsan Sludge Collection
Problem
Figure 1 - Sludge was mounding in the sedimentation basin due to
compression problems and inadequate head pressure.
Ulsan Sludge Collection
Proposed Solution
Figure 2 - Sludge collects in the sludge collector at the bottom of the
sedimentation basin and is pulled into the collection pipe by a the newly
added pump and the air injectors.
21

-------
F. Ammonia Concerns
1.	Ammonia Control (Breakpoint Chlorination vs. Ozonation)
Ammonia-Nitrogen is found in the raw water supplies of South Korea. Its
presence leads to a potential taste and odor (T&O) concern and a nutrient issue
which can lead to bacterial regrowth in the distribution system. It is currently
being treated by breakpoint chlorination. One of the proposals for ozonation is to
destroy the ammonia-nitrogen and eliminate the need for breakpoint chlorination
and the possibility of creating chlorine by-products such as THMs.
In and of itself, breakpoint chlorination for the removal of ammonia-
nitrogen should not produce a significant amount of THMs if the chlorine dose is
at the breakp jint level. The EAT believes that this practice is safe and is probably
the most cost-efficient method to remove the ammonia-nitrogen. Breakpoint
chlorination can be complimented with dechlorination by sulfur dioxide (SOi),
hydrogen peroxide (H2O2), activated carbon, or aeration.
The increasing use of ozone at drinking water treatment plants in South
Korea should address this problem. Proper application of ozone should even
remove the high concentrations of ammonia-nitrogen seen at most of the water
treatment facilities the EAT visited in South Korea.
It should be noted that this contaminant can and should be controlled in the
source water by the control of agricultural run-off and the development and
implementation of a stringent standards for regulatory wastewater and industrial
discharges. This would allow for the control of ammonia-nitrogen inputs into raw
water supplies.
2.	High pH Air Stripping to Remove Ammonia
A significant amount of information on this subject is found in the
following reference documents:
a.	Powers, S.E., Collins, A.G., Edzwald, J.M., and Dietrich, J.M. "Modeling
an Aerated Bubble Ammonia Stripping Process." Journal Water Pollution
Control Federation, Volume 59, Number 2, February 1987.
b.	Gonzales, G.J. and Culp, R.L., "New Developments in Ammonia Stripping
Part 1." Public Works, 104, 5, 78 (1973).
c.	Gonzales, G.J. and Culp, R.L., "New Developments in Ammonia Stripping
Part 2." Public Works, 104, 6, 82 (1973).'
22

-------
d.	Tsai, K.C., et al., "Air and Steam Stripping of High Strength Ammonia
Wastes" Proceedings 3(?h Industrial Waste Conference, Purdue University,
1982.
e.	Srinath, E.G., and Loehr, R.C., "Ammonia Desorption by Diffused
Aeration." Journal Water Pollution Control Federation, Volume 46, 1974.
f.	Culp, R.L., et al, "Handbook of Advanced Wastewater Treatment", Van
Nostrand Reinhold Company, New York, N.Y., 1978.
g.	Stumm, W., and Morgan, J.J., "Aquatic Chemistry." 2nd edition, John
Wiley and Sons, New York, N.Y., 1991.
G. Miscellaneous
1.	Planning for Future Growth in Plant Design
Most public water systems in the United States do take future growth into
account when they design and build expansions to existing water treatment plants
or build new plants to accommodate areas of rapid growth. Water treatment
officials utilize census data to try and predict future population growth over a 20
year period and size the plant accordingly to serve the projected population.
Although a new plant or expansion will usually take projected growth into
account, actual construction of the plant will often include hardware to meet
current population or short-term growth. In this way, space is left in the water
plant for the addition of extra treatment processes which can be added as the long-
term population growth occurs. This method, of designing the overall plant
"footprint" to accommodate the future growth but only building the necessary
treatment trains for current and short-term service needs is one of the most cost-
effective methods of plant design.
2.	Distribution System Management
The EAT was able to obtain scant information concerning the drinking
water distribution systems in South Korea. Recent reports of extremely high water
loss rates and the lack of information from water system officials concerning their
own distribution systems lead the EAT to conclude that distribution system
management may be better managed at some South Korean water systems. Other
observations which enforce this concern are: low finished water pH; lack of
corrosion control practices, infrastructure deterioration, and red water complaints.
This is a serious issue, since the vast majority of waterborne disease
outbreaks in the United States result from distribution system problems. It is
23

-------
interesting to note that no matter how advanced water treatment may be, the
quality of drinking water delivered to the customer can be adversely impacted by
the distribution system.
In order to address this serious problem, the EAT recommends
implementation of the following actions:
a.	Microbial Monitoring and Control -
Microbial testing throughout the distribution system should occur
on a monthly basis. In the United States, total coliform bacteria are the
indicator organisms which are monitored for this purpose. When positive
coliform samples occur, repeat testing for fecal coliform is done to
determine whether a threat is posed to human health.
Maintenance of a strong residual disinfectant is key to distribution
system management. Disinfectants such as chlorine, chloramines, or
chlorine dioxide should be maintained at concentrations no less than 0.2
mg/L. When distribution systems are lengthy, it may become necessary to
add disinfectant booster pumping stations to maintain an adequate residual.
This disinfectant residual should be checked on a daily basis throughout the
distribution system.
b.	Corrosion Control -
Care should be taken in the treatment process to ensure that a non-
aggressive finished water is delivered to the distribution system. This
means that the finished water pH should range from 7.5 - 8.0 and that
adequate alkalinity concentrations be maintained. Many plants feed lime,
soda ash, caustic soda, or sodium bicarbonate to the filtered water to
increase the alkalinity and pH and thus stabilize the water. When industry
needs a low hardness water or the pH level is adequate, corrosion
inhibitors such as sodium hexametaphosphate, zinc orthophosphate,
polyphosphates, or activated silica may be added to protect distribution
system piping.
c.	Distribution System Preventative Maintenance -
Development and implementation of a distribution system
maintenance plan is the only way to ensure the delivery of safe drinking
water to the customer. This plan should include ar.rv.al flushing of the
entire distribution system at high flow rates to scour the pipes; leak
detection and repair (assuming a 15% water loss rate is normal due to
firefighting, calibration of meters, and unexplained losses); meter
24

-------
calibration and replacement program; cross-connection control; and
maintenance of a minimum distribution system pressure of 20 psi to each
customer.
3.	Public Perception of Drinking Water Quality
EPA, along with State regulatory agencies and the regulated industry have
gone to great lengths to educate the public concerning the safety of tap water.
Examples of this include: celebration of National Drinking Water Week (the first
week of May each year); environmental education programs in schools; tours of
drinking water treatment plants; and availability of compliance monitoring data for
public inspection.
Although the bottled wau.r and point-of-use treatment industries
continually attack the safety of tap water, the aforementioned measures taken by
regulatory agencies have served to educate the public. The vast majority of the
public trusts the safety of tap water, and utilizes it for their primary source of
drinking water.
4.	Fluoride - Guidelines on Selection and Methods of Addition
Information concerning fluoridation can be found in Appendix J. This
information is from the American Water Works Association's book entitled Water
Quality and Treatment. Additionally, a Centers for Disease Control and
Prevention Handbook entitled "Water Fluoridation - A Manual for Engineers and
Technicians" has been provided to USAEP as a reference.
5.	Scum Formation Issues
A common problem at many of the drinking water treatment facilities
visited by the EAT was scum formation in the rapid mix process. This problem
seemed to be caused by gases in the water rising to the surface and lifting the floe.
These gases may be associated with sediments from the raw water sources which
are brought in through the raw water supply. This problem was being solved well
at plants such as Taegu, where a mechanical scum scraping system had been
installed and seemed to be working quite effectively. However, other plants
seemed to be having operational difficulties in manual scraping to control the
scum. The EAT would recommend several different methods to control this
problem.
a. The affected plants could attempt to limit the amount of sediment entering
the plant by changing gates on the raw water intake (higher up in the river,
away from the river bottom).
25

-------
b.	The systems could enhance their existing sand/grit removal process to
obtain a higher removal efficiency.
c.	Aeration could be utilized before the rapid mix to remove most of the
gases.
d.	Utilization of efficient mechanical scum removal devices similar to ones
observed at Taegu or install high pressure spray nozzles in the rlume to the
flocculator to break down the scum.
V. CONCLUSIONS
The EAT observed sound engineering principles in South Korea's drinking water
treatmeni. ~acilities. A good regulatory program operated by MOE along with sound
research from KICT and progressive water resource planning from KOWACO contribute
to the solid foundation of a drinking water program progressing into the twenty-first
century.
It must be noted that as South Korea progresses with the G-7 project which
dictates the development and use of AWT, a concurrent effort to address the
aforementioned source water protection issues must be developed and undertaken. The
potential gains in drinking water quality from AWT development are threatened by the
continued deterioration of source water quality. No single problem, such as drinking
water quality, can completely be addressed without a holistic approach to the problem.
The point is made that addition of AWT to the existing drinking water treatment
facilities will not solve all of the problems faced by the South Korean drinking water
industry. Conventional treatment requires upgrading and enhancement in order to allow
for optimal water treatment through the AWT treatment process. Also, distribution
system management becomes even more crucial with the use of ozonation and BAC due
to the threat of microbial regrowth from high AOC levels from the ozonation process and
the possibility of bacterial slough off from the BAC process.
It should be noted that the Composite Correction Program of Drinking Water
Treatment Optimization is a very effective way to optimize treatment for removal of
particulate matter with relatively small costs involved. This type of program would lend
itself well to the South Korean drinking water industry, where many conventional
treatment plants have been observed that could benefit from process optimization.
The AWT research program could be expanded and enhanced to include predictive
modeling for contaminants of concern. This could also include disinfection by-products
specific to any type of disinfectant utilized (including OBPs and especially bromate). This
would allow AWT and conventional treatment optimization for greatest contaminant
removal and minimize DBP risk to the public.
26

-------
Renewed attention should be given to regulatory oversight of drinking water.
This should include increased inspections; development of a training regimen for plant
operators and regulatory inspectors; review and consideration of additional drinking water
standards; and increased availability of drinking water treatment facility operating and
monitoring data.
The EAT was made aware of concerns about diminished public trust of tap water
safety in South Korea. This can and should be addressed through public awareness
campaigns about drinking water, along with environmental education, public involvement,
and most importantly, availability to the public of each water systems regulatory
compliance monitoring data to calm fears over drinking water safety.
VI. ACKNOWLEDGMENTS
The EAT would like to thank all of the organizations and individuals whose hard
work and dedication allowed this Action Team visit to be completed in a successful
manner. Special recognition goes to the US-AEP organization in Seoul, specifically to
Director Chi-Sun Lee and Deputy Director Je-Ha Yang for organizing the visit. Special
thanks goes out to the Ministry of Environment, especially Director General Kuk Hyun
Chung, Director Chong Chun Kim, and Deputy Director Sung Soo Kim all of the Water
Supply and Sewage Policy Division, along with Dr. Jong-Suk Kim, Director General of
MOE's National Institute of Environmental Research. Thanks also to the Korea Institute
of Construction Technology, especially Executive Director Dr. Sang-Eun Lee, and
Director Dr. Chiasung Gee, and Senior Researcher Lee of the Environmental Engineering
Division. The EAT is also appreciative of Dr. Seongho Hong of Soongsil University, in
his service as a technical translator.
The EAT would also like to thank all of the City administrators and Water Treatment
officials which we had the pleasure to interact with in Taejon, Pusan, Ulsan, Taegu, and
Kimhae. Their hospitality was most gracious and their company always enjoyable.
The EAT looks forward to continued interaction and cooperation with the Government of
the Republic of Korea concerning drinking water treatment and related subjects.
27

-------
| G-/ Proiect j
Development of
Advanced Water Treatment System
Division of Environmental Engineering
KOREA INSTITUTE OF CONSTRUCTION TECHNOLOGY

-------
OBJECTIVES
° Production of Good Quality Water from Contaminated Raw Water
0 Development and Practical Application of Advanced Water Treatment
System Appropriate for the Country
° Development and Practical Applicaion of Proper Groundwater
Treatment Technologies

-------
RESEARCH STRUCTURE
° Cooperated Research of Industries, Universities, and Laboratories
•	Main Research Institute:
Korea Institute of Construction Technology
•	Participating Companies:
Daewoo Corporation, LG Construction Co., Ltd.,
Samsung Engineering and Construction Co., Ltd.
Korea Water Resources Corporation
•	Participating Universities:
Yonse' University, Chungnam National University.
0 Individual Ro.es
•	KICT : Planning and Coordination of Project
Optimization of AWTS
Methods Development for Practical Use
•	University : Basic Research
Optimization of Advanced Unit Processes
Problem Solving on Pilot Plant Operation
•	Industry : Practical Use
Selection of Unit Processes for Practical Use
Direction of Research for Practical Use
Accimulation of Technical Know-How

-------
RESULTS FROM FIRST STAGE RESEARCH
DEVELOPMENT
oGAC
° Fluidized-Bed and
Fixed-Bed BAG
o Photocatalytic(TiOj)
I Oxidation
o AOP
oDAF
° Biological Pretreatment
° Removal of Micro
Pollutants in
Groundwater
Construction and Operation of Advanced Water Treatment Demonstration Plant
Establishment of Basis for Pragmatization	 	
OPTIMIZATION
° Optimization of GAC
Design Technology
° Optimization of BAC for
the Removal of DOC
and NH3-1N
° Optimization of AOP
Using Ozone, H2O2 and
UV
o Optimization of
PhotocataJysis. DAF,
Biological Pretreatment
oOptimization of
Groundwater Pollutant
Removal
	
SUGGESTION OF
PROPER AWTS
o Characterization of Raw
Water Quality for Rivers
o Selection of Optimum
AWTS
1
o Evaluation and
Optimization of
Applicable AWTS for
Rivers
1
o Suggestion of AWTS
Proper to Korea

-------
DEMONSTRATION PLANT
~	Perspective
o Location : Bupyung Water Treatment Plant, Incheon
o Capacity : 200CMD
o Raw Water : Lake Palciang
~	Description
o Combination of Ozonation, AOP and Activated Carbon Adsorption
(GAC & BAC) with Conventional Water Treatment Processes
(Coagulation/Flocculation—> Filtration—'Chlorination)
o 2 Basins for Rapid Mixing, Coagulation and Flocculation & Rapid
Sand Filters, 2 Ozone Contactors, 5 Carbon Adsorption Columns
o 4 Types of System Assembled around Ozonation and Activated
Carbon Adsorption Using Individual Design and Operation
Parameters

-------
SECOND STAGE RESEARCH
DIRECTIONS
o System Automation
o Scale-Up Technology
o Development of Advanced Unit Process

-------
CONTENTS AND SCOPES
FIRST YEAR(1995) : Practical Application for Advanced Water Treatment
Technology
o Development of Practical Application for Ozone/AOP and Activated
Carbon Processes
o Development of Economic Regeneration Technology for GAC
Adsorption Process
o Development of Practical Application for AOP and Photochemical
Oxidation Processes
SECOND YEAR(1996) : Evaluation and Improvement of Practical Application
for Advanced Water Treatment
o Evaluation and Improvement of Practical Application for Ozone/AOP
and Activated Carbon Processes
o Development of Technology for the Improvement of Activated
Carbon Regeneration Efficiency
° Practical Application for AOP and Advanced Unit Processes
THIRD YEAR(1997) : Standardization and Commercialization of Practical
Application Technology for Advanced Water Treatment
o Standardization and Commercialization of Ozone/AOP and Activated
Carbon Processes
o Development of Application Technology for Activated Carbon
Regeneration
° Settling of AOP and Advanced Unit Processes

-------
J CI', r- *'J tW /} , *'formation
will allow a direct comparison of DBF formation among diffejent-v' :
~ 'Tb1

ntil recently, two approaches
assess the formation of chlorina-
y-products (DBPs) in drinking
on potential (FP) test and the sim-
tion system (SDS) test. Under the FP
test, waters are chlorinated at high doses and long
incubation times, which maximize DBP formation,
thus providing an indirect quantification of DBP pre-
cursors.1 This was originally a site-specific test that
would measure DBP precursors at a given location.
When the test was incorporated as a standard method,
standard incubation conditions were added.2
Typically, FP is tested under the following con-
ditions: seven-day incubation, 25°C, pH 7, and a
chlorine residual of 3-5
mg/L at the end of the
incubation period.2
These standardized con-
ditions allow a similar
mixture of the DBP pre-
cursors to be reacted and
thus allow comparisons
of results from site to site.
However, these high
chlorine dosages and
long incubation times
lead to the formation of
80 lOUBNJAI A\aaa/a

-------
higher concentrations of DBPs and more chlorine-
substituted over bromine-substituted DBPs3-5 than
occur under conditions common to a treatment plant
and distribution system. Moreover, a utility cannot
directly use the FP test results to evaluate how
changes in a given treatment will affect compliance
with DBP regulations.
To more accurately represent DBP formation in
a specific distribution system, utilities can perform
CT^S tests, in which samples are chlorinated under
specific conditions of time, temperature, pH,
chlorine dose or residual that reflect the con-
ditions of the distribution system. This test was ini-
tiated by the US Environmental Protection Agency
(USEPA) for five field investigations in 1978.6 Results
from SDS tests have been shown to well represent
DBP formation found in
distribution systems.7 A
limitation of the SDS test is
that chlorination conditions
vary among distribution
systems, complicating DBP
formation comparisons
between different waters.
As a site-specific test, dif-
ferent fractions of the pre-
cursors would be expressed
in each location. The SDS test approach also limits
the capability of a utility to assess how DBP pre-
cursor removal changes plant performance, because
formation conditions—especially temperature—may
vary from season to season.
A test has been proposed in Germany in which the
DBPs are measured after 0.5- and 24-h incubation
times.8 In this test, chlorine is dosed to yield a free
residual of 0.1 mg/L after 0.5 h, which is the mini-
mum regulatory level in Germany. Samples are incu-
v ' at 10°C and at ambient pH (7.3-7.5 for the
5 tested). The test, termed the minimal DBP
Cv.„^entration test, has been applied to water from
DOX MttMfyMlM MMtMi a wide range
^ofMogeo-tubstftuted organic
¦fiom^Mmthrformed during chlorination.
91n» Moqptefr'adsorbed onto activated
carbon and Introduced into a pyrolysis
l^ye'JGlrmari utilities and reported
ponly for trihalomethanes (THMs).
use of distribution system chlo-
jeiiesidual requirements in the
United States, the applicability of
this approach is limited.
A new test has been developed
in an effort to provide a means for
direct DBP formation comparisons
among different waters. Using
chlorination under constant con-
ditions that represent average con-
ditions in US distribution systems,
the test measures the amount of DBPs formed. In
this approach, termed the uniform formation con-
ditions (UFC) test for DBP formation,910 different
fractions of the precursors are reacted for waters
with different concentrations of organic and inor-
ganic precursors. The UFC test can also be used to
study how treatment conditions affect subsequent
DBP formation.
The objectives of this article are to (1) describe
the UFC test conditions and the rationale for their
selection, (2) present the results of DBP formation
sensitivity analyses performed on each UFC parame-
ter for three waters that represent the range of DBP
precursors and alkalinity commonly found in fin-
ished waters, and (3) present applications of the test
to a range of waters. Previous studies have shown
f the major advantages of the UFC
is the opportunity to compare the
DBP formation of different waters under
similar conditions.
how time, temperature, pH, and chlorine dosage
affect the formation of DBPs. With few exceptions,
however, these studies were conducted at chlorine
dosages 3-10 times higher than those typically used
in practice.
UFC test
The chlorination conditions selected for the UFC
test are incubation time—24 ± 1 h, incubation tem-
perature—20.0 ± 1.0°C, incubation pH—8.0 ± 0.2,
and 24-h free chlorine residual—1.0 ± 0.4 mg/L. The
process of selecting the conditions for the UFC test
began with the distribution of a comprehensive sur-
JUNE 1996 81

-------
Uniform formation conditions (UFC)
pH: 8.0 ± 0.2
Temperature: 20.0 ± 1.0°C
Incubation time: 24 ± 1 h
Chlorine residual: 1.0 ± 0.4 mg/L as free chlorine after 24 h
Preliminary study
A 24-h chlorine demand study of the water sample may be required before dosing under uniform for-
mation conditions to determine the applied dosage that will yield a chlorine residual of 1.0 mg/L after
24 h (procedure described below).
Materials
•	Chlorine demand-free glassware
•	pH 8 borate buffer
•	pH 8 combined hypochlorite-buffer dosing solution
Methods
•	Chlorine demand-free glassware
Incubation bottles (amber with PTFE-faced caps): soak in detergent* at least overnight, rinse
four times with hot tap water, two times with distilled, deionized (Dl) water. Place in 10-20 mg/L
chlorine solution (made with Dl water) for at least 24 h. Rinse four times with Dl water and then
one to two times with laboratory clean water (reverse osmosis/ion exchange/granular activated
carbon [RO/IX/GAC]); dry in 140°C oven at least overnight. Store dosing pipettes in ~50 mg/L
Cl2 (made with laboratory clean water). Rinse three times with dosing solution before use, and store
pipettes in chlorine solution after use.
•	pH 8 borate buffer
Before dosing, water samples are buffered to pH 8.0 with 2 mL/L borate buffer: 1.0 M boric acid
(American Chemical Society [ACS] grade) and 0.26 M sodium hydroxide (ACS grade) in boiled lab-
oratory clean water.
•	pH 8 combined hypochlorite-buffer dosing solution
A combined hypochlorite-buffer solution (based on the method described in reference 11) is made
by buffering the hypochlorite solution to pH 8.0 with pH 6.7 borate buffer.
To make pH 6.7 borate buffer: 1.0 M boric acid (ACS grade) and 0.11 M sodium hydroxide (ACS
grade) in boiled laboratory clean water (RO/IX/GAC).
Add pH 6.7 borate buffer to chlorine solution (1,000-3,000 mg Cl2/L) to yield a pH 8 dosing
solution. (A 4-5:1 volume ratio of pH 11.2 hypochlorite solution to pH 6.7 borate buffer yields
a pH 8 combined hypochlorite-buffer solution with about a 20 percent drop in chlorine strength.)
The dosing solution (combined OCh-buffer) chlorine strength should allow for a dosing volume of
< 0.5 percent of the water sample volume (e.g., 2.5 mL dosing solution in a 1-L bottle).
•	Preliminary study
Perform a 24-h chlorine demand study (sample buffered at pH 8.0 and incubated in the dark
at 20.0°C as described in the dosing procedure) using a series of three chlorine dosages
based on CI2:T0C ratios of 1.2:1, 1.8:1, and 2.5:1 after adjusting for inorganic demand. From
the results of these tests, the chlorine dose for UFC is selected to yield a 24-h residual of 1.0
mg/L free chlorine.
•	Dosing procedure
1.	Add 2.0 mL/L pH 8 borate buffer to water sample.
2.	Adjust to pH 8 with H2S04/Na0H (if necessary).
3.	Fill incubation bottle three quarters full with buffered water sample.
4.	Dose with combined hypochlorite-buffer solution holding pipette just above water surface.
5.	Cap bottle: invert twice.
6.	Fill to top with buffered water sample and cap headspace-free.
7.	Invert 10 times.
8.	Incubate in dark at 20.0°C for 24 h.
9.	After incubation, measure chlorine residual and pH, and sample for disinfection by-products.
•Fisher a-70, 4 percent. Fair Lawn. NJ.

-------
Simulated distribution system chiorination conditions: survey results
SDS Chtortnatkxi Comfitioni
.Utility
Laboratory
	 . .. ..
: S Targat CMnfaiiS
" ' - IM^
o
vey to utilities and researchers who perform the SDS
and FP tests. A range of conditions for the SDS test
was obtained from data generated by the surv ey and
is summarized in Table 1. The survey also provided
information about methods of chiorination and buffer-
ing. The AWWA Water Industry Data Base (WIDB)
nade data available about 318 large utilities that used
;e chlorine as the only residual disinfectant.
On the basis of these two
data sets, a preliminary set of
UFC conditions was devel-
oped. After discussions with
several researchers and utility
representatives, including a
meeting of 15 drinking water
experts at the 1993 AWWA
Annual Conference in San
Antonio, Texas, the proposed
conditions for the UFC test
were agreed on. After that
meeting, the window for the
24-h chlorine residual was increased from ±0.3 mg/L
to the current level of ± 0.4 mg/L to facilitate appli-
cation of the appropriate chlorine dosage and to yield
a three-to-five-day detectable residual for waters high
in total organic carbon (TOC).
The following rationale was used to select the
conditions for the UFC test.
Incubation time. An incubation time of one to
three days was considered. WIDB data showed that
the average maximum time in the distribution system
is three days, and the average mean time is 1.3 days.
Data from the SDS survey indicated that an average
of 3.3 days with a range of 1 h to seven days was
>d. However, these times were meant to represent
h maximum and average times in the distribu-
uon systems simulated. Initially, an incubation time
of three days was pro-
posed, representing aver-
age maximum conditions
in the distribution system.
From a practical view-
point, however, one-day
incubation yields faster
results and allows for
more tests (four tests in
a five-day work week) to
be conducted. A three-
day incubation time
would allow for three
tests per week. Addition-
ally, if a chlorine demand
study was required to set
the appropriate chlorine
dosage for a given sam-
ple, then the complete
test could be finished in
two days using a one-day
incubation time, whereas
the complete test would
require six days if a three-
day incubation time was used. An investigation of
DBP formation kinetics (described later) showed that
the increase in DBP formation from one to three days
averaged less than 20 percent for the three waters
examined and three DBPs analyzed. An incubation
time of one day was chosen for the UFC test.
Incubation temperature. Initially, a temperature
of 22 ± 2°C was proposed, allowing laboratories to
f the primary objectives in the
ablishment of the UFC test is that the
chiorination conditions be representative
of average distribution system
conditions.
conduct the test at ambient temperatures without an
incubator. The SDS survey indicated an average of
20°C with a range of 3 to 30°C, which reflects seasonal
variation. After further discussion, it was found that
temperature extremes commonly found in laborato-
ries were greater than those acceptable for the UFC
test and that many laboratories would require a tem-
perature-controlled incubator. In view of this, a lower
temperature with a narrower range of acceptable
deviation was proposed: 20.0 ± 1.0°C.
Incubation pH. The pH value for the test has
not changed throughout the discussions. A pH of
8.0 was chosen to reflect the influence of the Lead
and Copper Rule on distribution system pH. The
SDS survey indicated a median pH value of 7.8 with
a range of 6 to >9. Both samples and the chlorine
JUNE 1996 83

-------
Summary of UFC DBP formation and specific yield
UFC DBP*
Concentration—iig/L

Bromide
TOC
Br-/TOC

Hg/L
mg/L
fg/mg
Water



Ohio River (Ohio)
35
1.3
26.9
Salt River Project
89
2.2
40.5
(Arizona)



Manatee Lake
110
4.1
26.8
(Florida)



Passaic River
80
3.2
25.0
(New Jersey)



Lake Gaillard
14
1.5
9.3
(Connecticut)



Florida groundwater
276
10.0
27.6
Harsha Lake (Ohio)
29
3.6
8.1
Miami Whitewater
32
4.0
8.0
Lake (Ohio)



Great Miami River
126
3.2
39.4
(Ohio)



Green Swamp (Florida)
79
4.9
16.1
Average
87
3.8
22.8
Standard deviation
76
2.5
12.1
FuNOM*
90
2.3
39.1
Treatment: Ohio River



Conventional
132
2.1
62.9
Enhanced coagulation
132
1.5
88.0
Ozone-biotreatment
80
1.4
57.1
GAC.A ; i;
132
0.3 •
528
GAC8 • V
132
0.6
236 ... J
GACC "\
132
. 1-1 ;
125- iA
Nanoflltration ¦''•) 1
' 79 .
0.4 ' 1
184 :'j
TTHM HAAS
| Seasonal variation:
Passaic River
Autumn - ¦¦¦•-
Winter
Spring
Summer
Average
Standard deviation
58
68
151
73
31
238
95
104
97
155
107
60
66
86
59
38
¦ 2 '
.
' 53- '
' 21-'!
27
37
82
70
29
DOX
120
201
628
277
131
142	1,046
78¦	396
62	; 345
55	"i 295
A


:i> .... . ,,
:i :• - ;.<•
A
90 -
i
3.2
" v - J
28.1
85 '

90
2.3
39.1
65-

70
3.7
18.9
101
X
78
2.6
30.0
77,.

82
3.0
29.0
82

10
0.6
8.3
f 15
107
69
36
32
26
19:.
13
?.V.
i >; "-7.0-t!
'U'«
Specific Yield
lig DBP/mg TOC
TTMM/
TOC
44.6
30.9
HAAS/
TOC
20.8
i6.a
DOX/
TOC
'-¦•k ' -
y .
-SK
36.8	20.0
22.8	21.9
20.7	19.3
V. - " .; ,
23.8	14.2
26.4	21.7
29.0 j 15.5
30.3 • I . 17.2
92.3
91.4
153
86.6
87.3; ;
105 : >
110 -1
86.3 ; t5
92.2 ;
444 .3
388
277
221
170 ;
92'
56
• ,4-.?
'
"31.6*3
29.4 :j
7.2
28.7
41.0 .
' 39.3
• 27.11
•' Tir. 9.6 A
-,t,26.8.d
"50.0. ;
;-^48:8-r1
£21.8
18.9
2.8
13.9
- .12.4
12.7
».3- !
;'14.8'.r.-
r :A5.5:v
>1312'.
15.3f4,i
90.6
99.4"^
20.4 •'
96.1 -


~NOM extracted, from a German
_______ v * ?-.. .4 ; '-£1 7.' iv '-1- •S'-iJ.'i

' 75-."
39 .-',1
84 .
55
20 - - •
. .if
¦J y&S. (?§L0l ftg&i
'	J
• 26.6 2 .23.4-.
283 17.0 '
300; ^
313 •:
317 r
.210 '
285 VI
.
i :
81.0
61.3
40.0-j
14.0'j
3zi.:|,
69.8 >:
- na X $
¦ V-: iV
93.8.
138.' -
85.7V? •*
80.8 "
' 99.1 h

.273^,	22.7
29.6 • s	21.2
27.9 .<-1	2i:i ";:
V:X3 25.3,
dosing solution were buffered at pH 8.0 using a
borate buffer. About 2 mL/L pH 8.0 borate buffer
(1.0 M boric acid and 0.26 M sodium hydroxide)
was added to each sample before dosing. The dosing
solution was buffered to pH 8.0 by the addition of pH
6.7 borate buffer (1.0 M boric acid and 0.11 M
sodium hydroxide) based on the method described
by Koch et al.u
Chlorine residual. A target free chlorine resid-
ual of 1.0 mg/L after 24 h was chosen, initially with
a rar.ge of 0.5-1.5 mg/L (1.0 ± 0.5 mg/L). These val-
ues were supported by the WTDB data, which reported
an average mean free chlorine residual of 0.90 mg/L
ard average minimum and maximum residuals of
0.39 and 1.57 mg/L, respectively. The SDS survey
yielded an average target residual of 0.62 mg/L at an
incubation time that averaged 3.3 days.
Tnree main concerns were raised in the discus-
sions about setting the UFC chlorine residual and
about the acceptable range. The first was that the
residual at 24 h should be high enough to yield
detectable residuals at three to five days. This criterion
is a reflection of distribution system conditions: a
detectable residual needs to be maintained at all
points. A second concern was that the acceptable
range of UFC chlorine residuals would be too nar-
row for appreciable differences in the formed DBPs to
be evident. This was thought to be especially impor-
tant on the lower end of the chlorine residual range,
i.e., <1.0 mg/L. The third issue was one of practical
importance in conducting the UFC test. Determin-
ing the correct chlorine dosage that will yield the
UFC 24-h chlorine residual is difficult if the acceptable
window is too narrow.
The acceptable range for the UFC test was nar-
rowed from 1.0 ± 0.5 to 1.0 ± 0.3 mg/L after con-
cerns were expressed that (1) DBP formation is very
sensitive to small changes in chlorine residuals at
these low residual concentrations and (2) a residual
of 0.5 mg/L after 24 h would not lead to detectable
residuals after three to five days. A 24-h chlorine
residual of 0.7 mg/L was thought to be high enough
to yield a detectable residual after three to five days
in most waters. Some concern was also expressed

-------
that for high-TOC waters a
residual would not be de-
tectable after five dav^. To
address these issues, chlorine
-demand studies were per-
formed: some results are given
in the next section. On the
basis of the chlorine demand
study and DBP formation
results of this article and the
chlorine demand results of
Dugan et ai.12 the acceptable
range for the 24-h residual
was increased to 0.6-1.4 mg/L
(1.0 ± 0.4 mg/L). This small
widening of the acceptable
window significantly facilitates
determination of the apnro-
priate chlorine dosage.
Material and methods
Three waters that repre-
sent the range of finished
water quality commonly
found in the United States
were used in this study: Ohio
River water (ORW) taken
from the Cincinnati (Ohio)
Water Works Miller Water
Treatment Plant, Salt River
Project water (SRPW) taken
from the Phoenix (Ariz.)
water treatment plant, and
Manatee Lake water (MLW)
taken from the Manatee
County (Fla.) Public Works
water treatment plant. All
three were sampled after con-
ventional treatment that
included alum coagulation,
sedimentation, and filtration.
The TOC and bromide con-
centrations are summarized
in Table 2. ORW is an indus-
trially polluted source water
with low concentrations of
TOC and bromide and me-
dium alkalinity (60 mg/L).
SRPW has a moderate TOC
concentration, high bromide
concentration, and high alka-
linity (110 mg/L). MLW has
a high TOC concentration,
high bromide concentration,
and low alkalinity (12 mg/L).
The bromide-to-TOC ratios
were similar for ORW and
MLW, and chlorination did
not produce appreciable
amounts of bromoform i Br).
The Br-to-TOC ratio of SRPW
Effect of chlorine dosage on chlorine decay kinetics for Salt River
Project water and Manatee Lake water
24-h chlorine residual—mg/L C 1.0	0.6
Salt River Project water
TOC = 2.2 mg/L
Effect of incubation time on THM formation for Ohio
Salt River Project water, and Manatee Lake water
River water,
JUNE 1996 85

-------
Effect of incubation time on HAA formation for Ohio River water,
Salt River Project water, and Manatee Lake water
MLW HAA6
Effect of Incubation time on DOX formation for Ohio River water,
Salt River Project water, and Manatee Lake water
600
400
200
Incubation Time—days
was 50 percent higher than the other two waters,
and chlorination led to significant concentrations of
bromo-substituted DBPs.
DBP formation was characterized by total THMs
(TTHMs), the sum of six haloacetic acids (HAA6),
and dissolved organic halide (DOX). The HAA
species included in HAA6 were monochloroacetic
acid (MCAA), dichloroacetic acid (DCAA), tri-
chloroacetic acid (TCAA), monobromoacetic acid
(MBAA), dibromoacetic acid (DBAA), and bro-
mochloroacetic acid (BCAA). Free chlorine residual
was measured according to method 4500-C1 D2
MLW
1
Results and discussion
Chlorine demand. A pri-
mary objective in the estab-
lishment of the UFC test is
that the chlorination condi-
j tions be representative of
] average distribution system
i conditions. As stated earlier,
the UFC test targets a 1.0-
mg/L free chlorine residual
after 24 h. Another concern
is the maintenance of a resid-
ual for periods that reflect
maximum times in a distribution system. Therefore,
chlorine decay kinetics were investigated for the
three waters—-ORW, SRPW. and MLW—to deter-
mine whether the selected chlorine residual of 1.0
mg/L after one day would still yield detectable resid-
uals after three to five days, as would be necessary in
a distribution system.
+Dohrmann DC-180. Santa Clara. Calif.
fVarian 3400. Sugarland. Texas
tTekmar LSC 2000, Cincinnati. Ohio
§Tracor 1000. Austin. Texas
'"Dohrmann DX-20. Santa Clara. Calif.
Incubation Time—days
with an amperometric titra-
tor. TOC analysis was per-
formed following method
5310 C2 using a persulfate-
ultraviolet oxidation TOC
analyzer.* THMs were ana-
lyzed according to USEPA
method 524.2, revision 3.0
using a gas chromatographf
with a purge-and-trap sys-
tem:): and an electrolytic con-
ductivity detector.§ DOX was
analyzed according to meth-
od 5320-2 using a DOX ana-
lyzer.** DOX results are often
reported as total organic
halide (TOX) in previous lit-
erature. HAAs were analyzed
by the Technical Support
Division laboratory of USEP-
A's Office of Ground Water
and Drinking Water accord-
ing to USEPA method 552.2.
The method detection limits
for THMs, HAAs, and DOX
were 1.0, < 0.6, and 3 pg/L,
respectively. Duplicate chlo-
rination and subsequent DPB
analyses were run for ORW,
SRPW, and MLW at 30, 42,
and 42 percent of the num-
ber of total experimental
points, respectively. The
operating procedure for the
UFC test is given on page 82.
86 JOURNAL AWWA

-------
SRPWTTHM
Figure 1 shows chlorine
decay kinetics for SRPW and
MLW that had TOC concen-
trations of 2.2 and 4.1 mg/L.
respectively. These values are
similar to those of the 40th
and 80th percentiles of the
national distribution of TOC
concentrations in finished wa-
ter. A measurable residual was
found after three and five days
for the SRPW for chlorine
dosages that yielded a 24-h
chlorine residual near 1.0 and
0.6 mg/L. The results for ORVV.
with a TOC of 1.3 mg/L (not
shown), also yielded five-day
chlorine residuals for dosages
that had 24-h residuals of 0.6
to 1.3 mg/L. Thus, the low end
of the window, 0.6 mg/L, will
likely yield a detectable resid-
ual after five days for finished
waters with TOC concentra-
tions below 2 mg/L.
For waters with higher
TOC concentrations, the 24-h
residual needs to be higher to
yield a five-day detectable
residual, as illustrated in Figure
1 for MLW. This water, dosed
to yield the target 24-h resid-
ual of 1.0 mg/L, yielded a
detectable residual after three
days but not after five days.
Increasing the dosage to yield
a 1.5-mg/L residual after 24 h
resulted in a 0.5-mg/L five-day
residual. Thus, the upper end
of the 24-h residual, 1.4 mg/L.
should be targeted for high-
TOC waters if a residual after
five days is required. Chlorine
decay studies performed on
finished waters in which a 1.0-
mg/L residual was measured
after one day have consistently
yielded detectable chlorine
residuals after three to five days.12 Experience in the
University of Cincinnati laboratory has shown that
chlorine dose-to-TOC ratios of 1.2:1 to 1.8:1 will yield
a 24-h residual in the UFC test's acceptable range. 1 ±
0.4 mg/L, for most finished waters without apprecia-
ble inorganic demand. Advanced treated waters—e.g..
activated carbon and nanofiltration effluents—tend
to require higher dosage ratios because of the stronger
influence of inorganic demand, which is not removed
to the same level as the organic demand by these
advanced treatment processes.
Sensitivity analysis. THMs. HAAs. and DOX were
assessed to determine the sensitivitv of DBP formation
Effect of incubation temperature on THM formation for Ohio River
water, Salt River Project water, and Manatee Lake water
MLW TTHU
incubation Temperature—"C
Effect of incubation temperature on HAA formation for Ohio River
water, Salt River Project water, and Manatee Lake water
to small and large deviations from UFC test condi-
tions. For the sensitivity analysis, one parameter was
varied over a defined range, and the remaining para-
meters were held constant at the proposed UFC test
conditions. This procedure was repeated for all para-
meters for all three waters. Results are shown in Fig-
ures 2-13. Detailed THM and HAA species results
from the ORW and SRPW are presented in Summers
et al9 and Hooper et al.10 respectively. All error bars
represent relative percentage differences based on
DBP analyses of duplicate chlorination tests. Pairs of
vertical dashed lines shown in the figures indicate the
limits of the UFC test window for each parameter.
JUNE 1996 87

-------
formation was already present
after the first 24 h of reaction:
24-h DBP formation averaged
84 percent of the three-day
formation, ranging from 77
percent for TTHMs to 94 per-
cent for DOX. Others have
reported higher rates of DBP
formation after 24 h: however,
these studies were conducted
at much higher (3 to 10 times)
ratios of chlorine dosage to
TOC. which would yield much
higher chlorine residuals after
24 h and thus more DBP for-
mation.1314 Krasner et al15
used low chlorine dosages,
yielding residuals similar to
those used in this study, and
reported similar kinetic effects
for THMs, as did Zhu and
Reckhow16 for HAAs. Stevens
et al13 showed the nonpurge-
able DOX (NPOX) fraction to
have much faster formation
kinetics than the purgeable
DOX (POX) fraction at a chlo-
rine dosage-to-TOC ratio of
about 3:1. They equated the
POX fraction to the THMs. As
shown in Figures 2 and 3, the
increased formation of both
TTHMs and HAA6 after 24 h
was largely influenced by the
formation of the more chlo-
rine-substituted species,
whereas the more bromine-
substituted species showed
fast initial rates of formation
within the first 24 h and
smaller increases after 24 h.
This phenomenon occurred
in all three water ?xamined
and has been reported by
others.3-51'
Effect of incubation tem-
perature. The effect of incu-
bation temperature on DBP
formation was examined over
a temperature range of <10°C
to >30"C, and the results are
shown in Figures 5-7.
Although all DBPs analyzed
Effect of incubation temperature on DOX formation for Ohio River
water, Salt River Project water, and Manatee Lake water
SRPW
Effect of pH on THM formation for Ohio River water, Salt River
Project water, and Manatee Lake water
SRPW TTHM
ORW TTHM
X CHBrCU
? CHBr2Cl
Effect of incubation time. Figures ? 3, and 4 show
the effect of incubation time on the formation of
TTHMs, HAA6, and DOX, respectively, for all three
paters. The THM and HAA species for SRPW are also
;hown in Figures 2 and 3, respectively. Under UFC
chlorination. the DBP formation kinetics are relatively
rapid. For all three waters, most of the three-day DBP
showed an increase in forma-
tion with increasing temperature, the effect was more
pronounced for TTHMs and HAA6, which increased by
more than 100 percent over this temperature range for
all three waters. DOX formation increased by only 10
to 75 percent. Stevens et al15 also showed that at neu-
tral pH. DOX formation—particularly the NPOX frac-
tion—was less affected by temperature than were the
88 JOURNAL AWWA

-------
700
POX or THM fraction results.
DOX formation after 24 h was
less influenced by the higher
temperatures because the reac-
tion rates are faster than those
of THMs and HAAs (Figures
2-4); thus the reaction was
more complete after 24 h. The
more chlorine-substituted
THMs and HAAs showed
larger increases in formation
with increasing temperature
than did the more bromine-
substituted species. Chloro-
form increased by a factor of
2-4 over this incubation tem-
perature range; the increase
in the formation of the
bromine-substituted THM
species was significantly less.
MCAA, DCAA, and TCAA for-
mation on average doubled,
and MBAA and DBAA forma-
tion showed only a 30-40 per-
cent increase over the same
temperature range. This influ-
ence of temperature on speci-
ation was probably caused by
the slower formation kinetics
of the chlorine-substituted
DBPs.3-15 As shown in Figures
5 and 6, most of the bromine-
substituted species were al-
ready formed after 24 h,
whereas the chlorine-substi-
tuted species were still forming
and thus were more sensitive
to temperature changes. These
results indicate that halogena-
tion from hypochlorous acid
is highly temperature-depen-
dent, whereas halogenation
from hypobromous acid occurs
rapiuiy even at low tempera-
tures. DBP formation at high
temperatures will have higher
concentrations of chlorine-
substituted DBPs.
Effect of pH. Figures 8-10 show the effect of pH on
TTHM, HAA6, and DOX formation, respectively, for
all three waters. The pH range examined was from
below 7 to about 9. A base-catalyzed reaction, TTHM
formation increased by 50-60 percent with a 2-unit
pH increase. For the three waters, this was largely
driven by an increase in chloroform. The increase in
THM formation with increasing pH can be attributed
to hydrolysis of chlorinated intermediates. Base-cat-
alyzed hydrolysis of chloral hydrate and 1,1,1-tri-
chloroacetone have been shown to yield chloro-
form.17-18 Similarly, Trussell and Umphres19 reported
an increase in THM formation when the pH was
Effect of pH on HAA formation for Ohio River water, Salt River
Project water, and Manatee Lake water
Effect of pH on DOX formation for Ohio River water, Salt River
Project water, and Manatee Lake water
m
•;-V'
• •z.tir.f. , .	v' .{t	"
I
... - - •/. i s»
I :
7.0
7.5

• ORW .
m*rji
raised from 6 to 9 even in the absence of a chlorine
residual, indicating that this type of hydrolysis may
occur even under low chlorine residual conditions
similar to that of UFC.
Conversely, HAA6 formation decreased somewhat
with increasing pH, especially for MLW. This is con-
sistent with the results of others. For all three waters,
the decrease in HAA6 formation was driven by a
decrease in TCAA formation, as illustrated in Figure
9 for ORW. The change in pH will affect the distrib-
ution of the chlorine species. At lower pH. HOCl is
favored, resulting in significant chlorine substitution,
as HOCl is a stronger chlorinating agent than hy-
JUNE 1996 89

-------
Effect of 24-h chlorine residual on THM formation for Ohio River
water, Salt River Project water, and Manatee Lake water
180
160
140

¦¦ -'J vf ;•	.
-5 MLW .TTHM ^ ~
Trs.'	¦¦¦ • -
' *	p

>;iv ;>.?>%• V	v, :> ;-A pr*
|	.»	-„.*K!,_v-	t;V
Effect of 24-h chlorine residual on HAA formation for Ohio River
water, Salt River Project water, and Manatee Lake water
90
!
o
o
o
5
z
80
70
. ,		 		 . ..
.	*;¦ .
IRPWHAA6M-:
... • •J" *'	,•«:• •
—ORW MAA6^4-iv
i	ifff/v* : >rN
¦> *"7v~ ¦¦;'	' '
,V:V'rV' '•	.
4 ;>T6:1),
extreme pH values (<4 to
>11), or both are used, DOX
production increases as the
pH values decrease.14-18
Effect of 24-h chlorine
residual. The effect of 24-h
chlorine residual on DBP
formation was investigated
in the residual range of
0.3-3.1 mg Cl2/L for ORW
and from 0.5 to 1.5 mg Cl2/L
for SRPW and MLW. The
results are shown in Figures
11-13. ForTTHM formation,
only the results from the
SRPW showed an increase
greater than 10 percent with
increasing residual in the
range shown, and this in-
crease was because of the
increase in chloroform for-
mation. HAA6 formation
was more strongly affected;
for SRPW and ORW it
showed a linear increase
over the entire residual
range. The increase in HAA
formation was largely driven
by increases in TCAA and
DCAA. DOX formation was
relatively constant for all
three waters over the resid-
ual concentration that was
examined. Most previous
research on this topic has
shown an increase in DBP formation with chlorine
dosage, but previous research was done at higher
chlorine dosage-to-TOC ratios and over a much
wider range of conditions than examined herein.
Krasner et al15 have shown that increasing the chlo-
rine residual from 0.5 to 1.4 mg/L yielded a 17 per-
cent increase in THMs, which is similar to the THM
increase found in the work described here: <5 to 20
percent. The increase reported by Krasner et al was
also driven by increases in the more chlorine-sub-
stituted species.
90 JOURNAL AWWA

-------
DBP formation sensitivuy
within UFC test window. The
objective of setting an accept-
able window for each UFC
test parameter was to allow
for some laboratory variabil-
ity v^ithout sacrificing the
goal of constant formation
conditions for DBP formation.
From the data collected for
the sensitivity analysis, the
percentage change in DBP
formation within the UFC
test window was calculated.
The DBP concentrations were
interpolated at the extremes
of the acceptable window and
compared with those formed
at the target UFC. Within the
acceptable window. DBP for-
mation was least sensitive to
incubation time (average 2.3
percent change) and temper-
ature (average 2.9 percent
change); it was most sensi-
tive to pH (average 5.5 per-
cent change) and the 24-h chlorine residual (aver-
age 5.0 percent change). The average change for
the 24-h chlorine residual was 3.8 percent when
the window was set at 1.0 ± 0.3 mg/L. Therefore,
increasing the acceptable range to 1.0 ± 0.4 mg/L did
not substantially increase the DBP formation vari-
ability. For ORW, the average change in DBP for-
mation within all parameter windows was estimated
at 3.4 percent, with a minimum of <1 percent and
a maximum of 11 percent. For SRPW and MLW. the
average, minimum, and maximum were 4. <1, and
8 percent and 4.2. <1. and 13 percent, respectively.
The overall average change in DBP formation within
four parameter windows for all three waters was
only 3.9 percent. This low level of variability within
the acceptable range of UFC test conditions allows
researchers to feel confident comparing the UFC
DBP results.
Applications of UFC test. One of the
major advantages of the UFC test is the
opportunity to compare the DBP forma-
tion of different waters under similar con-
ditions. The UFC-DBP formation results of
10 conventionally treated waters from river,
lake, and groundwater sources are shown
in Table 2.22"26The range of TOC (1.3-10
mg/L) and bromide concentrations (14-276
pg/L) are representative of those commonly found
in the United States. Chlorination of these waters
with the UFC test yielded a wide DBP concentration
range, as expected. However, the specific DBP-to-
TOC mass concentration ratios of the different waters
were surprisingly similar. The coefficient of variation
(standard deviation-to-mean ratio) averaged less than
20 percent for these three DBPs. The average UFC
Effect of 24-h chlorine residual on DOX formation for Ohio River
water, Sait River Project water, and Manatee Lake water
I	I
05	1.0
24-h Chlorine Residual

specific yields were 29.4, 18.9, and 99.4 pg/mg TOC
for TTHM. HAA6. and DOX. respectively. As expected,
these values are much lower than the formation
potential specific yields reported in the literature
because of the higher chlorine dosages and longer
reaction times of the FP test.27 The average
UFC-TTHM specific yield is higher than that of the
minimal TTHM used in Germany because of the lower
chlorine dosages used in that test.8 The UFC test also
allows an evaluation of the representativeness of an
isolated natural organic matter (NOM). An NOM
extracted from a German groundwater, termed
FuNOM, has been used in the University of Cincin-
nati laboratory for several studies. As can be seen in
Table 2. the DBP formation results of this isolated
NOM are similar to the average of the 10 treated
water sources.-
2S

s in incubation pH and 24-h
rine residual had the most
effect on DBP formation.
Another use of the UFC test that is especially
germane to a specific utility is the comparison of
DBP formation alter different types of treatment.
Table 2 lists results that illustrate the effect of treat-
ment on the UFC-DBP formation of a batch of Ohio
River water.--25 Both the effect on formed DBP
concentrations and specific yields can be seen from
these data. DBP precursors were removed by all of
JUNE 1996 91

-------
these advanced treatment processes, and thus the
formed DBP concentrations decreased. However,
the DBP specific yields were not uniformly affected,
which indicates a change in the nature of the pre-
cursors— Enhanced coagulation decreased the
UFC-DOX specific yield but had little effect on that
for TTHM and HAA6. The effluent after ozonation
and biofiltration resulted in a lower specific yield for
all three DBPs The results for granular activated
carbon (GaC) A, GAC B, and GAC C represent
three effluent points on a breakthrough curve—
UFC test will be used for some
pects of the Information
Collection Rule.
nonadsorbable, 2 5 percent TOC breakthrough, and
50 percent TOC breakthrough. The specific yields for
TTHM and DOX were initially lower and rose with
time, whereas the specific yield for HAA6 showed
little change. The specific yields after nanofiltra-
tion showed a slight increase.
A third potential use of the UFC test is the assess-
ment of seasonal variability of DBP precursors, as
illustrated in Table 2 for conventionally treated Pas-
saic River water.26 Although the concentration of
DBPs formed varied by a factor of 2, the specific yields
were very similar from season to season, indicating
that the nature of the DBP precursors does not
change.
Summary
The goal of this study was to develop a method to
facilitate DBP formation comparisons between dif-
ferent waters under constant chlorination conditions
that are representative of distribution systems. The
UFC test conditions were developed for the incuba-
tion time, temperature, pH, and 24-h free chlorine
residual based on data from surveys of the distribu-
tion systems of 318 utilities and of 24 laboratories. In
addition to developing conditions representative of dis-
tribution systems, the acceptable parameter value
range was chosen to be narrow enough to minimize
variations in the formed DBPs but wide enough to
allow the UFC test to be easily run.
After the UFC test conditions were set, chlorine
demand studies and a DBP formation sensitivity analy-
sis were run using three conventionally treated waters
with water quality representative of that commonly
found in finished waters. Chlorine demand tests from
this and other research have shown that targeting
the UFC test 24-h free chlorine residual of 1.0 ± 0.4
mg/L should vield a three-to-five-day detectable chlo-
rine residual in most treated waters. For finished
waters with a TOC value above 2 mg/L, the upper
end of the acceptable 24-h chlorine residual range,
1.0-1.4 mg/L. should be targeted. Future work should
include chlorine demand studies on finished waters
with TOC concentrations above 4 mg/L. because the
occurrence of a three- or five-dav residual will
decrease as the TOC concentration increases.
DBP formation under UFC test conditions was not
drastically affened by vanations wuhin the acceptable
windows given for the three waters examined. The
average change m DBP formauon at the extremes of all
UFC parameter windows for all three DBPs and all
three waters was 3.9 percent. For any given
DBP. formation at the limits of the UFC para-
meter windows did not exceed 13 percent.
Changes in incubation pH and 24-h chlo-
rine residual had the most effect on DBP for-
mation. Withm the UFC test window. TTHM
formation was most sensitive to changes in
pH. HAA6 formation was most sensitive to
changes in 24-h chlorine residual, and on
average DOX was the least sensitive to all
changes. Overall, for the three waters exam-
ined, DBP formation was not overly sensitive to the
range of conditions chosen for the UFC test.
UFC test results also allow a direct comparison of
DBP formation among different waters. The average
UFC specific yields of 10 conventionally treated waters
found in this study were 29.4, 18.9. and 99.4 pg/mg
TOC for TTHM. HAA6. and DOX, respectively. As
expected, this is lower than the specific yields mea-
sured under formation potential chlorination condi-
tions. The UFC specific yields give a better estimate of
the expected production of DBPs than does that for
the FP test because of the more representative chlo-
rination conditions. The UFC test can also be used
to investigate the effect of seasonal variability and
treatment changes on subsequent DBP formauon for
a specific water. The UFC test has been successfully
used by several projects sponsored by the AVVVVA
Research Foundation. USEPA-ORD Drinking Water
Research Division, and USEPA Office of Ground Water
and Drinking Water (OGWDW) Technical Support
Division and applied to water from more than 20
utilities It will also be used for some asnects of the
Information Collection Rule
Acknowledgment
The authors thank the many utilities and labora-
tories that responded to the SDS-FP survey and sub-
sequent discussions, in particular Richard Miltner,
Stuart Krasner. and Jim Svmons for their comments.
The authors also acknowledge the assistance of Ali-
son Gusses and Nicholas Dugan. This work was partly
funded by the AWWA Research Foundation and the
USEPA OGWDW, Technical Support Division (CX-
820659-01)
References
1 Stevens. A.A. &¦ S'i.mons. J.M. Measurement of
Trihalomethane and Precursor Concentration
Changes Jour /HVH'4. 69:10:546 (Oct. 1977)
92 JOURNAL WWA

-------
2.	Standard Methods for the Examination of Water
and Wastewater. 1992. APHA. AWWA, and WEF,
Washington. D.C. (18th ed.).
3.	Symons. J.M. et al. Measurement of THM and
Precursor Concentrations Revisited: The Effect
of Bromide Ion. Jour /UVHM, 85:1:93 (Jan.
1993).
4.	Symon's, J.M. et al. Influence of Bromide Ion on
Trihalomethane and Haloacetic Acid Formation.
Disinfection By-products in Water Treatment the Chem-
istry of Their Formation and Control (R.A. Minear
and G.L. Amy. editors). CRC Press Inc.. Boca
Raton. Fla. (1996).
5 Shukairy, H.M Er Summers, R.S. DBP Speciation
and Kinetics as Affected by Ozonation and Bio-
treatment. Disinfection By-products in Water Treat-
ment: the Chemistry of Their Formation and Control
(R.A. Minear and G.L. Amy. editors). CRC Press
Inc., Boca Raton. Fla. (1996).
6.	Miller. R. et al. Feasibility Study of Granular
Activated Carbon Adsorption and On-Site Regen-
eration. EPA-600/S2-82-087 (1982).
7.	Koch, B. et al. Predicting the Formation of DBPs
by the Simulated Distribution System. Jour
AmVA, 83:10:62 (Oct. 1991).
8.	Mueller, U. et al. Trihalomethane Formation in
Drinking Water Treatment (in German). Vom
Wasser, SO: 193 (1993).
9.	Summers, R.S. et al. Development of Uniform
Formation Conditions for the Assessment of Dis-
infection By-product Formation. Proc. 1994
AWWA Annual Conf., New York.
10.	Hooper. S.M. et al. Development of a New Test
for the Assessment of Disinfection By-product
Formation: Uniform Formation Conditions. Proc.
1994 AWWA WQTC, San Francisco, Calif.
11.	Koch, B.. Chinn, R.; 8- Davis. M.K. A Simulated
Distribution System Trihalomethane Formation
Potential Method Proc. 1987 AWWA WQTC,
Baltimore, Md (1987).
12	Duga.n. N. et al. An Alternative Approach to
Predicting Chlorine Residual Decay. Proc. 1995
AWWA WQTC. New Orleans, La.
13	Stevens, A. cr al. Organic Halogen Measure-
ments: Current Uses and Future Prospects. Jour
AmVA. 77:4:146 (Apr. 1985).
14.	Fleischaker, S.J. & Randtke. S.J. Formation of
Organic Chlorine in Public Water Supplies Jour
AWWA. 75:3:132 (Mar. 1983).
15.	Krasner. S.W et al. The Impact of Chlorine
Dose, Residual Chlorine. Bromide, and Organic
Carbon on Trihalomethane Speciation Proc.
1992 AWWA WQTC. Toronto. Ont.
16.	Zhu, Q. & Reckhow. D.A. Investigation of Preo-
zonation and Chlorammation on Haloacetic Acid
Formation. Proc 199 3 AWWA Annual Conf..
San Antonio, Texas
17.	Reckhow, D.A. &• Singer. P.C. Mechanisms of
Organic Halide Formation During Fulvic Acid
Chlonnation and Implications with Respect io
Preozonation. Water Chlonnation Chemistrv. Envi-
ronmental Impact and Health Effects, vol. 5. Lewis
Publ , Chelsea, Mich. (19S5).
IS. Stevens, A.A.: Moore, L.A.; & Miltner, R.J. For-
mation and Control of Non-Trihalomethane Dis-
infection By-products. Jour. ,4lVtV4, S1:S:54 (Aus
1989)
19.	Trussell. R.R. & Umphres. M.D. The Formation
of Tnhalomethanes. Jour .4tVtV.4, 70:11:604 (Nov.
1978).
20.	Miller, J.W. &- Uden, P.C. Characterization of
Nonvolatile Aqueous Chlonnation Products of
Humic Substances. Envir Set. e2 Techno!.. 17:150
(1978).
21.	Urano, K. Er Takemasa, T. Formation Equation
of Halogenated Organic Compounds When
Water Is Chlorinated. Water Res., 20:12:1555
(1986).
22.	Summers. R.S. et al. Control of DBP Precursors
by GAC Adsorption. Draft report. AWWA. Den-
ver, Colo. (1996).
23.	Miltner, R.J. et al. The Control of DBPs by
Enhanced Coagulation. Proc. 1994 AWWA
Annual Conf., New York.
24.	Dryfuse. M.J. An Evaluation of Conventional
and Optimized Coagulation for TOC Removal
and DBP Control in Bulk and Fractionated
Waters. Master's thesis, Univ. of Cincinnati.
Cincinnati. Ohio (1995)
25.	Allgeier, S.C. &¦ Summers, R.S. Evaluating NF
for DBP Control With the RBSMT. Jour. /llvlvy,
87:3:87 (Mar. 1995).
26 Solarik. G. et al. ICR Bench- and Pilot-Scale
GAC Treatment Studies Conducted for the Pas-
saic Valley Water Commission. Proc. 1995 AWWA
WQTC. New Orleans, La
27. Reckhow, D.A.: Singer, PC.; & Malcolm, R.L
Chlonnation of Humic Materials: Bv-produci
Formation and Chemical Interpretations. En\ir.
Sci t? Techno!. 24:1 1-165 5 (1990).
2S Koechling, M.T.: Summers, R.S.; Er Shukairy, H.M
Elfeci of Ozonation and Biotreatment on Mole-
cular Size and Hydrophilic Fractions of Natural
Organic Matter Presented at ACS Annual Meet-
ing, Chicago. Ill (1995).
About the authors: R Scott Sum-
mers is associate professor in the
Department of Civil and Environ-
mental Engineering, POB 210071,
University of Cincinnati. Cincinnati.
OH 45221-0071. He received A WW A's
Best Paper Award from the Research
Division in 1990 and 1996. Stuart
M Hooper and Gabriele Solarik are graduate research assis-
tants. also at the University of Cincinnati Hiba M Shukairy
is a research associate at Oak Ridge Institute for Science and
Education. US Environmental Protection Agency, Drink-
ing Water Research Division, 26 IV Martin Luther King
Drive. Cincinnati, OH 4526S Douglas Owen is a vice-pres-
ident with Malcolm Pirnie Inc. 703 Palomar Airport Road.
Carlsbad, C.4 92009
JUNE 1995 93

-------

i I-c	' v.	Chapter
hi L y~ / f)C /•/'-?a, -
¦1	' •¦	/'
/-fiv. 
-------
mechanical generators. Depending on generator design, the oxidizing
agent is usually (1) gaseous or aqueous chlorine alone, (2) a mineral acid
by itself or with chlorine, or (3) acid in combination with a hypochlorite
salt solution. The main by-products of generating and using CIO2 are
chlorite ion (C102~), chlorate ion (CIO;,-), and chloride ion (CI")
Government Regulations
The US Environmental Protection Agency (USEPA) has established
several rules to address various issues of water treatment It promulgated
the Surface Water Treatmer' Rule (SVVTR) in 1989 to protect the public
fiom waterborne disease-causing organisms. The treatment goals of the
SWTR may be partially met by disinfecting drinking water. The
effectiveness of a disinfectant in killing microorganisms may be specified
in terms of the CxT value. This value represents the combination of the
residual concentration of a disinfectant C (in milligrams per litre),
multiplied by the time of contact T (in minutes). Acceptable CxT values
vary depending on the type of disinfection a treatment plant uses
Chlorine, the traditional disinfectant, is used by more public water
systems than any other disinfectant. Howev er, free chlorine is known to
react with certain organic substances found in most raw water to produce
trihalomethanes (THMs), a type of halogenated disinfection by-product.
The USEPA limits the amount of total trihalomethaneb (TTHMs) that can
be found in drinking water It is necessary for water treatment plants to
prevent the formation of elevated levels of THMs while at the same time
making sure that water is adequately disinfected. One of the advantages
of chlorine dioxide is that, if used properly, it produces lower levels of
chlorinated THMs than free chlorine alone
The current recommended level for CIO2, CIO2-, and CIO3"" is
1 mg/L for the sum of all three compounds (Lykins, Goodrich, and Hoff
1990). The Microbial-Disinfection Bv-Product Rule Cluster (M-DBP)
proposes an individual maximum contaminant level (MCL) of 1 mg/L
for CIO2 and a maximum residual disinfectant level (ivIRDL) of 0.8 mg/L
for CIO2, but these may be subject to change
Historical Overview
Chlorine dioxide was first used in the United States during the 1940s
in plants along the Niagara River in New York to control phenol-related
taste and odor (Aston 1947). Chlorine dioxide was used in these cases
because other disinfectants resulted in unpleasant medicinal tastes in the
water. The tastes were found to be associated with chlorophenols present
in heavily contaminated raw water sources. Subsequently, CIO2 was
observed to have the capacity to oxidize chlorophenols once they had
been formed either upstream of plant intakes or by immediate
-------
INTRODUCTION Health Effects
pretreatment with chlorine (CI;) Its disinfection capabilities were also
recognized at the time, but its major advantage was in taste-and-odor
control
Many facilities abandoned the use of CIO2 during the 1960s and
1970s because of equipment or installation design problems or high
chemical costs Some equipment failed because of the corrosive nature of
highly alkaline concentrated sodium chlorite precursor solutions. Other
reasons these systems were shut down include poor conversion of the
precursoi chemicals into chlorine dioxide, gross contamination of the
CIO2 solutions with high chlorine levels, or inadequate dose control.
Some CIO2 generation systems needed twice the theoretical
(stoichiometric) amount of chlorine for acceptable conversion, and the
resultant excess chlorine levels were high enough to negate any beneficial
effects of the CIO2.
Other systems were turned off because of the effect the CI2-CIO2
mixtures had on the aesthetic quality of finished water. Reports as early
as the 1940s of odors resembling kerosene, diesel fuel, bleach, swimming
pools, or medicine were associated with the use of CIO: in the treatment
process Inefficient systems that generate chlorine-contaminated CIO2
solutions can still be found in drinking water plants. Although such
characteristic tastes and odors are still attributed to the use of CIO2, it is
generally the purity of the CIO2 that remains in question (Hoehn et al.
1990, Dietrich et al. 1992). These impure mixtures are likely to be
common because optimization and monitoring programs are not
universally practiced or enforced m the United States or Canada.
Despite numerous setbacks, there has been an increase in interest in
CIO2 for the treatment of drinking water since the 1970s As many as 500
water utilities currently use CIO2 full time in the United States, and up to
900 plants throughout North America may use it on a short-term or
seasonal basis (Hoehn 1993)
Health Effects and Regulatory Concerns
A discussion of potential health and toxicity issues surrounding
oxychlorine compounds (CIO2, ClCh", and CIO;,-) is included in appendix
A of this handbook because of the importance of these issues in the
regulatory context of CIO2. Recent toxicological studies specifically
address many of the key regulatory issues. Concern about the health
effects of the chlorite ion is the most important reason the use of CIO2 in
public drinking water is not more widespread throughout North America
and Europe.
Concerns about the use of CIO2 in drinking water will continue until
the regulatory agencies formally resolve these questions and regulate
scientifically based levels for C102~ and CIO2. The health-related
information presented in this handbook may also serve as an up-to-date
3
-------
technical resource guide Final results or a two-generation study of rats
fed sodium chlorite are expected by the end of 1996 and may
significantly affect the regulatory levels for the chlorite ion and possibly
chlorine dioxide (CMA 1994; Gates and Harrington 1995).
Overall, there is strong evidence that exposure to mutagenic (i.-e.,
mutation-causing) and potentially carcinogenic (i.e., cancer-causing)
material in drinking water is reduced if CIOz is used in the treatment
process (Miltner 1976, Lykins and Griese 1986; Patterson, Lykins, and
Richardson 1995) Chlorine dioxide by itself will not form THMs or
haloacetic acids (HAAs) from natural humic substances to the same
degree as free chlorine. Total organic halides (TOX) can be reduced up to
20-fold when CIO2 is used in place of free chlorine for disinfection
(Lykins, Goodrich, and Hoff 1990; Lykins et al. 1991).
As with any of the disinfectants used in drinking water, there must
be a balance struck among the potential risks from acute chemical
exposure, possible long-term carcinogenic potential, and the realistic need
for greater disinfection capabilities to reduce microbial infection or
disease among humans.
Advantages and Disadvantages of Using Chlorine
Dioxide in Drinking Water
Some advantages of using chlorine dioxide rather than chlorine are
•	decreased formation of halogenated DBPs (THMs, TOX, and
HAAs), assuming well-maintained generators
•	improved taste-and-odor control for algal-related compounds, as
well as reduced color problems
•	improved C x T credits over a broader pH range
•	better Giardia and Cryptosporidia control
Some of the disadvantages of using chlorine dioxide are
•	aggressive regulatory monitoring requirements related to health
issues
•	the need for more complicated analysis methods and more
sKilled operators
^ specialized mechanical equipment to generate chlorine dioxide
on-site
•	higher cost than chlorine (but storage of gaseous chlorine may
still be required; see White 1993)
•	the need to store two or three chemicals, not just one as with
chlorination
« the possibility of taste-and-odor episodes in homes with new
carpet (such homes have chemicals in the air that can react with
chlorine dioxide)
-------
INTRODUCTION References
Summary
The use of CIO2 mav significantly reduce exposure to halogenatcd
DBFs in drinking water, especially if used in conjunction with
chloramines Many drinking water treatment facilities are capable of
using CIO2 for disinfection given the appropriate equipment
modifications. Tne use ot CIO2 as a primary disinfectant, or the use of
both CIO2 and chloramines, may enhance the disinfection process,
producing a microbiologically cafer water. Chlorine dioxide is not
intended to replace free chlorine or chloramines in drinking water
disinfection, but it does possess special properties that can complement
the more traditional disinfectants and minimize microbial risks associated
with water supplies. The judicious use of CIO2 can contribute to an
effective disinfection and/or oxidation regime in treatment plants. Many
installations have used CIO2 successfully because of an improved
understanding of CIO2 generation chemistry and recent improvements in
analvtical methods for the related oxychlorine compounds.
References
Aston, R N 1947 Chlorine Dioxide Use in Plants on the Nia^nia Border jinn
/UVIWI, 40 120/
Chemical Manufacturer's Association (CMA) 1994 Comments of the Chlonne
Dioxide Panel of the Chemical Manufacturer's Association on FPA's
Proposed D-DI3P Rule Filed USEPA D-DBP Docket Clerk. Dec 22, 1994
Appendix 13 Piolocol tor Dunking Water Rat Two Generation
Reproductive lo\icit\ Study Sodium Chlorite
Dietrich, A M , T D I.edder, D L Gallagher, M N Grabeel, and R C. I loehn 1992
Determination of Chlorite and Chlorate in Chlorinated and Chloraminated
Drinking Water by Flow Injection Analysis and Ion Chromatography
Ari'h/hc'il Chcinisti 1/, 64 496-502.
C.ates, DJ , and R.M Harrington. 1995 Nevroreproductive Toxicity Issues
Concerning Chlorine Dioxide and the Chlorite Ion in Public Drinking
Water Supplies In Proc. AWWA Water Quality Technology Conference
Denver, Colo. American Water Works Association.
Floehn, R C. 1993 Key Issues In Chlorine Dioxide Use m Water Treatment Pioc
CMA. USEPA, and AWWARF, 2nd International Symposium Houston, Texas
May 1992
Hoehn, R C , A.M Dietrich, W5. Farmer, M P Orr, R G Lee, E M. Aieta, D W.
Wood, and G Gordon 1990. Household Odors Associated With the Use of
Chlorine Dioxide, jour AWWA, 80(4)-166—172
L\ kins, B.W., J A. Goodrich, and J.C. Hoff 1990 Concerns of Using CIO2
Disinfection in the USA. /. Water SRT—Aqua, 39.366-376
L\ kins, B.W., Jr., J.A. Goodrich, W.E Koffskey, and M H Griese 1991 Controlling
Disinfection By-Products With Alternative Disinfectants. In Proc 1991
AWWA Annual Conference Denver, Colo : American Water Works
Association.
Lvkins, B W, and M H. Griese 1986 Using Chlorine Dioxide for Trihalomethane
Control jour A WW/1, 78(6) 88-93
-------
Miltner, R J 19/6. The Ettect ot Chlorine Dioxide on IYihalomethanes m Drinking
Water M.Sc. thesis, University ot Cincinnati, Cincinnati, Ohio.
Patterson, K.S., B.VV Lykins, Jr., and S.D. Richardson 1995. Mutagenicitv'ot
Drinking Water Following Disinfection. / Water 5RT—Aqua, 441-9.
White, G C. 1993 Handbook of Chlonnation. New York' Van Nostrand Reinhold.
-------
doses are adjusted to account for the organic matter ozone demand.
Case studies describing the use of ozonation for iron and manganese removal are
reported (2). Optimization of the use of ozonation for metal oxidation may become more
important as more utilities use ozonation for primary disinfection and/or DBP control.
3.2.2 Oxidation of Bromide
Bromide ion (Br") can be oxidized to yield bromate. The interest in bromate formation
stems from the fact that it is one of the cont?Tiinants in drinking water that will be regulated
by the proposed D-DBP Rule. An MCL of 0.010 mg/L for bromate is proposed.
The interest in controlling bromate formation is founded on studies that indicate that it
is carcinogenic. Potassium bromate. which is also used as an oxidizing agent in making bread,
is nephrotoxic in animals (26) and in humans (27). It is a genotoxic carcinogen and possesses
both initiating and promoting potential for renal carcinogenesis (26). Based on a multistage
model used by the USEPA to extrapolate the results of carcinogenesis experiments with high
doses in animals to low doses in humans, the expected MCL at the 10"6, 10°, and 10"1 cancer
risk levels are 0.08, 0.8 and 8 /xg/L, respectively. The MCL of 0.010 mg/L that is proposed
in the D-DBP Rule is based on analytical limitations and not on the risk level (10). The
minimum detection level (MDL) range reported by analytical laboratories is in the range of 3
to 10 fj.g/L. A selective anion concentration method has been proposed to decrease the MDL
for bromate. Although the method is tedious, the reported MDL is 0.3 /zg/L (28).
Bromide ion in source waters may originate from natural sources, from anthropogenic
processes, and/or as an impurity in chlorinating solutions. The most common natural
contribution is from saltwater intrusion in aquifers. The average bromide concentration in
20
-------
source waters in the US is about 0.1 mg/L; the median in rivers and groundwaters is about
0.04 and 0.06 mg/L. respectively, and 0.03 mg/L in lake waters (10. 29).
The oxidation of bromide by ozone is of particular importance in drinking water
treatment. Br" reacts with ozone with an appreciable rate constant of 160 M"1 s"1 (14). A
thorough description of the reaction of bromide with ozone, specifically the kinetics of
formation of hypobromous acid (HOBr) and bromate (Br03 ) is presented in Haag and Hoigne
(30). Ozoi""; oxidizes bromide to form HOBr which reacts further with ozone, only when it
exists in its ionized form, as hypobromite ion (OBr). Hypobromite is oxidized to bromate and
to an unidentified species that regenerates bromide. The following scheme of reactions
describe the formation of bromate (30).
O3 + Br 	 0-> + OBr	(3.3)
03 + OBr" 	 20, + Br"	(3.4)
203 + OBr" 	 202 + Br03	(3.5)
Equations 3.3 and 3.4 represent the catalytic decomposition of ozone. HOBr/OBr"
concentrations are pH dependent, and the equilibrium shifts towards the protonated form at
lower pH. The dissociation constant (pIC,) is 8.66 at 25 °C. The reaction of ozone with
HOBr is negligible, however, in the presence of organic matter, HOBr reacts to form
bromine-substituted DBPs.
Krasner and co-workers surveyed bromate formation in the US. They estimated that
the 20th, median, and 80th percentile for bromate occurrence in surface waters using ozone for
predisinfection is in the range of 0.5-0.8, 1-2 and 3-5 fig/L, respectively. The 90th and 95th
percentile are in the range of 5-20 ^g/L (31). Therefore, because of the proposed regulations,
21
-------
some utilities would not meet the MCL for bromate. These utilities would either choose not to
employ ozone in their DWT or operate the ozonation system such that bromate formation is
decreased. Bromate formation is affected by the reaction conditions.
Impact of pH: Because of the distribution of the species of free bromine (HOBr/ OBr"), the
pH effect on the formation of ozonation by-products is significant. At low pH, HOBr
dominates and the reactions with the organic matter to form bromine-substituted by-products
become the most important reactions. At higher pH, the equilibrium shifts to OBr" and this
ion can be further oxidized to bromate. Therefore, decreasing the pH of ozonation will
decrease bromate formation, but may enhance the formation of bromine-substituted DBPs.
The impact of pH on bromate formation and control has been investigated by many researchers
(32, 33).
The effect of bromide concentration and pH on bromate formation in raw Ohio River
water can be seen in Figure 3.1 (32). An ozone residual of 1 mg/L was used for each pH and
bromide level evaluated. As expected, increasing bromide concentration results in an increase
in bromate formation. The effect is more pronounced at higher pH. Although bromate
concentration decreases with decreasing pH, organo-bromine compounds are enhanced. This
effect can be seen in a study that investigated the effect of operational parameters such as pH
and temperature on DBP formation during ozonation (33). Table 3.1 depicts the increase in
total organic bromide (TOBr), a surrogate measure of the bromine-substituted organic
compounds, as a function of decreasing pH and increasing temperature for Contra Costa Water
District (CCSW) water source in California, at a constant bromide concentration of 1 mg/L,
and an ozone to dissolved organic carbon (DOC) concentration of 3 mg per mg. Although
-------
decreasing the pH. at a certain ozone residual, decreases bromate formation, there is an
increase in TOBr. A trade-off exists between the control of bromate and the formation of
TOBr.
Effect of Temperature: An increase in temperature will increase the rate of ozone
decomposition as the activation energy increases. At the same time, an increase in
temperature will increase the reaction rates. Siddiqui and Amy (33) report an increase in
TOBr formation with an increase in temperature from 20 to ^0 °C, as seen in Table 3.1. The
authors also report a small increase in bromate formation which they attribute to higher OBr"
concentrations. This is a consequence of an increase in the dissociation constant of HOBr.
Effect of Ammonia Addition: TOBr and bromate formation decrease upon the addition of
ammonia. This occurs because free bromine, a result of the oxidation of bromide by ozone,
reacts very rapidly with ammonia, faster than the reaction of bromide with ozone. The
following reactions explain the reaction of HOBr and OBr" with ammonia.
HOBr + NH3	 NH2Br + H20	(3.6)
OBr' + NH3	 NH2Br + OH"	(3.7)
The formation of dibromamine and tribromamine is also possible. Combined bromine reacts
with ozone to form nitrate and bromide (34).
NH2Br 4- 303 + 2H20	 N03" + Br" + 2H30+ 4- 302	(3.8)
Equation 3.8 indicates that ammonia appears to compete with other reactions that involve free
bromine, resulting in ozone consumption and an enhancement of its decomposition.
Consequently, a decrease in bromate formation occurs because of enhanced ozone
23
-------
decomposition. Both ammonia and bromamines can scavenge OH radicals and inhibit bromate
formation through the inhibition of free radical reactions (35). A decrease of about 30 to 36
percent in bromate formation upon ammonia addition was observed (33).
Although ammonia addition can decrease bromate and TOBr formation, it does
enhance ozone decomposition and may form other unknown by-products. In DWT plants that
employ ozonation for disinfection, the presence of ammonia should be accounted for in
determining the required ozone concentration for the purpose of disinfection.
Effect of Contact Time: In the presence of bromide, ozone contact time has a significant
effect on bromate formation. For a given ozone residual, increasing the contact time results in
an increase in bromate formation. This can be seen in Figure 3.2 (32). For a bromide
concentration of 258 ng/L, ozone was applied to meet the demand in raw Ohio River water
and produce a 1 mg/L ozone residual after 7 minutes contact time. After the initial contact,
the ozone residual was allowed to dissipate over time. As the contact time increased from 7 to
30 minutes, ozone residual decreased by 70 percent and bromate formation increased by about
70 percent. Therefore, in the presence of a residual, as the contact time increases, so does
bromate formation.
Effect of Ozone Dose and Residual: Increasing ozone dose results in an increase in bromate
formation. For the reaction to take place, the ozone demand of the water should be satisfied
and a residual ozone should be available. The reactivity of ozone with organic compounds
present in water is significant. In fact, the reaction of phenolic compounds with ozone is
known to be more rapid that with bromide (14). Figure 3.3 shows a correlation between
ozone residual and bromate formation for three bromide levels (36). Raw Ohio River water
24
-------
with an initial TOC concentration of 1.68 mg/L was ozonated. The ambient bromide
concentration was 50.8 ^g/L; the other two bromide levels were attained by spiking bromide
into the water prior to ozonation. Other reaction conditions, such as pH, organic matter
concentration and characteristics, and ozonation temperature were held constant. As ozone
residual increased at any bromide level, bromate concentrations increased. The correlation
coefficients show the strong relationship between ozone residual and bromate formation at all
bromide levels tested.
The relationship between ozone dose and residual can be seen in Table 3.2, which
shows the effect of ozone dose on bromate formation for the same study (36). As the ozone
dose is increased, by a factor of 5, the ozone demand increases and eventually reaches a
plateau. At low ozone doses, ozone is consumed in fast reactions with easily oxidizable
substances, and only very small residuals can be detected. As the demand is satisfied, higher
ozone doses result in higher residuals and consequently higher bromate concentrations. In this
study, TOBr was not measured. The table shows that the ozone demand that may be exerted
because of the ten-fold increase in bromide concentration is minimal in comparison to the
organic demand. The table also shows the log inactivation of Giardia cysts for the calculated
CT (concentration x time) values (based on T10 and a contact time of 7.4 min.). The USEPA
has specified the CT concept to assure adequate disinfection, where C is the concentration of
dissolved disinfectant (in mg/L) and T is the nominal contact time in minutes. T is equal to
T10, which is determined by tracer studies and equals the time it takes for 10 percent of the
tracer mass to exit the contactor (37).
As can be seen in Table 3.2, higher ozone concentrations result in more inactivation of
25
-------
the cysts, but also increase bromate formation. The increased formation of bromate is
accompanied by a simultaneous decrease in bromide concentration. Therefore, a trade-off
exists between the ability of ozone to inactivate microorganisms and the formation of bromate.
Krasner and co-workers have shown that staging ozone application into two or three
contactors may be beneficial in providing adequate inactivation concentrations and CT credit
while controlling bromate formation (31).
3.3 Reaction of Ozone with Organic Compounds
Ozone reacts with the components of microbial cells and with NOM present in source
waters. The reaction of ozone with NOM proceeds by two ozonolysis pathways: the cyclo-
addition direct pathway which takes place at electron rich sites; and the less specific free
radical pathway. The mechanism of inactivation of microorganisms is proposed as occurring
because of the oxidation of sulfhydryl groups contained in membrane proteins and by damage
to nucleic acids (38).
In aqueous media, water generally participates in the reactions of ozone with organic
compounds. Under ozonation conditions generally used for microbial inactivation in DWT,
the ozone doses are not high enough to cause mineralization of the organic matter to carbon
dioxide, but some degradation of the organic matter occurs. Because of the nature of these
types of oxidation reactions, ozonation has been used to control taste and odor compounds,
color, and some synthetic organic compounds. Recent studies have shown the advantage of
using ozonation to control the halogenated DBPs because some sites that would react with
chlorine are destroyed by preozonation. The reaction of ozone with organic matter also results
in an increase in the biodegradability of the organic matter. Smaller, more hydrophilic and
26
-------
oxygen rich compounds are generated by the ozonation process. Such compounds, if not
controlled by subsequent biological filtration, will serve as substrate and promote regrowth in
the distribution system.
3.3.1 Reaction of Ozone with NOM
Natural waters contain a variety of organic compounds, termed NOM. that are mostly
of unknown composition. NOM results from biological, chemical and photochemical reactions
that occur during the degradation and polymerizatic of algal, animal, and vegetable products.
It is thought to be composed of polymerized organic acids, polymerized carbohydrates, free
and combined amino acids and nucleobases in various concentrations. The sum of the
dissolved, suspended and colloidal organic matter is generally measured by the surrogate
parameter TOC.
Many researchers have tried to isolate the organic matter from water to try to
characterize it and identify some of its components. The dissolved organic compounds are
measured as DOC, which is the organic carbon that passes through a 0.45 /xm pore-diameter
membrane filter; or by absorbance of UV radiation at a wavelength of 254 nm. The
concentration of the organic mater is very variable in surface waters, and is generally higher
than in groundwaters.
The reaction of ozone with NOM occurs initially with electron rich sites. Therefore,
aromatic and unsaturated sites in the complex molecules that make up NOM, are very reactive
towards ozonation. Unsaturated organic compounds are characterized by a it cloud that
promotes electrophilic ozonolysis by 1,3-dipolar cyclo-addition. Splitting of the double bond
results in the formation of aldehydes and carboxylic acids. Glaze and Weinberg propose the
27
-------
formation of two moles of carbonvls per mole of oxidized substrate (39). Glaze and co-
workers postulate that higher molecular weight aldehydes, such as n-heptanal. result from the
ozonation of unsaturated fatty acids such as hexadecanoic acid (40). In the presence of
enough ozone, aldehydes may be further oxidized to yield carboxylic acids. The reactivity of
these sites is highly dependent on substituents: those that withdraw electrons hinder the
reaction, and those that supply electrons increase the reactivity of the double bond (Section
3.0).
The reaction of ozone with aromatic compounds will result in the formation of
dialdehydes and mixed aldehyde-acid compounds. An electrophilic reaction occurs on part of
the ring, and this is followed by further oxidation and degradation to smaller molecules
containing carbonyl functions. Initial ozonolysis on the ring is selective and degradation
pathways are more free-radical in nature.
Changes in NOM: The effect of increasing ozone dose on NOM in raw Ohio River water, can
be seen in Figure 3.4 (36). The source water is a surface water (DOC0 = 1.68 mg/L). The
ozone dose is increased to 2.5 mg/mg DOC, and under these ozonation conditions, the DOC
concentration remains unchanged, but the UV absorbance decreases with increasing ozone
dose; i.e., NOM is not completely mineralized (no conversion to carbon dioxide), but its
nature changes as apparent from the decrease in UV absorbance, which is an indication of the
loss of unsaturated and aromatic sites. Most of the effect on UV absorbance occurs at low
ozone doses indicating that these functional groups are very reactive and represent the easily
oxidizable portion of the organic matter. Increasing the ozone dose results in a further
28
-------
decrease in UV absorbance. but the effect is not as pronounced. The initial attack by ozone on
the aromatic nuclei is the rate determining step. Conjugated carbonvls and other functional
groups undergo slower stepwise degradation.
The change in UV absorbance for other organic matter sources can be seen in Table
3.3. Ozone doses between 2.0 and 2.5 mg ozone per mg DOC results in no change in DOC
concentration but the UV absorbance decreases by about 50 percent, in two soil-derived fulvic
acids, one soil-derived humic acid and for a dissolved organic matter derived from a natural
water source, the Biscayne Aquifer (41).
UV absorbance is a good surrogate parameter for humic and fulvic acids, subsets of
NOM. The ratio of UV absorbance to TOC is generally an indicator of the humic content of
the water. For a colored groundwater in Orange County, LA., a decrease in UV absorbance
and color with increasing ozone dose occurs. The decrease in color is because of the oxidation
of the chromophores in the organic matter (42).
Oxidation Bv-Products: The reaction of ozone with NOM leads to the formation of ozonation
by-products such as aldehydes, keto acids, carboxylic acids and AOC. Ozonation also changes
the characteristics of the organic matter in that it converts larger molecular size compounds to
smaller compounds (43, 44) and increases the hydrophilic fraction of the organic matter (44).
This results in an increase in biodegradability, that is measured as BDOC (45-47).
Although many aldehydes may be formed upon ozonation, current analytical techniques
can quantify only the low molecular weight aldehydes and dialdehydes. These compounds are
not regulated in the D-DBP Rule, but because of the health risks associated with them, they
are on the DWPL and may be regulated in the future. The most reported aldehydes formed
29
-------
during ozonation in DWT are: formaldehyde, acetaldehyde. glyoxal and methyl glyoxal.
These low molecular weight aldehydes are hardly present in untreated source waters but
increase in concentration with increasing ozone dose. This effect can be seen in Table 3.4
(48). Weinberg and co-workers report these aldehydes as the major DBPs in a survey of
pilot- and full-scale plants that use ozonation in both the US and Canada (49). Increasing
levels of higher molecular weight aldehydes (hexanal through tridecanal) with increasing ozone
dose have been reported by others (2, 50).
Similar to the aldehydes, another class of compounds, the keto acids, have been
reported by Xie and Reckhow (51). The concentrations of these compounds increase with
increasing ozone dose and they are found at concentrations that may be higher than the
corresponding aldehydes. Keto acids correlate well with the AOC, both increase with
increasing ozone dose (51). AOC concentrations represent the easily assimilable organic
carbon that can be utilized by Pseudomonas fluorescens (PI7) and Spirillum NOX strain.
The effect of ozone dose on AOC, BDOC and formaldehyde can be seen in Figure 3.5. The
concentrations of these ozonation by-products increase with increasing ozone dose (48).
The biodegradability of aldehydes is well documented (45, 47-49, 52). If these and
other biodegradable oxidation by-products are not well controlled during DWT, they will
serve as substrate and promote regrowth in the distribution system. Therefore, it is
advantageous for utilities that use ozonation to follow it with BAF to remove the biodegradable
fraction of the organic matter (Section 2.1).
Low molecular weight carboxylic acids have been reported and are quantifiable (53,
54). These carboxylic acids may be the dominant quantifiable oxidation by-products in
30
-------
ozonated waters (55. 56). Other formed by-products are peroxides and epoxides (50). Glaze
and co-workers report a qualitative formation of carboxylic acids, aliphatic and alicvclic
ketones, and hydrocarbons by gas chromatography/ mass spectrometry (50). More research
on the formation and biodegradability of ozonation by-products is needed. Development of
methods to accurately determine the concentrations of these compounds in aqueous matrices is
necessary.
3.3.2 Reaction of Ozone with C"II Constituents
Knowledge of the reactions of ozone with cell constituents is important to attempt to
describe the disinfection efficiency of ozone. Bacterial cells have been used as models to
locate the target sites for the oxidation reactions which lead to inactivation. The outer
boundary of a bacterium is composed of the cell wall which is often surrounded by exo-
polysaccharides. The cell wall is tough and hard to break by mechanical means. It gives the
bacterium its shape. It is composed of polysaccharides, sugars and amino acids. Inside the
wall, the cytoplasmic membrane surrounds the cytoplasm which contains the genetic material
in a single chromosome and the protein forming ribosomes. The cytoplasm pH is usually near
neutral and contains a high concentration of bicarbonate ions. It is therefore highly unlikely
that free radical ozone mechanisms take place in this medium, given that bicarbonate ions are
inhibitors of free radical reactions. Cell constituents are made up of carbohydrates, fatty
acids, amino acids and nucleobases (57). The following is a summary of the reactivity of the
various precursor molecules with ozone. For more details, the reader is referred to Langlais
and co-workers (2).
Carbohydrates: The reactivity of carbohydrates with ozone is very similar to that of aliphatic
-------
Table 3.1 The Effect of pH and Temperature on TOBr and Bromate Formation for
CCSW. (Adapted from Ref. 33).
Parameter
Concentration (^g/L)
TOBr
Bromate
pH
6.0
75
21
7.5
66
43
8.5
56
56
Ozonation
Temperature °C
20
64
42
25
71
59
30
78
62
Bromide concentration = 1 mg/L
Ozone to DOC ratio = 3 mg/mg
DOC0 = 3.4 mg/L
-------
Table 3.2 The Effect of Ozone Dose and Bromide Concentration on the Conversion of Bromide to Bromate (Ref. 36).
Transferred
Residual
Ozone
Bromate Concentration (/u.g/L)
CP
Log
Ozone dose
Ozone
Demand



(mg.min/L)
Inactivation1'
Br0c
Br0*
Br0-
(mg/L)
(mg/L)
(mg/L)
50.7 Oxg/L)
258 (fxg/L)
550 (/xg/L)

Giardia cysts
0
0
0
<0.2
<0.2
<0.2
0

0.89
0.28
0.61
1.1
7.6
14.2
0.32
1.3
1.37
0.66
0.71
4.1
25.4
24.4
0.74
3
1.93
1.16
0.77
10.5
45.2
58.8
1.32
>3
3.02
2.15
0.87
24.1
103
145
2.44
>3
4.32
3.27
1.05
40.7
198
303
3.71
>3
a CT = CavJ{X T10in which Cavg = 0.5 residual ozone and T10 = 2.27 min.
bTaken from Reference (37).
c Initial bromide concentration.
-------
Table 3.3 The Effect of Ozone* on UV Absorbance (Ref. 41)
Organic Matter (OM)
DOC (mg/L)
Absorbance
Absorbance
Percent
Source

Pre-03
(cm"1)
After 03
(cm"1)
Removal
Contech Fulvic Acid
4.62
0.314
0.114
51
Peat Fulvic Acid
4.92
0.188
0.106
44
Aldrich Humic Acid
4.56
0.391
0.189
52
Biscayne Aquifer
6.17
0.202
0.092
54
Dissolved OM




* Ozone dose range = 2.03 - 2.43 mg 03 per mg DOC
-------
Table 3.4 Changes in Aldehyde Concentration by Ozonation in Ohio River Water (Ref.
48).
Ozone-to-DOC
Ratio (mg/mg)
Concentration (/xg/L)
Formaldehyde
Acetaldehyde
Glyoxal
Methyl Glyoxal
Raw Water*
(no ozone)
0.8
0.35
2.7
0.15
0.42
8.0
1.3
3.9
3.9
0.87
16.8
2.4
7.4
9.9
1.64
23.5
3.4
10.9
14.9
* DOC0 = 2.66 mg/L.
Detection limit = 0.1 jxg/L.
-------
Figure 3.1 Effect of pH and bromide concentration on bromate formation (Ref. 32).
-------
400
350
300
250
200
150
100
50
_ 7 minute
detention time
0
Ohio River water
Initial bromide = 258 ug/l.
pH = 8.5
t-|o = 1 min
10 15 20 25
Incubation time (min)
30
35
Figure 3.2 Bromate formation as a function of incubation time (Ref. 32).
-------
Ozone residual (mg/L)
Figure 3.3 Correlation between ozone residual and bromate concentrations at three
bromide levels (adapted from Ref. 36).
-------
0	1	2
Transferred ozone I DOC (mg/mg)
Figure 3.4 Effect of ozone dose on the organic matter in Ohio River water (Ref. 36).
-------
500
400
< 300
CP
3
o
§ 200
CD
100
0
200
V FORMALDEHYDE
A BDOC
O AOC
1	2
TRANSFERRED 0Z0NE/T0C (mg/mg)
150
- 100
- 50
0
Figure 3.5 Effect of ozone dose on the formation of formaldehyde, AOC and BDOC
(Ref. 48).
-------
treatment for the worst possible condition. Initial costs, as well as operating and maintenance
costs play an important part in designing ozonation facilities.
5.1 Important Parameters Used for Design Purposes
Location of Ozone Application: Ozone may be applied in one or more locations in the
treatment train. The best location is determined from treatability studies and is the location
that allows the utility to meet its objectives at the least overall cost. In DWT plants, ozone is
typically applied to raw water, or to clarified water prior to filtration, or both.
Ozone Dose Requirements: Ozone generation and contacting are affected by the ozone
dose requirements. The ozone demand has to be taken into account and will depend on the
water quality and may vary seasonally. The decomposition of ozone should also be taken into
account. Many utilities will select ozone doses to meet the demand and provide a residual for
disinfection purposes, to obtain CT credit (Section 3.2.2).
Contact Time: Treatability studies are used to determine the minimum time necessary
to accomplish the targeted objectives. Because oxidation reactions of ozone with organic and
inorganic compounds are generally faster than disinfection reactions, the contact time may be
chosen based on disinfection requirements, if that is the treatment objective.
Both the contact time and the residual ozone concentration are important
because of the CT concept (Section 3.2.2). Determination of CT values can be found in
Appendix O of the Surface Water Treatment Rule Guidance Manual (37); both tracer studies
for determining TI0 and T100 and approaches for determining the average concentration in the
contactor (Cavg) are described.
45
-------
The amount of ozone that is transferred during the selected contact time is important
both from a reactivity point of view and from an economic perspective. Optimizing the
transfer of ozone into the treated water is more cost effective and affects the reaction rates.
5.2 Treatability Studies: Bench-Scale Versus Pilot-Scale Systems
Treatability studies are done to provide information for the design of ozonation
facilities. Studies are usually developed to account for ozone application objectives, and are
expected to take into account the seasonal and annual variation in the water quality that is to be
treated. Each treatability study is unique, but there is a common approach to testing. The
water quality should be evaluated and oxidation characteristics, such as stoichiometry and
kinetics, as well as ozone decay need to be assessed. There are three phases to be considered
in treatability studies. The first and initial testing phase is primarily to define the relationship
of the water to be treated to ozone. It is generally done by evaluating a 3 x 3 x 3 matrix: three
ozone doses, three contact times and another parameter such as gas concentration or pH. This
initial testing will also evaluate ozone demand and residual, the transfer efficiency and the
ozone decay rate. The second phase is a confirmation phase, where the selected ozone dose
and location from the first phase is used, and the effect of ozonation on downstream processes
and on meeting the treatment goals can be evaluated. The third phase of the treatability
studies involves the assessment of seasonal variations on the performance of ozonation in
meeting the primary objectives.
Because treatability studies are important in attempting to incorporate ozonation into a
DWT, they should be carefully planned and executed, taking into account all the treatment
objectives. Treatability studies can be classified by scale: bench- or pilot-scale.
46
-------
Bench- Scale Reactors: Bench-scale studies can be utilized to develop preliminary
estimates of design parameters. The most common reactors used for such studies are the batch
reactor, the semi-batch reactor and the continuous flow-through reactor.
Batch reactors are generally used to determine the stoichiometry, kinetic constants and
mechanisms in the ozonation reactions. Reported contactors vary in size from less than 1 L to
70 L. Generally, two solutions, one containing ozone at a known concentration, and the other
containing the reactants are brought together under controlled temperature and mixing
conditions. The extent of the reaction is monitored by sampling the reactor and measuring the
concentration of reactants and products over time. However, it is difficult to translate results
of bench-scale studies into design parameters.
Ozone is continuously introduced into semi-batch reactors while the volume of the
treated water is constant. The semi-batch system can be used to estimate required doses,
obtain information on residual decay, and for disinfection studies. The reactor is generally
well mixed, the feed gas is transferred into the liquid phase, and the concentration of the
reactants is uniform in the semi-batch reactor. The disadvantage of the semi-batch reactor is
that it can overestimate the extent of the reactions. The easily oxidizable compounds will react
immediately because there is no continuous inflow of water into the reactor, ozone will start to
react with the more oxidation-resistant compounds. Therefore, more competitive oxidation
reactions occur in the flow-through systems than in the semi-batch systems. This allows for
the build up of reaction intermediates in the semi-batch reactor and this may affect the end
products and ozone consumption. Another disadvantage is having to monitor low gas flow
rates.
47
-------
Continuous flow reactors have a continuous flow of both liquid and ozone feed gas. In
these systems, it is important to optimize the gas-to-liquid ratio to obtain good transfer of
ozone into the liquid phase. The problems with these type of reactors are having to control
and monitor very low liquid and gas flow rates and handling large volumes of water.
Bench-scale systems are practical, cost-effective and easy to construct and maintain.
The most common bench-scale systems use a countercurrent contactor with bubble diffusion
devices for the dissolution of ozone into the water. The hydrodynamics of the reactor vary
from plug-flow to completely mixed stirred tank reactors. The main problem with them is the
difficulty in determining the effect of the ozonation process on the downstream processes, and
monitoring applied and transferred ozone doses when utilizing semi-batch and flow-through
systems.
Pilot-Scale Systems: These systems should be designed with liquid-phase hydrodynamics
similar to those used in the full-scale reactors. Good gas-to-liquid ratios should be maintained.
The reactor characteristics can be described by tracer studies (77). These studies are
necessary to relate the pilot-scale results to the full-scale systems.
The most common pilot-scale reactor is the bubble column reactor, which is typically
operated with the gas flow countercurrent to the liquid flow. The hydrodynamics of the
reactors may vary from plug flow reactors to completely mixed stirred tank reactors.
In designing treatability studies, the selection of reactor depth and volume, the liquid
flow and gas flow rates are important. For flow-through studies, the choice of reactor volume
is determined by the liquid flow rate. For fast oxidation reactions, contact time can be short,
while for ozonation that is primarily for the purpose of disinfection, sufficient CT values have
48
-------
to be obtained. Another purpose of reactor depth is for obtaining high transfer efficiencies and
minimizing production costs. Taller contactors will behave more like plug-flow reactors,
while the shorter ones will behave as completely mixed stirred-tank reactors. Typical
contactor diameters are 0.1 - 0.3 m (4 to 12 in) and a reasonable height is 4.6 m (15 ft). Glass
columns are usually preferred because it is an advantage to monitor the ozonation process.
The liquid flow rate is selected based on the process needs and the hydrodynamics of
the reactor, and a general rule to piloting is to use a higher flow rate than that required for the
subsequent processes and waste the excess flow. The gas flow rate is also important. Ozone
dose rate can be changed by either changing the gas flow while maintaining a constant gas
phase ozone concentration, or changing the gas phase concentration while maintaining a
constant gas flow. The ratio of the gas flow to the liquid flow determines the gas-to- liquid
ratio and impacts the transfer efficiency. This ratio is important in assessing reactor
performance and affects the hydrodynamics of the reactor.
6.0 ANALYTICAL DETERMINATION OF OZONE
6.1 Common Terminology Used for Flow-Through Ozonation Systems
In evaluating ozonation systems, the minimum required information is the applied gas-
phase ozone concentration, off gas ozone concentration, temperature, pressure, gas flow rate,
liquid flow rate, liquid depth, liquid volume, dissolved ozone residual concentration and an
assessment of ozone decay.
Water Flow Rate: (QL) (L3/T). The water flow rate is measured to calculate hydraulic
detention time and may be adjusted to obtain the selected gas-to-liquid ratio. It is needed to
49
-------
calculate ozone dose in flow-through systems.
Gas Flow Rate: (QG) It is generally measured as L3/T and is reported at NTP. The
reference pressure is usually near one atm. (760 mm Hg), and the reference temperature is
variable.
Ozone Dose Determination: In order to determine the transferred ozone dose, the
applied gas and the off gas should be measured.
Applied ozone (AO) = mg 03 / L0
Off gas ozone (00) = mg 03 / LG
Gas-to-liquid ratio = G/L = Qg/Ql
Applied ozone (AO) = mg 03 / LG x Qg^Ql = mg 03 / Lw
Off gas ozone (00) = mg 03 / LG x Qg/Ql = mg 03 / Lw
Transferred ozone (TO) = AO - 00 = mg 03 / Lc
Transfer efficiency (TE) = (TO/AO) x 100 (percent)
Ozone Demand: The ozone demand is the difference between the transferred dose
(TO) and the dissolved ozone liquid-phase residual (DO). For a given water, the contact time,
the ozone dose and the contactor design will affect the ozone demand (DO).
Ozone demand (OD) = TO - DO
If the objective of ozonation requires a given concentration for residual disinfection (to
meet CT, for example), then the applied dose should satisfy the demand and exceed it to obtain
the necessary residual. The demand of the water is a function of the reactivity of ozone with
various compounds in water, Sections 3.2, 3.3. The determination of ozone decay for each
50
-------
water is important, particularly for disinfection purposes. It indicates the type of competing
reactions that occur after the immediate demand is satisfied.
6.2 Gas Phase Measurements
UV absorbance, iodometry and chemiluminescence are the three most common
methods for the measurement of ozone in the gas phase (78). These methods have been used
to measure ozone that has been stripped out of the solution to the gas phase (such as the off-
gas) and to measure the gas gener?red so as to determine the applied gas.
UV absorption is a common method for the measurement of ozone in the gas phase,
provided the molar absorptivity is known (79). For full-scale DWT plants, commercial UV
monitors typically require a gas flow rate of at least 1 L/min.
For the applied gas, excess applied gas is generated during the process and therefore,
commercial monitors can be used. If the flow rates are lower than what is required for the
monitors, then wet chemistry analytical methods, like iodometry, are employed. It is
important to use the same gas flows and pressures to the generator during applied gas
concentration measurements and during the actual ozonation of the water. For the off gas, or
for the applied gas where low gas flow rates are employed, iodometry or lab-scale gas-phase
spectrophotometers can be used.
The principle of operation of UV absorption is used in commercial monitors and on a
laboratory scale. In the laboratory, a UV spectrophotometer, at a wavelength of 253.7 nm,
which is where the maximum absorbance of ozone occurs, is used. The gas flow is passed
through a flow-through cell that is in the path of the incident light.
51
-------
Absorption of the radiation is governed by the Lambert-Beer Law:
l0
Absorbance = Log (—) = £bc
(6.1)
Where:
I0 = intensity of the incident light beam with no sample present.
I = intensity of the light beam through the sample,
e = specific molar absorptivity at the selected wavelength,
b = length of light path.
c = concentration of the absorbing material (in this case, ozone).
The amount of light is measured when ozone is present and compared to the reference
gas which can be air, or oxygen depending on the gas used for ozone generation. The meters
can be calibrated using other known methods such as iodometry.
Measurement of ozone in the gas phase can also be accomplished by wet chemistry
methods. A review of disinfectant residual measurement methods is presented in Gordon and
co-workers (80). The most common method is iodometry (80, 81). This involves the
oxidation of iodide by ozone to form iodine, which in turn is titrated with sodium thiosulfate.
The iodometry method is based on one mole of ozone liberating one mole of iodine. The
following reactions describe the chemistry:
03 + 21" + H20 	 I, + 02 + 20H"
(6.2)
I2 + 2S2032"
21" + S4062"
(6.3)
52
-------
There are some disadvantages to this method, particularly that the measured oxidant is
not only ozone, but a combination of all oxidants capable of oxidizing iodide.
The method is easy to use, involves bubbling of ozone into a gas-wash bottle
containing a known volume of a 2 percent potassium iodide solution. The liberated iodine,
after acidification with sulfuric acid, is titrated with sodium thiosulfate (0.1 N) solution, which
is standardized using potassium dichromate as a primary standard. A starch indicator solution
is used near the end point to simplify the visual determination of the end point.
The concentration of ozone can be calculated as follows:
mg Ozone /L = (mi of thiosulfate) x N (thiosulfate) x 24 mg x 1000 / (sample vol. ml)
where 24 is the conversion factor to convert units of mg ozone/meq.
Another problem with the method is that the flow rate of ozone into the potassium
iodide solution affects the measurement (79). The method is robust and may be used as a
check on the UV spectrophotometry method. In a study that compared many disinfectants and
the analytical methods to measure them, determination of ozone by iodometry was classified as
having a working range of 0.5 to 100 mg/L, and a detection limit of 0.002 mg/L (80).
Therefore the use of iodometry is acceptable for gas phase ozone measurements, but
not for liquid residual.
6.3 Liquid Phase Residual Measurement
The determination of the ozone residual in the aqueous phase is very important in light
of the CT concept. Historically, iodometry was used as the reference method, against which
other methods were standardized. However, because of all the problems associated with it, it
has been stated that iodometry is not a good method to determine the concentration of ozone in
-------
the liquid phase. There are other methods to measure ozone in the liquid phase among them:
Arsenic (III) direct oxidation; the FACTS procedure; the DPD method; o-toluidine; carmine
indigo; amperometry and other electrochemical methods (80). All of these methods have
advantages and disadvantages and will not be discussed herein.
Indigo trisulfonate is the method of choice for liquid-phase ozone residual. It is the
most common and is now the accepted method (68). It was developed by Bader and Hoigne
(82). It is very sensitive, fast and subject to fewer interferences than most colorimetric
methods or iodometry methods. The method is selective for ozone, hence compounds like
chlorite, chlorate, hydrogen peroxide and others do not oxidize indigo at pH 2. Chlorine,
bromine and iodine are believed to cause come interferences, as well as the oxidized forms of
manganese. Adding malonic acid will mask the interference of chlorine.
The stoichiometry of indigo oxidation by ozone is 1:1. The indigo complex absorbs
light at 600 nm. When oxidized by ozone, the color disappears as a function of ozone
concentration. The change in absorbance by ozone oxidation can be followed
spectrophotometrically. The decrease in absorbance is linear over a wide concentration range.
The slope of the calibration curve at 600 nm is 0.42 ±0.1 cm"1 per mg/L, which is based on
a molar absorptivity of 2950 M"1 cm"1.
The indigo spectrophotometric method has been adapted to accommodate three working
ranges of ozone concentrations: 0.01 - 0.1 mg/L with an MDL of 0.001; 0.05 - 0.5 mg/L with
an MDL of 0.006 mg/L; and greater than 3 mg/L with an MDL of 0.1 mg/L (80, 82). It is
important to do a preliminary assessment of ozone residuals, because the reagent
concentrations used for the test should correspond to the ozone concentration ranges.
54
-------
An indigo stock solution (1 mM potassium indigo trisulfonate) is prepared using
distilled deionized (DI) water (500 mL), 1 mL of concentrated phosphoric acid, and 770 mg of
potassium indigo trisulfonate. The solution is stable when kept in the dark for about 4
months. The absorbance of a 100-fold dilution should be checked periodically and the solution
discarded if the absorbance drops below 0.16 absorbance units per cm. The stock solution is
then used to make up working solutions for the ozone concentration ranges listed above. For
example, a working solution (for ozone concentrations in the range of 0.05 to 5 mg/L) is made
from the stock as follows: 100 mL stock indigo solution, 10 g sodium dihydrogen phosphate
and 7 mL phosphoric acid are added to DI water and the solution made up to 1 L. The
absorbance of the solution is monitored and a 10-fold dilution should have an absorbance equal
to or greater than 0.17 absorbance units per cm.
To determine the residual ozone concentration, 10 mL of the working indigo solution is
placed in a 100 mL volumetric flask. This is diluted with to 100 mL with DI water. This is
the blank solution. To a second flask, 10 mL of indigo are added and the solution containing
ozone is added with immediate mixing into the indigo solution. It is important to ensure no
degassing of ozone occurs because of bubble formation to prevent any ozone loss. The
decolorization of the indigo solution is compared to the absorbance of the blank (no ozone)
solution.
The residual ozone concentration is calculated as:
Ozone concentration mg/L = (AAbs x 100)/(f x b x V)
where AAbs is the difference in absorbance between the blank and the sample; b is the length
of the cuvette in cm; V is the sample volume (normally 90 mL) and f = 0.42. The factor f is
55
-------
based on a sensitivity factor of 20,000 M'1 cm"1 for the change in absorbance at 600 run per
mole of added ozone. The factor f = 0.42 corresponds to a molar absorptivity for aqueous
ozone e258 = 29 50 M^cm"1.
The advantage of the method is its ease, reproducibility, and accuracy. The sample
containing ozone should be collected directly into the indigo solution, while swirling the flask
to ensure mixing. Vigorous shaking should be avoided to prevent bubble formation and ozone
loss.
For details on other methods of analysis the reader is referred to other references (2,
68, 79, 80, 82).
Table 6.1 presents standard or EPA methods for analytical measurements used for the
ozonation system in DWT. The standard methods for some DBP analyses are also given.
7.0 ADVANCED OXIDATION PROCESSES
Advanced oxidation processes are chemical processes that generate free radicals by any
mechanism that accelerates the decomposition of the ozone molecule. As was explained in
Section 3.0, the decomposition of ozone can be accelerated by a variety of water contaminants,
including hydroxide ion, humic material and some transition metals. The decomposition of
ozone was described in detail in Section 3.1. The processes that are used to accelerate ozone
decomposition to form the more powerful and less selective free radicals, particularly the
hydroxyl radicals, are peroxone, which uses hydrogen peroxide with ozone, UV radiation with
ozone, elevated pH levels with ozone, and UV with hydrogen peroxide. The advantage of
such processes is that the oxidation of chemical contaminants can be achieved more effectively
56
-------
Tnhle 6.1 Analytical Methods
Parameter
Method
Reference
TOC or DOC
5310 C
68
UV254 absorbance
5910 B
68
Bromide
IC, Method 300, 4500-Br C
87, 68
Bromate
Method 300
87
03 residual
45OO-O3 B
68
03 dose (gas phase)
422
81
Aldehydes
6252 (proposed)
68
Keto acids
NA
51
Carboxylic acids
NA
53, 54
Total THMs
Method 551, 6232 B
88, 68
HAA6
6251 B
68
TOX
5320 B
68
NA = not applicable
-------
9.0 BIBLIOGRAPHY
(1)	A. Vosmaer, Ozone, Its Manufacture, Properties and Uses. Van Nostrand
Publishers, New York (1916).
(2)	B. Langlais, D.A. Reckhow and D.R. Brink (editors), Ozone in Water Treatment
Application and Engineering, Lewis Publishers, Chelsea, MI. (1991).
(3)	G.C. White, Handbook of Chlorination and Alternative Disinfectants, 3rd edition,
Van Nostrand Reinhold, New York (1992).
(4)	R.G. Rice, Ozone Sci. & Engrg., 17:5 (1995).
(5)	C.H. Tate, Journal AWWA, 83:5:40-47 (1991).
(6)	R.M. Clark and W.A. Feige, "Meeting the Requirements of the Safe Drinking Water
Act," in Strategies and Technologies for Meeting SDWA Requirements, (R.M. Clark
and R.S. Summers, editors), Technomic Publishing Co., Lancaster, PA. pp. 1-46,
(1993).
(7)	F.W. Pontius, Journal AWWA, 85:9:18 (1993).
(8)	F.W. Pontius, Journal AWWA, 85:11:22 (1993).
(9)	F.W. Pontius, Journal AWWA, 88:3:36 (1996).
(10)	USEPA, National Primary Drinking Water Regulations, Disinfectants and
Disinfection Byproducts. Proposed Rule, Federal Register, 40, Parts 141 and 142,
July 1994, EPA 811-Z-94-004.
(11)	S.W. Krasner, J.J. Westrick and S. Regli, Journal AWWA, 87:8:60 (1995).
(12)	J. Hoigne and H. Bader, Water Research, 17:173 (1983).
(13)	J. Hoigne and H. Bader, Water Research, 17:185 (1983).
(14)	J. Hoigne, H. Bader, W.R. Haag and J. Staehelin, Water Research, 19:8:993 (1985).
(15)	P.S. Bailey, Ozonation in Organic Chemistry, Academic Press, Inc., New York
(1978).
(16)	M.D. Gurol and P.C. Singer, Envir. Sci. & Technoi, 16:7:377 (1982).
62
-------
(17)	J. Staehelin and J. Hoigne. Envir. Sci. & Technol., 19:1206 (1985).
(18)	F.iM. Clark and E. Bone. Journal AWWA. 59:8:1036 (1967).
(19)	T.D. Waite and F.M.M. Morel, Envir. Sci. & Technol., 18:860 (1984).
(20)	A. Stone and J.J. Morgan. Envir. Sci. &. Technol., 18:450 (1984).
(21)	A. Stone and J.J. Morgan, Envir. Sci. & Technol., 18:617 (1984).
(22)	W.R. Knocke and co-workers, Alternative Oxidants for the Removal of Soluble Iron,
AWWARF Report (1990).
(23)	T.L. Theis and P.C. Singer, "The Stabilization of Ferrous Ion by Organic
Compounds in Natural Waters," in Trace Metals and Metal-Organic Interactions in
Natural Waters, (P.C. Singer, editor), Ann Arbor Science Publishers, Ann Arbor,
MI (1973).
(24)	B. Legube and co-workers. "Removal of Humic Substances and Manganese in
Slightly Mineralized Water by Means of Iron Salts and Ozone, Proceedings
AWWA Annual Conference, Orlando, FL (1988).
(25)	H. Paillard and co-workers, Ozone Sci. & Engrg., 11:93 (1989).
(26)	Y. Kurokawa and co-workers, Environmental Health Perspectives, 87:309 (1990).
(27)	Ben-Ezer.D. Gradus and co-workers, American Journal of Nephrology, 4:188 (1984).
(28)	D.P. Hautman, "Analysis of Trace Bromate in Drinking Water Using Selective
Anion Concentration and Ion Chromatography," Proceedings, AWWA, WQTC,
Toronto, Ont. Canada (1992).
(29)	G.L. Amy and co-workers, "Bromide Occurrence: National Bromide Survey,"
Proceedings AWWA Annual Conference, San Antonio, TX (1993).
(30)	W.R. Haag and J. Hoigne, Envir. Sci. & Technol., 17:5:261 (1983).
(31)	S.W. Krasner and co-workers, Journal AWWA, 85:1:73 (1993).
(32)	S.C. Allgeier, R.S. Summers and H.M. Shukairy, "Formation and Control of
Bromate," Proceedings, AWWA Annual Conf., Anaheim, CA (1995).
(33)	M.S. Siddiqui and G.L. Amy, Journal AWWA, 85:1:63 (1993).
63
-------
(34)	W.R. Haag and co-workers. Water Research, 18:9:1125 (1984).
(35)	G.V. Buxton and co-workers, J. Phys. Chem. Ref. Data, 17:2:513 (1988).
(36)	H.M. Shukairv, R.J. Miltner and R.S. Summers, Journal AWWA, 86:6:72 (1994).
(37)	USEPA, Appendix O to Guidance Manual for Compliance with the Filtration and
Disinfection Requirements for Public Water Supply systems Using Surface Water
Supplies, Office of Drinking Water, Washington, DC. (Dec. 1990).
(38)	M.H. Stewart and B.H.Olson, "Bacterial Resistance to Potable Water Disinfectants,"
in Modeling Disease Transmission and Its Prevention by Disinfection, (C.J. Hurst,
editor), Cambridge University Press, Cambridge, UK. pp. 140-192 (1996).
(39)	W.H. Glaze and H.S. Weinberg, Identification and Occurrence of Ozonation By-
products in Drinking Water, AWWA Research Foundation, (1993).
(40)	W.H. Glaze, M. Koga and D. Cancilla, Envir. Sci. & Technoi, 23:7:838 (1989).
(41)	G.L. Amy, C.J. Kuo and R.A. Sierka, Ozone Sci. & Engrg., 10:39 (1988).
(42)	L. Tan and G.L. Amy, Journal AWWA, 83:5:74 (1991).
(43)	H.M. Shukairy, M. Koechling and R.S. Summers, "Fractionated Organic Matter
DBP Formation: Kinetics, Reactivity and Speciation," Proceedings, AWWA Annual
Conf., New York, NY (1994).
(44)	M.T. Koechling, H.M. Shukairy and R.S. Summers, "Effect of Ozonation and
Biotreatment on Molecular Size and Hydrophilic Fractions of Natural Organic
Matter," in Water Disinfection and Natural Organic Matter - Characterization and
Control, (R.A. Minear and G.L. Amy, editors) ACS Symposium 649 Series,
pp. 196-210 (1996).
(45)	H.M. Shukairy, R.J. Miltner and R.S. Summers, Journal AWWA, 87:10:71 (1995).
(46)	G.E. Speitel Jr. and co-workers, Journal AWWA, 85:5:86 (1993).
(47)	H.M. Shukairy, "The Control of Disinfection By-Product Formation by Ozonation
and Biotreatment," Ph.D. dissertation, University of Cincinnati, Cincinnati, OH.
(1994).
(48)	R.J. Miltner, H.M. Shukairy and R.S. Summers, Journal AWWA, 84:11:53 (1992).
64
-------
(49)	H.S. Weinberg and co-workers. Journal A WWA, 85:5:72 (1993).
(50)	W.H. Glaze and co-workers. Journal AWWA, 81:8:66 (1989).
(51)	Y. Xie and D.A. Reckhow, "A New Class of Ozonation By-Products: The Keto
Acids,1' Proceedings AWWA Annual Conf. Vancouver, BC. (1992).
(52)	S.W. Krasner, M.J. Sclimenti and B.M. Coffey, Journal AWWA, 85:5:62 (1993).
(53)	S.W. Krasner and co-workers, "Ion Chromatographic Determination of Short- Chain
Carboxylic Acids in Ozonated Drinking Water," in Water Disinfection and Natural
Organic Matter - Characterization and Control, (R.A. Minear and G.L. Amy, editors)
ACS Symposium 649 Series, pp. 350-365 (1996).
(54)	S. Peldszus, P.M. Huck and S.A. Andrews, J. of Chromatography A, 723:27 (1996).
(55)	S.W. Krasner and co-workers, "Characterization of the Components of BOM,"
Extended Abstracts, 4th International BOM Conference, Waterloo, Canada, p. 1,
(1996).
(56)	W.B. Anderson and co-workers, "Evolution of BOM Components at a Full-Scale
Treatment Plant Employing Ozone," Extended Abstracts, 4th International BOM
Conference, Waterloo, Canada, p. 26, (1996).
(57)	F.C. Neidhardt and co-workers, Physiology of the Bacterial Cell. A Molecular
Approach, Sinauer Assoc., Inc. Publ., Sunderland, MA. (1990).
(58)	M. Dore and co-workers, "Ozonation of Molecules Constituting Cellular Matter,"
Proceedings 9th Ozone World congress, IOA, New York (1989).
(59)	W.L. Current and co-workers, New Engl. J. Med., 308:1252 (1983).
(60)	I. Campbell and co-workers, Vet. Rep., 111:414 (1982).
(61)	D.G. Korich and co-workers, Appl. Env. Microbiol., 56:5:1423 (1990).
(62)	J.E. Amoore, Journal AWWA, 78:3:70 (1986).
(63)	I.H. Suffet, C. Anselme and J. Mallevialle, "Removal of Tastes and Odors by
Ozonation," in Ozonation: Recent Advances and Research Needs, AWWA Annual
Conf., Denver, CO (1986).
65
-------
(64)	C. Anselme. J. Mallevialle and I.H. Suffet, "Removal of Taste and Odors by the
Ozone-Granular Activated Carbon Water Treatment Process," presented at the 7th
Ozone World Congress, IOA, Tokyo, Japan (1985).
(65)	S.W. Krasner, M.J. McGuire and V.B. Ferguson, Journal AWWA, 77:3:34 (1985).
(66)	W.H. Glaze and co-workers, Journal AWWA, 82:5:79 (1990).
(67)	W.H. Glaze, "Chemical Oxidation," in Water Quality and Treatment. 4th Edition,
(AW^A, F. Pontius, editor), McGraw-Hill, Inc. Publ., New York, p. 747, (1990).
(68)	Standard Methods for the Examination of Water and Wastewater, APHA, AWWA
and WEF, Washington, DC. (18ili Edition, 1992).
(69)	R.S. Summers and co-workers, Journal AWWA, 88:6:80 (1996).
(70)	D.A. Reckhow and P.C. Singer, Journal AWWA, 76:4:151 (1984).
(71)	J.G. Jacangelo and co-workers, Journal AWWA, 81:8:74 (1989).
(72)	J. Hoigne and H. Bader, Water Research, 22:313 (1988).
(73)	R.J. Miltner and R.S. Summers, "A Pilot-Scale Study of Biological Treatment,"
Proceedings AWWA, Water Quality, Vancouver, B.C. (1992).
(74)	R.J. Miltner, "Transformations of NOM during Water Treatment: Oxidation,
Formation of DBPs, Biodegradation," Proceedings AWWA and Lyonnaise des Eaux
Dumez workshop: NOM in Drinking Water, Chamonix, France (1993).
(75)	H.M. Shukairy, R.S. Summers and R.J. Miltner, "the Impact of Ozonation and
Biological Treatment on Disinfection By-Products," Proceedings 4th Workshop on
Drinking Water, Environment Canada, Montreal, Canada (1992).
(76)	H.M. Shukairy and R.S. Summers, Biological Treatment for the Control of
Disinfection By-Products: A Review," in Strategies and Technologies for Meeting
SDWA Requirements, (R.M. Clark and R.S. Summers, editors), Technomic
Publishing Co., Lancaster, PA. pp. 381-390, (1993).
(77)	O. Levenspiel, Chemical Reaction Engineering, 2nd Edition, John Wiley and Sons,
Inc., New York (1972).
(78)	J. Grunwell and co-workers, Ozone Sci. & Engrg., 5:203 (1983).
66
-------
(79)	G. Gordon and co-workers. Journal A WWA, 81:6:72 (1989).
(80)	G. Gordon and co-workers. Disinfectant Residual Measurement Methods.
AWWA Research Foundation Report, AVVWA, Denver CO. (1987).
(81)	Standard Methods for the Examination of Water and Wastewater, APHA, AWWA
and WPCF, Washington, DC. (16th Edition, 1985).
(82)	H. Bader and J. Hoigne, Water Research, 15:449 (1981).
(83)	J.P. Duguet and co-workers, Ozone Sci. & Engrg., 7:241 (1985).
(84)	L. Berglind and co-workers, "Removal of Organic Matter in Water by UV and
Hydrogen Peroxide," in Oxidation Techniques in Drinking Water Treatment, (W.
Kuhn and H. Sontheimer, Editors), EPA-570/9-79-020, EPA, Washington (1979).
(85)	J. Staehelin and J. Hoigne. Environ. Sci. Tech., 16:676 (1982).
(86)	W.H. Glaze, J.W. Kang and E.M. Aieta, "Ozone-Hydrogen Peroxide Systems for
the Control of Organics in Municipal Water Supplies," The Role of Ozone in Water
and Wastewater Treatment, Proceedings 2nd International Conference, Edmonton,
Alberta (1987).
(87)	USEPA, Methods for the Determination of Inorganic Substances in Environmental
Samples, (1993), EPA/600/R/93/100.
(88)	USEPA, Methods for the Determination of Organic Compounds in Drinking Water,
Supplement 1, (1990), EPA/600/4-90/020.
67
-------
FILTRATION
In the filtration process, filter efficiency is approximated by determining
-overall plant reduction in turbidity. The best way to assure high filtration
efficiency is to select an effluent turbidity goal (level) and stay below the
target value (such as 0.2 NTU).
Lowering turbidity is important in protecting public health and preventing
operational problems in the distribution system. Turbidity interferes with the
disinfection process, because the suspended particles shield microorganisms
from the disinfectant. The particles also combine chemically with the
disinfectant and leave less disinfectant to combat the microorganisms.
Turbidity also causes deposits in the distribution system that contribute to
taste, odor and bacterial growth problems. Therefore, lower turbidity results
in more effective disinfection and better public health protection.
The U.S. EPA currently requires a composite filtered water turbidity of 0.5 NTU
or lower 95 percent of the time as an indication of proper filtration. In low
turbidity water, the EPA Guidance Manual suggests a removal of 70 percent of
the raw water turbidity. These composite turbidity samples are to be made from
the water prior to the addition of post chemicals.
The major purpose of filtration is to remove the suspended material (measured
as turbidity) from water. This suspended material can include floe from the
coagulation/flocculation/sedimentation processes and microorganisms, both of
which are removed when water passes through a bed of granular material known as
the filter media.
AFO&M-2C-(11/15/1991)
2031/DM/W
2-65
-------
The filter is usually sand or some combination of sand, anthracite coal, garnet
and/or activated carbon. The filter used in the filtration process is commonly
thought of as a sieve or strainer that traps suspended material between the
grains of filter media. However, straining is the least important process in
filtration since most suspended particles can easily pass through the spaces
between the grains of the filter media.
FILTRATION - DESCRIPTION
Filtration depends on a combination of complex physical and chemical
mechanisms, the most important being adsorption. As water passes through the
filter bed, the suspended particles contact and adsorb (stick) on to the
surface of the individual filter grains or onto previously deposited material.-
The forces that attract and hold the particles to the grains are the same as
those at work in coagulation and flocculation. Some flocculation and
sedimentation occurs in the filter bed, a fact that points out the importance
of good chemical coagulation before filtration. Poor coagulation can cause
serious operating problems with filters as discussed later in this manual.
In all gravity filtration systems, the water level or pressure (head) above the
media forces the water through the filter media. The rate at which water
passes through the granular filter media may vary from 2 to about 10 gpm/sq ft.
(commonly referred to as filtration rate). The rate depends on the type of
filter media and pretreatment. A rate above 3 gpm/sq ft. is usually designated
high-rate filtration. At the AFC Plant, average filtration rates will vary
from about 6 gpm/sq ft. to about 3 gpm/sq ft. when the plant is producing
45 MGD.
AFO&M-2C-(11/15/1991)
2031/DH/U
2-66
-------
There are 12 filters in the plant. Each filter has dimensions of 20 feet by
32 feet, yielding an area of 640 square feet. At an average plant output of
45 MGD, there will be an average of about 1.5 MGD of recycled backwash water,
•giving a flow through the filters of 46.5 MGD. When one filter is being
backwashed, the total area available of filtration is 11 x 640, or
7,040 sq. ft. The average filtration rate with one filter being backwashed is
calculated as follows:
Filtration rate =	46,500.000 gal/day	= 4.59 gpm/sq ft.
1,440 min/day x 7,040 sq ft.
Normally, filter runs will be about 48 to 72 hours long, so that the filters
will be operating at an average rate of about 4.5 gpm/sq ft.
Filter control systems regulate flow rates through the filter. Rapid flow rate
increases or fluctuations can force previously removed particles through the
media. These fluctuations can be caused when a filter is taken out of service
for backwashing causing the filters remaining in operation to pick up the
non-operating filter's load.
In declining-rate filters such as these, flow rate varies with head loss. Each
filter operates at the same water surface level. This level varies with total
plant flow and with the number of filters in service. This system is
relatively simple, but requires an effluent control structure or master
throttling valve to maintain adequate media submergence.
AFO&N-2C-0 1/15/1991)
2031/DM/U
2-67
-------
When the filter is clean and filtering at the maximum rate of 6 gpm/sq ft., the
media head loss will be about 2 feet and the orifice plate/filter effluent-
piping losses will be about 8 feet. As the media clogs, its loss of head
increases. Since, the head available is relatively constant, the filtration
rate decreases. The piping losses decrease proportionally with the square of
the filtration rate. When the loss of head in the media reaches eight feet,
the loss in the piping will be two feet. To accomplish this balance, the
filtration rate will have reduced automatically and gradually to one half the
starting rate or 3 gpm/sq ft. The average rate will have been 4.5 gpm/sq ft.
The media in each filter consists of 12 inches of anthracite over 18 inches of
sand. The anthracite is the coarse larger-sized granular material with an
effective size of about 0.9 mm after washing and scraping; it has a specific
gravity of about 1.65. The sand is the fine media with an effective size of
0.5 mm and a specific gravity of 2.65. By using a fine media with a higher
specific gravity and a coarse media with a lower specific gravity, the layer:
tend to maintain their respective positions even after the agitation caused by
backwashing. Near the end of the backwash cycle, the backwash rate is slowly
reduced to allow the sand to settle more rapidly due to its higher specific
gravity and return to the lower portion of the filter. As described later, it
is important to reduce the backwash rate slowly to accomplish this regrading of
media. Some mixing will occur between the layers but this mixing actually
improves the effectiveness of the filter as the fine media tends to fill in the
pore space of the coarse media.
In filter operation, the coarse layer on top removes the larger suspended
solids while the finer particles pass through this layer and are removed at the
AFO&M-2C-(11/15/1991)
2031/DH/VI
2-68
-------
interface of the two media. The sand layer acts as a safety barrier. The deep
penetration allows greater solids loading on the filters prior to clogging or
breakthrough. Running the filters too long between backwashes pushes floe into
%the fine media and requires longer backwashing.
An underdrain system is provided under the dual media. Directly underneath the
dual media are 12 inches of various layers of gravel consisting of hard,
rounded stones with a specific gravity of at least 2.5. The gravel is placed
larger to smaller then back to larger from the bottom upwards. The two upper
layers of gravel will have sand mixed in it, but these top two layers of gravel
help keep the fine gravel below from moving. The fine gravel in the middle
layer prevents the sand from being washed into the underdrain system and helps
to distribute the backwash water flow across the entire filter area.
Underdrain blocks are located underneath the gravel layers. They consist of a
single layer of plastic dual lateral blocks shaped to form internal channels to
collect filtered water and distribute backwash water and air. These channels
hydraulically distribute the washwater and air evenly. The blocks cause little
loss of head at 3 to 6 gpm/sq ft filtration rates.
FILTRATION - OPERATION
The three major operations which influence or cause most filtration problems
are chemical treatment before the filter, control of filter floirf rate, and
backwashing the filter. If these three operations are not performed
effectively, the quality of the filtered water will suffer and additional
maintenance problems will occur.
AFO&W-2C-(11/15/1991)
2031/0M/U
2-69
-------
The importance of proper coagulation has been discussed earlier in this
manual. Proper adjustment of chemicals and mixing speed are only possible ifi
the filtration process is closely monitored. Since many raw water
characteristics, such as turbidity and temperature, are not constant, coagulant
dosage changes are necessary. Consequently, the continuous readout instruments
that keep track of filter turbidity and flow rate are very important. The
trends for these characteristics should be checked at least hourly.
Rate increases or rapid fluctuations in the filter flow rate can force
previously deposited filtered material through the media. At this plant,
however, the filter surface is large enough to dampen flow surges
hydraulically, so that this problem should not occur.
In this plant, the upstream filter level will be the same over all the
filters. Each filter therefore accepts the portion of flow that the relative
I
cleanliness of its filter bed will allow. When a filter is removed from
service for backwashing, the flow automatically redistributes to the others.
The downstream control is a throttling valve which will be adjusted each time
the plant throughput changes. This valve is called the Master Control Valve
and the plant operator sets the position of the valve at the plant operator's
console.
The hydraulic conditions at this plant are controlled by individual filter
orifice plates, individual filter effluent valves and a master control valve.
The head losses in these plates and valves are designed to maintain the proper
balance between the loss of head build up in the filters and the piping losses.
AFO&H-2C-(11/15/1991)
2031/DM/U
2-70
-------
The filter hydraulics are designed to reduce the possibility of the media
becoming air bound. The operator can prevent this by maintaining the levels on
the filters above Elev. 946. If the head loss in the media is always less than
the depth of water over the sand, there will not be any air released from the
water.
Any time a filter water level is low, it should be filled from the bottom, by
opening the filter effluent valve or washwater supply valve. The water will
replace the air and gradually fill the media without upsetting the gravel.
Filter run lengths should be selected such that the starting filtration rate is
between 2 and 3 times the filtration rate just before the filter is
backwashed. Maintaining these start to finish filtration rate ratios will
normally result in using the least amount of backwash water. The maximum run
length can be realized by maintaining a water level on the filters between
Elev. 946 and 947.5. The level on the filters can be raised or lowered by
adjusting the Master Control Valve which is located downstream of the filter
weir chamber. The valve adjustment should be made each time the plant
throughput is changed. As described earlier when the plant throughput is to be
changed, the raw water throttling valves are adjusted so that the raw water
meters reflect the new plant flow. This adjustment will result in the water
level changing on the filters. Once the filter level has reached equilibrium
the filter master control should be adjusted so that the water level on the
filters is about mid way between Elev. 946 and 947.5. Small incremental
adjustments will more quickly reach equilibrium on the filter land than drastic
changes.
AFO&H-2C-(11/15/1991)
2031/DM/VI
2-71
-------
There is a tendency for operators to allow filters to run beyond the 3 to 1
start vs. end of run filtration rate ratio since longer filter runs result.
While the runs are longer, the dirty filters are contributing very little
"Water, and the increased head on the filters will drive the floe further into
the media. This practice results in very little additional water being
filtered as a result of the longer runs but, and at the same time, mors water
is required for backwashing to properly clean the media. There is also an
indication that longer filter runs result in high filtered water particle
counts even though turbidities do not increase.
BACKWASHING - EQUIPMENT
In order to produce optimum cleaning of the filter media during backwashing,
either "surface wash" or "air scour" is required. These systems provide
additional scrubbing action to remove attached floe and other entrapped solids
from the filter media. The system utilized at this plant uses air scour]
introduced at the beginning of the backwash cycle.
The air supply system consists of two positive displacement blowers, each with
a capacity of 2,800 scfm at a discharge pressure of 4.5 to 6.0 psi, which will
supply 4.4 scfm of air/sq. ft of filter surface during the air scour phase of
the backwash. One blower is the backup for the other. There is no flow or
pressure control. The units can be operated at the blower or automatically
during backwash.
The washwater supply system consists of two pumps, a storage tank and a
backwash flow controller. The pump capacity is not adequate to successfully
backwash a filter, so sufficient water must be pumped to the tank for a
AFO&M-2C-(11/15/1991)
2031/DH/VI
2-72
-------
backwash before the cycle is started. The controller keeps the backwash rate
at a preset level. Each pump is a two stage vertical turbine pump that will
pump 4100 gpm at 83 feet TDH. The 1200 rpm pumps are driven by 125 HP motors,
and are mounted on- the clear wells, with one pump over each clear well. The
pumps are controlled by the level in the washwater tank. One pump will start
when the tank is four feet below the overflow and pump until the level rises to
two feet below the overflow. Under normal conditions the selected pump will
rotate after each pumping cycle, but both pumps will not operate at the same
time. When a clear well is taken out of service, the corresponding pump should
be switched off.
The storage tank is about 44 feet in diameter and 25 feet tall with a storage
capacity of about 250,000 gallon. The backwash flow controller consists of a
30-inch butterfly valve and a venturi tube. The rate of backwash is adjustable
at the plant operator's work station.
BACKWASHING - DESCRIPTION
As the filters continue to process water, solids will eventually saturate or
clog the filter, and must be removed by taking the filter out of service and
backwashing it. The filter must be backwashed if any of the following
conditions occur:
¦	Filter clogs (flow decreases),
¦	Breakthrough occurs (increase in effluent turbidity), or
¦	A specified time period has passed.
AFO&M-2C-0 1/15/1991)
2031/CM/V
2-73
-------
Backwashing consists of a sequence of steps which involves shutting down the
filtering process for the filter to be backwashed, feeding air through the^
underdrains, adding backwash water low rates with the air, turning off the air
and increasing the backwash rate; then gradually decreasing the backwash rate,
and finally putting the filter back into a filtering mode.
When one filter is taken out of service for backwashing (its inlet valve is
closed), the flow through each of the five remaining filters on that bank does
not change immediately. Therefore, the total flow from the filters is reduced
to 5/6 of the previous flow. Since the flow to the plant remains constant, the
water level on the remaining filters gradually increases to force more water
through them. At the end of a backwash, the clean filter starts and the levels
on the other filters drop to a slightly lower level than existed before
backwash started. It should not be necessary to adjust the Master Control
Valve for these short term fluctuations.
The operation and control of the backwash sequence can be conducted manually or
semi-automatically from the backwash control console, or automatically by the
computer at the Plant Control Center. The backwash water used at this plant is
water that has been previously filtered and chlorinated, pumped from the clear
wells and stored in a 250,000 gallon washwater tank. Figure 2.8 shows a
typical filter piping arrangement with the valve numbers used in the Contract
Drawings. All valve numbers which relate to a specific filter are shown on
Figure 2.9.
AfO&M-2C-<11/15/1991)
2031/DH/U
2-74
-------
ft* id rt *vr
an ti t*s(Atv*
P-«O0 * ^ P-40F
p-«oo	4oe
FILTER
NO. 7
i±
I . rrf
atfUr
NO A
FtLTCR

At2±
FtLTUf

HQ tO


Nan
FILTER

(TYPJ











»a rtircfy











{Ty#j

m





1*1


f
AJO./Z.
C
P • 400-^^^^ r • 4C
PARTIAL PLAN
SEC SHEET P-40J
P-«OQ
* iTent inciudco *nrn Am
wa*m tnreM only
6 ft rrcHs tvcojoeo with surrACc
WA*H +Y+7CM OJL Y
SOL >P P( 4W
&KAT!W» (Air*)
f— -rff rrr**&r-s*f r-
^Mcrre • /» rue m- titvAjnn or*jecur
ok ca&jcxero jf no gkhjtr* *np*jnt*£>
ajotc th/4 sneer suohjs itchs
COMMON TO GOTH tU+SJJ SVSTCH9
A&rf @ mi m svAmrrs ** srA/vu** *M4t
rvr s*au #*rf a *ruw	at/
^ _ ah f**K3 /M»r j»*f /* aw/A CMIAIN
itftfM IM OA1I NOVtMf
IOAII !/•" • t 0'
321
to 2
-------
BACKWASH FLOW
RATE CONTROLLER
(MOV 69200)
FROM
WASHWATER
STORAGE
TANK
FROM
WASHWATER
HEADER
1	
FROM
AIR
HEADER
f	
NOTE
SEE CONTRACT DWG. 1-6
FOR FULL VALVE NUMBERS
MOV
9

MOV
I 3
30
SETTLED
WATER
FLUME
MOV
15
T--HH-35-
I	MOV
17
-£-c5—| |
I—£	35
UNDERDRAIN header
GULLET
n
TROUGHS
MOV
1 |
MOV
7

II
—-ch—

TO
_4 FILTERED
WATER
FLUME
TO
-f REWASII
DRAIN
FILTER
DRAIN TO
DECANT
TANK
FIGURE 2.8
SCHEMATIC OF TYPICAL FILTER PIPING
-------
Before a filter can be backwashed, several conditions must be met:
¦	The filter must meet one of the three conditions that require backwash.
¦	No other filter may be in a backwash mode.
¦	The water level in washwater storage tank must be above Elev. 1,003.
¦	The water level in one of the decant tanks must be below Elev. 930.
The time interval between filter washes is measured from the beginning of one
wash to the beginning of the next wash. The queuing order	of the filters has
been selected so that headloss in the settled water flume	and filtered water
flumes will be about equal for all filters. The	queuing order is
1-7-2-8-3-9-4-10-5-11-6-12.
There are four operating modes for the filters which can	be selected at the
filter consoles or at the Plant Control Center:
Mode 1 - Off. Filter is taken out of queuing order. All valves can be
operated manually, but their operation will not initiate any automatic
operation.
Mode 2 - Manual backwash. Filter is taken out of queuing order and
individual valve operations are controlled by operators. Opening or
closing certain valves will initiate air wash and rate of flow control
automatically.
AFO&M-2C-(11/15/1991)
2031/DM/U
2-77
-------
Mode 3 - Semi-automatic backwash. Filter is kept in queuing order, but
operator must be present during backwash to initiate certain steps in thd
backwash sequence.
Mode 4 - Automatic backwash. Filters are washed by central control system
without action from operators.
The manual backwash mode allows the operators to adjust the filter valves from
the individual filter consoles and to initiate the actual backwash. The air
scour sequence and the backwash rate controller sequence are initiated when the
air valve is opened, and the backwash rate controller sequence is initiated at
the low rate (3,200 gpm) when the washwater supply valve is opened.
Under the semi-automatic backwash, a light will illuminate indicating that it
is time to wash the next filter in queue. The operator must acknowledge thfd
Request To Start at the filter console. The filter will then be shut down
automatically by the individual valves being adjusted in a pre-set sequence.
The operator will then be asked to initiate the backwash. An acknowledgement
signal will start the backwash sequence including the air scour and water
backwashing sequences under the semi-automatic mode, after a preset time the
operator will be asked to stop the backwashing. The backwashing at the high
rate will continue until the run is stopped; the washwater tank goes dry; or
the decant tank fills up. The operator must initiate the shutdown of the
backwash program, and the controls will automatically ramp down the backwash
rate, close the rate controller valve and put the filter into a standby
condition. To start the filter process again, it is necessary for the operator
to request that the filter be put back in service at which time the settle^
AFO&H-2C-(11/15/1991)
2031/DM/U
2-78
-------
water valve and rewash valve open. After an adjustable time, the operator must
initiate the opening of the filtered water valve and closing of the rewash
valve.
Under the automatic backwash mode, the filters will be backwashed in order of
the queuing at the time intervals set by the plant manager and, following
backwash, returned to service. No action is required by the operators.
The general sequence of a backwash is detailed in Table 12.5. An effective
manual backwash must duplicate the steps outlined for an automatic backwash in
this table.
BACKWASHING - OPERATION
Backwashing is the process of reversing the flow of water through the filter
media to remove the entrapped solids. In order to remove the trapped solids
from the filter media, the filter media must be expanded or fluidized by
reversing the flow of water. Backwash flow rates ranging from 10 to 25 gpm/sq
ft. of filter media surface area are usually required to clean the filter
adequately. Insufficient backwash rates may not completely remove trapped
solids from the filter media, while too high a backwash rate may cause excess
loss of filter media and gravel disturbance (mounding). Higher backwash rates
are usually required at higher temperatures and with larger media than at lower
temperatures and with smaller media. Backwashing at too high a rate is much
more destructive than backwashing at too low a backwash rate. The ideal
backwash rate is the rate which causes particle motion at the sand/coal
interface.
AFO&M-2C-(11/15/1991)
2031/DM/U
2-79
-------
There are a number of decisions which must be made by the operator regarding
the backwash sequence. They involve the various settings which can be made ati
the Plant Operators Console in the control room. Some of the settings will be
determined from initial plant operating experience and be left untouched
thereafter; while others will be changed seasonally or during radical changes
in raw water quality. These adjustments are discussed below.
Adjust the ramping rate and the water backwash rate during the last part of
the air scour. Too fast a ramping rate could cause the bed to become upset by
evacuating any accumulated air too fast. The ramping rate should be at least
one minute. The best backwash rate is one-half the velocity required to
fluidize the media, which can change with water temperature. The concurrent
air scour and backwash is the most effective part of the backwash sequence.
Two minutes of these concurrent actions is all that is available before the
water level reaches the trough and media loss occurs. If it is determined that|
the media is not being properly cleaned (individual particles show an
accumulation of slime) the filter run cycle should be reduced. Initially, this
rate is set at 5 gpm/sq ft or 3200 gpm, but may vary from 4-6 gpm/sq ft
depending on water temperature.
Adjust the length of time and the rate of the high rate water backwash. The
high rate may need to be adjusted seasonally. The intent of this high rate
wash is to flush out material loosened by the concurrent air scour and backwash
phase. There is a tendency for operators to extend this flushing stage beyond
the most efficient washwater usage period. It is not necessary to remove all
the material flushed out. It takes approximately twice as much backwash water
to remove the last 10% of the material as the first 90%, and the 10% remainina
AcC£jV2C- (11/15/1991)
2031/DM/U
2-80
-------
will not substantially shorten filter runs. Typically, operating the high rate
long enough to replace the volume of the filter should be adequate.
^Adjust the timing of the backwash polymer pump. The intent is for polymer to
be added to the backwash water so that much of the water remaining in the
filter media at the end of the backwash contains polymer. Initially polymer
will not be used.
Adjust the rate of ramp down. The ramping down is very important, since it
is during this stage that the sand and coal separate and the media regrades.
The 2 minute period has been selected since it will allow a reasonable length
of time at the critical backwash rate for separation to take place.
The backwash sequence will be interrupted if both decant tanks fill up, the
washwater tank empties or a power failure occurs. If a filter is being
backwashed and an interruption occurs, the washwater supply valve will close
and the washwater sequence will stop. Once the interruption is corrected, the
acknowledge button will light. When it is pushed, the backwash procedure will
go back to the beginning and repeat the previous steps. It will be necessary
for the operator to put the filter into the proper mode after the backwash is
completed, to put it back in the proper queuing position.
As discussed above, the filters can be backwashed using several operational
modes. For all modes, it is the time between backwashes of each filter (filter
runs) that is the important process control variable.
AFO&H-2C-(11/15/1991)
2031/DM/V
2-81
-------
Backwashes should be spaced evenly throughout the day. The timing should be
selected and adjusted with each raw water quality change to keep the start to
finish filter flow ratios between 2:1 and 3:1. The time interval between
washes should be determined by the plant manager and should not be changed by a
shift operator. The length of run should be based upon the filtration rate
ratio or limited by early breakthrough -which can not be eliminated by use of a
reasonable amount of coagulant/filter aid.
The plant manager should evaluate filter performance from season to season
comparing the filter run length, in hours, for each season. As the filter run
gets shorter, the amount of water used during the backwash cycle becomes
increasingly important when compared to the amount of water produced during the
filter run. Percent backwash water statistics should also be maintained and
evaluated, preferably in graph form.
During low plant outputs, less than 15 MGD, the headloss available for buildup
in dirty media is greatly reduced. There are two low output operating options
- increase the head loss, or shut off some of the filters. The head available
can be increased by increasing the start to finish filtration rate ratio to
3:1. Ratios higher than 3:1, in reality, mean that the dirty filters are
producing very little, and are nearly stopped. Alternately, filters can be
removed from service to keep about the same flow per filter as 15 MGD with
12 filters. Whichever option is selected, a backwash schedule which washes the
filters at equal intervals should be maintained. The interval selection should
be based upon head losses, and avoiding breakthroughs as under high plant
outputs. Filter runs should not exceed 72 hours.
AFOSM-2C-C11/15/1991)
2031/DM/W
2-82
-------
If short filter runs are occurring because of turbidity breakthrough, perhaps
more coagulant, better mixing or more filter aid is needed. If short runs are
due to rapid buildup of head loss (rapid decreases of flow), perhaps more or
less coagulant, better mixing or less filter aid is needed. It is the
operator's job to recognize these types of problems and choose the proper
remedial actions.
The backwash should be in automatic or semi-automatic except for emergencies or
periods of maintenance work when the operator can supervise the operation.
During manual control from the filter consoles, the length of wash can be
selected by the operator. One limit will be the amount of washwater
available. The washwater tank holds about 250,000 gallons. A normal wash with
2 minutes at the high rate will use about 45,000 gallons. Each additional
minute at the high rate adds an additional 10,000 gallons.
FILTER OPERATING PROBLEMS
When the filter is drained, its surface should appear smooth. If cracks
develop separating media from filter walls, or mudballs or ridges appear,
problems with backwashing exist. The observations should be recorded and be
reported to the plant manager.
Periodically, the operator - should check the condition of each filter by
observing the backwash and filter's surface. This includes watching for media
boils (uneven flow distribution) during backwashing, media carryover into the
washwater trough and the clarity of the waste washwater as the end of the
backwash cycle nears. A filter in good condition with even backwash
distribution will appear very uniform with the media moving laterally on the
AFO&H-2C-(11/15/1991)
2031/DM/U
2-83
-------
surface. Violent, upswelling boils of water indicate the presence of
problems. If some areas do not appear to clear up as rapidly as others, uneven
distribution of backwash flow is indicated.
Effective backwashing is essential to consistent production of high quality
water. Ineffective backwashing can cause problems as discussed below:
Mudball Formation - During filtration, grains of media become covered with
sticky, floe material. Unless backwashing removes this material, these grains
clump together and form mudballs. Mudballs will usually form near the top of
the media, at the dual media interface or at the perimeter of the filters. As
the mudballs become larger, they can sink into the filter bed during
backwashing and clog those areas where they accumulate. These clogged areas
become inactive, causing higher than optimum filtration rates in the remaining
active areas and unequal distribution of backwash water. Additional problems,
such as cracking and separation of media from the filter walls, may also
result. Mudballs are usually seen on the surface of the filter after
backwashing, particularly if the problem is severe. A periodic check for
mudballs should be made. Mudballs can be prevented by using adequate backwash
flow rates and filter agitation (air scour), which scrubs the filtered material
from the media grains. Filter agitation is essential for dual filters, since
mudballs can form deep within the bed.
Every 12 months, the filters may need to be "lance cleaned". A 1/2 inch pipe
with a single surface wash nozzle is used. The water pressure should be
40-50 psi. While maintaining a low rate backwash, the lance is forced
18 inches into the anthracite every 12 inches in all directions. Care must be
AF04H-2C-(11/15/1991)
2031/DM/V
2-84
-------
exercised not to exceed 18-inch penetration, or the gravel bed may be
disturbed. It is best to paint a stripe on the lance 68 inches from its end
(distance from trough lip to sand level) as a reference point for optimum
penetration.
Filter Bed Shrinkage - Filter bed shrinkage or compaction can result from
ineffective backwashing. Clean media grains rest directly against each other
with little compaction even at terminal head loss. However, dirty media grains
are kept apart by the layers of soft filtered material. As the head loss
increases, the bed compresses and shrinks, resulting in cracks and separation
of the media from the filter walls. The water then passes rapidly through the
cracks and receives little or no filtration.
Gravel Displacement - If the backwash valve is opened too quickly, the
supporting gravel bed can be washed into the overlying filter media. This
gravel displacement can also occur if part of the underdrain system is clogged,
causing unequal distribution of the backwash flow. Eventually, the increased
velocities displace the gravel and create a sand boil. When this boil occurs,
the media starts washing into the underdrain system.
Since some gravel movement always occurs, filters should be probed at least
once a year to locate the gravel bed. This operation can be performed with a
1/4-inch metal rod while the filter is out of service. By probing the bed on a
grid system and keeping track of the depths at which the gravel is located, it
can be determined if serious displacement has occurred. If displacement has
occurred, the media must be removed and the gravel regraded. At this plant,
AFO&M-2C-(11/15/1991>
2031/DM/U
2-85
-------
potential gravel displacement has been minimized by placing coarse gravel on
top of the fine gravel.
Media Loss - At least once a year, the filter media should be examined and
its overall condition evaluated. The depth of media should be measured for an
indication of media loss during the backwashing process.
Loss of some media is unavoidable, especially with air scours. However, if
considerable media is being lost, backwashing procedures should be examined.
Since the bed is usually completely fluidized at 25% expansion, further
expansion may not be needed and avoiding it will reduce media loss. If serious
media loss problems continue, lower backwash rates concurrent with air scour
should be used.
AFO«M-2C-<11/15/1991)
2031/DM/U
2-86
-------
Process alternatives
Direct filtration. Where applicable, direct filtration can produce water
of a quality equal to that obtained through the use of flocculation, sed-
imentation, and filtration. Because all solids removal in direct filtra-
tion takes place in the filter, this process should be employed only for
high-quality water. Recommended raw-water quality criteria for the
direct-filtration process are presented in the section, "Design Crite-
ria." Pilot filter operations are recommended before a decision is made
to use direct filtration.
Cost savings is the principal advantage of direct filtration. Cost sav-
ings of up to 30 percent can be realized as a result of the elimination of
sedimentation (and possibly flocculation) basins and equipment. A 10
to 30 percent reduction in chemical costs is possible, as generally less
coagulant is required to produce a filterable floe than to produce a floe
that will settle. Polymer, acting as a filter aid, is often added ahead of
the filters. Operation and maintenance costs may be reduced, and the
quantity of sludge produced is less than that produced by conventional
treatment. The sludge is relatively dense and more easily dewatered
than conventional filtration plant sludge.9
Disadvantages include shorter filter runs prior to backwash and the
inability of the process to handle large variations in suspended solids
loadings. Generally, more operator attention is required to maintain a
high-quality effluent because the retention time in the plant is sub-
stantially shorter, and wash-water requirements may be higher than
for conventional filtration facilities.
Filter media designs for direct filtration are generally similar to
those for filters preceded by flocculation and sedimentation. However,
sand filters generally cannot provide the pore space that is required
for storage of solids removed from the water; therefore, dual and
mixed-media beds are more frequently used.
Pilot testing and previous experience have indicated that filter
rates of 1 to 8 gpm/ft2 can yield a high-quality effluent.9 A common
design value is 4 to 5 gpm/ft2, which provides operational flexibility
and a margin of safety against variations in raw-water quality. Re-
cent pilot testing indicates that deep-bed uniformly graded anthracite
filters, when preceded by ozonation, can operate reliably at rates of 10
to 15 gpm/ft2. All filter effluent lines should be equipped with a
turbidimeter for continuous monitoring of effluent quality, which will
permit optimization of coagulant and polymer dosages. Additional in-
formation concerning process monitoring is presented in the section,
"Control and Monitoring."
While the filtrate quality produced by direct filtration will not ex-
ceed that of a conventional filtration plant, the lower cost of direct fil-
-------
tration may permit its implementation in locations where filtration is
desirable but is not presently provided.
Automatic backwash control. Automatic control systems are availab
that can interpret a triggering signal indicating high filter head loss
or filter effluent turbidity, remove the filter from service, backwash it,
and return the filter to service. The controls normally are designed to
permit the plant operator to optimize backwash sequencing and the
rate and duration of each sequence.
Continuous backwash. An alternative to the automatic control of
standard filters would be the use of continuously backwashing filter
beds which eliminate the need to remove the beds from service for
washing. The beds are divided into a series of narrow, contiguous
cells, each containing its own underdrain system which allows it to be
backwashed independently from the remaining cells. Backwashing is
accomplished by means of a traveling hood suspended above the bed.
As the hood travels across the bed, each cell is isolated, and a small
backwash pump draws clean water from the filter effluent and re-
verses the flow through that particular cell. The water is removed by
a second backwash pump located in the traveling hood and discharged
to waste. The backwash cycle time is controlled by preset adjustable
timers to permit optimization of the automatic operation feature. Me-
dia depth will vary with each application, but the depth is typicallyi ~
to 36 in.
In addition to the automatic backwashing features, these filters
have the capability of producing relatively constant wash-water flow.
In a properly sized system, this constant flow can eliminate the need
to provide wash-water equalization facilities and permits direct recy-
cle to the plant headworks.
Granular media alternatives
In potable water filtration applications in the United States, the most
commonly used filter media are natural silica sand and crushed an-
thracite coal. Other materials, less widely used, include garnet,
ilmenite, and granular activated carbon. Selection of appropriate me-
dia involves a number of design decisions concerning filtration config-
uration, raw-water quality, pretreatment, and desired filtrate quality.
Backwashing requirements and underdrain system options are depen-
dent upon the media chosen.
The media variables over which the designer has control include
bed composition, bed depth, grain size distribution, and, to a lesser ex-
tent, specific gravity. (The designer has a very limited number of ma-
terials with different specific gravities from which to choose.) In ,
tion to media design characteristics, some control can be exeru—
over media quality through specifications covering, where applicable,
-------
hardness or abrasion resistance, grain shapes, and acid solubility, im-
purities, moisture, adsorptive capacity, manner of shipment, and
other such factors. Suggested criteria and a discussion of the applica-
bility of these parameters can be found in the AWWA Standards
for Filtering Material (BlOO-72)10 and Granular Activated Carbon
(B604-74).11
Traditionally, in the United States, granular media have been de-
scribed in terms of effective size and uniformity coefficient. The effec-
tive size is that dimension exceeded by all but the finest 10 percent (by
weight) of the representative sample; it is also referred to as the "10
percent finer" size. The uniformity coefficient is the ratio of the "60
percent finer" size to the effective size.
Common practice in Europe is to express media sizes as the upper
and lower limits of a range. These limits may be expressed either
as linear dimensions or as passing and retaining sieve sizes (i.e., 1.0
to 2.0 mm or -10 + 18 mesh). Conversions between the standard
U.S. sieve series and dimensions in millimeters are presented in
Table 7.1.
As indicated previously, filter beds may be classified as upgraded,
graded fine-to-coarse, graded coarse-to-fine, or uniformly graded, de-
pending upon the distribution of grain sizes within the bed during fil-
tration. The transition from the upgraded slow sand filter to the fine-
to-coarse rapid sand filter resulted from dissatisfaction with the low
loading rates and laborious cleaning procedure characteristic of the
slow filter. Filters with uniformly graded or coarse-to-fine beds are
now being operated at higher loading rates and longer run times than
are feasible with the conventional rapid sand filter.
Ungraded media. The primary example of an ungraded bed is the slow
sand filter. Because slow sand filters are not backwashed, no hydrau-
lic grading of the media occurs. Distribution of the various grain sizes
in the bed is thus essentially random. Typical slow filter beds contain
table 7.1 U.S. Sieve Series
Sieve designation Size of opening, Sieve designation Size of opening,
number*	mm	number*	mm
200	0.075	20	0.850
140	0.106	18	1.00
100	0.150	16	1.18
80	0.180	14	1.40
70	0.212	12	1.70
60	0.250	10	2.00
50	0.300	8	2.36
30	0.600	4	4.75
'Approximately the number of meshes per inch.
-------
2 to 4 ft of sand with an effective size of 0.2 to 0.35 mm and a unifor-
mity coefficient not exceeding 3.0.
Fine-to-coarse media. Fluidization and expansion of rapid sand fill
beds during backwashing results in accumulation of the fine grains at
the top of the bed and the coarse grains at the bottom. Consequently,
filtration occurs predominantly in the top few inches, and head loss
increases relatively rapidly during operation. Rapid sand beds typi-
cally have effective sizes of 0.35 to 0.60 mm and uniformity coeffi-
cients of 1.3 to 1.8. Smaller effective sizes are seldom practical because
of the shortened run times. Grains passing a number 50 sieve (0.3
mm) or captured on a number 16 sieve (1.18 mm) are normally limited
by specifications to very small portions of the media. Bed depths are
typically 24 to 36 in.
Single-medium anthracite beds have been used in the same basic
configuration as rapid sand beds. Because anthracite is more angular
than sand, the porosity of an anthracite bed is higher than that of a
sand bed containing media with the same effective size. A typical an-
thracite bed has a porosity of 50 to 55 percent. The porosity of a single-
medium sand bed is generally 40 to 45 percent. Consequently, anthra-
cite will not perform in exactly the same manner as sand of equivalent
size. Because of the lower specific gravity, anthracite beds are also
easier to fluidize and expand than sand beds.
Coarse-to-fine media. In a coarse-to-fine bed, both small and large
grains contribute to the filtering process. The presence of fine media
in a filter is desirable because of the relatively large surface area per
unit volume they provide for particle adhesion. Fine media are instru-
mental in achieving high filtrate quality. Coarse media, when placed
before fine media in filtering sequence, decrease the rate of head loss
buildup and increase the available storage capacity in the bed.
Tests by Oeben, Haines, and Ives12 and Craft13 demonstrated that
sand media placed in a downflow reverse-graded (coarse-to-fine) align-
ment exhibit filtering performance superior to that of the same media
in the conventional fme-to-coarse alignment. Better utilization of the
entire bed, manifested as lower head loss and longer run time, was
achieved by the reverse-graded beds without decline in filtrate qual-
ity. Reverse grading of the beds used in these studies was accom-
plished, however, by physically transferring backwashed media to an-
other filter vessel, a method not applicable to full-scale operations.
Attempts in the United States to approximate coarse-to-fine filtration
have been directed almost entirely toward the use of dual- and mixed-
triple-) media beds.
-------
Dual-media beds normally contain silica sand and anthracite coal.
Triple-media beds contain an additional layer of garnet or ilmenite
sand. Beds with three or more media types which intermix after
backwashing have been patented as "mixed-media" filters and are
proprietary technology.14 Specific gravities of materials used in filtra-
tion are roughly as follows: silica sand, 2.55 to 2.65; anthracite coal,
1.5 to 1.75; garnet, 4.0 to 4.3; and ilmenite, 4.5. A typical dual-media
bed contains 6 to 12 in of silica sand (effective size 0.4 to 0.55 mm) and
20 to 27 in of anthracite (effective size 0.8 to 1.1 mm). A typical mixed-
media filter contains 3 to 4 in of garnet (effective size 0.15 to 0.35
mm), 6 to 9 in of silica sand (effective size 0.35 to 0.5 mm), and 18 to
24 in of anthracite (effective size 0.8 to 1.1 mm). Figure 7.5 displays
media grain distribution in a typical mixed-media bed.
TOP
a.
UJ
a
a:
uj
bottom
		 SAND
i
\ /

\
\
1
COAL
4
/
V
s
¦y* -
/ 1
/
/
/
/
/
/
\
\
\
"y^ARNET
\
\
\
>
25
50
75
100
PARTICLE DISTRIBUTION - PERCENT
Figure 7.5 Graphic depiction of media distribution in a typical mixed-media filter.
(Courtesy of MicroFloc Products, CPC Engineering Corp.)
-------
The degree to which media layers are intermixed in the bed depends
upon the sizes and shapes of the media used, the nature of th°
backwashing procedure, the rate of valve closure, and the specil
gravities of the different media. There is disagreement over whethv
distinct layers or intermixed layers are most desirable. If layers mix
completely, the purpose of using more than one medium would be de-
feated. If no mixing occurs, individual fine-to-coarse layers would re-
sult, and the possibility of rapid clogging at interfaces is raised. Pro-
ponents contend that in a properly designed mixed-media filter, a
gradual decline in pore sizes from the top to the bottom of the bed is
established after backwashing. The original argument can be traced
to Conley10,16 and Camp17,18 in the early 1960s. More recently,
Brosman and Malina19 concluded that a slightly mixed bed was supe-
rior to a distinctly layered bed in terms of head loss development, fil-
ter run time, and effluent turbidity. Cleasby and Sejkora,20 however,
disagree that superior performance can be attributed to interfacial in-
termixing in and of itself; rather, it is a result of differences in the
media sizes required to construct mixed and separated beds. They
found that to provide a relatively sharp interface in a dual bed, fairly
coarse sand was required. The resulting bed would not provide the
same filtrate quality as a bed using finer sand which mixed more
readily with the coal. The sizes of sand and coal used by Cleasby
and Sejkora to achieve separate and mixed beds are presented
Table 7.2.
Provision of anthracite and silica sand in sizes commonly used in
dual-media filters inevitably results in some intermixing of layers. In
a triple-media bed, intermixing of silica sand and garnet sand nor-
mally occurs more readily them mixing of silica sand and coal. Cleasby
and Woods21 suggest that, as a rule of thumb, the ratio of the average
particle size of coarse silica grains to the size of coarse garnet grains
should not exceed 1.5 to ensure that some garnet remains at the bot-
tom of the bed. They also suggest that a ratio of coarse coal grain size
to the fine silica sand grain size of about 3 will result in a reasonable
degree of mixing in dual- or mixed-media beds. Brosman and Malina19
TABLE 7.2 Media Sizing for Layered and Intermixed Dual Beds

Layered bed
Intermixed bed
Effective size (D10), sand
0.85 mm
0.46 mm
Uniformity coefficient, sand
1.29
1.49
Effective size (D10), coal
0.91 mm
0.92 mm
Uniformity coefficient, coal
1.45
1.60
Ratio of bottom coal to top sand,
1.93
4.05
{D9
-------
found that "anthracite-sand filter media with a size ratio at the inter-
face of less than 3:1 will exhibit little mixing" and that "the zone of
mixing increases linearly as the size ratio increases above 3:1."
In a few cases in the United States, taste and odor removal and fil-
tration have been combined in a single filter using granular activated
carbon.22,23,24,20 The granular carbon is sometimes added to existing
rapid sand units from which some sand has been removed. Granular
carbon depths of 12 to 36 in over silica sand layers of 6 to 18 in have
been reported. Typically, 0.5 to 0.65 mm effective size carbon has been
used. This technique is applicable only where taste and odor and not
turbidity are of primary concern. The use of activated carbon rather
than anthracite just to provide a coarse upper layer is not economical.
Also, if turbidity levels are high, activated carbon pores will become
plugged fairly rapidly, and carbon life will be greatly reduced. If both
turbidity and taste and odor are significant problems, activated car-
bon beds should be preceded by conventional granular filtration. If
carbon adsorption is desired for removal of organics, the depth of car-
bon which can be provided in a converted gravity filter is likely to be
too shallow to provide adequate contact time.
In Europe, upflow filters are frequently employed to achieve coarse-
to-fme filtration. Hamann and McKinney26 reported that the upflow
filters commonly used in the Soviet Union are relatively deep (6.5 to
8.5 ft of sand) and that recommended media sizes range from 0.5 to 2.0
mm. The primary difficulty associated with upflow filters is break-
through resulting from bed lifting as head loss increases. Grids are
frequently installed above the sand to discourage lifting. An upflow
filter tested in the United States contained a 6-ft sand bed.27 As used
in a demonstration program, the bed consisted of 3 ft of 0.95-mm ef-
fective size sand with a uniformity coefficient of 1.26 above 3 ft of 1.8-
mm effective size sand with a uniformity coefficient of 1.11. Available
literature contains no evidence of the use of upflow filtration for pota-
ble water production in a permanent installation in the United States.
Upflow filters have received greater acceptance in nonpotable than
potable water applications in the United States.
An alternative solution is the biflow filter, in which water is intro-
duced at the top and the bottom of the bed simultaneously. Russian
biflow filters typically contain 5 to 5.5 ft of sand. Filtrate is withdrawn
at an intermediate point 1.5 to 2 ft below the top surface. During fil-
tration, the head on the upper bed aids in preventing expansion of the
lower bed. The loadings on the upper and lower portions of the filter
are not equivalent. The lower bed is a coarse-to-fme bed, although the
finest grains in the upflow filter are as coarse as the coarsest grains in
the downfiow bed. Variations observed in filtrate quality between the
upper and lower portions of the filter are attributable to differences in
-------
grain sizes and media depth. Potable water treatment plants using
biflow or upflow sand filters are apparently common in Europe am
the Soviet Union."5
Uniform media. The uniformly graded deep-bed Filters used in Europe
utilize relatively coarse media, ranging from 0.5 mm to as much as 6.0
mm. (These extreme sizes would not be found in the same filter.) Me-
dia uniformity coefficients are typically 1.2 to 1.3, but values as high
as 1.5 may be found. Greater media depth is substituted for the lack of
fine media in the bed. Such a substitution requires more vigilant op-
eration and increased chemical usage to avoid breakthrough. Depths
of 4 to 6 ft are common, and in some cases media depths reach 8 ft.
Filters of this type are not expanded during backwash, and stratifica-
tion of grain sizes does not occur.
Design criteria. A granular filter designed for potable water produc-
tion should produce a high-quality filtrate. The National Primary
Drinking Water Standards set by the U.S. EPA in 1977 require
monthly average turbidities to be less than 1.0 TU for surface sup-
plies, and impending regulations may result in further reductions in
allowable turbidity levels.
Additional interrelated criteria, including loading rate, run t1'*^?.
head loss, and applied water quality, are less predetermined
creased loading rates accelerate head loss and decrease run time~. ^
some point, increased throughput will also result in declining filtrate
quality. A low loading rate, however, will not guarantee a high fil-
trate quality. If coagulation and flocculation are inadequate, a high-
quality filtrate cannot be achieved at any loading rate. With proper
media selection and pretreatment, low effluent turbidities can be
readily achieved at rates much higher than the once-standard 2
gpm/ft2. Loading rates, type of media, media depth, and backwash de-
sign are largely determined by the designer's experience with similar
applications. In some cases, pilot-plant testing is needed prior to de-
sign. Specialized media are available for some applications, such as
iron and manganese removal (see Chap. 11).
The amount of water required to backwash a filter using a given
washing technique is relatively constant regardless of run time. Con-
sequently, long runs are desirable. Runs are ended when either the
terminal head loss is reached or filtrate quality declines beyond a pre-
determined level. The maximum head available to operate a gravity
filter is dependent upon the depth of the filter box and the hydraulic
configuration of the plant. Filter design should preclude the apr
tion of sufficient head to produce shear forces which will strip
well-coagulated floe from the bed during filtration. For maximum uti-
-------
lization of bed storage capacity, terminal head loss should ideally oc-
cur immediately before filter breakthrough. The quality of the water
to be filtered naturally affects the rates at which head loss develops,
and thus, the length of runs between washes. The desired filtration
rate is that which, for a given applied water quality, results in the
maximum net production of water meeting filtrate quality require-
ments.
Filtration rate
Slow sand filters, designed for loadings of 3 to 6 million gallons per
acre per day (0.05 to 0.10 gpm/ft2), were replaced by rapid sand filters
loaded at 1 to 2 gpm/ft2. The 2-gpm/ft2 rate became widely accepted as
an upper limit in U.S. water supply practice. Subsequent investiga-
tions have demonstrated that dual-media and mixed-media, as well as
single-medium, filters can be successfully operated at higher rates.
The overall performance of multi-media units is, however, superior to
that of single-medium filters. Table 7.3 presents operating experience
with plant-scale rapid sand and multi-media filters. Laughlin and
Duvall2S conclude that mixed-media beds can be successfully operated
at an average rate of 5 gpm/ft2 and a maximum rate of 8 gpm/ft2. In
side-by-side comparison with single-medium filters, mixed-media beds
achieved longer runs with the same head loss, required less backwash
water per unit of water filtered, and produced higher filtrate quality.
Tuepker and Buescher29 concluded that a dual-media bed, while oper-
ating at a higher hydraulic loading (3.5 vs. 2.0 gpm/ft2), could yield
longer runs and lower percent usage for backwash than a sand filter,
without reducing the quality of the product water. Westerhoff32 found
the performance of a mixed-media bed operating at 5 to 6 gpm/ft2 su-
perior to that of a sand bed at 2 gpm/ft2 in terms of run length, percent
usage for backwash, and filtrate quality. Additional sources, cited in
Table 7.3, found dual- and mixed-media filters to operate successfully
at rates from 3 to over 6 gpm/ft2 in a variety of locations. Similar re-
sults regarding the feasibility of high-rate filtration have been re-
ported as a result of pilot studies by Conley16,35 Robeck et al.,36 Dostal
and Robeck,37 Kirchman and Jones,38 and Rimer.39 The quantity of
evidence of the practicality of high-rate filtration is such that in 1972
the AWWA Committee on Filtration Problems concluded, "It has been
amply demonstrated that filters can be designed and operated to pro-
duce water of acceptable quality at flows substantially higher than
the rate of 2 gpm/ft2 once considered the maximum."40
Average filtration rates of roughly 2 to 7 gpm/ft2 are reported for
the upflow, biflow, and deep-bed filters discussed previously, but data
are limited.26,41 Pilot testing conducted in the southwestern United
-------
TABLE 7.3 Plant-Scale Performance Results





Aver-
Aver-
Aver-
Aver-






Aver-
age
age
age
age
Termi-





age
raw
applied
filtered
run
nal
Per-
Refer-

Water

rate,
turbid-
turbid-
turbid-
length,
head
cent
ence
Media
source
Pretreatment*
gpm/ft2
ity, TU
ity, TU
ity, TU
h
loss, ft
backwash
28
Single
River
C(A), F, S
2
nr
4.9
0.28
45
8
2.11

Mixed
River
CCA), F, S
2
nr
4.9
0.24
63
8
1.62

Single
River
C(A), F, S
4
nr
6.3
0.35
19
8
3.15

Mixed
River
C(A), F, S
4
nr
6.3
0.29
24
8
2.11

Single
River
C(A), F, S
6
nr
4.5
0.45
10
6.8
2.93

Mixed
River
C(A), F, S
6
nr
4.5
0.3
15
6.8
2.07
29
Single
River
LS, C(FE, P),
2
nr
2
0.1
42
3.0
2.1



CL, S








Dual
River
LS, C(FE, P),
3.5
nr
2
0.1
60
2.2
1.6



CL, S







30
Dual
River
C, F,S
6.87
25-300
nr
0.06
51
nr
0.58
31
Mixed
nr
nr
3-6
nr
0.5-10
0.1-0.5
15-80
nr
1-3

(16 plants)









32
Single
Lake
AE, C(A), F,
g
2
5
1
0.06
40
8
2.5

Mixed
Lake
AE, C(A), F,
<3
5-6
5
1
0.05
23
8
1.8
33
Mixed
River
u
C (A, P)
5
2-15
nr
0.3
8-12
nr
nr
34
Dual
Reservoir
CL, C (A, P)
3.3
0.5
1-3
0.07
80
nr
nr
nr = not reported.
•C (A, FE, P) = Coagulation (alum, ferrous sulfate, polymer); F = flocculation; S = sedimentation; LS = lime softening; AE = aeration;
CL = chlorination.
-------
States has shown that deep-bed uniformly graded anthracite filters,
when preceded by ozonation, can operate reliably at rates of 10 to 15
gpm/ft2.
Applied water quality
The nature as well as the quantity of suspended material in the ap-
plied water is critical to the performance of the filter. Unflocculated
water can be very difficult to filter regardless of the type of media in
use.7,42 However, the work of Robeck et al. with dual-media filters
showed that if the applied water is properly coagulated, filtration at 4
or 6 gpm/ft2 will produce essentially the same filtrate quality as fil-
tration at 2 gpm/ft2. Subsequent investigations have shown similar re-
sults for mixed-media filters.08,31,32
The use of chemicals in conjunction with filtration is limited prima-
rily to metal salts used as coagulants and to polymers. Coagulants are
ideally fed into mechanical mixing basins preceding flocculation.
Whether sedimentation is also required depends on the quantity of
suspended solids in the influent water. Coagulants are intended to
produce agglomerations of natural and chemical solids. Polymers
added with coagulants aid in the strengthening and growth of these
agglomerations during flocculation. Cationic polymers are sometimes
used as primary coagulants, eliminating the need for two chemical
feed systems. Anionic and nonionic polymers are used as coagulant or
filter aids. Filter-aid polymers are used to increase the strength of ad-
hesion between media grains and floe in coarse-to-fine filters.
The use of a filter-aid polymer can result in improved floe capture,
better filtrate quality, and longer filter runs and higher head loss
prior to turbidity breakthrough. Filter-aid polymers are not generally
used with fine-to-coarse filters because they promote rapid surface
clogging. Filter aids are often fed in dilute liquid form to allow disper-
sion without mechanical agitation just prior to filtration. Filter-aid
polymer doses to gravity filters are usually low (0.03 to 0.05 mg/L).
Doses required for pressure filters may be higher if a higher operating
head loss is employed. Because the viscosity of water increases with
decreasing temperature, breakthrough as a result of floe shearing is
more likely at lower water temperatures. Consequently, increased
polymer doses may be required in cold weather. A longer contact time
prior to filtration may also be necessary in cold weather.
Assuming that adequate coagulation is feasible, the designer must
decide whether sedimentation is desirable. In the past, settling has
been provided prior to rapid sand filtration if turbidities exceeded
roughly 10 TU.43 The increased storage capacities of dual- and mixed-
media filters have made filtration of water with higher turbidities
-------
practicable. The primary advantage of providing direct filtration ij^
the elimination of the capital and operating costs associated with S|
imentation. The higher solids load on the filter will, ho%vever, short
run times and increase the portion of the product water required for
backwashing. Although the point at which advantages outweigh dis-
advantages will vary with local conditions, a number of investigators
have suggested conditions which would justify consideration of direct
filtration. Culp9 lists the following as alternative conditions under
which direct filtration is likely to be feasible: turbidity and color both
less than 25 units; low color and maximum turbidity less than 200
units; and low turbidity and maximum color less than 100 units. He
adds an additional general qualification that diatom levels in excess of
500 to 1000 areal standard units (asu)/mL can make direct filtration
impracticable and that lesser levels will affect media selection. Based
on experience in the Great Lakes region, Hutchison44 states as condi-
tions for direct filtration that diatoms be less than 1000 asu/mL and
that the alum, dose required for coagulation be less than 15 mg/L.
Conley35 indicates that sedimentation is probably required if raw-
water turbidities often exceed 100 units. Robeck et al.36 concluded that
direct filtration using dual-media beds is probably feasible if turbidity
is less than 25 units. The variations among these recommendations
reflect the fact that they are intended only as guidelines. Differen'
in local conditions at each site under consideration require that pi
studies be conducted to determine the feasibility of using direct filtra-
tion.
Head loss
Terminal head loss through rapid sand, dual-media, or mixed-media
gravity beds is typically 4 to 10 ft. Because of coarse top grains, how-
ever, the rate at which head loss increases in a well-designed multi-
media bed will be much less than in a sand bed under the same load-
ing conditions. Terminal head losses in pressure filters in potable
water applications are usually in the same range as in gravity beds,
but may be as high as 25 ft. Outlet pressures are sometimes deter-
mined by the needs of the distribution system. Rated tank pressures
can be as high as 150 psi (roughly 350 ft), but 50 psi is more common.
If the head loss at any level in the filter bed exceeds the static head,
a vacuum can result. This situation is referred to as negative head and
can result in air binding of the filter. When the pressure in the filter
bed drops below atmospheric levels, dissolved gases will be released,
from the water being filtered. Gas bubbles trapped in the bed will f
ther increase the head loss and aggravate the problem. They may
-------
result in displacement of media during backwashing. This problem is
particularly acute when filtering with insufficient water depth over
the media or when surface waters are saturated with atmospheric
gases because of rising temperatures in the spring. Remedies for air
binding in gravity filters include increased frequency of backwashing,
maintenance of adequate static head above the media surface, or a
clearwell water level above the surface to keep the filter media sub-
merged. Pressure filters normally discharge well above atmospheric
pressure and are not subject to air binding.
Filter run length
Rapid sand filters are generally operated with run lengths between 12
and 72 h, with 24-h runs being typical. Plant-scale tests of dual- and
mixed-media filters listed in Table 7.3 reveal run lengths of 8 to 80 h.
The 17 direct-filtration plants surveyed by Culp9 had similar average
run lengths. Pressure filters may have somewhat longer run lengths
than gravity filters if they can be operated at higher head losses with-
out breakthrough.
Long run lengths are desirable in that they result in a reduction of
the portion of total water production used for backwashing. Run
lengths may, however, be influenced by the need to maintain reason-
able operating shifts. Control of biological growth in the filter may
also be a factor in determining run times in some locations if
prechlorination is not provided.
Design Details
In addition to selection of media and loading rate, a number of factors
contribute to the success or failure of filter design. The filter box (or
tank) must be constructed to permit effective and reliable filter per-
formance over the ranges of loading rate and head loss desired. Fea-
tures which must be considered include piping and valving, and place-
ment of underdrains and media within the filter.
Influent
Raw or flocculated water is usually delivered to a gravity filter
through the wash-water gullet. Influent to a pressure filter is gener-
ally distributed by a tapped pipe serving as a manifold, or by a baffle
plate. Influent conduits should be designed to deliver water to the fil-
ters with as little disturbance as possible. Free fall or turbulence
which can disturb the media surface is undesirable. Delivery of influ-
ent beneath the water surface in the filter or baffling of the incoming
-------
stream may be employed to prevent bed disturbance. Necessary mea-
sures will depend upon the control strategy used.
Piping and valves
Typical piping serving a gravity filter is shown in Fig. 7.6. Influent
piping is often sized to limit velocities to about 2 ft/s. This may result
in the use of an influent flume rather than a pipe in large plants. Hy-
draulic considerations generally result in velocities of 3 to 6 ft/s in
wash-water and effluent piping. At higher velocities, head losses often
become excessive, and undesirable effects such as water hammer are
more likely to occur. Flanged and cement-lined cast iron, ductile iron,
or steel pipe is commonly used for filter piping. Where flexibility or
removability for maintenance is required, harnessed mechanical
joints or couplings are cften used.
A typical filter is equipped with five valves: influent, effluent, wash-
water supply, wash-water drain, and surface wash or air wash supply.
A filter-to-waste valve may also be included; however, waste lines con-
Flgure 7.6 Typical filter piping. (Source: Black & Veatch, Engineers-Architects./
-------
stitute a potential cross-connection and must be equipped with air-gap
protection against backflow from the drain to the filter. The filter ef-
fluent line configuration should provide positive pressure on the
downstream side of the effluent rate controllers. Rubber-seated but-
terfly valves are most common in filter valving.
Placement of piping and valves should permit easy access for main-
tenance. Valves should be equipped with easily distinguishable posi-
tion indicators.
Wash-water troughs
In U.S. practice, wash-water troughs are suspended at even spacings
above gravity filter beds to provide uniform removal of wash water
during backwashing. This limits the horizontal travel required and
equalizes the static head on the underdrainage system. In contrast,
European designs often feature narrow beds with wash-water conduits
on one or both sides, but not suspended over the media. In the
European designs, tilting side weirs, horizontal water jets, and a pro-
cedure allowing influent water to enter the filter on the side opposite
the wash-water trough at the end of the wash cycle are sometimes
used to aid the movement of scoured solids to waste. Spacing of
troughs in U.S. practice is usually at 5- to 7-ft centers to limit hori-
zontal travel disturbances to 2.5 to 3.5 ft. Media loss may result if
troughs are placed too close to the surface of the unexpanded filter
bed. The design elevation of the weir edge of the troughs may be de-
termined by adding the depth required for maximum bed expansion
(usually 50 percent) and the overall depth of the trough, plus a small
margin of safety. If air scour is installed, additional care must be
taken in locating the troughs. Simultaneous use of backwash water at
6.2 gpm/ft2 and air at 4 scfm/ft2 can toss 1- to 2-mm sand grains 15 to
16 in. If anthracite is placed above sand, the danger of media loss is
increased because of the lower density of the coal.
The flow in a rectangular trough with free discharge can be deter-
mined by an equation of the form
Q = Cbhza
where Q = rate of flow, cfs
C = a constant
b = trough width, ft
h = maximum water depth in the trough, ft
For horizontal, rectangular channels of such length that friction losses
are negligible, the theoretical value of C is 2.49. Values of C as low as
1.72 are used in practice. Capacities of nonrectangular channels may
be approximated by using the dimensions of a rectangular section of
-------
equivalent area in the formula. A more rigorous result can be ob-
tained by referring to derivations provided by Fair, Geyer,
Okun46 and Brater and King.47
Troughs are usually made of reinforced concrete or fiberglass-
reinforced plastic (FRP). Concrete troughs usually have V-shaped bot-
toms and FRP troughs semicircular bottoms. Typical cross sections
are shown in Fig. 7.7.
After troughs are installed in a filter, the weir edges should be
smoothed and leveled to uniformly match a still water surface at the
desired overflow elevation.
Media placement
Careful placement of media is critical to subsequent filter operation.
AWWA publication B-lOO-7210 describes procedures for placing,
washing, scraping, and disinfecting media prior to filter operation.
If the underdrains and media are not uniformly laid, uneven flow
patterns which may affect filtrate quality will develop. Uneven solids
deposition in the bed and unequal distribution of backwash water may
be aggravated with each successive filter cycle until serious disrup-
tion of the bed occurs.
Washing and scraping of the filter media after placement are
quired to remove excessively fine material. In order to match de!
media elevations after fines are removed, excess material must be
a. REINFORCED CONCRETE	b. FIBERGLASS REINFORCED
PLASTIC
Figure 7.7 Typical wash trough cross sections: (a) reinforced concrete
(b) fiberglass-reinforced plastic. (Source: Black & Veatch, Engineers-Architects.)
-------
placed in the bed initially. If too much material is removed, however,
the resulting media will have a higher effective size than desired.
Underdrain systems
Underdrain systems are provided to support and retain filtering me-
dia, distribute backwash water, and collect filtrate. Uniform distribu-
tion of wash water and collection of filtrate are essential to proper fil-
ter operation. The three general types of underdrain assemblages are
pipe laterals and gravel, blocks with or without gravel, and false bot-
tom with or without gravel.
Traditionally, the great difficulty in underdrain design has been to
provide a barrier to the finest media which will not clog during filtra-
tion or backwash. Early attempts to use fine screens or strainer., were
largely unsuccessful, leading to the use of gravel layers below filter
sand. The position of gravel layers may, however, be disrupted during
backwashing. Jet action, which is discussed in greater detail
elsewhere,7 causes sand and gravel mixtures to be more easily dis-
rupted than gravel alone. If auxiliary air wash is used, even greater
gravel disturbance may occur. Fine gravel, which is usually placed at
the sand-gravel interface, is most easily dislocated. A possible solution
to this problem is the use of gravel in a coarse-to-fine-to-coarse grada-
tion, which has been shown to be very stable at high water backwash
rates.7 Fine media penetrate the upper coarse gravel layer without ap-
parent ill effect.
Mixed-media beds, which have very fine garnet at the bottom of the
bed, are generally constructed with a layer of coarse garnet on top of
the silica support gravel. The coarse garnet prevents leakage of the
fine garnet and also helps to stabilize the underlying silica gravel.
European-type deep-bed filters utilize relatively coarse and uni-
formly graded media. As a result, bed stratification is not required
and air scour presents less of a hazard to proper bed operation. Also,
the use of strainers is more likely to be feasible because of the larger
permissible openings. Consequently, false-bottom underdrains with
nozzles designed for both air and water distribution and without sup-
port gravel are commonly used in deep-bed filters.
Pipe laterals and gravel. Pipe-lateral underdrains were once popular
because of their relatively low cost and their adaptability to use in
pressure filters. Problems with relatively high head loss and poor
wash-water distribution have resulted in a general decline in their
use. They are still encountered, however, when older filters are up-
graded.
-------
Pipe underdrain systems generally consist of a centrally located
manifold pipe to which smaller, equally spaced laterals are attaclj
The lateral pipes usually have one or two rows of W to 3/t-in-diamf
perforations on their bottom sides. They may be fitted with nozz.
Guidelines for lateral design include the following ratios:
Orifices are normally spaced at 3 to 12 in and laterals at roughly the
same spacings as the orifices. Roughly 18 in of gravel is required to
cover a lateral network. Usually three to five graded layers are in-
volved, with sizej varying from IV2 to Ya in. The bottom layer should
extend 4 in above the highest wash-water outlet.
Blocks with gravel. A commonly used block underdrain consists of vit-
rified clay blocks which are grouted in place. The size and arrange-
ment of these blocks and typical support gravel layers are shown in
Fig. 7.8. In mixed-media applications, the third gravel layer is re-
placed by garnet of similar size.
The underdrain system described above is intended for use wit
auxiliary air-scour backwash. Air scour is usually limited to fL_
with underdrains which do not require gravel. Recently, a polyethyl-
ene underdrain block which is designed for use with overlying gravel
and an air/water wash has been introduced (see Fig. 7.9). Dispersion
of the air through a relatively large number of closely spaced orifices
is intended to reduce the possibility of upsetting the gravel layers.
False bottom with gravel. One of the most widely used false-bottom
underdrains is constructed of precast or cast-in-place reinforced con-
crete supported on concrete sills. Each system contains uniformly
spaced inverted pyramidal depressions. Unglazed porcelain balls are
placed in the depressions to distribute flow. Each depression is filled
and leveled with 1- to lVb-in gravel before placement of overlying
gravel layers. A typical arrangement including the gravel layers is
shown in Fig. 7.10. The last silica gravel layer should be replaced by
coarse garnet in a mixed-media bed.
False bottom without gravel. False-bottom underdrains which do not
require gravel have impervious bottoms penetrated by strainer^
porous bottoms. Fine openings eliminate the need for support gi
Filter-box depth is thus reduced.
Total area of orifices (surface area of bed)
Cross-sectional area of lateral (total area of orifices served)
Cross-sectional area of manifold (total area of laterals served)
0.0015 to 0.005:1
2 to 4:1
1.5 to 3:1
-------
A variety of false-bottom and strainer underdrains are available.
Three typical units are shown in Fig. 7.11. (All are shown equipped
with plunge pipes for air wash.) The false bottoms are usually con-
crete, polyethylene, or tile blocks; monolithic concrete; or a steel plate.
Strainers are usually constructed of stainless steel, plastic, or brass.
Plastic strainer orifices are sometimes smaller on the filter side to pre-
vent clogging during filtration, even though such a configuration can
contribute to clogging during backwash. Strainer-type underdrains
are used primarily in filters employing air/water wash systems; fail-
ures resulting from plugging and breakage have been experienced,
however. Acceptance of these systems is less widespread than accep-
tance of the two systems described in the previous sections.
Porous-bottom underdrains. Porous-bottom underdrains with a porous
aluminum oxide plate have been used in both block and false-bottom
-------
Figure 7.9 Plastic block underdrain designed for use beneath gravel lay-
ers and with air/water wash. (Source: F. B. Leopold Co.)
-------
Figure 7.11 Typical strainers used in false-bottom
underdrains without gravel. (Source: (a) Walker
Process Corp; (b) Eimco Process Equipment Co.)
configurations. They are constructed of plates mounted on concrete or
steel piers or on clay tile saddles to form blocks. The latter type is
shown in Fig. 7.12. Very small pore sizes make porous-bottom
underdrains susceptible to plugging and therefore unsuitable for use
in softening or iron-and-manganese removal plants or other plants
where plugging by chemical deposition may occur. They may also clog
with rust or debris during backwashing. Additional problems which
may occur include breakage because of the brittle nature of the porous
material and failure of caulked joints between plates. Porous bottoms
have been used successfully in a few locations, but are less widely ac-
cepted than the block or false-bottom and gravel underdrains dis-
cussed previously.
Backwashing
As the amount of solids retained in a filter increases, bed porosity de-
creases. At the same time, head loss through the bed and shear on cap-
tured floe increase. Before the head loss builds to an unacceptable
level or filter breakthrough begins, backwashing is required to clean
the bed.
Failure to clean the filter adequately can lead to a multitude of
problems. Initially mudballs will form and accumulate in the bed,
causing clogging. Clogged areas contract as head loss increases.
Shrinkage opens cracks in the filter surface and sometimes at the fil-
ter walls. Cracks can cause short circuiting of the bed during filtra-
tion, with subsequent decline in filtrate quality. Clogged areas also
contribute to channeling of backwash water, which can lead to bed up-
-------
X
V
t




"""/--ALUNDUM
- J?
V POROUS PLATE
J	h-
\ 	,
I REE —
FACTORY
BONDED-
k'/4
CLAY SADDLE—^
'l" » sV. H7/a "t 177a" or 237/8"
Figure 7.12 Porous filter bottom (ALUNDUM®). (Source: Norton Company, Industrial
Ceramics Diuision.)
set. The mechanisms whereby washing problems lead to filter failures
are discussed in greater detail elsewhere.7
The selection of a washing technique is closely tied to media and
underdrain design. In current potable water filtration practice,
backwashing invariably includes upflow water flushing. The rate and
duration of water flushing are variable, however, and may be supple-
mented with air scour or surface water wash. The operational se-
quencing of dual washing systems and the source of the wash water
introduce additional variations.
Water source
Common backwash water source options include the following
bled from high-service discharge and used directly for washing
-------
fill an above-ground wash-water tank prior to gravity washing; grav-
ity flow from above-ground finished-water storage; gravity flow from a
separate above-ground wash-water tank; or direct pumping from a
sump or below-ground clearwell.
Bleeding flow from high-service discharge results in energy loss as a
result of the pressure reduction required prior to washing. If direct
washing is employed, a pressure-reducing valve or orifice is placed in
the wash-water supply line. If bleeding is used to fill a wash-water
tank, an altitude valve is used to control the water level in the tank.
In either case, the wash-water supply line is often sized to restrict the
maximum amount of water which can be delivered. Both options avoid
provision of separate wash-water pumps. Direct washing also avoids
construction of a wash-water tank, but presents greater difficulty in
controlling wash-water flow. Because of the large pressure drop often
involved in supplying wash water by high-service bleeding, the poten-
tial for cavitation in or following head-dissipating devices in the sup-
ply line is significant.
If an above-ground clearwell is not available to provide head for fil-
ter washing, wash water may be pumped to a separate tank or directly
to the filters. Use of a wash-water tank permits pumping at a lower
rate. Tank storage must be sufficient to permit washing of the filters
at the maximum backwash rate while they operate at minimum run
times.
A number of proprietary filters are available which obtain
backwash water by means other than those listed previously. These
include several filters which utilize vertical steel tanks divided into
upper and lower compartments. Sufficient filtering head is provided so
that following downflow filtration in the lower compartment, filtered
water flows through a pipe into the upper tank. When terminal head
loss in the filter bed is reached, wash water flows from the upper tank
back through the filter.
Some filter-control systems permit gravityTflow backwashing of a
filter utilizing the effluent from the filters remaining in service. Un-
usually deep filter boxes are required to provide filtering Head in such
systems. Backwashing head is obtained by placing an effluent weir
several feet above the level of the wash troughs.
Washing method
A thorough review of backwashing methodology has been published
by the AWWA Subcommittee on Backwashing of Granular Filters.45
Three basic methods are available: upflow water wash without auxil-
iary scour, upflow water wash with surface wash, and upflow water
wash with air scour. The application will normally dictate the method
-------
to be used. Filter-bed expansion during upflow water washing resif
in media stratification. Air washing results in bed mixing. If strat
cation is desired, air scour must be avoided or must precede fluidiz.
tion and expansion with water.
In current U.S. practice, upflow water wash with surface wash is
commonly used wherever mudball formation may occur. Manual
cleaning can sometimes be substituted for surface wash in small
plants. Upflow water wash alone may be sufficient in some filters
which receive low solids loadings. Use of auxiliary air scour is becom-
ing more common in U.S. potable water plants.
Upflow water wash without auxiliary scour. In the absence of auxiliary
scour, washing in an expanded bed occurs as a result of the drag forces
on the suspended grains. Grain collisions do not contribute signifi-
cantly to washing.45,48,49
Maximum shear on the grains theoretically occurs (for typical filter
sand) at a bed expansion of 80 to 100 percent.45 The increase in shear
with increasing bed porosity, however, is relatively slight beyond the
point at which expansion begins. Optimal expansion may be less than
20 percent.50 Normally when water wash is applied exclusively, an ex-
pansion of 20 to 50 percent is used. Water wash at a sufficient rate to.
substantially expand (10 percent or more) a granular bed is often
ferred to as high-rate water wash. Wash rates incapable of fully ^
idizing a bed (less than 10 percent expansion) may be referred to as
low-rate. Experience in the United States with high-rate water wash
used alone is extensive. It is generally successful for applications in
which iron precipitates have been filtered from groundwater or color
has been removed from otherwise high-quality surface water. The rel-
atively weak cleaning action of water wash without auxiliary scour of
some type, however, generally renders it unsuitable for filters remov-
ing large quantities of suspended solids.
High-rate water wash tends to stratify granular media. In dual- and
mixed-media beds, this action is essential and beneficial, but it is not
required for uniformly graded single-medium beds. In rapid sand fil-
ters, it results in movement of the fine grains to the top of the bed,
which has a negative effect on head loss and run length.
Upflow water wash with surface wash. Surface-wash systems have been
widely applied to supplement high-rate upflow washing where"
mudball formation is likely to be a problem. Either a fixed-nozzle or
rotary wash system may be used. Fixed systems distribute auxilr
wash water from equally spaced nozzles in a pipe grid. Most
plants utilize rotary systems in which pipe arms swivel on centra
-------
bearings. Nozzles are placed on opposite sides of the pipes on either
side of the bearing, and the force of the jets provides rotation.
Rotary systems are generally preferred because they provide better
cleaning action, usually lower water requirements, and less obstruc-
tion for filter access. Possible problems with rotating surface wash
units include failure to rotate, failure to clean in corners, abrasion of
concrete walls near the point of closest passage of the arm, and locally
high velocities caused when passing under wash troughs. Either type
of system may fail to provide auxiliary scour where it is most needed.
This can be especially true in layered multi-media beds if substantial
removals are occurring at media interfaces.
Surface-wash systems are typically suspended about 2 in above the
surface of the unexpanded filter bed. Systems have been placed in the
unexpanded bed, however, and dual-arm rotary systems which have
one arm above and one arm below the unexpanded surface are avail-
able. Plugging of nozzles with media has been a problem with the sub-
merged units. Rotary systems may have either straight (Fig. 7.13) or
curved pipe arms. Nozzle diameters of Vs to V4 in are common. Single-
arm units typically operate at 50 to 100 psi and discharge from 20 to
over 200 gpm depending on length. Standard units are available up to
approximately 20 ft in diameter. Some models induct air into the
wash-water jets.
Advantages of auxiliary water wash include proven effectiveness in
alleviating dirty filter problems, improved cleaning (in comparison
with water wash alone) without a great change in system complexity,
and possibly lessened danger of gravel upset if the quantity of wash
water introduced through the underdrain is reduced.
Because surface-wash systems constitute a possible connection be-
tween filtered and unfiltered water, backflow prevention devices must
be provided in supply lines.
Upflow water wash with air scour. Approaches to the use of auxiliary
air scour in backwashing filters are numerous. Air scour has been
used alone and with low-rate water backwash in an unexpanded bed
or slightly expanded bed. Each procedure is utilized prior to either
low- or high-rate water wash.
Air scour provides very effective cleaning action, especially if used
simultaneously with water wash. Cleaning is attributable to high in-
terstitial velocities and abrasion between grains. On the other hand,
air wash presents substantial potential for media loss and gravel dis-
ruption if not properly controlled.
If more than one filtering medium is used and stratification of the
bed is desired, high-rate water wash must follow air scour. In a single-
-------
-e:
CENTER NOZZLE
LATERAL, REO 9RA33



END CAP
AGITATOR NOZZLE
SURFACE AGITATOR HEAOER-
STABILIZER ANGLE
h

WASH TROUGH


< 1

-SURFACE AGITATOR

-SURFACE AGITATOR HEADER
1
!, O \ i

)

^ JL f	SURFACE t
XGITA
r°R J
t



Figure 7.13 Typical surface agitator and arrangements. (Source: F. B. Leopold Co.)
medium bed, if a low-rate wash can adequately remove scoured solids,
high-rate wash can be avoided.
If air scour occurs simultaneously with water wash, the air flow
must usually be stopped prior to wash-water overflow to prevent ex-
cessive media loss. Thus, the permissible duration of air washing will
be short unless the concurrent water-wash rate is low or the bed is
very deep.
Alternatives to the use of a strainer-type underdrain which may
still permit the use of air wash include use of a coarse-to-fine-to/^T,ce
gravel gradation or the provision of a separate air distribution
-------
located above the gravel. The latter approach has been implemented
at Contra Costa, Calif.,30 but it is intended primarily to aid in move-
ment of solids to the wash troughs rather than for scour of the media
grains.
Experience in Europe indicates that air scour essentially eliminates
mudball formation. Difficulties have arisen, however, from failure to
remove scoured solids from filter surfaces. Probable contributing fac-
tors include low water-washing rates, long horizontal-travel distances
to wash-water troughs, and a necessary lag between termination of air
scour and initiation of higher-rate water wash.
Backwash rates
In the United States, wash rates are expressed as volumetric flow per
unit surface area (gpm/ft2) or as the equivalent water rise velocity
(h/s, ft/min, or in/min). Required water-washing rates are variable
and depend on water temperature, filter type, and washing method.
Water viscosity decreases with increasing temperature. Consequently,
as wash-water temperature rises, drag forces on media grains are re-
duced and higher wash rates are required to achieve bed expansion.
Each degree Celsius increase in water temperature requires roughly a
2 percent increase in wash rate to prevent a reduction in bed expan-
sion. Backwash systems should be designed for the warmest wash-
water temperature which will be encountered.
Media selection also affects washing rate. Rate requirements in-
crease with increasing grain size and media density. Angular grains
are more easily expanded than round grains. In beds using more than
one type of medium, sizes of different media must be chosen carefully
to ensure proper positioning after backwash. Recommended size ratios
for dual and mixed beds were discussed in the section, "Granular Me-
dia Alternatives." Figure 7.14, taken from Cleasby and Baumann,51
displays the effect of media size on the wash rate required to achieve
10 percent bed expansion for three common media. Figure 7.15, from
the same source, shows the effect of water temperature on the wash
rate for silica sand and coal and on the viscosity of water.
Characteristic washing rates and durations vary for each of the
washing methods discussed previously. The suitability of a washing
method is related to influent water quality, filtering media and bed
configuration, and underdrain design. Consequently, not all washing
methods are applicable in all cases, and different methods may or may
not yield similar results in a particular case.
Upflow water wash without auxiliary scour. When water wash is used
alone, a high-rate wash is employed. Generally a rise rate of 15 to 23
-------
MEAN SIEVE SIZE, mm
Figure 7.14 Minimum fluidization velocity (Vm/) needed to achieve 10 percent bed
expansion at 25°C (from Cleasby & Baumann61). {Source: USEPA Technology
Transfer.)
0.04
0.03
0.02
0.01
_
m Viscosity	*-
-
-
	 Cool 	*

-
		Sand
-
-
1 1
1
1.0
0.5
10
20
TEMPERATURE
30
Figure 7.15 Effect of water temperature on of sand and coal, and on absolute vis-
cosity of water (from Cleasby & Baumann61). (Source: USEPA Technology Transfer.)
-------
gpm/ft2 is applied. After the water level in the bed has been lowered to
near trough level, backwash duration may be 3 to 15 min.
Upflow water wash with surface wash. Combined surface and water
wash usually involves three phases. After the water surface is lowered
in the bed, surface wash is activated and operated alone for 1 to 3 min.
Low-rate water wash is then applied simultaneously for an additional
period of roughly 5 to 10 min. Termination of surface wash precedes a
final phase (1 to 5 min) during which a higher wash-water rate is used
to expand the bed 20 to 50 percent. This usually requires a backwash
use rate of 15 to 23 gpm/ft2. Wash-water flow during surface agitation
is usually limited to that required to expand the bed only slightly.
However, if anthracite makes up the t°p filtering layer, bed expansion
above the surface-wash system may be desirable to reduce the likeli-
hood of media loss. Rotary surface-wash systems typically add 0.5 to
2.0 gpm/ft2 to the wash-water flow. Fixed-nozzle systems deliver 2 to 4
gpm/ft2.
Upflow water wash with air scour. Three variations of air and water
wash were discussed in the previous section. The first, air scour alone
followed by low-rate water wash, is commonly applied in Great
Britain to single-medium sand filters with 0.6- to 1.2-mm media. After
the water in the filter is lowered to below the wash-water overflow, air
is injected at 1 to 2 scfm/ft2 for 3 to 5 min. Water wash of 5 to 7.5
gpm/ft2 follows. Bed expansion and stratification are not achieved, al-
though relatively cool water temperatures may result in fluidization
of the upper sand layers. Problems with gravel disruption have not
been experienced if air and water are applied separately and at levels
not exceeding 1.5 scfm/ft2 and 7.5 gpm/ft2, respectively.45
Air scour alone followed by high-rate water wash can be applied to
rapid sand or dual-media filters, since bed stratification occurs during
water wash. This method has been used in the United States with air
scour at 3 to 5 scfm/ft2 followed by water wash at 15 to 23 gpm/ft2.
Simultaneous air scour and water wash is generally limited to the
deep, coarse-grained filters common in Europe. For 1- to 2-mm media,
air-scour rates of 2 to 4 scfm/ft2 are used with a water flow of 6.3
gpm/ft2. For 2- to 6-mm media, 6 to 8 scfm/ft2 and 6.3 to 7.5 gpm/ft2
are used. Simultaneous air and water wash typically lasts 5 to 10 min
and is followed by water wash alone for another 5 to 10 min. The rate
of final water wash is generally one to two times that used with air
scour. In the U.S. installation mentioned previously30 in which air is
distributed above an expanded bed from a fixed piping network, 2
scfm/ft2 of air accompanies 15 gpm/ft2 of wash water. As noted, how-
-------
ever, the air in this case serves to improve transport of solids to
troughs rather than to increase scour in the bed.
Wash-water disposal/recovery
Following filter washing, disposal or recovery of the backwash water
is required. In the past, wash water has been discharged to a surface-
water body (which may have been the raw-water source) or to a drain-
age ditch. Disposal in this manner was both simple and relatively in-
expensive. In areas where water conservation was important, wash-
water recovery was used to reduce the amount of water withdrawn
from the water source. Recent regulations governing discharges of wa-
ter treatment wastes have made the use of wash-water recovery more
widespread.
In conventional plants, wash water can be returned to the head of
the treatment facilities. Because wash water is generated in slugs,
however, equalization is usually required. Treatment of the wash wa-
ter prior to return is employed in some conventional plants and is a
design consideration. In direct-filtration plants, separate wash-water
treatment must be provided. Additional coagulation and flocculation
of wash water may not be required, but standby polymer feed fa^i
may be included. If land area is available, lagoons can be used t<
vide both equalization and sedimentation.
Wang et al.52 conducted pilot-scale tests of wash-water recovery at a
plant using alum, activated carbon, and chlorine in conjunction with
sedimentation and filtration of water from a reservoir. Advantages of
recycling included improved settling of basin sludge, lower filter head
losses, elimination of discharge to a stream, and recovery of much of
the wash water. A small increase in alum use was required. Wash-
water recovery may not, however, be feasible at some locations with
significant taste and odor problems because of the possibility of inten-
sification of the problem.
Control and monitoring
Control and monitoring of the filtration process are critical to success-
ful operation. Decisions covering control and monitoring methods and
facilities must be made during initial design because each affects the
physical layout of the filtering facilities.
Filter control may be predicated either on head loss through
ter bed or on the rate of filtration. In either case, smooth tra
during changes in filtration rate is highly desirable. The deleic ,uf
effects of sudden flow surges on filtrate quality have been ampl}
-------
documented.29,53 Control methods are generally tested most severely
by removal of filters for backwashing and their return to service.
A variety of control methods are in use in the United States. Control
methods are often distinguished in the literature as constant-rate,
constant-level, influent flow splitting, declining-rate, and so on. In ac-
tual practice, however, control systems may incorporate the character-
istics implied by more than one of those terms.
Constant-rate filtration can be achieved by placing a rate controller
on the filter effluent line. The rate controller consists of a metering
device (flow tube), a variable rate-setting control mechanism, and a
throttling valve. In the absence of additional controls, the valve re-
sponds to the pressure differential in the flow tube, opening to in-
crease flow or closing to restrict flow. Filtration remains at a set rate
regardless of the condition of the bed until the number of filters in ser-
vice is changed. When head loss in the bed reaches the point at which
the desired flow cannot be achieved with the valve wide open, back-
washing is required. The sum of the rate settings on the filters in ser-
vice must match the influent flow. If the controls are well designed,
the depth of water in each filter box will be fairly stable during most
of the run because the rate controller will compensate for the addi-
tional head loss as solids accumulate in the bed. The water level in the
filter box, however, is not actually regulated (except for limits on ex-
treme levels) and may fluctuate.
Constant-level filtration, in its simplest form, can be achieved by ty-
ing operation of the filter effluent valve to a level-sensing device in
the filter box. The level in the box varies within a relatively narrow
control range. At the top of the range the valve is fully opened, and at
the bottom the valve is closed. Such a control scheme serves only to
regulate the water level in the filter and prevent dewatering. The rate
of filtration is determined by the rate at which water is delivered to
the filter.
A method of filter control which has been successfully applied in a
large number of plants utilizes both level and rate control. In this
scheme, level sensing occurs at the influent header or flume rather
than in each filter box. Effluent valves are controlled individually to
maintain the same rate through and level in each filter despite possi-
ble differences in the conditions of the beds. The controls also limit the
maximum rate through each filter. As long as the total influent flow
remains constant, a constant filtration rate is achieved.
The total influent flow may also be equally divided using weirs
rather than rate controllers. Filters of this type normally discharge
over an effluent weir, eliminating the possibility of bed dewatering.
The relatively high discharge elevation requires an unusually deep
-------
filter box to provide filtering head. From 5 to 6 ft of additional depth is
typical.54 The filtration rate is determined by the plant influent /
The level in each filter rises as necessary to accept an equal porti
the influent and indicates the head loss. When the level rises a
fixed upper limit, backwashing is initiated.
Filters equipped with effluent weirs rather than rate controllers
may also operate without weirs or other devices which equally split
the influent. These filters are usually termed declining-rate filters.
Flow is distributed on the basis of the relative conditions of the beds.
Assuming that influent piping losses are roughly the same for all the
filters, a uniform operating water level in all the filters will be
achieved. The filtration rate then will be highest in the cleanest bed
and lowest in the dirtiest bed. In each bed the filtration rate will de-
crease as solids accumulate. An orifice or other flow-limiting device is
used on each filter to control the maximum flow. To determine which
bed is in greatest need of backwashing, some type of effluent rate in-
dication must be provided. Advantages claimed for declining-rate fil-
ters include higher water production for a given run length7 and im-
proved filtrate quality.65
Filters may also be operated without rate controllers by pumping
directly from the underdrain. If each filter is connected to a single
pump, approximately constant-rate filtration is achieved (althoui"
head loss builds, pumping may decrease slightly). If more than on
ter is connected to each pump, the cleanest bed will pass the most
water.
Valve operating systems may be hydraulic, pneumatic, or electric.
Hydraulic systems were developed first, but these are generally no
longer installed in new plants because of problems with leakage and
with plugging of orifices in the lines by deposition from the fluid. Cur-
rently either pneumatic or electrical systems are used. Pneumatic sys-
tems are generally less expensive, but they require oil- and moisture-
free air. Electrical systems offer greater reliability, but usually at
higher cost. In the event that maintenance is required, however, the
technical skill required to service electrical controls may be much
greater than that required for pneumatic controls.
Monitoring
Filter design should include instrumentation to monitor filtrate tur-
bidity, filtration rate, head loss, and backwash rate. If auxiliary air
scour backwash is installed, air-flow monitoring should also b£_in-
cluded. Pilot filters may be used as an aid in determining coa t
dosage.
-------
Turbidity. Continuous monitoring of overall plant effluent turbidity
should be practiced, with alarms tied to high levels. A recording-type
turbidimeter is valuable because variances from normal operation are
displayed graphically. Observations of the effects of fluctuations in
raw-water quality, rate changes, equipment malfunctions, chemical
feed variations, filter backwashing, and other such occurrences con-
tribute to the operator's understanding of the plant and increase his or
her ability to deal with such situations. Continuously reading
turbidimeters are available from a number of manufacturers. Piping
to such units should be designed to preclude the presence of air bub-
bles, which can distort readings. Sample taps should be provided at
the effluent of individual filters to permit periodic turbidity monitor-
ing, although continuously reading units may not be required. A sin-
gle turbidimeter is sometimes connected by manifold to all the filters
to facilitate periodic monitoring. Turbidimeters are provided for each
filter if backwashing is to be automatically activated by high turbidity
in the filtrate.
Filtration rate. Flow tubes are generally used for monitoring flow
through individual filters. Design consideration should be given to the
need for straight lengths of pipe preceding the meters. Recorders or
totalizers are sometimes included. If flow splitting is employed, the fil-
tration rate may be determined from monitoring of the total plant
flow.
Head loss. Head loss in a filter bed is a valuable indicator of filter
condition and may be used to automatically activate backwashing.
The head loss through the bed is normally monitored by differential
pressure-cell devices.
Backwash rate. Because backwash flow requirements may vary with
the seasons and with other conditions, operator knowledge of the rate
in use at a particular time is essential. Flow tubes are usually em-
ployed as the monitoring device. Recording meters are generally not
necessary. A totalizing device is desirable to determine the overall
volume of water used in backwashing. An alarm can be provided that
is triggered if the wash rate exceeds a predetermined maximum.
Pilot filters. Pilot filters are bench-scale models of plant filters that
are used to determine coagulant dosage. Coagulated water is diverted
to pilot filters from the full-scale treatment units. Monitoring of the
pilot-filter effluent turbidity provides an indication of the adequacy of
the coagulant feed. The effect of the use of pilot filters is to greatly
-------
reduce the time lag in coagulant feed system adjustment and thus im-
prove plant qperation. Parallel pilot filters are provided to ensur^
tinuous control. Because of the nature of connecting piping to t
filters, higher filter-aid polymer doses than would be used in the
scale plant are usually required. The effect of increasing the polymer
dose is to shorten run times. Consequently, pilot filters are generally
not used to predict run lengths or polymer dosage. In most cases, how-
ever, these variations do not affect determination of the optimum co-
agulant dosage.
Operating facilities
Several types of filtration operation and control have been employed
in practice. These include local and remote manual, and semi- and
fully automatic operation.
Most filter plants currently in operation in the United States have
control consoles on the filter floor from which the filters are operated.
Because observation of each filter during backwashing provides a
check against malfunction and contributes to the operator's under-
standing of the filters, remote automatic operation of backwash
sequences is not usually practiced. The filter control consoles arp "°.u-
ally located immediately adjacent to the filters they serve. Ne\
less, remote manual operation, with the remote filter controls lucked
on the plant control console, is used in some plants. This allows a sin-
gle operator to wash filters and still observe the plant processes indi-
cated on the control console.
In the past, all major valves were controlled by individual manual
controls, and all filter operations were operator-directed. However, ad-
vances in sensing and control equipment have made the use of remote
automatic or semiautomatic control more popular. In semiautomatic
operation, backwashing is initiated by the operator but consists of a
predetermined sequence which requires no additional attention. Fully
automatic filters are backwashed without operator input on the basis
of loss of head in the bed, filtrate turbidity, or a fixed maximum run
time. Automatic systems permit operation of all filters from a central
location, thus reducing personnel requirements. However, remote au-
tomatic operation may not permit the operator to observe the
backwash cycle directly. Unless they are properly designed, auto
mated systems may be difficult to alter when changing conditions re
quire modification of backwash rate or sequencing. Labor-saving sys
tems should not be installed unless operational flexibili nc
effective filter performance can be assured.
-------
Energy considerations
Energy consumption in filtration is attributable primarily to the
pumping required to provide filtering head and backwashing. If air
scour is provided, the compressors will also contribute to the power de-
mand.
Available head for filtering is fixed by design and, for most filters
being installed today, falls within a relatively narrow range. The head
loss attributable to clean media is unavoidable. Depending on the con-
trol method used, however, the head loss resulting from piping and ap-
purtenances can be reduced. Tuepker and Buescher29 investigated fil-
ter control without rate controllers at St. Louis and concluded that
energy savings can be realized. Selection of a control strategy should
not, however, be based solely upon energy considerations.
Energy savings can also be achieved by increasing run times. Max-
imization of water production between washes will result in a reduc-
tion in the energy required per net unit volume of water produced. A
design which permits utilization of the entire filter bed is thus desir-
able. The type of bed chosen may also affect the amount of energy re-
quired for each backwash. Deep, uniform beds do not require full ex-
pansion for stratification, as do multimedia beds, but are more
difficult to fluidize and are usually equipped with air scour.
If power charges are assessed on the basis of peak usage, energy
costs can be reduced through judicious selection of the backwash wa-
ter source. Backwashing by gravity flow from an elevated tank rather
than by pumping from storage substitutes transfer pumping (at a
lower rate) for backwash pumping and thus reduces peak demand. If
pumped-flow washing is used, peak energy demands can be reduced by
providing a larger number of smaller filters. Another alternative is
tho use of "split" filters which provide for backwashing of. one-half the
filter media at a time.
Conservation of water within the plant will result in lower raw-
water pumping requirements and may result in a significant net en-
ergy savings. Wash-water recovery may possibly result in savings of
this type. Direct filtration may result in energy savings if the effects
of higher loadings and shortened run times are less significant than
the energy reductions resulting from elimination of clarification and
sludge-handling facilities.
Design trends
Dual- and mixed-media filters are likely to remain the prevalent types
of filters being constructed for potable water production in the United
States for some years to come. As a result of recent legislation and the
-------
capabilities of these filters, the proportion of direct-filtration plants
being constructed will probably increase. Wash-water recovery will
receive increased attention as energy costs rise and discharge limita-
tions become more stringent. Automation and computer applications
in filtration plants will also increase.
An additional concern for the future is that there will be more cases
in which filtration requirements will result from or be aggravated by
environmental pollution. An example of this is the discovery of asbes-
tos fibers, a suspected carcinogen, in Lake Superior water used for
public supply. Pilot studies conducted in Duluth, Minn, in 1974 indi-
cated that removal of amphibole asbestiform fiber by filtration is
feasible.56 For the water studied, direct mixed-media filtration was
determined to be the most economical treatment. Alum was found to
be more effective than ferric chloride as a coagulant, and nonionic
polymer proved most effective as a filter aid. Successful filtration was
achieved at loadings of 4 to 6 gpm/ft2. Based on the findings of the pi-
lot study, a 30-mgd direct-filtration plant has been built at Duluth.57
One dual-media and three mixed-media filters, designed for 5 gpm/ft2,
are included in the plant. Wash water is recycled following chemical
addition, flocculation, and clarification. Sludge containing asbestiform
fibers is subjected to freeze-thaw treatment in lagoons. Decant from
the lagoons is returned to the plant headworks with reclaimed wash
water.
Water quality concerns may also increase interest in particle count-
ing as an index of quality in lieu of or in addition to turbidity mea-
surements. Turbidimeters are widely used in water quality control be-
cause they are convenient to use and give instantaneous results.
Turbidimeters indicate only light-scattering properties, which are not
directly related to the quantity of suspended material present in a wa-
ter sample. The poor correlation between turbidity and total micro-
scopic count has been amply documented.58,59
Automatic particle sizing and counting equipment offers the conve-
nience of turbidimeters and also provides an indication of the quantity
(size and number) of particles present. The particle counter thus gives
a more accurate measure of water quality than the turbidimeter. Au-
tomatic counters cannot, however, be calibrated over an unlimited
particle size range and are also unable to distinguish the type of par-
ticle being counted and sized.
The use of deep-bed uniformly graded filters is expected to increase
because of their ability to operate at hydraulic loading rates 2 to 3
times that of conventional dual- and mixed-media filters. One diredj
filtration facility recently placed in service in the southwest Unitec
States utilizes deep-bed (6-ft depth) anthracite filters operating at
rates of up to 13.5 gpm/ft2. A primary factor in the successful opera-
-------
tion of these filters is the addition of ozone prior to flocculation and
filtration. Ozonation "preconditions" raw-water particulate matter,
thereby enhancing the flocculation process and increasing allowable
filter run lengths. Increased attention to variations in raw-water qual-
ity and maintenance of appropriate coagulant and filter-aid feed rates
is required for successful operation at the higher loading rates.
Other Filtration Techniques
This chapter has been devoted primarily to granular-media filters be-
cause they are the prevalent means of filtration in the potable water
industry both in the United States and elsewhere in the world. Other
types of filtering or straining technology are, however, encountered in
water treatment practice. These include microscreens, diatomaceous
earth filters, and cartridge filters.
Microscreens
Microscreens are used in water treatment to remove algae or other
plankton which may occur in lake and reservoir supplies. In some
cases, microscreens are used as pretreatment for granular or
diatomaceous earth filtration, but they may also serve as the sole fil-
tration step. Prescreening may be applicable where plankton levels
are such that clogging of granular filters is a concern or run lengths
are impaired. In some locations, the nature of the plankton present
may be such that microscreens are more effective than granular filters
in removing them. The use of microscreens without further filtration
may be feasible if the raw water is essentially free of color and colloi-
dal turbidity but contains distinct floating or suspended organisms.
Such applications are not common.
Microscreens consist of a rotating drum placed in a rectangular
tank (see Fig. 7.16). The drum supports a straining fabric, which may
be either stainless steel wire cloth or woven polyester cloth. Polyester
media are preferable if the water being screened has been chlorinated.
Influent enters at one end of the drum and flows outward into the
tank. Spray nozzles and a wash-water trough at the top of the drum
are provided for washing the media. Commercially available units
have drums 4 to 12 ft in diameter and 1 to 16 ft in length. Nominal
cloth openings of 17, 21, 23, 35, and 60 ^.m are used in potable water
practice. Coarser mesh sizes are available for industrial and wastewa-
ter applications.
Normally, two-thirds to three-fourths of the drum surface area is
submerged during operation. Loadings are expressed in terms of flow
per submerged unit surface area. In the absence of prior experience,
-------
the permissible loading at a given location is usually determined by
pilot testing. Rates of 5 to 25 gpm/ft2 of submerged area are typical in
water supply applications.
Diatomaceous earth filters
Diatomaceous earth, a natural silicious fossil material, has beer? 1
as a filtering medium in potable water plants. Filters utilizing uiuoj-
mite consist of a pressure vessel containing a number of porous tubu-
lar structures on which the diatomite is supported (see Fig. 7.17). An
initial layer of roughly Vs in of diatomite is applied before water pro-
duction begins. Because the diatomite layer will develop cracks as
pressure drop increases, continuous feed of additional diatomite is re-
quired throughout the run. Hydraulic loading rates of 0.5 to 2.0
gpm/ft2 of media are typical of current operating practice.
Diatomaceous earth filters were initially used during World War II
because they could be adapted to portable water plants. Subsequent
attempts to utilize this technology in full-scale plants were unsuccess-
ful in a number of cases because of operational deficiencies which
have been described elsewhere.60 Although the operation of diatomite
filters is now more adequately understood ahd guidelines for proper
operation are better defined,61 serious drawbacks remain. These in-
clude rigorous operating requirements, high operating costs, in-
creased sludge production, and a lack of proven experience in dealing
with raw water which is not of relatively high quality. The primary
advantage of diatomaceous earth filtration is a low initial capita
Consequently, it is best suited either to very small plants for l
the comparable cost of a granular media plant is relatively high, oi to
-------
Figure 7.17 Typical cylindrical element. (Source: Manville Sales
Corp.)
plants which Eire used only intermittently, such as emergency units
and those for swimming pools. In most locations, diatomite filtration
is not economically or technically competitive with granular filtration
in full-scale potable water supply applications.
Cartridge filters
Cartridge filters are employed prior to membrane desalting processes
(reverse osmosis or electrodialysis) to provide protection against foul-
ing. They are available in a range of nominally rated pore sizes and
may require either disposal or cleaning when a terminal head loss is
reached. The filters most commonly installed before membrane pro-
cesses are 5 to 25 ^m; they are generally provided by equipment sup-
pliers as a part of the total desalting system. Media configurations in-
clude wound fiber, bonded fiber, sintered metal, and woven metal
mesh. Available wound-fiber cartridges include polypropylene, cotton,
acrylic, or modacrylic fibers on polypropylene, stainless steel, or
-------
tinned-steel cores. Bonded fibers may be cellulose, fiberglass, wool,
acrylic, or viscose rayon with a phenolic or melamine resin binder.
Sintered metal and woven metal-mesh cartridges are of stainless steel
construction. Wound-fiber cartridges are normally used in potable-
water plants.
Multiple cartridges are usually arranged in pressure vessels (see
Fig. 7.18). Duplicate pressure vessels allow replacement of cartridges
without interrupting water production. Capacities and operating pres-
sures vary widely. Individual cartridges are typically 2 to 3 ii 5
and 4 to 30 in long.
Use of these filters in potable water supplies in the United Stat, *s
common only in areas where brackish water supplies are used, prima-
rily Florida and the Southwest.
-------
American Water Works Association
ANSI/AWWA B100-96
(Revision or ANSI/AWWA B100-89)
AWWA STANDARD
FOR
FILTERING MATERIAL
[AMERICAN NATIONAL)
IB STANDAfiOBBP
Effective date: Dec. 1, 1996.
First edition approved by AWWA Board of Directors Nov. 15, 1948.
TTiis edition approved Feb. 4, 1996.
Approved by American National Standards Institute Sept. 6, 1996.
AMERICAN WATER WORKS ASSOCIATION
6666 West Quincy Avenue, Denver, Colorado 80235
-------
AWWA Standard
This document is an American Water Works Association (AWWA"' standard. It is not a specification.
AWWA standards describe minimum requirements and do not contain all of the engineering and
administrative information normally contained m specifications. The AWWA standards usually contain
options that must be evaluated by the user ot the standard Until pach nnrinn?1 fo-irrina
by the user, the product or service is not fully defined. AWWA publication of a standard does not
constitute endorsement of any product or product type, nor does AWWA test, certify, or approve any
product The use of AWWA standards is entirely voluntary. AWWA standards are intended to represent a
consensus of the water supply industry that the product described will provide satisfactory service.
When AWWA revises or withdraws this standard, an official notice of action will be placed on the
first page of the classified advertising section of Journal AWWIA. The action becomes effective on the
first day of the month following the month of Journal AWWA publication of the official notice.
American National Standard
An American National Standard implies a consensus of those substantially concerned with its scope
and provisions. An American National Standard is intended as a guide to aid the manufacturer, the
consumer, and the general public. The existence of an American National Standard does not in any
respect preclude anyone, whether that person has approved the standard or not, from manufacturing,
marketing, purchasing, or using products, processes, or procedures not conforming to the standard.
American National Standards are subject to periodic review, and users are cautioned to obtain the
latest editions. Producers of goods made in conformity with an American National Standard are
encouraged to state on their own responsibility in advertising and promotional materials or on tags
or labels that the goods are produced in conformity with particular American National Standards.
Caution Notice The American National Standards Institute (ANSI) approval date on the front
cover of this standard indicates completion of the ANSI approval process. This American National
Standard may be revised or withdrawn at any time. ANSI procedures require that action be taken
to reaffirm, revise, or withdraw this standard no later than five years from the date of publication.
Purchasers of American National Standards may receive current information on all standards by
calling or writing the American National Standards Institute, 11 W. 42nd St., New York, NY 10036;
(212) 642-4900.
Copyright © 1996 by American Water Works Association
Printed in USA
ii
-------
Committee Personnel
The following AWWA subcommittees, which developed this standard, had the
following personnel,at the time'
Py-r\Mam c •
John W. Guptill, Chair
R.H. Eggleston	L.K. Noble
T.L. Gloriod	F.W. Pogge
Subcommittee on Granular Activated Carbon:
Stephen L. Bishop, Chair
R.P. Beverly	R.H. Moser
E.A. Bryant	F.W. Pogge
S.L. Butterworth	K.J. Roberts
P.H. Kreft
Subcommittee on High Density Media:
W. Kirk Corliss Jr., Chair
M.S. Chopra	R.H. Moser
L.E. Gorrill	G.F. Stolarik
Subcommittee on Quality Control to Delivery:
Thomas P. Walter, Chair
R.P. Beverly	C.E. Stringer
E.F. Morey
Subcommittee on Physical Testing:
Larry K. Noble, Chair
R.P. Beverly	J.B. Hambley
S.L. Bishop	P.H. Kreft
W.S. Caton	I.M. Markwood
J.E. Durrant	David Verona
David Gittleman	T.P. Walter
L.E. Gorrill
in
-------
Subcommittee on Anthracite:
Richard H Moser, Cha
ir
W.K. Corliss
R.L. Roberts
The AWWA Standards Committee on Filtering Materials that reviewed and
approved this standard had the following personnel at the time of approval:
Richard H. Moser, Chair
R. Lee Roberts, Vice-Chair
W. Kirk Corliss, Secretary
Consumer Members
J.E. Durrant, Philadelphia Water Department, Philadelphia, Pa.	(AWWA)
T.L. Gloriod, St. Louis County Water, St. Louis, Mo.	(AWWA)
J.R. McQueen, Connecticut Water Company, Clinton, Conn.	(NEWWA)
R.H. Moser, American Waterworks Service Company, Voorhees, N.J.	(AWWA)
F.W.	Pogge, Water & Pollution Control Department, Kansas City, Mo.	(AWWA)
G.F.	Stolarik, Los -Angeles Department of Water & Power,
Los Angeles, Calif.	(AWWA)
C.E. Stringer, Dallas Water Utilities, Dallas, Texas	(AWWA)
General Interest Members
S.L. Bishop, Metcalf & Eddy Inc., Wakefield, Mass.	(NEWWA)
E.A. Bryant, Consultant, New York, N.Y.	(AWWA)
J.L. Cleasby, Iowa State University, Ames, Iowa	(AWWA)
W.K Corliss Jr., Gannett Fleming Inc., Harrisburg, Pa.	(AWWA)
K.R. Fox, USEPA-RREL, Cincinnati, Ohio	(USEPA)
G.L. Hoffman,* Council Liaison, Finkbeiner, Pettis & Strout, Akron, Ohio (AWWA)
P.H. Kreft, Montgomery Watson, Portland, Ore.	(AWWA)
I.M. Markwood, Aivord, Burdick & Howson, Springfield, 111.	(AWWA)
K.J. Roberts, W20 Inc., Mississauga, Ont.	(AWWA)
J.H. Wilber," Standards Engineer Liaison, AWWA, Denver, Colo.	(AWWA)
Producer Members
M.S. Chopra, Minerals Research & Recovery, Tucson, Ariz.
J.W. Guptill, EW2 Environmental Inc., Charlotte, N.C.
L.K. Noble, Northern Gravel Company, Muscatine, Iowa
R.L. Roberts, Roberts Filter Company, Darby, Pa.
T.P. Walter, Carbon Sales Inc., Wilkes-Barre, Pa.
(AWWA)
(AWWA)
(AWWA)
(AWWA)
(AWWA)
"Liaison, nonvoting
iv
-------
Contents
All .4W'1V!4 standards follow the general format indicated subsequently Some ccr.cf.onj from this format may be
found in a particular standard.
SEC.	PAGE	SEC.	PAGE
Foreword
I	Introduction	 vi
IA	Background	 vi
I.B	History	 vi
I.C	Acceptance	 vi
II	Special Issues	 vii
IIA	Source of Supply	 vii
II.B	Filter Media	 vii
II.C	Filter Gravel	 ix
II.D	Acid Solubility	 x
II.E	Anthracite Quality Tests	 x
II.F	Bulk Shipment	 xi
II.G	Media Records	 xi
III	Use of This Standard	 xi
III.A	Purchaser Options
and Alternatives	 xi
III.B	Modification to Standard	 xii
IV	Major Revisions	 xii
V	Comments	 xii
Standard
1	General
1.1	Scope	 1
1.2	Purpose	 1
1.3	Application	 1
2	References	 1
3	Definitions	 2
4	Requirements
4.1	Physical Requirements	 3
4.2	Chemical Requirements	 5
4.3	Impurities	 5
4.4	Placing Filter Materials	 5
4.5	Preparing Filter for Service	 6
5	Verification
5.1	Approval Samples	 8
5.2	Sampling	 8
5.3	Test Procedures—General	 9
6	Delivery
6.1	Marking		13
6.2	Packaging and Shipping		13
6.3	Affidavit of Compliance		14
Appendixes
A Bibliography	 15
B Calibration of Sieves
B.l Precision of Sieves	 19
B.2 Glass Spheres	 19
Figure
1 Specific Gravity Test Apparatus	 11
Tables
F.l Gravel Layers for Two Sizes
of Fine Media and Two Sizes of
Underdrain Orifices	 x
1	Physical Characteristics
of Filter Media	 3
2	Maximum Backwash Rates	 7
3	Minimum Size of Composite
Sample	 8
4	Sampling of Bagged Media	 9
5	Minimum Sample and Acid
Quantities for Acid-Solubility
Tests	 10
6	Minimum Sample Size for
Sieve Analyses	 12
B.l Nominal Dimensions, Permissible
Variations for Wire Cloth
of Standard Test Sieves	 20
V
-------
Foreword
This foreword is for irformc'ion only and is not c pert o/~.4UrU'".A BI00
I. Introduction
I.A. Background The purpose of ANSI/AWWA B100 is to provide purchasers
with a standard for the purchase and installation of filtration materials.
A wealth of information on innovations in filter design is available from various
sources, including Journa/AWWA and Water Treatment Plant Design.' These sources
include design parameters for filters using single and multiple media. As a result,
ANSI/AWWA B100 makes reference to filter design only as the design relates to the
filtering materials used (see Appendix A). ANSI/AWWA B604 Standard for Granular
Activated Carbon should be consulted when using granular activated carbon (GAC)
as a filter medium, because GAC is not specifically covered in B100. NOTE:
ANSI/AVvWA Standard B604-96 covers GAC as a filter media.
I.B. History. The AWWA Standard for Filtering Material was approved as tenta-
tive by the AWWA Board of Directors on Nov. 15, 1948, and as standard on Jaa 16,
1950. Revisions were approved on June 2, 1953, Jaa 31, 1972, June 20, 1980, and
Jan. 29, 1989. The original standard was approved and promulgated in the course of
activities of the Water Purification Division and under jurisdiction of the Committee
on Water Works Practice. This edition was approved by the AWWA Board of Direc-
tors on Feb. 4, 1996.
I.C. Acceptance. In May 1985, the US Environmental Protection Agency
(USEPA) entered into a cooperative agreement with a consortium led by NSF Interna-
tional (NSF) to develop voluntary third-party consensus standards and a certification
program for all direct and indirect drinking water additives. Other members of the
original consortium included the American Water Works Association Research Foun-
dation (AWWARF) and the Conference of State Health and Environmental Managers
(COSHEM). The American Water Works Association (AWWA) and the Association of
State Drinking Water Administrators (ASDWA) joined later.
In the United States, authority to regulate products for use in, or in contact
with, drinking water rests with individual states.7 Local agencies may choose to
impose requirements more stringent than those required by the state. To evaluate
the health effects of products and drinking water additives from such products, state
and local agencies may use various references, including
1.	An advisory program formerly administered by USEPA, Office of Drinking
Water, discontinued on Apr. 7, 1990.
2.	Specific policies of the state or local agency.
3.	Two standards developed under the direction of NSF, AN5I:/NSF§ 60,
Drinking Water Treatment Chemicals—Health Effects, and ANSI/NSF 61, Drinking
Water System Components—Health Effects.
'Water Treatment Plant Design, AWWA, ASCE, and CSSE, Denver, Colo. (1989).
Tersons in Canada, Mexico, and non-North American countries should contact the
appropriate authority having jurisdiction.
-American National Standards Institute. 11 \\ 42nd St., New York, NY 10036.
§NSF International, 3475 Plymouth Rd., Ann Arbor, MI 48106.
vi
-------
4 Other references, including AWWA standards. Food Chemicals Codex,
Water Chemicals Codex, and other standards considered appropriate by the state or
local agency.
Various certification organizations may be involved in certifying products in
accordance with AN'Sl/N'SF 61. Individual states or local agencies have authority to
accept or accredit certification organizations within their jurisdiction. Accreditation
Vl	uiiLai.un	luay • k j nuui jui	i,y juhju«v.».uu.
Appendix A, 'Toxicology Review and Evaluation Procedures.'' to ANSI/NSF 61
does not stipulate a maximum allowable level iMAL) of a contaminant for sub-
stances not regulated by a USEPA final maximum contaminant level (MCL). The
MALs of an unspecified list of ''unregulated contaminants" are based on toxicity
testing guidelines (noncarcinogens) and risk characterization methodology (carcino-
gens). Use of Appendix A procedures may not always be identical, depending on the
certifier.
AWWA B100 does not address additives requirements. Thus, users of this
standard should consult the appropriate state or local agency having jurisdiction in
order to
1.	Determine additives requirements including applicable standards.
2.	Determine the status of certifications by all parties offering to certify products
for contact with, or treatment of, drinking water.
3.	Determine current information on product certification.
II. Special Issues
II.A. Source of Supply. Filtering materials, such as silica sand, high-density
sand, granular activated carbon, or anthracite, as well as support gravel, should be
obtained from sources that are expressly qualified to produce and furnish these materi-
als for water treatment plants.
II.B. Filter Media. Filter media is the portion of the filter bed that removes
particulate matter from the water during the filtration process. This standard covers
anthracite, silica sand, and high-density sand. Properties of granular activated carbon
when used as a filter medium will be covered in a pending revision to ANSL'AWWA
B604. Standard for Granular Activated Carbon. Properties of media used in precoat
filters (such as diatomaceous earth) can be found in ANSI/AWWA Standard B101,
Standard for Precoat Filter Media.
Sand or anthracite filter media used in a wide range of bed depths and particle
sizes have produced satisfactory results. Selection of the bed depth or particle size to
be used in any particular filter is the responsibility of the designer and should be
executed with careful consideration of raw water conditions and plant pretreatment
facilities.
In general, for a given pretreatment of raw water at a given filtration rate,
coarse media will permit longer filter runs between washings than allowed by fine
media. With good pretreatment facilities and close technical control, coarse media
will yield water of satisfactory quality. With all other conditions fixed, removal of
particulate matter is a function of both media size and filter bed depth, and removal
generally improves with greater filter depth or with smaller media size, or both
Dual- or multiple-media filters have been used instead of single-medium filters
in many water treatment applications. The dual or multiple media are selected to
"Both publications available from National Academy of Sciences, 2102 Constitution Ave.
N W., Washington. DC 204 IS.
vri
-------
maintain coarse media in the upper portion of the bed ar.d Fine media in the lower
portion of the bed The coarse-to-fine grading tend? to combine longer filter runs,
characteristic of coarse media, with superior Filtration. characteristic of fine media,
for improved overall performance Proper selections of particle size range and specific
gravity for the different layers of media are necessary to maintain the coarse-to-fine
gradation during filtration and after repeated backwashing
GAC is suitable for use as a filter medium either alone or as a dual media with
sand. Long-term experience indicates that GAC performs effectively in a dual role as a
filter medium and as an adsorber for control of taste and odors. A planned revision
of ANSI/AWWA B604 will provide information on the use of GAC as a filter medium
including its properties, sampling, testing, shipping, placement, and preparation for
service.
Where anthracite is used in dual- or multiple-media fdters, the size of the
anthracite depends on the size and specific gravity of the sand or other material
used beneath the anthracite. If the anthracite grains are too small, excessive losses
will be incurred during the minimum backwash required to clean the sand effec-
tively. If the anthracite grains are too large, excessive mixing of the two materials
will occur at the interface.
High-specific-gravity (high-density) filter media consisting of garnet, ilmenite,
hematite, magnetite, or associated minerals of those ores are used by some utilities
in an attempt to remove more suspended solids at higher filtration rates. This small,
high-density media remains as a layer under the silica sand as a result of particle
size and specific gravity differences in the same way that silica sand remains separated
from overlaid coal in a dual-media filter.
Garnet refers to several different minerals (mostly almandite and andradite)
that are silicates of iron, aluminum, and calcium mixtures. However, garnet could
also be grossularite, spessartite, and uvarovite, the latter being a chromium mineral.
Ilmenite is an iron titanium mineral, which invariably is associated with hematite
and magnetite, both iron oxides.
Particle size distribution. There are two methods of classifying particle size distri-
bution; either method may be used. The first method assigns limiting sizes to stated
percentages by weight. For example, 10 percent, by weight, of the total lot of filter
media shall measure between X mm and Y mm, 60 percent shall measure between
A mm and B mm, and 90 percent shall measure between S mm and T mm. Because
sieves will not separate the media into fractions exactly equal to 10 percent, 60 percent,
and 90 percent of the total weight, the sizes corresponding to the percentages must be
interpolated from a plot of the percentage of sample passing each sieve against the
separation size of that sieve. The plot should be made on log-probability paper or
semilog paper.
The second method of classifying particle size distribution defines the percentage
of media that shall be finer than a stated particle size. For example, the percentage
of media finer than 0.4 mm shall be between X percent and Y percent of the total lot
of filter media. By fixing percentages X and Y that correspond to the separation
sizes of standard sieves, the results of a sieve analysis can be used directly without
plotting.
In addition to classifying particle size distribution as described above, media
gradation may also be described in terms of effective size and uniformity coefficient
as defined in Sec. 3.3 and Sec. 3.3 of ANSI/AWWA B100, respectively. In 1392, Hazen
found that the permeability of sand in a loose state correlates with the effective size
viii
-------
and uniformity coefficient, and subsequent practice has indicated that these terms
are useful for characterizing filter media gradations.
When specifying filter media size, the purchaser should use either < the effective
size and uniformity coefficient or (2) one of the t-.vo methods of classifying particle
size distribution previously discussed. Attempting to specify media size by both tech-
nioup'' m,iv rpsuln in = nf*ri?vin!? .1 nnrtir!^ ;i7p Hi;rnhntinn r'-mt- i-in-.-.- ho
by media producers
Anthracite sizes. Effective sizes of anthracite generally range from a low of
0 6 mm to a high of 1.6 mm, and uniformity coefficients are generally 1.7 or lower.
Silica sand sizes. Effective sizes of silica sand generally range from a low of
0 35 mm to a high of 0.65 mm, and uniformity coefficients are generally 1.7 or lower.
High-density sand sizes. Effective sizes for high-density sand generally range
from a low of 0.18 mm to a high of 0.60 mm, and uniformity coefficients are generally
2.2 or lower.
II.C. Filter Gravel. If the openings in the underdrain system are larger than
the filter medium, c. system of supporting layers of gravel is required to prevent the
filter medium from entering and blocking the underdrain system and to help distribute
backwash water evenly. The size and depth of the gravel layers must be selected to
achieve both objectives and to ensure that the gravel will not be displaced by the
rising wash water.
The following guidelines can be used to select the sizes and depths of gravel
layers for a conventional gravel system.
The grains of each layer should be as uniform in size as possible, with the ratio
of maximum particle size to minimum particle size not greater than 2. The minimum
particle size of the top layer of fine gravel should be four to four-and-a-half times the
effective size of the finest filter medium to be retained. From layer to layer, the ratio
of maximum particle size of the coarser layer should not be greater than four times
the minimum particle size of the finer layer. The gravel of the bottom layer should
be coarse enough to prevent its displacement by the jets of air or water emerging
from the orifices of the underdrain system. The minimum particle size of the lowest
layer should be two to three times the size of the orifices.
The thickness of each layer of gravel should be at least three times the maxi-
mum particle size of the gravel in the layer, but not less than 3 in. in any case. In
the case of irregular underdrain bottoms, such as pipe laterals, the lowest layer
should completely surround or cover the underdrain to provide a uniform upper
gravel surface on which the next gravel layer is placed.
Many combinations of gravel size and layer thickness have been used. Table F.l
describes two typical series of gravel layers that generally meet the aforementioned
guidelines. The top layer gradation is controlled by the fine filter-medium size to be
retained, and the bottom layer gradation is controlled by the underdrain orifice
sizes. The examples use commercially available gravel sizes indicated by their ASTM
Ell sieve designations.
In some designs, a high-density filter gravel is used as a replacement for, or in
addition to, the top layer in the gravel system to give added stability to the gravel
system during backwashing. The range in size and thickness of the high-density
fdter gravel layer must be closely coordinated with the other gravel layers and the
overlying media. Generally, at least 92 percent by weight shall pass through a No. 4
sieve and no more than 8 percent by dry weight shall pass through a No. 16 sieve.
The layer thickness normally ranges from 2 in. to 4 in.
-------
Table F. I Gravel layers for two sizes of fine media and two sizes of underdrain
orifices
Fine Media
Effective Size
0 40 mm-0.50 mm
Underdrain Orifice Size
6 min i u zd in.)
Gravel Layers




From Top to
Gradation of
Thickness of •


Bottom
Gravel7
Layer
Graduation
Thickness
1st"
3.35 mm-1.70 mm
76 mm
4.75 mm-2.0 mm
76 mm

(No. 6-No. 12)
(3 in.)
(No. 4-No.lO)5
(3 in.)
2nd
6.3 mm-3.35 mm
76 mm
9.5 mm-4.75 mm
76 mm

(ki in.-No. 6)
(3 in.)
(3/8 in.-No.
(3 in.)
3rd
12.5 mm-6.3 mm
76 mm
19.0 mm-9.5 mm
76 mm

(V-i in.-1/.* in.)
(3 in.)
(3/i in.-3/g in.)
(3 in.)
4th
25.0 mm-16.0 mm
76 mm-102 mm
37.5 mm-19.0 mm
76 mm-127 mm

(1 in.-5/s in.)"
(3 in.—4 in.)
(1V2 in.-34 in.)
(3 in.-5 in.)
5 th
None

63 mm-37.5 mm
127 mm-203 mm



(2Vo in.-lVo in.)
(5 in.-8 in.)
These examples do not apply when air scour is delivered through the gravel layers.
'Standard sieve sizes from ASTM Ell (Standard designation and alternative designation. See Table
B.l, column 1, subcohimns 1 and 2.)
'This layer may be replaced or supplemented by high-density gravel. Gradation and thickness of layer
must be coordinated with the other gravel layers and the filter media.
^No. 4-No. 8 preferred, if available.
3/4-in. to Vj-in. size may be considered as an alternate.
For special applications, high-density gravels are available for all layers. These
applications are not described in this standard. Special provisions are required when
air scour delivered through the gravel layers is used to assist the backwashing.
These special provisions are not described in this standard.
II.D. Acid Solubility. An acid-solubility test is included in this standard to
provide a means of measuring acid-soluble minerals or other impurities that may be
present in the filter material. The limits for acid solubility given in this standard
are based on tests of filter materials with proven performances in a wide range of
water treatment applications. Acid-solubility limits are necessary to ensure against
substantial quantities of detrimental minerals or other substances in the filter material
and to ensure against substantial solution of filter material in acidic waters or during
an acid cleaning. In many cases, the principal acid-soluble impurity in filter silica
sand and gravel is calcium carbonate (limestone).
II.E. Anthracite Quality Tests. Based on some utility experiences of high anthra-
cite loss during use in filters and the problem with Mohs' scale of hardness not
accurately defining the hardness of coal, other abrasion tests were investigated.
Samples of anthracite (new and used, soft and hard, good and poor performing) were
subjected to a battery of tests for abrasion (Mohs' scale of hardness, paint shaker
friability, and Hardgroves' Grindability Index [HGI]). These data were correlated to
other characteristics (volatiles, ash, carbon content). The committee also arranged
for presentations by a major filter equipment supplier, who extensively studied various
r ine Media
Effective Size
0.50 mm-0 60 mm
Underdrain Orifice Size
i'Z.T mm tu 5 in.)
x
-------
sources of anthracite, and an anthracite expert, who has significant experience
specifying anthracite for other industries. Both outside experts concluded that HGI
and other characteristics were also valuable in defining a high-quality coal.
Despite the consensus on the value of these new parameters, the committee
could not agree to change the standard at this time because more data is needed.
The committee .vas also concerned that many current coal suppliers might not be
able to meet the new standard and. therefore, supply to the entire water industry-
might be jeopardized. More data may allow precise limits for these new parameters
to be set. Therefore, discussion of these new anthracite characteristics will be con-
fined to the foreword only. Users of anthracite are encouraged to request data on
these new characteristics from their suppliers beginning immediately. These data
will be crucial to the preparation of the next revision of this standard. Following is a
list of suggested additional tests: HGI, percent volatiles (dry ash-free), ash percent
(dry), carbon percent, and washability characteristics (percent material with specific
gravity below 1.4, and percent material with specific gravity above 1.95).
These new characteristics can be tested by using the following standards:
ASTM D409—Standard Test Method for Grindability of Coal by the Hardgrove-
Machine Method.
ASTM D3174—Standard Test Method for Ash in the Analysis Sample of Coal
and Coke from Coal.
ASTM D3175—Standard Test Method for Volatile Matter in the Analysis Sample
of Coal and Coke.
ASTM D4371—Standard Test Method for Determining the Washability Charac-
teristics of Coal.
II.F. Bulk Shipment. The issue of protecting media from contamination during
shipment has been addressed in this revision.
Bulk shipment is not recommended; however, when trucks or railcars are specified
for hauling a bulk shipment of filter material, it is recommended that an impermeable
plastic liner be used because these trucks or railcars may be contaminated from
hauling previous bulk material.
Vibration during transit will result in media separation with the coarser material
migrating toward the top. If one compartment of the bulk shipment is divided between
two or more filters or filter halves, the filter media is likely to have different size
gradations and consequently perform differently. Therefore, if bulk shipment is
allowed, the container should be required to be compartmentalized so that each com-
partment fills no more than one filter cell. If it is specified, representative media
samples for analysis can be obtained at the point of production or loading. If the
purchaser requires sampling at the point of installation, this requirement should be
stated in the specifications.
II.G.	Media Records. Users are encouraged to maintain records of the physi-
cal characteristics and chemical composition of all media installed in filters. For
limits on undesirable impurities, refer to NSF Standard No. 61 and Section I.C in
the foreword.
ni. u se of This Standard. AWWA has no responsibility for the suitability or
compatibility of the provisions of this standard to any intended application by any
user. Accordingly, each user of this standard is responsible for determining that the
standard's provisions are suitable for and compatible with that user's intended
application.
III.A.	Purchaser Options and Alternatives. The following items should be covered
in the purchaser's specifications:
xi
-------
1.	Standard used—that is. ANSI/'AWWA BlOO, Standard for Filtering Material,
of latest revision.
2.	Method of measurement and payment, and whether this project covers the
furnishing of filter materials only, or the furnishing and placement of the materials
and preparation for service.
3.	Method of disinfecting (Sec. 4.5.3) and who will perform the disinfection
procedure
4.	Whether an affidavit of compliance is required or whether the purchaser
will select a representative to inspect the supply for compliance with this standard.
5.	Whether representative approval samples are required before shipment
(Sec. 5.1) or are in place (Sec. 4.5.2.4).
6.	Sizes, types, and characteristics of filter materials required and quantities
of each (Sec. II.B, Sec. II.C, Sec. 4.1.1, and Sec. 4.1.2). If the supplier, manufacturer,
or constructor is to be held responsible for meeting a specification regarding particle
size for media in place, the spc ifications should require that the supplier, manufac-
turer, or constructor supervise the transportation, handling, on-site storage, place-
ment, and field preparation of the media for sampling. This includes all
backwashing of media prior to sampling.
7.	Method of placing the material, if there is a preference (Sec. 4.4.2).
8.	Method of checking elevation of top surface of each layer, if there is a
preference (Sec. 4.4.3).
III.B.	Modification to Standard. Any modification to the provisions, definitions,
or terminology in this standard must be provided in the purchaser's specifications.
IV.	Major Revisions. Major changes made to the standard in this revision
include the following:
1.	The format has been changed to AWWA standard style.
2.	The acceptance clause (Sec. I.C) and definitions (Sec. 3) have been revised
to approved wording.
3.	Discussion of a potentially new anthracite standard is included in the
foreword.
4.	Addition of reference to a planned revision of ANSI/AWWA B604 for use of
GAC as filter medium.
5.	Addition of high-density sand and gravel media.
6.	Revision to delivery and bulk shipment requirements.
7.	Adoption of an alternative test for large-aggregate specific gravity.
8.	Addition of in-place media sampling.
9.	Substitution of ASTM C128 for ASTM C188 as the standard specific gravity
test of fine aggregates.
V.	Comments. If you have any comments or questions about this standard,
please call the AWWA Standards and Materials Development Department,
(303) 794-7711 ext. 6283, FAX (303) 795-1440, or write to the department at 6666
W. Quincy Ave., Denver, Colorado 80235.
xii
-------
American Water Works Association
*5
ANSI/AWWA B100-96
(Revision of ANSI/A WWA 3100-39)
A WWA STANDARD FOR
FILTERING MATERIAL
SECTION 1: GENERAL
Sec. 1.1 Scope
This standard covers gravel, high-density gravel, silica sand, high-density media,
anthracite filter materials, and the placement of the materials in filters for water sup-
ply service application. A planned revision to ANSI/AWWA B604 Standard for Granular
Activated Carbon will address the use of GAC as a filter medium and as an adsorbent.
Sec. 1.2 Purpose
The purpose of this standard is to provide purchasers with a standard for purchas-
ing and installing filtration materials; it is not intended as a guide for filter design.
Sec. 1.3 Application
This standard can be referenced in specifications for purchasing and receiving
filtering material and can be used as a guide for testing the physical and chemical
properties of filtering material samples. The stipulations of this standard apply
when this document has been referenced and only to filtering materials used in the
treatment of drinking water supplies.
SECTION 2: REFERENCES
This standard references the following documents. In their latest editions,
these documents form a part of this standard to the extent specified within the
standard. In any case of conflict, the requirements of this standard shall prevail.
1
-------
2 AWWA B100-96
AJSTM C40—Standard Test Method for Organic Impurities in Fine Aggregates
for Concrete.
ASTM C117—Standard Test Method for Materials Finer Than 75-um (No. 200)
Sieve in Mineral Aggregates by Washing.
.ASTM C123—Standard Test Method for Lightweight Pieces in AsrTesrate.
w	o	OO O
.ASTM C127—Standard Test Method for Specific Gravity and Absorption of
Coarse Aggregate.
ASTM C12S—Standard Test Method for Specific Gravity and Absorption of
Fine Aggregate
" ASTM C136—Standard Test Method for Sieve Analysis of Fine and Coarse
Aggregates.
ASTM C702—Standard Practice for Reducing Samples of Aggregate to Testing
Size.
ASTM D75—Standard Practice for Sampling Aggregates.
ASTM Ell—Standard Specification for Wire Cloth and Sieves for Testing
Purposes.
MIL-STD-105D—Sampling Procedures for Inspection by Attributes.
AN SI/AWWA C653—Standard for Disinfection of Water Treatment Plants.
SECTION 3: DEFINITIONS
The following definitions shall apply in this standard:
1.	Bag: A plastic, paper, or woven container generally containing approxi-
mately 1 ft3 or less of filter material.
2.	Constructor: The party that furnishes the work and materials for place-
ment or installation.
3.	Effective size: The size opening that will just pass 10 percent (by dry
weight) of a representative sample of the filter material; that is, if the size distribu-
tion of the particles is such that 10 percent (by dry weight) of a sample is finer than
0.45 mm, the filter material has an effective size of 0.45 mm.
4.	Manujkcturer: The party that mar>"fartures, fabricates, or produces materials
or products.
5.	Purchaser: The person, company, or organization that purchases any materi-
als or work to be performed.
6.	Semibulk container: A large plastic or woven bulk container generally con-
taining approximately 1 ton or more of filter material. It is commonly referred to as
a sack.
7.	Supplier: The party that supplies materials or services. 'A supplier may or
may not be the manufacturer.
8.	Uniformity coefficient: A ratio calculated as the size opening that will just
pass 60 percent (by dry weight) of a representative sample of the filter material
divided by the size opening that will just pass 10 percent (by dry weight) of the
same sample.
"American Societv for Testing and Materials, 100 Barr Harbor Dr., West Conshohocken, PA
1942S-2S59.
-------
FILTERING MATERIAL 3
SECTION 4: REQUIREMENTS
Sec. 4.1 Physical Requirements
4 1.1 Filter media. Filter media of anthracite, silica sand, and hish-density
sand shall conform to the following requirements.
4 111 Anthracite.
1.	Filter anthracite shall consist of hard, durable anthracite coal particles of
various sizes. Blending of non-anthracite material to meet a_ny portion of this standard
is not acceptable.
2.	The anthracite shall have specific gravity, Mohs' scale of hardness, and
acid solubility levels as indicated in Table 1.
3.	The anthracite shall be visibly free of shale, clay, and other extraneous
debris.
4.1.1.2	Silica sand.
1.	Silica sand shall consist of hard, durable, and dense grains of predomi-
nantly siliceous material that will resist degradation during handling and use.
2.	The silica sand shall have specific gravity and acid solubility levels as indi-
cated in Table 1.
3.	The silica sand shall be visibly free of clay, dust, and micaceous and organic
matter.
4.1.1.3	High-density sand.
1.	High-density sand shall consist of hard, durable, and dense grain garnet,
ilmenite, hematite, magnetite, or associated minerals of those ores that will resist
degradation during handling and use.
2.	The high-density sand shall have specific gravity, Mohs' scale of hardness,
and acid solubility levels as indicated in Table 1.
3.	The high-density sand shall be visibly free of clay, dust, and micaceous and
organic matter.
NOTE: Testing for clay, dust, and micaceous and organic matter is normally not
necessary, but if deleterious materials are noticeable, the media shall be within the
following limits: (1) a maximum of 2 percent minus No. 200 (0.074 mm) material by
washing, as determined by ASTM CUT; and (2) a color not darker than the stan-
dard color in ASTM C40 for organic impurities in fine aggregate.
4.1.1.4	Media size.
1. The media size is commonly specified in terms of effective size (ES) and
uniformity coefficient (UC) or in terms of particle size range. Only one of the following
shall be used:
Table 1 Physical characteristics of filter media
Filter Media
Specific Gravity
Characteristics
Hardness
(Mohs' Scaled
Acid Solubility
Anthracite"
Silica Sand
High-Density Sand
> 1.4
>2.5
>3.8
>2.7
NA
>5
-------
4 AWWAB100-96
a.	The effective size, as defined in Sec. 3 3. and the uniformity coefficient, as
described in Sec 3.3. shall be as specified by the purchaser
b.	The particle size range, including allowable percentage, by weight, of
undersize and oversize particles, shall be as specified by the purchaser The size
range shall state the 90 percent, 60 percent, and lu percent sizes passing by dry
weight, or other information pertinent to special applications.
4.1.2 Filter gravel. Filter gravel, including silica gravel and high-density filter
gravel, shall meet the following requirements.
4.1.2.1	Silica gravel.
1.	Silica gravel shall consist of coarse aggregate in which a high proportion of
the particles are round or equidimensional in shape. It shall possess sufficient
strength and hardness to resist degradation during handling and use, be substantially
free of deleterious materials, and exceed the minimum specific gravity requirement.
2.	Silica gravel shall have a saturated-surface-dry specific gravity of not less
than 2.5, unless a higher minimum specific gravity requirement is specified meet
a design requirement for a particular layer or filter.
3.	Not more than 25 percent, by dry weight, of the particles shall have more
than one fractured face (Sec. 5.3.2).
4.	Not more than 2 percent, by dry weight, of the particles shall be flat or
elongated to the extent that the longest axis of a circumscribing rectangular prism
exceeds five times the shortest axis (Sec. 5.3.2).
5.	The silica gravel shall be visibly free of clay, shale, or organic impurities.
Note Testing for clay, shale, or organic impurities is not normally necessary,
but if deleterious materials are noticeable, the gravel shall be within the following
limits: (1) a maximum of 1.0 percent minus No. 200 (0.074 mm) material by washing,
as determined by ASTM C117; and (2) a maximum of 0.5 percent coal, lignite, and
other organic impurities, such as roots or twigs, as determined by ASTM C123 for
lightweight pieces in aggregate using a liquid with a 2.0 specific gravity.
4.1.2.2	High-density filter gravel.
1.	High-density filter gravel shall be a coarse aggregate consisting of garnet,
ilmenite, hematite, magnetite, or associated minerals of those ores in which a high
proportion of the particles are either round or equidimensional in shape. It shall
possess sufficient strength and hardness to resist degradation during handling and
use, be substantially free of deleterious materials, and exceed the minimum density
requirement.
2.	High-density filter gravel shall have a specific gravity of not less than 3.8,
meaning that at least 95 percent of the material shall have a specific gravity of 3.8
or higher.
3.	Not more than 2 percent, by dry weight, of the particles shall be flat or
elongated to the extent that the longest axis of a circumscribing rectangular prism
exceeds five times the shortest axis (Sec. 5.3.2).
4.	The high-density gravel shall be visibly free of clay, shale, or organic impurities.
4.1.2.3	Gravel size. Filter gravel shall be furnished in the particle size ranges
stated in the purchaser's specification. For each size range of gravel specified, not
more than 8 percent by dry weight shall be finer than the lowest designated size
limit, and a minimum of 92 percent by dry weight shall be finer than the highest
designated size limit.
4.1.2.4	Acid solubility. Acid solubility shall not exceed 5 percent for sizes
smaller than No. 8 (2.36 mmi, 17.5 percent for sizes larger than No. 8 (2.36 mm) but
smaller than 25.4 mm (1 in.), and 25 percent for sizes 25.4 mm (1 in.) and larger. If
-------
FILTERING MATERIAL 5
gTavels contain materials larger or smaller than the specified size, and if the total
sample does not meet the specified solubility limit for the smaller material, the
gravel shall be separated into two portions and the acid solubility of each portion
must meet the appropriate designated percent solubdity.
Sec. 4.2 Chemical Requirements
This standard has no applicable information for this section.
Sec. 4.3 Impurities
Refer to acceptance section (Sec. I.CI in the foreword.
Sec. 4.4 Placing Filter Materials
4.4.1	Preparing filter cell. Filter cells shall be prepared according to the fol-
lowing procedure.
4.4.1.1	Cleaning filter cells. Each filter cell shall be cleaned thoroughly before
any filter materials are placed. Cleaning shall include the underdrain plenum, which
may need to be vacuumed. Each cell shall be kept clean throughout placement
operations.
4.4.1.2	Marking each layer. Before any materials are placed, the top elevation
of each layer shall be marked by a level line on the inside of the filter cell.
4.4.1.3	Storing and handling materials. Filter materials shall be kept clean. If
material cannot be placed immediately into the filter, the bulk materials shall be
stored on a clean, hard, dry surface and covered at the water utility site to prevent
contamination. Materials shipped in bags or semibulk containers shall be covered
with a durable opaque material to block sunlight and to provide protection from
weather. Bags and semibulk containers shall be stored on pallets or dunnage. Each
size and type of filter material shall be stored separately. Materials shipped in bags
or semibulk containers shall not be removed from the bags or semibulk containers
before placement in the filter under any circumstance, except for sampling.
4.4.2	Placing materials.
4.4.2.1	Caution. The bottom layer of gravel shall be carefully placed to avoid
damaging the filter underdrain system. For materials smaller than V2 in., workers
shall not stand or walk directly on the filter material. They shall walk on boards or
plywood that will support their weight without displacing the material. The same
care should be taken when an air wash system is installed above the gravel.
4.4.2.2	Placing layers. Each layer shall be completed before the layer above it
is started. Each layer of filter material shall be deposited in a uniform thickness.
Care shall be exercised in placing each layer to avoid disturbing the integrity of the
layer beneath. The top surface shall be screeded level.
4.4.2.3	Alternate method of placement. Bulk materials may be placed dry by
using a chute or conveyor to discharge the materials onto a platform from which
they may be distributed with a hand shovel. Alternatively, bulk materials may be
placed hydraulically by pump or ejector.
"In new filter construction, the placement of filter media should follow operational testing
of the backwash system and assurance that the filter box is watertight. See Table 2 for
maximum backwash rates.
-------
6 AWWAB100-96
For filter sand or anthracite placed using the wet method, the materials shall
be added through the water and then backwashed for leveling Pneumatic handling
of anthracite is not recommended
4.4.2.4	Placing material from bags or semibulk containers. When filter material
is shipped in'bags or semibulk containers and hydraulic placement is not used, the
bags or semibulk containers shall be placed in the filter and the material distributed
directly irom mem. iuaCTIOX Uo not disturb any layers already in place.) For the top
media layer, only 90 percent of it? intended depth should be added, then the initial
backwashing shall proceed. Following this, the additional 10 percent or whatever is
necessary to reach the finished elevation shall be added.
4.4.2.5	Layer elevatioa The elevation of the top surface of each layer shall be
checked by filling the filter with water to the level line previously marked on the
inside of the filter cell.
4.4.2.6	Washing gravel layer. After all filter gravel is placed, and before any
filter sand or anthracite is placed, the filter should be washed for 5 min at the
maximum available rate, not to exceed 25 gpm/ft2 of filter area. Care shall be taken
not to disturb the graded gravel, especially if air is present in the underdrain. Any
gravel that becomes disturbed by the wash shall be removed and replaced with clean
material of the proper type and size.
4.4.2.7	Washing other material. With a dual- or multiple-media filter bed,
each material shall be washed and scraped or skimmed as the purchaser requires to
remove excess fine materials before the next material is installed.
4.4.3	Top Surface Elevation. The top surface of the filter material after initial
washing (Sec. 4.5.1.1) shall have an elevation equal to the finished elevation plus
the thickness of material to be removed by scraping.
4.4.4	Contamination. Any filter media that becomes contaminated after place-
ment shall be removed and replaced with clean material of the proper type and size.
Sec. 4.5 Preparing Filter for Service
4.5.1	Washing.
4.5.1.1	Initial wash After all filter materials have been placed, wash water
shall be admitted slowly upward through the underdrain system until the entire bed
is flooded. The bed shall be allowed to stand for as long a period as the purchaser
requires to saturate the media before the initial wash. This period shall not be less
than 12 h if the bed has been installed dry or allowed to stand dry. The wash rate
shall be increased gradually during the initial wash to remove air from the bed.
4.5.1.2	Backwash rate. During each backwash, the water shall be applied at
an initial rate of not more than 2 gpm/ft2 of filter area. The backwash rate shall
then be increased gradually over a period of 3 min to the maximum rate indicated in
Table 2, and maintained at the maximum rate for not less than 5 min.
4.5.2	Scraping. After the initial wash, the filter shall be partially drained and
a layer of fine material approximately 3/i6-in. thick shall be removed from the surface of
the filter by scraping.
4.5.2.1	Repetitions. The scraping operation shall be repeated as many times
as necessary to remove all fine material (these fines will be visible, giving a smooth
appearance rather than the desired rough surface texture) and, in the case of anthra-
cite, to remove all fiat particles.
4.5.2.2	Number of washes. The filter shall be washed at least three times
between scrapings. Each wash shall last at least 5 min and shall be at an appropriate
rate as listed in Table 2.
-------
FILTERING MATERIAL 7
Table 2 Maximum backwash races
Water Temperature
¦5 rr
Maximum Backwash Rate"
/p,v: ft -
50 or less
15
51-55
1G
56-60
17
61-65
IS.5
66-70
20
71-75
21
Above 75
22.5
"These maximum backwash rates are a guideline for 0.45 mm to 0.65 mm sand and 1 0 mm anthracite. The rates should
be adjusted as necessary for other filter materials. The lowest maximum backwash rate should be that which fluidizes the
bed and attains sufficient velocities to bring fines to the surface.
4.5.2.3	Additional material. If additional material is required to bring the top
surface of the filter to the specified finished elevation, sufficient material shall be added
before the final scraping operation. Adequate material shall be added to anticipate
the final scraping.
4.5.2.4	In-place media sampling. If in-place samples of media are required by
the purchaser, composite samples shall be prepared from a minimum of four filters
after they have been backwashed and drained. Core samples shall be taken using a
2-in diameter core sampler. It shall be inserted to the elevation just above the
gravel interface, and then removed by excavating around it in order to extract a
complete profile of material above that elevation Composite samples from each filter
shall consist of equal portions from a minimum of five cores distributed over each
media surface.
1.	Sample preparation Upon receipt of the samples, the laboratory shall pre-
pare them in the following manner:
a.	Place 0.25 L to 0.5 L of media sample in a 1-L or 1-gal bottle.
b.	Fill the bottle to within 1 in. of the top with clean water.
c.	Place cap on bottle and shake for 2 min using two or three forward and
backward motions per second.
d.	Allow the media to settle, then decant the supernatant liquid into a clean
container.
e.	Repeat steps b through d until supernatant is clean.
f.	If coal or granular activated carbon is used as the top layer, then separate
that media from the sand by using the technique described in ASTM D4371.
2.	Testing. Test samples in accordance with Sec. 5.3.
4.5.3 Disinfecting. After all work related to placement of media has been com-
pleted, and before the filter is placed in service, the entire filter shall be disinfected
by chlorination in accordance with ANSI/AWWA C653, unless otherwise specified in
the purchaser's specifications. The procedure for disinfection of granular activated
carbon will be stipulated in a planned revision to ANSI/AWWA B604.
-------
8 AWWA B 100-96
SECTION 5: VERIFICATION
Sec^S.l Approval Samples
When .-Decified a renre^entative ?.amnlp nf p-trh ^70 nr" filr->- rrrira^il iSnll ho
submitted for approval before shipment. The sample shall be submitted in clean,
n c; f _ f i  f r,Anrnno>*; n 111 n 1 * ¦ r^< <-* Q ,-3 "U f V>n »-«o »~>n ^ ^ ^ ^ ~ • U - - .. — — 1. ~ ^
the size or grade of the contents. After approval of the samples, shipments shall be
of a quality equal to the sample. Approval samples shall meet the requirements of
Sec. 5.2.
Sec. 5.2 Sampling
Sampling of filter materials shall be performed in accordance with ASTM D75
as modified and supplemented herein. The size of the composite samples shall be as
indicated in L. ble 3.
5.2.1	Bulk shipments. Bulk shipments are not recommended (see foreword,
Sec. II.F). Representative media samples in a bulk shipment are obtained most eas-
ily at either the production or loading point. When a truck or railcar is filled at the
production site, sampling across the cross section of flow of the material being
loaded is recommended. The composite sample shall be prepared in accordance with
Sec. 5.2.4, with the weight of the sample as given in Table 3. A composite sample
shall be taken as each railcar or truck is filled. It is not recommended that filter
materials be sampled on receipt at the jobsite. However, if the purchaser specifies
sampling on receipt, samples shall be taken from 10 locations in the railcar or truck.
The railcar or truck shall be sampled near, but not in, each corner, at the center, and
at five other random locations.
5.2.2	Bag shipments. When material is shipped to the jobsite in bags, repre-
sentative samples shall be collected using a core sampler. The representative samples
from each bag shall be combined to produce the required composite sample. The
minimum size of the composite sample is provided in Table 3. The number of bags to
be sampled is indicated in Table 4.
5.2.3	Semibulk container shipments. While semibulk containers are filled at
the production site, sampling across the cross section of the material being loaded is
recommended. The composite sample shall be prepared in accordance with
Sec. 5.2.4, with the weight of the sample as indicated in Table 3. The number of
semibulk containers to be sampled during filling shall be as indicated in Table 4. At
Table 3 Minimum size of composite sample
Maximum Size of

Minimum

Particle
in Sample

Sample Size

mm
(in.)
kg

fib)
63.0
(2V2)
45.0

(100)
37.5
(1V2)
32.0

(70)
25.4
(1)
23.0

(50)
19 0
(3/4)
14 0

(30)
12.5
(V2)
90

(20)
9 5
(3/s) and smaller
4 5

(10)
-------
FILTERING MATERIAL 9
Table 4 Sampling of bagged media
Lot Size
( number of bags shipped)
Minimum Sample Srzes
(number of bags)
2-S
9
16-25
26-50
51-90
91-150
151-2S0
231-500
501-1,200
8
13
20
32
50
80
125
200
315
500
1,201-3,200
3,201-10,000
10,001-35,000
35,001-150,000
'Refer to Military Standard MIL-STD-I05D (1963).
least one composite sample shall be generated for each size and type of material per
railcar load or truckload.
5.2.4 Composite sample. The composite sample shall be reduced to repre-
sentative samples for testing in accordance with ASTM C702. Samples shall be
tested by the methods indicated in Sec. 5.3.
If filter materials testing is not witnessed at the shipping point by the pur-
chaser, the material should be tested at the jobsite. The material shall be sampled in
accordance with ASTM D75 and reduced to testing size in accordance with ASTM
C702. A portion of the reduced sample should be retained for possible independent
analysis.
5.3.1 Acid solubility. The acid-solubility test is performed by immersing a
known weight of material in 1:1 hydrochloric acid (HC1-) (made by combining equal
volumes of 1.18 specific gravity HC1 and HoO) until the acid-soluble materials are
dissolved, then determining the weight loss of the material. The minimum sample
size and the minimum quantity of concentrated HC1 diluted one-to-one with distilled
water are indicated in Table 5.
5.3.1.1 Procedure. The procedure for testing acid-solubility shall include the
following:
1.	Wash sample in distilled water and dry at 110°C ± 5°C to constant weight.
2.	Allow sample to cool in a desiccator. Weigh dried sample to the nearest
0.1 percent of the weight of the sample.
3.	Place sample in beaker and add enough 1:1 HC1 to immerse the sample
completely, but not less than the quantity indicated in Table 5.
4.	Allow to stand, with occasional stirring, at room temperature for 30 min
after effervescence ceases.
5.	Wash sample several times in distilled water and dry at 110°C ± 5°C to
constant weight.
Sec. 5.3 Test Procedures—General
-------
10 AWWAB100-96
Table 5 Minimum sample and acid quantities for acid-solubility tests
Maximum Size of
Minimum
Minimum
Panicle
in Sample
Sample Weight
Quantity 1' 1 HC1
mm
fin !
a
0
rnL
63 0
(2 4;
4.000
7,U00
37.5
(I'A1
250
SOO
25.4
(1)
250
SOO
19.0
(3/4)
250
SOO
12.5
(»/2)
250
SOO
9.5
(3/3) and smaller
100
320
6.	Allow sample to cool in a desiccator and weigh to the nearest 0.1 percent
of the weight of the sample.
7.	Report the loss in weight as acid-soluble material.
5.3.1.2 Calculation.
To calculate acid-soluble material, the following equation shall be used:
acid solubility (%) = ^oss of weight ^	^ ^
original weight
Duplicate tests shall be made on each size of the material and the two results aver-
aged. If the two results do not agree within 2 percent of the total sample weight, then
two additional tests shall be made and the four determinations averaged.
5.3.2	Gravel shape. The following definitions and tests shall be used in identi-
fying fractured, flat, or elongated pieces of gravel. Identification of fractured, flat, or
elongated particles is to be done by visual separation.
5.3.2.1	Fractured face definition. A fractured face is defined as a surface sur-
rounded by sharp edges, such as those produced by crushing, that occupy more than
approximately 10 percent of the total surface area of the particle. This is intended to
exclude a surface with small nicks and chips from classification as a fractured face.
5.3.2.2	Shape determination. The ratio of the longest axis to the shortest axis
of the circumscribing rectangular prism for a piece of gravel shall be determined
using a caliper or a proportional divider. Suspected elongated pieces can be checked
by comparing the minimum thickness of the particle, as measured at its approxi-
mate midpoint, with the maximum length dimension.
5.3.3	Specific gravity. The specific gravity of filter silica gravel shall be deter-
mined in accordance with ASTM C127 and shall be reported as saturated-surface-diy
specific gravity or the Noble Large Aggregate Test. The specific gravity of high-density
gravel, high-density sand, silica sand, and filter anthracite shall be determined in
accordance with ASTM C128 and shall be reported as apparent specific gravity. Anthra-
cite may also be tested for float/sink in accordance with ASTM D4371.
5.3.3.1 Noble Large Aggregate Test Procedure.
1.	Soak the sample in water at room temperature (approx. 73°F) for 24 h.
2.	Set the water reservoir on a level surface with the cylinder valve closed.
3.	Fill the reservoir with room temperature water to a depth where the valve
opening is totally submerged.
4.	After 5 min, open the valve and allow the excess water to drain. Close the
valve after the last drop has drained.
-------
FILTERING MATERIAL 11
5.	Remove the presoaked sample from the water and pat the sample dry with
a dry cloth or paper towels to a saturated surface dry (SSD) condition.
6.	Immediately weigh the sample to the nearest 0.1 g.
7.	With a funnel, or by hand, carefully drop the preweighed sample into the
water reservoir as indicated in Figure 1. Leave the sample submerged for 15 min
while tapping on the sides of the reservoir and stirring to free the entrapped air.
8.	Place the graduated buret (with valve closed) under the transparent vinyl
plastic (tygon) tubing. Open the cylinder valve to allow the displaced water to drain
into the graduated buret to its last drop. Allow the buret tip to fill before taking a
final volume reading.
9.	Read the water volume in millilitres.
10. Perform calculation.
Bulk specific gravity (saturated surface dry) = Item 6/Item 9
5.3.4 Sieve analyses. Sieve analyses for filter materials shall be performed in
accordance with ASTM C136, as modified and supplemented herein.
5.3.4.1	Principle. Particle sizes shall be determined by screening through
standard sieves conforming to ASTM Ell. Particle size shall be defined in terms of
the smallest sieve opening through which the particle passes.
5.3.4.2	Sample size. The minimum sample size for sieve analyses shall be as
indicated in Table 6.
5.3.4.3	Procedure. The sieving procedure shall be in accordance with
ASTM C136. Care shall be taken to avoid breaking anthracite particles when
sieving. Generally, sieves require machine shaking times of 10 min ± 0.5 min for
sand or gravel and 5 min ± 0.5 min for anthracite. All standard sieves used for testing
filter materials shall conform to the tolerances required in ASTM Ell. If questions of
-------
12 AVAVA B 100-96
Table 6 Minimum sample size for sieve analyses
Max.-.!
Size of
Minimum
Particle in Sample
Sample Weight
mm
(m )
kg fib)
63.0
12 (A)
23 0 (50)
37 5
( L VJ.I
16 0 i35)
25.4
(1)
11.0 (25)
19.0
(3/t)
6.S (15)
12.5
(V'2)
4.5 (10)
9.5
(3/s)
2 3 (5)
No. 4 (4 751

500.0 g
No. 8(2.36)

100.0 g
compliance to specifications arise when nominal standard sieve openings are used,
standard reference materials (glass spheres) certified by the National Bureau of
Standards should be used in accordance with their calibration procedure to accurately
determine the effective opening size of each sieve. If standard reference material for
calibration is not used, then the data shall be replotted using both the maximum
and minimum permissible variation of average opening from the standard sieve desig-
nation as shown in Table 1, column 4 of ASTM Ell. (Sections of ASTM Ell, column 4,
are reprinted in Appendix B, Table B.l.) The materials shall be in compliance if either
of the plots agrees with the specifications.
To avoid excessive interpolation when determining the effective size (the size
opening that 10 percent of the particles can pass) and the D60 (the size opening that
60 percent of the particles can pass), the sieves used on a particular sieve analysis
shall have openings such that the ratio between adjacent sizes is the fourth root of
2, or 1.1892. The sieves shall be chosen so that the nominal opening of only one
sieve is smaller than the smallest allowable effective size so that the greatest range
of particle size distribution can be measured in one standard nest of six sieves. If
the media specification limits the quantity of fines, an additional sieve shall be
added for a total of seven sieves, so that there are two sieve measurements taken
below the effective size.
5.3.4.4	Calculation. The cumulative percent passing each sieve shall be calcu-
lated and plotted on log-probability paper or semilog paper, with the sieve opening
on the log scale and the cumulative percent passing on the probability scale or linear
scale. A smooth curve shall be drawn through the points plotted.
5.3.4.5	Uniformity coefficient. Read from the curve the sieve size corresponding
to the 10 percent size, which is the effective size in millimetres. Read the 60 percent
size and divide this by the 10 percent size. This ratio is the uniformity coefficient.
5.3.4.6	Mohs' scale of hardness. No standard test method has been found;
however, all commercial laboratories follow the same procedure.
5.3.5 Rejection. If the filter materials do not meet the applicable requirements
of this standard, they shall be removed from the site. An independent laboratory
deemed acceptable by the purchaser may be employed by the constructor, manufac-
turer, or supplier to sample and test the disputed material before its removal. Once
media has been placed in filters, every filter must meet the size specifications.
-------
FILTERING MATERIAL 13
5 3 5 I Additional held teats. At the option of the purchaser, constructor,
manufacturer, or supplier, two additional tests shall be conducted using two additional
representative samples and a mutually acceptable independent laboratory. Unless
otherwise agreed on between the purchaser and constructor, the results of all tests
shall be averaged arithmetically. If the independent laboratory reports that the material
complies with the applicable requirements of this standard, the purchaser shall accept
luc uiaiciiua n cue inaicuai uuco uuc mcc. uic iC4UH erneuco ui liui oiallQdlU, Cue
constructor shall promptly remove the material from the jobsite.
5.3.5.2 Alternative to removal. As an alternative to removing the rejected material,
the constructor may, with the purchaser's approved and control, reprocess the material
at the jobsite to meet the applicable requirements.
SECTION 6: DELIVERY
Sec. 6.1 Marking
6.1.1	Required. Each package and container shall have marked legibly on it
the name of the material, the gradation, the filling date, the net weight of the contents,
the name of the manufacturer, the lot number, and the brand name, if any, and shall
bear such other markings as are required by the US Department of Transportation
and other applicable regulations and laws. When shipped in bulk, this information
shall accompany the bill of lading.
6.1.2	Optional. Packages may also bear the statement, "This material meets
the requirements of AWWA B100, Standard for Filtering Materials," provided that
the requirements of this standard are met and the material is not of a different
quality in separate agreement between the supplier or constructor and purchaser.
Sec. 6.2 Packaging and Shipping
Shipment shall be made in bags or semibulk containers or in clean railcars or
trucks with tight closures to avoid loss or contamination of material in transit.
6.2.1	Bags. When specified, shipment shall be made in suitable new and unused
heavy-duty cloth, paper, woven polypropylene, or polyethylene bags that contain ultra-
violet (UV) light inhibitors and shall contain not more than 1 ft3 of material. Each
bag shall be marked in an appropriate manner so that its contents are identified.
6.2.2	Semibulk containers. When specified, shipment shall be made in suitable
new, unused, heavy-duty, woven, polypropylene semibulk containers, treated with
UV light inhibitors, and having a safety factor of at least 5:1. Each container shall
hold one or more tons of material. To aid in handling, semibulk containers should
have attached straps or sleeves strong enough to support their entire weight when full.
Each semibulk container shall be marked so that its contents are identified.
6.2.3	Bulk.
1.	Bulk shipment is not recommended for reasons described in the foreword.
2.	When truck shipment is specified, and where a liner is not used, shipment
shall be made in clean truck containers. Truck containers shall be cleaned before
loading by washing with water that is 1S0CF or hotter. Provisions for tight covering
shall be made to avoid loss and to prevent contamination. The trucks shall be exclu-
sively dedicated to hauling potable water filtering materials.
3.	When railroad hopper car shipment is specified, shipment shall be made in
clean cars lined with an impermeable plastic liner and tight closures to avoid loss
-------
14
AVAVA B100-96
and contamination. If open-top cars are used, they shall be tightly covered. The
purchaser is cautioned that potential contamination of the product is possible
because of the absence of hopper cars dedicated solely to filter materials.
6.2.4 Shipping notice. When a shipment of material is being loaded, the con-
structor shall notify the purchaser of the railcar number and the date of shipment.
The shipping notice shall contain a certification of the particle size distribution of
the material in the shipment.
Sec. 6.3 Affidavit of Compliance
When specified by the purchaser, the manufacturer, supplier, or constructor
shall provide an affidavit of compliance stating that the filter materials furnished
comply with the applicable provisions of this standard.
-------
APPENDIX A
Bibliography
This appendix is for information only and is not a part of AWWA BI00
Adin, A , and M. Reb'nun. 1974. High-Rate Contact Flocculation-Filtration With
Cationic Polyelectrolytes, :Jour. AWWA, 66(2)'109.
American Water Works Association. 1990. Water Quality and Treatment, A Handbook
of Public Water Supplies. 4th ed. New York, N.Y.: McGraw-Hill Book Company.
	. 1990. Water Treatment Plant Design. 2nd ed. Denver, Colo.: ASCE, AWWA,
CSSE.
Amirtharajah, A. 1978. Optimum Backwashing of Sand Filters, Jour. Enuir. Engrg.
Div., ASCE, 104(Oct):917.
Amirtharajah, A., and J.L. Cleasby. 1972. Predicting Expansion of Filters Bering
Backwash, Jour. AWWA, 64(1):52.
Arboleda, V.J., and J.L. Cleasby. 1979. Velocity Gradients in Granular Filter Back-
washing, Jour. AWWA, 71(12):732.
Baylis, J.R. 1950. Experience With High-Rate Filtration, Jour. AWWA, 42(7):687.
	. 1959. Review of Filter Bed Design and Methods of Washing, Jour. AWWA,
51( 11): 1433
	. 1960. Two-Layer Filter Media, Jour. AWWA, 52(2):215.
Bellamy, W.D., et al. 1985. Removing Giardia Cysts with Slow Sand Filtration,
Jour. AWWA, 77(2):52.
Berkeley, W.H. 1952. Experience With Filter Underdrains at Lewiston, Idaho, Jour.
AWWA, 44(6):491.
Bishop, S.L. 1981. Methods for Evaluating Performance of Filter Media, Jour.
NEWWA, 95(9):193.
Black, A.P. 1966. Better Ibols for Treatment, Jour. AWWA, 53(2):137.
Braidech, T.E., and R.J. Karlin. 1985. Causes of a Waterborne Giardiasis Outbreak,
Jour. AWWA, 77(2):48.
Cleasby, J.L. 1981. Filtration—Back to the Basics, AWWA Seminar Proc. 20155.
	. 1982. Unconventional Filtration Rates, Media, and Backwashing Tech-
niques, Proc. Public Water Supply Engineers Conference.
	. 1984. Unconventional Filtration Rates, Media and Backwashing Tech-
niques, Innovations in Water & Wastewater Fields. Stoneham, Mass.:
Butterworths. Proc. Seminar on Innovations in the Water and Wastewater
Fields, Univ. of Michigan, Feb. 2-4, 1983.
Cleasby, J.L., and R.E. Baumann. 1962. Selection of Sand Filtration Rates, Jour.
AWWA, 54(5):579.
Cleasby, J.L., et al. 1977. Backwashing of Granular Filters, Jour. AMA, 69(2): 115.
Cleasby, J.L., and K.S. Fan. 1981. Predicting Fluidization and Expansion of Filter
Media, Jour. San. Engrg. Div., ASCE, 107(June):455.
Cleasby, J.L., D.J. Hilmoe, and C.J. Dimitracopoulos. 1984. Slow Sand and Direct
In-line Filtration of a Surface Water, Jour. AWWA, 76(12):44.
Cleasby, J.L., and J.C. Lorence. 1978. Effectiveness of Backwashing of Wastewater
Filters, Jour. Emir. Engrg. Div., ASCE, 104(Aug):749.
Cleasby, J.L., and G.D. Sejkora. 1975. Effect of Media Intermixing on Dual Media
Filtration. Jour. Enuir. Engrg. Div., ASCE, 101(Aug):503.
15
-------
16 AWWAB100-96
Cleasby, J L., E.W. Stangl, and G.A Rice. 1975. Developments in the Backwashing
cf Granular Filters. -Jour Enuir Engrg Diu., ASCE, 10 L(Oct):713.
Cleasby, -J L., M.M. Williamson, and R.E. Baumann. 1963 Effect of Filtration Rate
Changes on Quality, Jour AWWA, 55(,7j:369
Cleasby, J.L' , and C.F. Woods 1975. Intermixing of Dual Media and Multi-Media
Granular Filters. -Jour AWWA. 67(41:197
Conley, W.R. 1961. Experience With Anthracite-Sand Filter, Jour. AWWA,
53^12 ,'.1473.
Conley, W.R. 1965. Integration of the Clarification Process, Jour. AWWA,
57U0V.1333.
Cosens, K.W. 1956. Design and Operation Data on Large Rapid Sand Filtration
Plants in the United States and Canada, Jour. AWWA, 48(7):819.
Craft, T.F. 1971. Comparison of Sand and Anthracite for Rapid Filtration, Jour.
AWWA, 63(1): 10.
Culbreath, M.C. 1967. Experience With a Multimedia Filter, Jour. AWWA,
59(8):1014.
Culp, G.L., and R.L. Culp. 1974. New Concepts in Water Purification. New York,
N.Y.: Van Nostrand Reinhold Company.
Dostal, KA., and G.G. Robeck. 1966. Studies of Modifications in Treatment of Lake
Erie Water, Jour. AWWA, 58(11):1489.
Eliassen, R., and E.A Cassell. 1957. How lb Design and Operate Rapid Sand Filter
Facilities, Wtr. Works Engrg., 110(Dec):1196.
Feben, D. 1960. Theory of Flow in Filter Media, Jour. AWWA, 52(7):940.
Fox, K.R., et al. 1984. Pilot-Plant Studies of Slow-Rate Filtration, Jour. AWWA,
76(12):62.
Ghosh, G. 1958. Mechanism of Rapid Sand Filtration, Wtr. & Wtr. Engrg., 62:147.
Grover, K 1980. Water Filter Design—What to Look For in the 80s, Amer. City &
Country, 95(June):39.
Hall, W.R. 1957. An Analysis of Sand Filtration, Paper 1276-1-9, Jour. San. Engrg.
Div., ASCE.
Hamann, C.L., and R.E. McKinney. 1968. Upflow Filtration Process, Jour. AWWA,
60(9):1034.
Haney, B.J., and S.E. Steimle. 1974. Potable Water Supply By Means of Upflow
Filtration (L'Eau Claire Process), Jour. AWWA, 66(2): 117.
Healy, G.D. Jr. 1965. Rapid Sand Filtration, S.W. Wtr. Works Jour., 46(Jan):23.
Heiple, L.R. 1959. Effectiveness of Coarse-Grained Media for Filtration, Jour.
AWWA, 51(6):749.
Hess, A.F. Ill, et al. 1982. Pilot-Scale Studies of the Treatment of the Susquehanna
River for Baltimore, Maryland. Proc. AWWA Annual Conference.
Hsiung, A.K. 1975. The Effect of Chemical Treatment and Filtration Variables on
Effluent Quality Proc. AWWA WQTC.
Hudson, H.E. Jr. 1959. Declining-Rate Filtration. Jour. AWWA, 51(11):145.
	. 1956. Factors Affecting Filtration Rates, Jour. AWWA, 48(9): 1138.
	. 1958. Factors Affecting Filtration Rates, Jour. AWWA, 50(2):271.
	. 1959. Operating Characteristics of Rapid Sand Filters. Wtr. & Sewage
Works, 106(9):R-261.
	. 1948. A Theory of the Functioning of Filters, Jour. AWWA, 40(8):86S.
	. 1981. Water Clarification Processes: Practical Design and Evaluation. New
York, N.Y.: Van Nostrand Reinhold Company.
-------
FILTERING MATERIAL 17
Hutchison. VVR 1975. Operational Variables and Limitations of Direct Filtration,
Res. Rept W54. Toronto, Ont.. Ontario Ministry of the Environment
Hutchison, W, and P.D Foley. 1974. Operational and Experimental Results of Direct
Filtration, Jour AWWA, 66i2):79.
Ives. K J 1964. Progress in Filtration, -Jour. AWWA, 56(9;' 1225.
Ives. K-J . and I. Sholji. 1965 Research of Variables Affecting Filtration, Jour. Sen..
Engrg. Diu., ASCE. 91 SA4, 1.
Jung, H., and E.S Savage. 1974. Deep-Bed Filtration, Jour. AWWA, 66t'2):73.
Kawamura, S. 1975. Design and Operation of High-Rate Filters, Part 1, Jour.
AWWA, 67( 10):535.
	. 1975. Design and Operation of High-Rate Filters, Part 2, Jour. AWWA,
67(11):653.
	. 1975. Design and Operation of High-Rate Filters, Part 3, Jour. AWWA,
67(12):705.
Kerrigan, J.E., and L.B. Polkowski. 1965. Experiments With Plastic Prefilter Media,
Jour. AWWA, 57(1):85.
Laughlin, J.E., and T.E. Duvall. 1968. Simultaneous Plant Scale Tests of Mixed
Media and Rapid Sand Filters, Jour. AWWA, 60(9): 1015.
Logsdon, G.S. 1979. Water Filtration for Asbestos Fiber Removal. EPA 600/2-79-206.
Logsdon, G.S., et al. 1985. Evaluating Sedimentation and Various Filter Media for
Removal oiGiardia Cysts, Jour. AWWA, 77(2):61.
Logsdon, G.S., and J.M. Symons. 1977. Removal of Asbestiform Fibers by Water
Filtration, Jour. AWWA. 69(9):9, 499.
McBride, D.G., et al. 1977. Pilot Plant Investigations for Treatment of Owens River
Water. Proc. AWWA Annual Conference.
McCormick, R.F., and P.H. King. 1982. Factors That Affect Use of Direct Filtration
in Treating Surface Waters, Jour. AWWA, 74(5):234.
O'Melia, C.R., and D.K. Crapps. 1964. Some Chemical Aspects of Rapid Sand Fil-
tration, Jour. AWWA, 56(10): 1326.
Oeben, R.W., H.P. Haines, and K.J. Ives. 1968. Comparison of Normal and Reverse
Graded Filtration, Jour. AWWA, 60(4):429.
Palmer, C.E. 1951. Anthrafilt and Rotary Surface Wash for Filters, Wtr. & Sewage
Works, 98(June):258.
	. Pitman, R.W. 1960. Test Program for Filter Evaluation at Hanford, Jour.
AWWA, 52(2):205.
Prindeville. P. 1983. Upgrading Water Filtration Plants, Civil Engrg., 53(Oct):64.
Qureshi, N. 1982. The Effect of Backwashing Rate on Filter Performance, Jour.
AWWA, 74(5):234.
Rae, F.C. 1958. Porous Plate Filter Bottoms—Are Now of Age. Wtr. & Sewage
Works, 105(Apr):157.
Rast, F.S. Jr. 1956. Combination Sand Anthrafilt Media Provides Longer Filter
Plant Runs, Wtr. Works Engrg., 109(Oct):934.
	. 1982. Recommended Standards for Water Works. Albany, N.Y.: Great Lakes-
Upper Mississippi Board of State Sanitary Engineers. Health Educ. Service.
	. 1953. Revision of Filtering Material Standard, Jour. AWWA, 45(8):872.
Robeck, G.G., K.A. Dostal, and R.L. Woodward. 1964. Studies of Modifications in
Water Filtration, Jour. AWWA, 56(2):198.
Sanks, R.L., ed. 1979. Water Treatment Plant Design For the Practicing Engineer.
Ann Arbor, Mich.: .Ann Arbor Science Publishers, Inc.
-------
18 AWWAB100-96
Sampling Procedures and Tables for Inspection by Attributes. 1963. Military Stan-
dard MIL-STD-105D.
Segall, B.A., and D.A. Okun. 1966. Effect of Filtration Rate on Filtrate Quality,
Jour. AWWA, 58(3):368.
Shepherd, H.H. 1965. Sand and Gravel Filter Media, Filtration and Separation,
ofv„../rw\. <-c
Shull, K.E. 1965. Experiences With Multiple Bed Filters, Jour. AWW.4., 57(3):314.
Slezak, L.A., and R.C. Sims. 1984. The Application and Effectiveness of Slow Sand
Filtration in the United States, Jour. AWWA, 76(12):38.
Stolarik, G. 1983. Ozonation and Direct Filtration of Los Angeles Drinking Water.
Proc. Sixth Ozone World Congress, International Ozone Association.
Stuppy, M.L., et al. 1954. Types of Filter Bottoms, Jour. AWWA, 46(6):548.
Tate, C.H., and R.R. Trussell. 1978. Use of Particle Counting in Developing Plant
Design Criteria, Jour. AWWA. 70(12):691.
Tentative Standard Specifications for Filtering Material-5C-T. 1949. Jour. AWWA,
41(3):289.
Toregas, G. 1983. Using Backwash Kinetics to Evaluate Attachment Mechanisms
and Forces During Filtration, Jour. AWWA, 75(1983):254.
Trussell R.R., et al. 1980. Recent Development in Filtration. System Design, Jour.
AWWA, 72(12):705.
Tuepker, J.L., and CA. Buescher Jr. 1969. Operation and Maintenance of Rapid
Sand Mixed-Media Filters in a Lime Softening Plant, Jour. AWWA, 60:1377.
Turner, H.G. 1943. Pennsylvania Anthracite as a Filter Medium, Indust. & Eng.
Chem., 35(Feb).
Ullrich, A.H. 1949. Rapid Sand Filter Design and Maintenance, Wtr. & Sewage
Works, 96(Oct):381.
Weber, W.J. Jr. 1972. Physiochemical Processes For Water Quality Control 1871. New
York, N.Y.: Wiley Inter-Science.
-------
APPENDIX B
Calibration of Sieves
Tkis appendix is for information only and is not a part of A VVIV'A BlOO.
Section B. 1 Precision of Sieves	
Although sieves are made from carefully selected brass wire cloth with meshes
that are as square and even-sized as possible, it is rare that they will have exactly
the same size openings, even when made from the same piece of material. For pre-
cise work, all sieves should be calibrated according to the procedures in ASTM" Ell,
Specification for Wire-Cloth Sieves for Testing Purposes. (For nominal dimensions
for wire cloth of standard test sieves, see Table B.l).
Section B.2 Glass Spheres
For routine checking of sieves and for determining the effective sieve openings,
a method that employs glass spheres is recommended. The glass spheres should not
be used to determine conformity to specifications. Glass spheres for use in sieve
calibration may be obtained from the National Institute of Standards.1" Four of these
standard reference materials are now available, including SRM 1019a for calibrating
sieves No. 8 to No. 35; SRM 1018a for calibrating sieves No. 20 to No. 70; SRM
1017a for calibrating sieves No. 50 to No. 170; and SRM 1004 for calibrating sieves
No. 140 to No. 400. Detailed instructions on the use of the glass spheres for calibrating
sieves are furnished with each sample.
"American Society for Testing and Materials, 100 Barr Harbor Dr., West Conshohocken, PA
19423-2959.
7N'arior.al Institute of Standards and Technology, Supply Division, BIdg. 301, Gaithersburg,
MD 20S99.
19
-------
20
AVTVA B100-96
Table B. 1 Nominal dimensions, permissible variations for wire cloth of standard test sieves
(USA Standard Series)



Permissible






Variation of
Maximum




Nominal
Average Opening
Opening Size for

Nominal


Sieve
From the
Not More Than
Maximum
Wire
Sieve Designation
Opening
Standard Sieve
3 Percent, of
individual
Diameter
Standard"
Alternative
in.'
Designation
Openings
Opening
mm§
125 mm
5 in.
5
±3.7 mm
130.0 mm
130.9 mm
8.0
106 mm
4.24 in.
4.24
±3.2 mm
110.2 mm
111.1 mm
6.40
100 mm"
4 in."
4
~3.0 mm
104.0 mm
104.8 mm
6.30
90 mm
3V2 in.
3.5
±2.7 mm
93.6 mm
94.4 mm
6.08
75 mm
3 in.
3
±2.2 mm
7S.1 mm
78.7 mm
5.80
63 mm
2V2 in.
2.5
±1.9 mm
65.6 mm
66.2 mm
5.50
53 mm
2.12 in.
2.12
±1.6 mm
55.2 mm
55.7 mm
5.15
50 mm"
2 in.
2
±1.5 mm
52.1 mm
52.6 mm
5.05
45 mm
1% in.
1.75
±1.4 mm
46.9 mm
47.4 mm
4.85
37.5 mm
lVo in.
1.5
-1.1 mm
39.1 mm
39.5 mm
4.59
31.5 mm
IV4 Ln.
1.25
±1.0 mm
32.9 mm
33.2 mm
4.23
26.5 mm
1.06 in.
1.06
±0.8 mm
27.7 mm
2S.0 mm
3.90
25.0 mm"
1 in."
1
±0.8 mm
26.1 mm
26.4 mm
3.80
22.4 mm
V8 in.
0.875
±0.7 mm
23.4 mm
23.7 mm
3.50
19.0 mm
3/4 in.
0.750
±0.6 mm
19.9 mm
20.1 mm
3.30
16.0 mm
5/8 in.
0.625
±0.5 mm
16.7 mm
17.0 mm
3.00
13.2 mm
0.530 in.
0.530
±0.41 mm
13.83 mm
14.05 mm
2.75
12.5 mm"
V2 in."
0.500
±0.39 mm
13.10 mm
13.31 mm
2.67
11.2 mm
7/ig in.
0.438
±0.35 mm
11.75 mm
11.94 mm
2.45
9.5 mm
^8 in.
0.375
±0.30 mm
9.97 mm
10.16 mm
2.27
8.0 mm
5/i6 in.
0.312
±0.25 mm
8.41 mm
8.58 mm
2.07
6.7 mm
0.265 in.
0.265
±0.21 mm
7.05 mm
7.20 mm
1.87
6.3 mm"
V4 in."
0.250
±0.20 mm
6.64 mm
6.78 mm
1.82
5.6 mm
No. 3V2n
0.223
±0.18 mm
5.90 mm
6.04 mm
1.68
4.75 mm
No. 4
0.187
±0.15 mm
5.02 mm
5.14 mm
1.54
'From ASTM Ell (Reprinted, with permission)
'These standard designations correspond to the value for test sieve apertures recommended by the International
Organization for Standardization (ISO), Geneva, Switzerland.
-Only approximately equivalent to the metric values in column 1.
5The average diameter of the warp and of the shoot wires, taken separately, of the cloth of any sieve shall not deviate
from the nominal values by more than the following:
Sieves coarser than 600 ;im	5 percent
Sieves 600-125 M-m	~.5 percent
Sieves finer than 125 ;un	10 percent
"These sieves are not in the standard series, but they have been included because they are in common usage.
"These numbers (3!/-2—100} are the approximate number of openings per linear inch, but it is preferred that the sieve be
identified by the standard designation in millimetres or micrometres.
:: 1,000 jim = 1 mm.
Table continued next page.
-------
*
filtration
Many factors could account for the
good performance of coal-based GAC
biofilters, including the quantity
of biomass attached to the media.
Jack Z. Wang, R. Scott Summers,
and Richard J. Miltner
~
br
tr
processes
^cumulating a sufficiently large mass
i bring about desired reactions is the
treatment.1 The accumulation of
large ai^^^^^nbiomass in biological treatment
processes ^^^WWneved through cell retention in
bioreaciors. Because substrate concentrations are low
in drinking water biological treatment, cell retention is
usually accomplished through natural attachment to
solid surfaces as biofilms.1 Biofilm-type biological
processes used for drink -
A phospholipid analytical technique was used to measure the
amount of biomass attached to the surfaces of drinking water
filter media. The method was reproducible and able to detect
significant differences in biomass concentrations in different
filters and at various depths within filters. The amount of
attached biomass decreased as filter depth increased, suggesting
that most removal of natural organic matter occurred at the top
of the biofilters. The results show that granular activated carbon
media were able to hold more biomass than were anthracite and
sand media and that concentrations of biomass in anthracite-
sand filters were lower with chlorine in the backwash water.
ing water treatment in-
clude slow sand filtration
and rapid rate bio<ration
employing anthracite and
granular activated carbon
(GAC) media.
Cell attachment to
solid surfaces is a physi-
cal-chemical process and
is affected by the surface
properties of the cells and
support media. In drink-
ing water biofiltration
DECEMBER 1995 55
-------
Pilot-plant Alter design and operating conditions

Cokii—i
Fitter


LB.
De**
niter
Rtter
In.
fc.
Medte
1
6
20
Anthracite


10
Sand
2
6
20
Anthracite


10
Sand
3
6
20
Antnracite


10
Sand
4
1.5
20
Anthracite


10
Sand
5
1.5
30
Sand
6
1.5
26
GACl


4
Sand -
7
1.5
26
GMC2


4
Sand >;.
8
1.5
2ft
a*C3ft


~
>in ;-*r. .
OtaaMtar of Top I
IS*
1.02
1.02
1-02
1.02
132..
1_32"
X3S
132
Tue—
:5***5r -XL ?
jT -
Reproducibility of phospholipid analysis


Anthracite
Blank
Sand blank
GAC1
Blank
PC ffitw
Anthracite
BWCffitar
Anthracite
pcnitai
processes, the filter media supply the surface for cell
attachment; different filter media have different sur-
face textures and accessible areas, resulting in differ-
ent capacities for biomass attachment. For example, a
GAC media can accumulate more biomass than can a
sand or anthracite media. As a result, biological acti-
vated carbon (BAC) filters perform better than sand or
anthracite filters in terms of biological removal of nat-
ural organic matter (NOM). In the past 20 years, many
researchers have evaluated the efficiency of BAC fil-
ters for drinking water treatment.2-11 Most of these
siudies showed that at the same empty bed contact
time (EBCT). the efficiency of the BAC filters for NOM
removal in drinking water was greater than that of
sand or sand-anthracite dual-media filters.
The objectives of this study were to assess the
attached biomass concentrations in drinking water
biofilters under different operating conditions and
to examine the relationship between the attached
biomass and the removal of
disinfection by-product (DBP)
precursors and ozone DBPs.
To relate the amount of bio-
mass to the removal of NOM,
DBP precursors, and ozone
DBPs, an accurate me-
ment of biomass in bit
is very important. I.
study, phospholipid analysis
was used for measurement
of the biomass attached to
the surface of drinking water
filter media. Phospholipids
are contained in cell mem-
branes and are common to
all cells. They are not stored
in cells but are turned over
relatively rapidly during
metabolism.1' With cell death, the cellular enzymes
hvdrolyze and thus release the phosphate group
within minutes to hours of cell death.13 The rela-
tionship between phospholipid and viable biomass
has been described under marine conditions.13
White is investigating the relationship under drink-
ing water conditions.14
The total lipids can be easily extracted from the
cells attached on the surface of support media and
measured quantitatively,15 which is advantageous
because the biomass need not be preseparated from
the media. The phospholipid technique has been used
largely for determining biomass in sediments and
soils by microbiologists.16 The modification made by
Findlay et al.17 based on the work of White et al.1' '
made this technique easier to use and accessib"
variety of applications.
The phospholipid analysis has also been applied to
measurement of biomass in drinking water biofil-
Pi OM4I iMA.Mi
-------
ters.'1- 'g Using this technique.
Wang and Summers13 and
Swertfeger et al!l> evaluated
the biomass distribution in
bio filters and the impact on
MOM. DBP precursor, and
ozone DBP removal.
The study presented in this
article is only a portion of a
project conducted at a US
Environmental Protection
Agency (USEPA) pilot plant.10
In this project, the particle
removal efficiency in biofilters
with different types of filter
media was also examined and
was reported elsewhere.20-21
The impact of backwashing on
filter performance is discussed
in pan 2 of this article.22
Materials and methods
Pilot plant. Raw Ohio River water was trans-
ported from the Cincinnati Water Works to the pilot
plant located at the USEPA's Drinking Water Research
Division (USEPA-DWRD) in Cincinnati. The raw
water was stored in a 5,000-gal (18.900-L) tank. A
submersible pump circulated the water and kept sed-
iment in suspension. The water was then ozonated in
a single 6-in.- tl5.2-cm-) diameter, glass, counter-
current contactor with an ozone dose near 0.8 mg
transferred 0}/mg total organic carbon |TOC). After
ozonation, alum coagulation, flocculation. and sedi-
mentation were applied prior to filtration. The alum
dose was typically in the range of 10 to 50 mg/L. The
detention time of the sedimentation tank was 6.1 h.
Settled water turbidities were typically in the range of
I to 3 ntu. the pH ranged from 6.9 to 7.7, and the TOC
ranged from 1.1 to 2.2 mg/L. Raw Ohio River water
quality is given by Goldgrabe et al.21
Following sedimentation, the flow was split into
eight parallel glass filters (Table 1). Filters 1-4 were
identical dual-media (anthracite-sand), and Filter 5
was a single media (sand). Filter 6 was bituminous
coal-based GAC* and is referred to as the GAC1 fil-
ter. filter 7 was lignite coal-based GACf and is referred
to as the GAC2 filter, and filter 8 was wood-based
GACt and is referred to as the GAC3 filter. The media
are described in Table 1. Filters 1-3 were 6-in.- {15.2-
cm-) diameter columns, and filters 4-8 were 1.5-in.-
(3.8-cm-) diameter columns. The design filter veloc-
ity and depth were the same for all filters: 2 gpm/sq
ft (5 m/h) and 30 in. (76 cm), resulting in a total
EBCT of 9.2 min.
Filter 1 received chlorinated water continuously
and is referred to as the prechlorinated (PC) filter.
Filter 2 was exposed to chlorinated water only dur-
ing backwashing and is referred to as the backwash-
chlorinated l BWC) filter. Filter 3 was not exposed to
chlorine and is referred to as the nonchlonnated iMC)
filter. Ail of the 1.5-in.- (3.3-cm-) diameter filters.
including anthracite-sand (referred to as the A-S fil-
ter). sand, and three GAC filters, were also not
exposed to chlorine. Chlorine was applied prior to
the PC filter and after the BWC filter to yield a free
residual near 0.2 mg/L after three days in samples
taken from the clearwells and held to simulate sam-
ples from the distribution system. To accomplish this,
chlorine doses were in the range of 2.5 to 3.0 mg/L,
resulting in free chlorine residuals in clearvvell efflu-
ents and in backwash water near 1.0 mg/L.
Details on pilot-plant design and operation are
reported elsewhere.10-21 On day 0 (Aug. 12. 1991).
new media were placed in each filter at its design
depth, backwashed with Cincinnati tap water to
remove fines, and placed in service, receiving ozonated
and settled water. Thus, the time to build up biomass
and to achieve steady-state control of MOM, DBP
precursors, and ozone DBPs could be assessed. On
day 0, water temperatures were near 27°C in filters.
Minimum temperatures (near 13°C) were reached
near day 180. and temperatures returned to 27°C
after day 300.
Sampling and analysis. Samples were collected
after sedimentation and after biofiltration to be ana-
lyzed for TOC. biodegradable dissolved organic carbon
(BDOC), assimilable organic carbon i AOC), aldehy-
des, the formation potential of the sum of six
haloacetic acids 1HAA6FP). trihalomethane forma-
tion potential (THMFP). and total organic halide for-
mation potential (TOXFP). Six HAAs, including bro-
mochloroacetic acid, were quantitated. Samples of
filter media at the top of each filter were collected
for biomass analysis, except for the distribution study-
in which samples of filter media at different filter
depths were taken for biomass measurement. Dry-
clean anthracite, sand, and GAC media were used as
the biomass blanks.
'Filtrasorb 400, Calgon Corp.. Pittsburgh. Pa.
f.Norit Hvdrodarco 4000. ICI Americas. Wilmington. Del
:PlCA8lOL. PICA. Lcfvailois. France
DECEMBER 1595 57
-------
T
¦f bi
Standard methods were used forTOC, THM, and
TOX analyses.23 and USEPA method 552 was used
for HAA analyses. Formation potential samples were
chlorinated at a dose of 12-15 mg chlonne/L and
held seven days at 25°C. A modified Joret-Levi
method was used for BDOC measurement.24 25 AOC
was measured at USEPA-DWRD according to the
method of van der Kooij et al26 with one modifica-
tion: instead of heating the sample water at 60°C
for 30 min before inocula-
tion. tne sample was
heated to 70°C for 45 min
to ensure inactivation of
any bacteria present. Both
Pseudomonas PI7 and Spir-
illum NOX were employed
for the inoculation. Only
data for AOC-N'OX are pre-
sented because AOC-P17
concentrations in the
ozonated water were very
low and changes through biological filters were dif-
ficult to assess. Aldehydes were also measured at
USEPA-DWRD using the method described by Milt-
ner et al.27 This method was a modification of the
methods of Sclimenti et al2S and Glaze et al.29
For measuring the biomass attached to the surfaces
of filter media, the phospholipid extraction method
was employed.17 About 1 g of sand or anthracite
media or 0.5 g of activated carbon media with
attached microorganisms were taken from the filters
for extraction in 20-mL scintillation vials. Prior to
extraction, the samples were washed with dechlorin-
ated tap water to remove the suspended solids in the
samples. Therefore, what was measured was the
attached biomass. Phospholipids contained in the cell
membranes were extracted by a chloroform-me-
thanol-water mixture with a volume ratio of 1:2:0.8.
The extraction period was 2 to 24 h. After extrac-
tion. the mixture was separated into a lipid-contain-
ing chloroform phase ; '
methanol-water aqu
phase by the addition of
chloroform and water so
the final ratio of chloroform-
methanol-water was 1:1:0.9.
A portion of the lipid-con-
taining chloroform was trans-
ferred into a 5-mL ampule
and evaporated using - tro-
gen gas; 0.9 mL of pot.' :m
persulfate reagent was :n
added into the ampule. s
reagent was prepared bs
solving 5 g of potassium p-
sulfate in 100 mL of 0.36
sulfuric acid solution. The
extracted lipids were digested
with potassium persulfate
reagent at 102°C for 2 h to
release phosphate. The re-
leased phosphate was complexed with reagents of
ammonium molyDdate [2.5 percent (NH4)6Mo7024 ¦
4H:0 in 5.72 /V H2S04] and malachite green (0.011
g of malachite green in 100 mL of 0.111 percent
polyvinyl alcohol solution) to form a colored com-
pound and then was measured colorimetrically at a
wavelength of 610 nm. The absorbance of each sam-
ple was compared with a standard curve to derer-
mine the concentration of lipid phosphate. The,
suggests that most of the removal
biodegradable NOM will occur at the
top of the biofilters or, equivalently,
within a short EBCT.
dard solution was 0.1 mAf KH2P04. The concentration
range of the calibration curve was 0.0 jiiVf to 10.0 p.v/
of phosphate. All samples were analyzed at least in
duplicate. The amount of biomass was reported as
nmol lipid-P/g dry filter media; 1 nmol lipid-P is
equivalent to about 10g bacteria of the size of E. coli. '¦'
The detection limit of this method was determined to
be 0.5 \iM (1 nmol phospholipid in 2 mL volume).
Results and discussion
Reproducibility of phospholipid analysis. Sam
pies of four filter media representing a wide range of
attached biomass were used for testing the repro-
ducibility of the phospholipid analytical method. The
samples were anthracite media taken from
BWC. and N'C filters and GAC media from the|
filter. Each of the collected samples was well
and divided into four subsamples with different
weights. The phospholipid contained in the biomass
-------
Bloma*» accumulation on fitter media after 95 days
Anthracite PC
2.0
0.5
Anthracite BWC
6.0
0.6
Anthracite NC
55.0
1.7
SarxS
90.6
1.3
GAC1
305
»
GAC2
465
9
GAC3
382
11
•SO—Mandard deotadon
Biomass depth distribution In fitters after
on* year of operation
r*.r .
ISA- j
29JJ-' "
i
«r?i»
U»^v
•jsy*

attached to the media surfaces was extracted and
determined by the phospholipid analytical technique.
A near 100 percent lipid recovery by this technique
has been reported by Findlav et al.17 In this study,
the blank values were measured to be 1.9 ± 0.1 nmol
1 ipid-F/g media for clean anthracite media, 0.5 ± 0.1
nmol iipid-P/g media for sand, and 0.3 ± 0.2 nmol
lipid-P'g media for GAC. Table 2 lists the means, stan-
dard deviations, and coefficients of variation (CV) of
the sample measurement results. The results indi-
cate that this method is highly reproducible, as the CV
values are less than 4 percent when the biomass val-
ues are >5 nmol lipid-P/g media. The biomass con-
centrations reported in Table 2 and elsewhere in this
article were not blank corrected.
Biomass concentration, prechlorination, and
preozonation. The concentration of biomass attached
to the anthracite at the top of the PC, BVVC, and NC fil-
ters was assessed over the first six months of operation.
The results are shown in Figure 1, with the error bars
representing the standard deviation of replicate sam-
ples The results on day 95 are shown in Table 3. Bio-
mass concentration in the PC filter was not detected
because the values were the same as clean anthracite,
about 2 nmol lipid-P/g media. This was expected
because the influent was chlorinated and biomass was
not expected to build up in the chlorinated environ-
ment. The biomass in the BWC filter was about 6 nmol
lipid-P/g media after two to four months of operation
and built up to about 10 nmol
lipid-P/g media after six
months of operation. This
indicates that backwashing
about twice a week with chlo-
rinated water impaired the
biomass growth, compared
with that in the N'C filter.
The N'C filter showed con-
tinuous biomass accumula-
tion for the first five months,
before reaching a steady state
of about 80 nmol lipid-P/g
media, indicating that accu-
mulation of biomass concentration is a slow process.
The absence of contact with chlorine in the backwash
water yielded about an order of magnitude increase in
biomass. The results of biomass in the three GAC fil-
ters are shown in Figure 2. Figure 2 shows that the bio-
mass in these filters Increased slowly but yielded nearly
a 50 percent increase during the time between 100 and
200 days. However, after the influent water was
switched from ozonated to unozonated Ohio River
water, the biomass in all three filters decreased by
about 50 percent. The influent BDOC also decreased
by about 50 percent during this time, from about 0.4
to 0.2 mg/L. This suggests that the substrate (BDOC)
loading affects biomass growth.
The increase of biomass with filter operation time
has also been reported by other researchers. Servais
et al50 reported that in biologically active GAC fil-
ters, 100 days were needed for accumulation of bio-
mass before steady state was reached. Collins et al31
found that in slow sand filters there was a strong
positive correlation between the age of the
schmutzdecke (top layer of slow sand filters) and bio-
mass density using the acriflavin direct cell count and
folin-reactive material methods. They showed that
the increase of biomass density most likely resulted
from cell growth in the schmutzdecke rather than
from filtration of bacteria from the source water. They
also found a strong positive correlation between bio-
mass and NOM loading rate, indicating that NOM
loading affects biomass growth.
Impact of Alter media. The NC. sand, and three
GAC filters were operated without chlorination but
with different filter media. The results in Table 3 and
Figures 1 and 2 show that the amount of attached
biomass in each of these filters was different, which
may be attributed to the difference in the surface
properties of the filter media. GAC is a porous media.
It has a larger surface area and better surface texture
for biomass attachment than does sand or anthracite.
Therefore, GAC has the potential to hold a greater
amount of biomass than does sand or anthracite. The
highest biomass was measured in the GAC2 filter. It
was typically 100 nmol lipid-P/g media higher than
that on the other GACs.
The differences in attached biomass among the
three GACs may have resulted from their different
effective surface areas (suitable for biomass attach-
CECSMSE3 1995 59
-------
Percent removal of ozone DBPs, NOM. and DBP precursors in dual-media filters at steady state
•*]
PC
Fitter
awe
Fitter
NC
rare
PanuiMter
Mean
SO*
Km
SO*
Meea
SO* P

m Level
Acetaldehyde
-11
23
45
27
54
21
Yes

"ormaidefiyde
3
9
74
16
88
5
Yes

Giyoxat
¦2
12
28
18
97
2
Yes

Methyl glyoxal
0
15
66
23
97
2
Ye*

AOC-NOX
-25
44
37
12
43
12
Yes

BDOC
-15
55
32
32
19
28
no

TOC
3
5
16
9
20
0
Yes

THMFP
1
11
13
6
21
a
Yes

TOXFP
11
14
17
3
25
a
Yes-

HAA6FP
9
9
23
7
37
4
Yes

¦SO—standa/d deviation








Percent removal of NOM and DBP precursor* In biofilters at steady state


A-a
FMer
Send
mter
(AC1 P


Number of






Parameter 0
fteei»¦Uum
Me—
SO*
Meea
so*
Meea

TOC
13
16
9
20
9
29
8
AOC-NOX
9
39
26
43
20
51
23
THMFP
9
23
6
23
7
40
. S-
TOXFP
8
23
4
25
5
52
5=.
SD—standard deviation







ment). The GAC2 filter is a mesoporous GAC. The
GAC1 filter has the largest overall surface area of the
three. However, it is a microporous GAC, and microor-
ganisms may not be able to enter the small pores for
attachment. The GAC3 filter is macroporous with the
smallest total surface area of the three. A decrease
in media diameter will increase the external specific
surface area, which may explain the biomass differ-
ence between the sand (ES = 0.44 mm) and anthracite
(ES = 1.02 mm) media. Other physical-chemical
properties of the surface, such as charge and surface
tension, may also play important roles in biomass
attachment.
Biomass depth distribution. Samples of the
attached biomass distribution with depth in the A-S
and three GAC filters were taken at the completion
of the study, i.e.. after 345 days. In the last two weeks
of the study, the influent water was not ozonated.
The sampling depths were 0, 15. and 30 cm. Table 4
lists the'mean biomass results at each filter depth
from six measurements. The results were very repro-
ducible because the standard deviations were nor-
mally less than 10 percent of the mean. Statistical
analysis of the six measurements tat each depth)
showed that the biomass concentrations at each fil-
ter depth were significantly different at the 95 percent
confidence level. The results show that attached bio-
mass was always highest at the top of filters and
decreased as filter depth increased. The biomass at
the 15-cm depth iEBCT = 1.4 mini was 70 percent of
that at the top of the filters: at 30 cm iEBCT = 2.9 mini
it was 60 percent of that at the top. The concentrations
of biomass at 15-cm and 30-cm depths would likely
be smaller had the filters not been regularly
washed because backwashing causes some mi.-j
media with its attached biomass. Studies32 _
shown that the trends were similar for total biomass.
Moll et al32 reported that both attached and sus-
pended biomass were highest at the top of biofilters.
This suggests that most of the removal of biodegrad-
able NOM will occur at the top of biofilters or, equiv-
alently, within a short EBCT.
Similar results were also reported in other stud-
ies.930-31 Servais et al9-30 showed that the biomass
in biological GAC filters was highest at the top of the
filters and that the biomass at the bottom part of the
filters remained relatively constant. Collins et al31
reported that in slow sand filters most of the biomass
was distributed in the top 5 cm.
Removal of NOM and DBP precursors. Table 5
lists the performance results of the PC, BWC. and
N'C filters for aldehydes, AOC-NOX. BDOC. TOC.
THMFP, TOXFP, and HAA6FP after steady-state
removal was achieved. The time required for achiev-
ing steady-state removal was different with different
parameters, ranging from 2 to 99 days for the N'C fil-
ter.25 The means and standard deviations listed in
Table 5 are calculated from the data obtained from the
steady-state operation, i.e.. up to day 200. The results
show that in the PC filter, which received chlorinated
influent continuously, the removal was very
TOC and DBP precursors and can be attrib
particulate matter removal because there w
biological activity in the filter. The increase in aul-
XOX. BDOC. and some of the aldehyde concentra-
tions mav have resulted from prechlorination, whicn
-------
is known to increase AOC.25
Removal of the ozone DBPs,
N'OM, and DBP precursors in
the BWC and NC filters was
much higher than that in the
PC filter and can be attributed
to the biological activity that
occurred in these two filters.
Results of statistical analy-
ses between the BWC and NC
filters are listed in Table 5 and
indicate that except for
BDOC, the mean removals in
the NC filter were different
from those in the BWC filter.
The NC filter outperformed
the BWC filter, and the
amount of biomass was an
order of magnitude higher in
the NC filter compared with that in the BWC filter.
However, the removal of NOM. DBP precursors, and
ozone DBPs was not always appreciably greater in
the NC filter. This suggests that above a critical level,
the amount of biomass required to partially biode-
grade this organic matter is not strongly related to
removal. Lack of a linear trend between biomass and
NOM and DBP precursor removals was also observed
in the NC filter. As shown in Figure 1, biomass con-
centrations gradually increased from day 70 to day
190. Yet the NOM and DBP precursor removals in
the NC filter, given in Table 5, were consistent over
BtomaM and removal* of TOXFP and glyoxal in flltaf 3 (NC)
*
GAC media contained three to eight times
ore biomass than either anthracite
or sand.
that time; i.e., they did not increase with increasing
biomass. This is seen in Figure 3 for TOXFP removal,
which reached steady state near day 79. However,
glyoxal and methyl glyoxal removal showed a better
correlation with biomass. Their control was signifi-
cantly better in the NC filter than in the BWC filter,
and glyoxal behavior in the NC filter over time was
more likely biomass-dependent (Figure 3). Its removal
did not reach steady state until day 93; during the
latter pan of this period, biomass concentration was
increasing, but the highest concentration of biomass
had not yet been achieved (Figure 3). Thus, although
there is an association between higher biomass con-
centration and better control and removal of DBP pre-
cursors and ozone DBPs, this relationship is not clear.
A complication is that the amount of biomass is not
directly related to the activity of the biomass, whereas
the removal of substrates like DBP precursors and
ozone DBPs is likely related to microbial activity.
Microbial activity may be affected by temperature.
exposure to chlorine during backwashing, type of sub-
strate, and other water quality parameters.
The performance of the nonchlorinated A-S, sand,
and three GAC filters for TOC, AOC-NOX, THMFP,
and TOXFP after steady-state removal was achieved
from days 155 to 330 is compared in Table 6. The
performance of the filters was compared by deter-
mining whether the mean percent removals were sta-
tistically different. With the exception of AOC-NOX
removal and the GAC 3 filter compared with the A-S
filter and the sand filter, the performance of most fil-
ters was statistically distinguishable. The filter perfor-
mance pairings that were
not statistically different at
an alpha level of 5 percent
are listed in Table 7.
The sand filter had 30
percent more biomass than
did the A-S filter and
removed 25 percent more
TOC, but its removal of the
other parameters was not
statistically different. The GAC1 and GAC2 filters
were the most effective, with the GAC1 filter per-
forming better for DBP precursor removal. Biomass
concentrations in the GAC2 filter were, however,
about 50 percent higher than those in the GAC1 fil-
ter (Figure 2, Tables 3 and 4), indicating that the bio-
mass may not be the sole indicator of BAC perfor-
mance. The GAC's ability to adsorb and release
substrates and its possible residual adsorption capac-
ity for these target compounds may explain the bet-
ter performance of the GAC1 filter. During the begin-
ning of the run, when adsorption dominated removal,
the GAC 1 filter also had a better adsorption capacity
compared with that of the GAC2 filter.10
Biomass in the GAC3 filter was comparable to
that in the GAC1 and GAC2 filters, but its perfor-
mance was more like that of the A-S and sand filters;
i.e., it did not exhibit classical breakthrough curves like
the GAC1 and GAC2 filters, implying that the GAC3
had a low adsorption capacity.10 Although the surface
DECEMBER 1395 61
-------
3.	Miller, G.W. 6- Rice. R.G. European Water Treat-
ment Practices—The Promise of Biological Acti-
vated Carbon. Civil Engrg.—ASCE. 2:81:83
(1983).
4.	DiGl^no, F.A. Influence of Microbial Activity on
the Performance of GAC. Proc. 1980 AWWA
Ann. Conf., Atlanta, Ga.
5.	Maloney, S.W. et al. Bacterial TOC Removal
on Sand and GAC. Natl. Conf. on Envir. Engrg.,
ASCE, Minneapolis, Minn., July 14-16, 1982.
6.	Li, A.Y. &- DiGiano. F.A. Availability of Sorbed
Substrate for Microbial Degradation on Gran-
ular Activated Carbon. Jour. WPCF. 55:392
(1983).
7.	Bablon, G.; Ventresque, C.; & Alm, R.B. Devel-
oping a Sand-GAC Filter to Achieve High-Rate
Biological Filtration. Jour. AWWA. 80:12:47 (Dec.
1988).
8.	Prevost, M. et al. Full-Scale Evaluation of Bio-
logical Activated Carbon Filtration for the Treat-
ment of Drinking Water. Proc. 1989 WQTC.
Philadelphia, Pa.
9.	Servais, P. et al. Microbial Activity in GAC Filters
at the Choisy-le-Roi Treatment Plant. Jour.
AWWA. 83:2:62 (Feb. 1991).
10.	Mhjner. R.J. Er Summers, R.S. A Pilot-Scale Study
of Biological Treatment. Proc. 1992 AWWA Ann.
Conf., Vancouver, B.C.
11.	LeChevallier, M.W. et al. Evaluating the Per-
formance of Biologically Active Rapid Filters.
Jour. AWWA. 84:4:136 (Apr. 1992).
12.	Whtte, D.C. In Situ Measurement of Microbial
Biomass, Community Structure, and Nutritional
Status. Phil. Trans. Royal Soc. Lond. A, 344:59
(1993).
13.	Whtte, D.C. et al. Determination of Sedimen-
tary Microbial Biomass by Extractable Lipid Phos-
phate. Oceologia, 40:51 (1979).
14.	Whtte, D.C. Personal communication (1995).
15.	Vestal, J.R. &¦ White, D.C. Lipid Analysis in
Microbial Ecology: Quantitative Approaches to
the Study of Microbial Communities. Biosci., 39:8
(1989).
16.	Federle, T.W. Microbial Distribution in Soils. Per-
spectives in Microbial Ecology. (F. Megusar and M.
Gautar. editors.) Slovene Society for Microbiol-
ogy, Ljubljana, Yugoslavia (1986).
17.	Findlay, R.H.; King, G.M.; & Watung. L. Efficacy
of Phospholipid Analysis in Determining Micro-
bial Biomass in Sediments. Appl. & Envir. Micro-
biol.. 55:11:2888 (1989).
18.	Wang, J.Z. & Summers, R.S. The Evaluation of
Organic Matter and Disinfection By-product
Controlling Biofilters With Biomass and Bioac-
tivity Analyses. Proc. 1993 WQTC Conf.,
Miami, Fla.
19.	Swertfeger, J. et al. The Control of Ozonation
By-products by Biological Filtration. Proc. 1993
AWWA Ann. Conf., San Antonio, Texas.
20.	Goldgrabe. J.C. Panicle Counting as a Method of
Evaluating Conventional and Biological Filter
Performance. Master's thesis, Univ. of Cincin-
nati. Ohio (1993).
21.	Goldgrabe. J.C.; Summers. R.S.: & Miltner. R.J.
Particle Removal and Head Loss Development
in Biological Filters. Jour. AWWA. 85:12:94 iDec
1993).
22.	Miltner, R.J.; Summers, R.S.: & Wang. J.Z.
Biofiltration Performance: Part 2. Effects of
Backwashing. Jour. AWWA. 87:12:64 (Dec.
1995).
2 3. Standard Methods for the Examination of Water and
Wastewater. APHA, AWWA, and WPCF, Wash-
ington D.C. (17th ed.. 1989).
24.	Joret, J.C. et al. Rapid Method for Estimating
Bioeliminable Organic Carbon in Water. Proc.
1988 AWWA Ann. Conf., Orlando, Fla.
25.	Miltner, R.J. et al. The Responses of Biological
Filters to Backwashing. Proc. 1992 WQTC,
Toronto, Ont.
26.	VAN DER Koou. D.; Visser. A; & Hijnen, W.A.M.
Determining the Concentration of Easily Assim-
ilable Organic Carbon in Drinking Water. Jour.
AWWA. 74:10:540 (Oct. 1982).
27.	Miltner, R.J.; Rice, E.W.; &• Smith, B.L. Ozone's
Effect on Assimilable Organic Carbon, Disin-
fection By-products, and Disinfection By-prod-
uct Precursors. Proc. 1991 WQTC, Orlando, Fla.
28.	Sclimenti, M.J. et al. Ozone Disinfection By-
products: Optimization of the Derivatization for
the Analysis of Aldehydes. Proc. 1990 WQTC,
San Diego, Calif.
29.	Glaze, W.H.; Koga, M.; &¦ Cancilla. D. Ozonation
By-products 2: Improvement of an Aqueous
Phase Derivatization Method for the Detection of
Formaldehyde and Other Carbonyl Compounds
Formed bv the Ozonation of Drinking Water.
Envir. Sci. & Technol., 23:7:838 (1989).
30.	Servais, P.: Billen, G.; &• Bouillot, P. Biological
Colonization of Granular Activated Carbon Fil-
ters in Drinking Water Treatment. Jour. Envir.
Engrg.. 120:4:888 (1994).
31.	Collins, M.R. et al. Removing Natural Organic
Matter by Conventional Slow Sand Filtration.
Jour. AWWA. 84:5:80 (May 1992).
32.	Moll, D.M.; Wang J.Z.; Er Summers, R.S. NOM
Removal by Distinct Microbial Populations in
Biofiltration Processes. Proc. 1995 AWWA Ann.
Conf.. Anaheim, Calif.
About the authors: Jack Z. Wang is a research engi-
neer at Louisville Water Company. 435 S. Third St..
Louisville, KY 40202. A member of AWWA, he has BS and
MS degrees from the University of Science and Technology
of China (Hefei, People's Republic of China) and a PhD
from the University of Cincinnati (Ohio). R. Scott Summers
is associate professor. Department of Civil and Environ-
mental Engineering. University of Cincinnati, POB
210071. Cincinnati. OH 45221-0071. Richard J. Miltner
is an environmental engineer in USEPA's Drinking Water
Research Division, 26 W. Martin Luther King Dr.. Cincin-
nati, OH 45263.
nprev liars • oo= «•*
-------
for BOM and particle removal:
a critical review
The authors review key parameters and engineering variables
influencing biological filtration and identify
areas requiring further research.
Daniel Urfer, Peter M. Huck,
Stephen D.J. Booth,
and Bradley M. Coffey
ore stringent water qual-
widespread use of
attention on biological
. Benefits of biofOtration
include decrease of the potential for bacterial re-
growth, reduction of chlorinated disinfection by-
products (DBPs) formed during secondary disinfec-
tion, reduction of chlorine demand, and decrease of
the corrosion potential.1
Additionally, biofiltration
has the potential to con-
trol taste and odor-caus-
ing compounds and other
micropollutants of health
and aesthetic concern.2-5
Several authors have ex-
amined and reviewed
biological treatment of
drinking water.16-9
The use of biologically
active rapid filters for the
production of biologically
stable water has been
recognized in Western
-------
{ Summary of studies examining different media and contact times in biological drinking water fitters
• 7.3 'v"
'isf&umxtsi.
v->	^
¦A •' ' • £& '•	- . amftratf
Anthiid**"'/ ?C 8.2
. - <66 i261>- •	'
v.	-^and.". • ; -si!
-•- • -:•••> '»Pt4»: . . >:> ji V
2.16 3.6 •;; ;san«: -4-20*. £
if
,	?. (tf.4-2j)^ %*&?¦ Z

sources).
,v '•-'"«"v3"ii ( imonti> •>"
experiment"	i^.v
•lpmy»q«,»2^«nVli.' .-» . . £f£fc \'CV • '•- jt *+3$'
tOronallon occurred prior to coapM^' fkxxtdttion, a«ia»c6nie«t»tiw>..; ..
tfrnthmerfte-J i?'i 'A'
"Fresh and spent FWrasorb300, Cafcotf Corp.. Pttstwijv Pa, waslnwaflfrta
GAC.fBtar oortalned fresh Fatrasort 300; the other 66 .	J v.ii"-- ¦
SSTt*4 stndyfttamlnBd eihausted GAD, PK*B*X H120. PICA," LavaHots- ftanea.
•••Fraeti media were pieced Si alt HMrcs at the start of the study. Three types of
4000, O Americas. Wilmington, Del; and 0CAB1OL, PICA, tevBflols. France.
Water,
Study
DwaUoa
Ratio of
0, to TOC
mg/mg
Df*k
(cm [la.])
LeChevallier Pilot,
et al15 Swimming
River,
21 months
Krasner
Shawl: -'-v
Project '
' Water (SPW).'
6 months
Phase 2:
pilot,
Colorado
River
Water,
6 months
0.5 to 1.0 Anthracite
(preozonation)t (46 [18]}-
sand
(23 [91)-
gamet
(7.6 (3D;
GACt
(51 [20D-
sand
{25(101; .
fiAT* v' ^ "
(122 [48])
7.5
7.5-15
14
Rat*
(m/hr
Dechlorinated TOC, AOC UPC and GAt-sand
tapwatef -i- ,Pi7 flnd. conforms better than
•i.jjV .^-r -.anthractte-
-rfi ' - sdkwnt ' sarid for
"v" ' • :'rf -v: • AOC and
•tor VUrtti
6-li . total (Mior^
5 ' weekly atp •*
If:# 4
..TOC removal
MPi- r-
V I*'
(31-PQJ- ? feiJ
i-:.." •' ¦ 8«p# «
(20 [8ft
SAC**
(51 {20D-
aand (20 [8]
0.4
(preozonation) Anthracite
(51 [20])-
sand
(2018]);
GACtt
(51 (20])-
¦ sand '
.(20(81)
l-ftand
" S3, '
J5
r:.- - t ' i.i'. ¦ - ;v-
bettt h.
GAC-sand
compared
with
anthraclte-
sand,.
especially at
colder temp-
eratures; • ¦
wood-Cased
UACand
Coat-based
GAG
Demonstration- 0.5
pilot SPW. (preoronation)
6 months \ .

Pilot,
Mllte ties
VfiheiV
0.4.mg/L
i In effluent
. of second
£ 't&Z1'1
-12 montte- ^ cdntactor
(iO (81);
anthracite
(51(201)-
sand
(20 [3D:
GACtt-':
(51 [20])-
sand
(20 (8»
GAC .	.	.		^
(200[79]- i. 12 : • iO usinftnpn- DOC- y 0ucose mocfvbettpr
sand	.- cWprtnflled'7. '. "8O0C,' • 'wspfra- foratdilvJe
(20 [gJ); ^ *' -	?'/ .-mauAef«8>;V •> -wheiiai'v "
Ohio River/ (praozonation) .-|SXI30fr". i'M 				
12 months	--~i ' '' -
(2#!105r i	' • ' iiM tontii* -AOONOK	I
' sand - .; v ' ;	,^th50 T-; *, „TMv>a J W£U[. pi
(76 [30]);,	-.	- -TOX. fr,- / - : ^
HozalsM Bench,
etet7' ~ ; synthetic
{various
; N6M '
-------
Europe for at least two decades.10-12 Prior to this,
traditional biological treatment steps such as slow
sand filtration, bank filtration, and ground passage
had been used for many years. Following the early
investigations by Sontheimer et al,10-12 several stud-
ies showed that biologically active rapid filters, both
single stage or in sequence with other granular media
filters for panicle removal (second stage), could pro-
duce biologically stable water with low turbidity.6-13-15
The most relevant components causing biological
instability in drinking water are biodegradable organic
matter (BOM), NH4+, Fe2+, Mn2+, N02-, dissolved H2,
and several reduced species of sulfur.1 This article focuses
on BOM because it represents the biodegradable sub-
strate of most relevance in North America.16 In bio-
logically active filters, heterotrophic bacteria attached to
the medium as biofilm oxidize BOM and use it as an
energy supply and a source of carbon. Although para-
meters such as assimilable organic carbon (AOC) and
biodegradable dissolved organic carbon (BDOC) are
useful surrogates for BOM, each suffers from some lim-
itations when used in this way.1718 For this reason,
attention is currently focusing on quantifying the major
components of BOM. These include humic substances,
amino acids, carbohydrates, and, where appropriate,
ozonation by-products. For certain BOM components,
the possible microbial formation in biofilters may con-
siderably complicate performance assessments.19
A review of European practice of biological drink-
ing water treatment identified a variety of ap-
proaches.16 In Germany and the Netherlands, bio-
logical treatment is often achieved through slow sand
filtration, bank filtration, or ground passage. If rapid
filters, granular activated carbon (GAC) contactors, or
both are incorporated in the process, they are typically
installed following ozonation. In France, on the other
hand, biological treatment for BOM removal is tra-
ditionally performed in second-stage rapid GAC filters.
However, for many North American water utilities
considering the biological retrofit of existing plants or
the construction of new biological filters, separate
and sequential particulate and BOM removal may
not be feasible. Therefore, the multiple objectives of
both particulate and BOM removal must be achieved
within the same filter, referred to here as single-stage
biofiltration. Recent waterborne outbreaks of giar-
diasis and cryptosporidiosis have underscored the
importance of particle removal by water treatment
plants. Clearly, water utilities that operate their filters
biologically cannot compromise finished water tur-
bidity and particle removal.
This article reviews the key parameters and engi-
neering variables affecting both biological perfor-
mance (BOM removal) and conventional perfor-
mance in biological rapid filtration. A review of
traditional filtration in nonbiological filters is outside
the scope of this article and is available elsewhere.20-21
Biological performance
Effect of ozonation. It is well established that
ozonation increases the fraction of natural organic
matter (NOM) that is biodegradable.22-24 The effects
of ozonation on NOM include formation of hydroxyl,
carbonyl, and carboxyl groups, increased polarity and
hydrophilicity. loss of double bonds and aromaticity,
and a shift in the molecular weight distribution toward
lower-molecular-weight compounds.25-28
In many cases, the substantial increase in BOM
upon ozonation may encourage bacterial regrowth
in the distribution system, especially if ozonation is not
followed by biofiltration.29-31 In all but a few drink-
ing water situations, carbon (C) rather than nitro-
gen or phosphorus is the limiting substrate for bac-
terial growth, although this limitation is normally in
the role of C as an electron donor, rather than as C per
se. Thus, the increase in BOM upon ozonation gen-
erally considerably enhances biological activity in fil-
ters following ozonation.32 Often, biofiltration fol-
lowing ozonation can reduce BOM concentrations
to approximately preozonation levels,15-3 3-36 although
this depends on the specifics of the biofilters and
water quality parameters, and the composition of
BOM may be different after biofiltration. Currently
identified organic ozonation by-products include alde-
hydes, carboxylic acids, aldo acids, and keto
acids,27-37^0 which are all relatively easily biodegrad-
able, and good removals, i.e., >75 percent, are nor-
mally observed in biofilters.35-41-43
Effect of filter media. The selection of filter media,
often a central question when biofiltration is used, is
important because of its major cost implications. Exper-
imental investigations have examined both an adsorp-
tive medium (GAC) and nonadsorptive media
(anthracite and sand) for biological BOM re-
moval.154144-45 Little biogrowth seems to occur in the
GAC micropores because their small diameter (1-100
rum) does not allow the penetration of bacteria, which
typically have a diameter >200 nm.46-47 Consequently
the specific surface area (unit surface per unit volume
of filter) available for biomass attachment might be
higher in a sand filter compared with a GAC filter,
because the effective size of sand is usually smaller
than GAC. However, the macroporous structure and
irregular surface of GAC offer suitable bacterial attach-
ment sites, providing increased protection from shear
stress. In addition, GAC has the ability to adsorb and
remove potentially inhibitory chemicals and to adsorb
and retain slowly biodegradable components that can
be biodegraded by the attached bacteria, leading to
continuous bioregeneration of the GAC.46-48-50
LeChevallier et al15 compared biological GAC-sand
and anthracite-sand filters in parallel using preozona-
tion (Table 1). AOC removal at an empty bed contact
time (EBCT) of 7.5 min and a temperature of 3.5°C
averaged 75 percent for anthracite-sand and 86 percent
for GAC-sand at an average filter influent AOC con-
centration of 780 pg/L C. The pilot facility was shut
down during weekends, which might have influenced
the media comparison; other studies have indicated
that GAC-sand filters reestablish biological BOM
removal faster after an out-of-service period than
anthracite-sand filters.41-44 Total organic carbon (TOC)
-------
removals averaged 51 percent
for GAC-sand and 26 percent
for anthracite-sand with a raw
water TOC of ~3 mg/L. This
major difference in the efflu-
ent TOC concentrations sug-
gests that part of the removal
of TOC and AOC might be at-
tributable to adsorption on the
initially fresh GAC, as sug-
gested by LeChevallier and col-
leagues15 and others.51-52 Le-
Chevallier et al15 concluded
that the GAC-sand filter per-
formed better than the an-
thracite-sand filter for the
treatment of AOC and TOC.
However, given that the pilot
plant was routinely out of ser-
vice and that part of the BOM
and NOM removals might be
attributed to adsorption and
not biodegradation, the actual
differences between GAC and
anthracite as biofilm support
media were likely not as sig-
nificant as reported.
A pilot study conducted by
Krasner et al41 (Table 1)
showed that GAC-sand filters
established aldehyde removal
sooner and were more resis-
tant to temporary perturbations such as intermittent
chlorination and out-of-service periods. Additionally,
GAC-sand outperformed anthracite-sand in removing
glyoxal, a less readily biodegradable aldehyde.41 Nev-
ertheless, long-term AOC removal was comparable
for both media configurations. This study also showed
that a carbon-based GAC* and a wood-based GACf
yielded similar biological removals of formaldehyde
and glyoxal. In another study. Coffey et al44 (Table 1)
found that percentage removals of formaldehyde and
glyoxal were somewhat better with GAC-sand, espe-
cially at lower temperatures: in terms of absolute alde-
hyde removal, however, the two media often differed
by <1-2 pg/L. During the first three months of the
study, the removal of dissolved organic carbon (DOC)
was about 0.24 mg/L higher in the GAC-sand com-
pared with the anthracite-sand filters, suggesting that
some adsorption of organics or continuous bioregen-
eration may have occurred in the GAC, which was
new at the beginning of the study.
Another investigation compared parallel deep-
bed anthracite-sand and GAC-sand filters55 (Table
1). Formaldehyde removals at 10°C were about 50
percent in the anthracite-sand filter, compared with
more than 80 percent in the GAC-sand filters con-
taining exhausted GAC. Glyoxal was totally removed
in the GAC-sand filters, whereas the anthracite-sand
filter showed no glyoxal removal. From these results,
Prevost et al53 concluded that filter media choice
Amount of viable biomass (measured as phospholipid) at the top
of several filters
strongly influenced biological aldehyde removal.
However, the reported aldehyde removals in the
anthracite-sand filter operated at an EBCT of 13 min
appear to be low compared with removals in other
studies.41-44 Possibly the ozone |03) residual main-
tained at 0.4 mg/L in the effluent of the ozone con-
tactor led to the presence of 03 residuals in the filter
influents. This would impair the biological perfor-
mance of the anthracite-sand filter,14-54 whereas GAC
would rapidly reduce the 03 chemically.55 Given
these possibilities, the substantially better perfor-
mance of GAC over anthracite observed by Prevost et
al53 might be partially attributable to the presence of
a residual concentration of 03 in the filter influents.
Merlet et a!56 compared a wood-based GACt with
a macroporous structure and a common coal-based
GACf with a more microporous structure at full scale.
The concentration profiles of both DOC and BDOC
were not different in the two GAC filters with an
EBCT of 25 min; ammonia removal, however, was sig-
nificantly improved in the wood-based GAC. Wang et
al45 have presented data on the relationships between
the amount of biomass attached to the filter medium
(measured as phospholipid57-58) and the removal of
DBP precursors and ozonation by-products (Table 1).
In this study, several pilot-scale filters were run in
•Filtrasorb 300, Calgon Corp.. Pittsburgh Pa.
+PICABIOL, PICA. Levallois. France
iFilrrasorb 400. Calgon Corp.. Pittsburgh Pa
-------
parallel, and after 95 days of operation, major differ-
ences in the amount of biomass per gram of medium
were observed at the top of the different filters (Fig-
ure 1). One filter was bituminous coal-based GAC
(G^C 1), the second was lignite coal-based GAC
(GAC 2), and the third was wood-based GAC (GAC
3).* The three GAC filters showed similar amounts of
biomass per gram of medium, despite the substan-
tially higher effective size of GAC 3 (1.62 mm) com-
pared with GAC 1 (0.64 mm) and GAC 2 (0.68 mm).
Figure 1 also compares the amount of biomass
per unit volume of filter, calculated by the authors of
this article. This parameter, which is a function of
the apparent density of the medium, may be more
meaningful in terms of filter performance. Figure 1
shows that, with two exceptions, all of the biofilters
operating without prechlorination or chlorine (Cl2l in
the backwash water contained similar amounts of
biomass per unit volume of filter at the top. The
exceptions were the anthracite-sand filter, which
yielded an average value about four times lower, and
GAC 3, which yielded an average value about two
times lower. BOM removal was highest in the micro-
porous, coal-based carbon (GAC 1); despite the dif-
ferences in attached biomass, similar removals of
AOC-NOX, trihalomethane formation potential
(THMFP), and total organic halide formation poten-
tial were observed in the anthracite-sand and the
wood-based carbon filter (GAC 3).
This suggests that measuring biomass alone, partic-
ularly at the top of the filters, does not adequately esti-
mate the BOM removal capacity of biofilters. In addition,
these data imply that as long as the amount of biomass
is above some minimum level, it does not determine
the removal of BOM and DBP precursors and therefore
is not the rate-limiting factor, as others have also noted.5"
The amount of biomass above which biomass is not the
rate-limiting factor is likely temperature-dependent and
BOM-component-specific and may be lower for rapidly
biodegradable components than for more slowly bio-
degradable components. Additionally, the formation of
soluble microbial products60-61 may be an important
factor in the weak relationship between the amount of
biomass and effluent BOM concentrations. The sus-
pended biomass in the filter appears to represent a sub-
stantial fraction of the total biomass in certain cases62 and
may be responsible for a significant portion of BOM re-
moval. Moll et al62 also showed that the microbial com-
munity structure in biofilters changes with filter depth
under certain conditions.
Effect of contact time. A number of researchers
have shown that contact time significantly influences
BOM removal within biological filters.32-43 63-66 Con-
tact time, usually expressed as EBCT, is therefore a key
design and operating variable. Recently, modeling
has provided additional justification for this experi-
mentally observed finding. Zhang and Huck66 and
Zhang67 have introduced the concept of dimension-
*GAC 1. Filtrasorb 400. Calgon Corp.. Pittsburgh. Pa.: GAC 2. Norn
Hvdrodarco 4000, ICI Americas, Wilmington. Del.; GAC 3. PICAJ3IOL. PICA.
Levallois. France
-------
Summary of studies examining the effect of backwashing on biological filters
Ahmad and GAC-sand
Amirtharaiah88
Duration
Variable
Air plus
subsidization
water followed •
i • •	by water ••
Servals et al87 QAC	Alrscour
(second stage) backwash with air
-	backwash	, .
Miltner et Anthracite-sand Water (no alr scour) 10
'¦	•• •." '.vS-i	;>; •.
KOttnerM*5® - Anthractte-sarxJ
2
NA
20
10
an

; -	rjggvj.v . backwash
GcfttffUfi*'- Aflftracl»-sand Backwash
X'. -i-- QAC-sand	with
¦#?V	®B»V ? &26S -
ifl^KalB backwash > jOjS ¦ ,	LiaEgf&jj &
iAS'""> "¦ Nwmafrwe a 15 25ft .	?'
ckwash -M? 'v: "• . ..	«
kwash	-• lOtt ' SSM3T£ ;*-20'c^
Ih surface w»sh V,'; ^	&|gp«p
.. .. ¦ ..¦¦:•	»®^aas«eas;asB^
less contact time, which incorporates EBCT, specific
surface area of the medium, and parameters related
to substrate biodegradability and diffusivity. A sim-
plification of this—the normalized EBCT, equal to
the product of EBCT and specific surface area—may
represent a useful design parameter. Zhang67 has
shown that percentage removal of AOC increases
with increasing contact time up to a maximum value.
In practice, either filter depth or hydraulic loading
can be changed to increase the EBCT.
Several researchers have shown that contact time
and not hydraulic loading is the key parameter for
biological BOM removal and that for a given EBCT,
BOM removal is independent of hydraulic loading in
the range typically used in rapid filtration.43-68-71 This
empirical finding suggests that external mass trans-
fer may play a minor role in BOM removal under
conditions typical for drinking water treatment because
for a given EBCT the increase in flow velocity did not
increase the rate of removal. Because influent BOM
concentration and the amount of BOM removed were
observed to be directly proportional, the removal of
BOM can be approximated by a first-order model.64~-
Consequently, increasing EBCT will improve removals
but less than proportionally. Thus the incremental
benefit of using very long contact times is likely small,
which has been shown both theoretically and exper-
imentally.66-67 Depending on the objectives of biofil-
tration. the required contact time may be substan-
tially different. In general, removal of chlorination
by-product precursors, chlorine demand, or both seems
to require longer contact times than does removal of
ozonation by-products and AOC.T5
Merlet et al56 reported an exponential decrease
in the concentrations of DOC and BDOC in the efflu-
ent of a GAC filter as EBCTs increased from close to
zero to 25 min. This supports the pattern of less than
proportional increases of BDOC and DOC removal
with increasing EBCT. Nevertheless, another investi-
gation found an essentially linear increase of BDOC
removal with increasing EBCT between 10 and 30
min.74 This result conflicts with conclusions from
others who predicted or measured a less than pro-
portional increase.56-64-66
In a study conducted by Huck et al14 involving
two-stage biofiltration, total EBCTs >25 min were nec-
essary' to consistently achieve effluent AOC levels <20
yg/L C. Data presented by Hacker et al36 suggested
higher AOC removals (90 percent) at higher filter influ-
ent concentrations (-550 pg/L C). compared with AOC
removals averaging 80 percent at lower influent levels
(-300 pg/L C). Similar findings, i.e.. higher percentage
removals of BOM at higher influent concentrations,
were obtained in modeling work for BDOC removal.75
These findings conflict with the work of Huck and co-
workers64-^ underlying the first-order model, which
is predicated on a constant percentage removal, regard-
less of influent concentration.
-------
Other authors have ob-
served minor effects of
EBCT on BOM removal.
Price30 and colleagues76
observed little or no effect
of EBCT on AOC removal
in the range of 4.5-22.5
min. These data, however,
had a fair amount of scatter, and the biofilters may not
have been fully acclimated during the whole study, as
suggested by poor removals of some BOM components.
Hozalski et al77 have reported data from a laboratory-
scale investigation on biological treatment (Table 1).
Sand filters were fed with ozonated NOM cocktails
(2-3.6 mg 0,/mg TOC), and the authors reported no sig-
nificant differences in TOC removal between filters
operated at 4, 10, and 20 min of EBCT. However, the
operational mode of the biofilters in this study may
not have simulated rapid filtration conditions: (1) the
filters were not backwashed, possibly leading to a high
amount of biomass attached to the media. (2) the 03-
to-TOC ratio was high, possibly leading to the formation
of large amounts of easily biodegradable material, and
(3) the hydraulic loading was about two orders of mag-
nitude lower than is common in full-scale rapid filtra-
tion, resulting in low shear stress, which is favorable to
the growth of a thick biofilm.
Effect of backwashing. The success of biological
filtration requires that the amount of biomass be care-
fully managed during the filtration cycle and that bio-
mass losses during backwashing be controlled. Because
biofilters accumulate both biological and nonbiological
particles, the difference in the detachment of these
groups of particles during backwashing will influence
the optimization of backwash strategies for biofilters.
Various authors have emphasized the importance
of backwashing on long-term performance of biofil-
ters.6-53-78-81 Although some treatment plants use
nonchlorinated backwash water for their biofilters,
others are operated with chlorinated backwash water
either for every backwashing or intermittently. There-
fore, it is important to understand how chlorinated
backwash water affects BOM removal and the amount
of biomass in the filters. The effects of chlorinated
backwash water, as well as determination of the
necessity of air scour for biofilters to control excessive
head loss buildup and to maintain long-term perfor-
mance. are important issues with respect to retro-
fitting existing plants biologically because of the cost
implications. Major studies examining backwashing
of biofilters are listed in Table 2.
For nonbiological filters, backwashing with water
alone has been shown to be an inherently ineffective
process because of the limited collisions and abrasions
among fluidized particles.82 More recent research on
nonbiological filters has established that the best
removal of panicles during backwashing is achieved by
simultaneous use of air and water at subfluidization
velocities to achieve collapse pulsing conditions.83-84
Ahmad and Amirtharajah85 investigated the strength
of attachment of different panicles in lab-scale biofil-
ters by including hydrophobic and hydration forces
(Table 2). Their data indicated that biological panicles
(measured as heterotrophic plate counts and cellular
adenosine triphosphate) are held with greater force
than nonbiological panicles (measured as turbidity).
This finding suggests that optimum backwashing con-
ditions for the removal of accumulated nonbiological
panicles might not lead to an excessive loss of biofilm
during backwashing. Preliminary experimental evi-
dence has shown that backwashing under collapse
pulsing conditions did not impair AOC removal.86
Servais et al87 reponed on backwashing full-scale
second-stage GAC filters (Table 2). They observed no
major losses of biomass (as measured by 14C-glucose
oxidation) upon backwashing with air scouring and
found that the venical stratification of the biomass was
essentially unchanged. Others have reported similar
findings, i.e.. no major biomass losses, for filters back-
washed without air scouring.70 Lu and Huck88 found
that biomass losses (measured as phospholipid) dur-
ing backwashing at collapse pulsing conditions were
significant; the values of biomass were very low, how-
ever, compared with other literature data.45 89
'Miltner et al89 conducted a pilot study focusing
on the effect of Cl2 in the backwash water of anthra-
cite-sand biofilters operated at an EBCT of 9.2 min
(Table 2). Two filters were operated in parallel; the
first was backwashed with chlorinated water (-1 mg/L
Ci2), whereas the second filter utilized nonchlorinated
backwash water. The authors reported that biomass
concentrations (measured as phospholipid) at the top
of the filter were approximately one order of magni-
tude lower for the first filter compared with the sec-
ond filter. Nonetheless, differences in the removal of
formaldehyde and acetaldehyde were much less than
what would be expected given the large difference in
biomass, indicating that the amount of biomass was not
the rate-limiting factor for removal of these aldehydes.
In a more recent pilot study, Miltner et al59 (Table
2) observed differences in the effects of chlorinated
backwash water on biofilters, depending on whether
the chlorine in the backwash water was present as free
or combined Cl2. Free Cl2 ( — 1.6 mg/L) showed a
stronger inhibition of the removal of several BOM
components and surrogates compared with combined
R«Mardwn should
document tho point In tho
tutor cydo at which
samples mo coHoctod and
conatdor tho poiontlai
dynamic prscMMs wfthtn
ono cydo*
-------
Cl2 (-2 mg/L). Others have observed relatively good
BOM removals in anthracite-sand36 and GAC-sand33
filters, despite backvvashing with chlorinated water.
- With regard to the effect of Cl2 in the backwash water
of biofilters, the duration of the backwashing proce-
dure—i.e.. the period of Cl2 exposure (similar to the
contact-time concept for disinfection 1—is likely a rel-
evant factor, as is the Cl2 concentration in the back-
wash water. Thus, vigorous backwashing for a short
period of time might be preferable to a longer, less
powerful backwashing procedure if the backwash
water contains Cl2.
Several authors have observed a decrease in BOM
removal, particularly in the top part of biofilters,
toward the end of a filter cycle.42'53-90 Carlson et al90
measured a major decrease of the real contact time in
the top pan of a biofilter toward the end of a filter
cycle. They attributed the observed decrease in DOC
removal to the lower actual contact time caused by
floe loading. Prevost et al53 suggested that back-
washing was beneficial to the efficiency of biofiltra-
tion, and they hypothesized that the accumulation of
panicles and floes inhibited bacterial activity. The
effect of backwashing was observed by comparing
BOM removals at various filter depths prior to and fol-
lowing backwashing (Table 2). These authors
advanced several hypotheses for the observed decrease
in BOM removals toward the end of a filter cycle,
including a decrease in the actual contact time and the
effects of precipitated metals.
Effect of temperature. Theoretically, BOM
removal in biofilters would be expected to increase at
higher temperatures, because both microbial kinetics
and mass transfer are more rapid. These effects, how-
ever, may be influenced by the media type. Similar to
the work of Krasner et al,41 Coffey et al44 showed that
the time needed to reach steady-state removal of gly-
oxal was shorter at 20-25°C than at 10-13°C, especially
for anthracite-sand filters. They demonstrated that in
the case of glyoxal, the apparent steady-state removal
in a GAC-sand filter was significantly higher at warmer
temperatures. Similar data were reported by Daniel
and Teefy,91 who found that total aldehyde removals
were lower at temperatures between 9-15°C com-
pared with 15-25°C, particularly for anthracite-sand
filters. Servais et al74 and Bouillot et al92 reported
increased removals of BDOC in GAC filters at higher
temperatures (18-20°C). Both studies presented data
supporting others' results,45 89 which indicate that the
amount of biomass is not the rate-limiting factor for the
removal of BOM, under certain conditions.
Effect of oxidants in filter influents. In some
cases, oxidant residuals, e.g., 03, hydrogen peroxide
(H202), Cl2, and monochloramine (NH2C1), may be
present in biofilter influents, and their potential effects
on BOM removal are of concern. In this regard, the
filter media is important; GAC decomposes Cl2 and
other oxidants through a redox reaction at the surface
of the GAC.55-93 Therefore, biological activity can be
established in GAC filters even if the filter influent is
chlorinated.94 However, Cl2 in the influent of GAC fil-
ters can lead to a structural deterioration of the GAC.55
Additionally, Cl2 reacts with the GAC and different
adsorbed organics, e.g., phenolic compounds and ani-
lines. to form chlorinated organics not formed in the
liquid phase.95-98 The ability to decompose oxidant
residuals might be part of the reason why GAC has
been thought to significantly enhance biofiltration.
In full-scale biofilters, measuiable 03 residuals
would typically not be permitted in the influent be-
cause this could lead to harmful 03 concentrations in
the ambient air of filter buildings. However, for
enclosed filters (e.g., pressure filters and many pilot-
scale filters), this restriction does not exist, and there-
fore 03 residuals might be present in the influent.
Several authors have observed that 03 residuals of
-0.1-0.2 mg/L inhibited bacterial development in
pilot-scale anthracite-sand filters and therefore reduced
biofiltration performance.14 54 In the case of GAC fil-
ters, the data from Weinberg et al54 showed that the
biological aldehyde removal in such filters seemed to
be essentially unaffected by the presence of 03 resid-
uals, likely as a result of the rapid reduction of the
03 on GAC. Kaur et al99 and Kaur and Bott100 showed
that approximately 90 percent of a P. fluorescens biofilm
grown on a glass surface was removed by exposure to
0.07 mg/L 03 for a period of 6 h.
H202 in combination with 03 is used for advanced
oxidation or to accelerate 03 decay,101-104 and influ-
ents of biofilters might contain H202 residuals in cer-
tain cases.105 H202 has been shown to be bactericidal
at high concentrations and bacteriostatic at concen-
trations in the low millgram-per-litre range; never-
theless, its disinfection effectiveness is low compared
with that of other disinfectants such as free Cl2.106-108
The continuous presence of ~1 mg/L H202 in the
influent of a lab-scale anthracite-sand filter did not
inhibit the removal of several easily biodegradable
components, and the H202 was rapidly destroyed in
the filter.105
Chloramine residuals of -1 mg/L in the influent of
anthracite-sand filters inhibited the biological removal
of aldehydes and AOC.109 About 3 mg/L free Cl2 in
the influent of a GAC filter had virtually no effect on
the biological processes because the Cl2 was rapidly
reduced, whereas NH2C1 at the same concentration
was more stable.15 Although filter effluent NH2C1
concentrations averaged 0.1 mg/L, AOC continued
to be removed. The biocidal effects of free Cl2 and
NH2C1 on biofilms appear to differ significantly because
NH2C1 does not react with extracellular polysaccha-
rides, whereas free Cl2 possibly does.110111 Cl2 used
as backup disinfectant in case of an ozone system
failure or to control algal growth in sedimentation
basins might be periodically present in biofilter influ-
ents; how this affects BOM removal is currently not
well understood.44
Modeling. In the last several years, researchers
have tried to model the removal of BOM or specific
components in biological filters to provide a frame-
work for interpretating and generalizating results,
which currently suffer from site-specificitv.5112 113
-------
Summary of selected mechanistic models applicable to drinking water biofiltration

Rlttmann and
McCarty114
Wanner and
Gojer113 .
Bfflen et'aP*
. 7i • jv ,
Wang and
Summers®
Zhang anda ,
Huck®6 'V '
: . • .r... ••
ft a&Wtttf pAyidbfttemiwfr
Mass transport of substrate to
blofilm, simultaneous dtffusion
and bioreaction of substrate
In Mofllm, definition of S^,
General model with steadystate
or dynamic cases possible,
mass transport of substrata' '
to blofilm, dlfftalon and' " • '
btofBaebon of substrate, jn.-
biofUm. biomass exchange'.,.
wftnbuKlkjukf •'	•;$»
-UOIPBKW aUIKTVOent OT . ^
bacteria tbiosynthesis ofEPSj."
bactsiM mortality and -v:"
grazing by protozoa	t
Mass transport erf substrate to
blofiim wtth substrate otfflzation
on biofSm surface, deBnitjon
of rapkSyi slcMrty. and . •
nonbiodegradable DOfi fractions
biomass dependence on fitter' '
Ao^a&* S ^ ^
• wtetrata." deflation of,
- ' 		r
Substrate must be "-; . .
grawtWImmng, . -
predation not tjlractty
considered -' .
BiofUm biomass
. treated as continuum,
""one spatial
•dtmeoslon .dJi'S?f£
dlmertstootesj EBCT

v,« /T vO	'
" ¦*" •??&'}' •
a ¦. ¦ ¦• • • -* •-•
? X io- •
¦'-hc-:
Ttyn blofltm. expression
of biomass
dapehdence ' 'i*:
^$p*,0ep#i,..
; b« kno*n ^ .;
¦ .. ' V.1K?'VX
ii ¦	a -xt '4 *Vs
ipCtegrowtMrnfOng
, prestation, not
'consHwod,
f... •;rw"*,<
Eg
The following section summarizes the practical util-
ity of the most widely known models for drinking
water applications (Table 3i Table 3 is not exhaustive,
however, and a detailed critical review of the differ-
ent approaches is outside of the scope of this article.
In a seminal study, Rittmann and McCarty114 in-
troduced a steady-state biofilm model that considers
microbial (Monod) kinetics and mass transport (Fick's
second law). Their approach assumes that all nutri-
ents are in excess except the rate-limiting substrate S.
An important contribution from these authors was the
introduction of the concept of Smin, which is defined
as the bulk substrate concentration below which a
steady-state biofilm cannot exist.
Several studies have shown that the principles of
the steady-state biofilm model are applicable to the low
substrate, i.e., oligotrophic. conditions typically en-
countered in drinking water biofiltration.66-67 1 15 1 16
Zhang and Huck66 and Zhang67 applied the steady-
state bicfilm approach to modeling AOC removal in
jbiofilters. As mentioned, the important development
in this model was the definition of an index for con-
tact time that incorporates actual contact time, specific
surface area of the medium, and the ease of diffusion
and biodegradation of BOM. This parameter allows
comparison of results among studies and for differ-
ent surrogates of BOM (AOC or BDOCl.
Several multispecies biofilm models with generally
similar structures117 have been developed.117-119 The
model of Wanner and Gujer"9 was initially devel-
oped for wastewater biofilms; it is very general, how-
ever, and may be adapted to many types of microbial
interactions. These authors divided biomass into three
subgroups: heterotrophs, autotrophs, and inert par-
ticulate material. They also noted that substrate
removal was strongly dependent on the phenomena
that limit biofilm growth: such phenomena include
reactor configuration or different biomass detach-
ment mechanisms (i.e., shear and sloughing).
Billen et ai75 developed the CHABROL model,
which predicts BDOC removal. The model introduced
the concepts of dividing BDOC into three subtrac-
tions with different biodegradabilities and took into
consideration the major microbiological processes.
For two of the twelve parameters required for this
model, no precise values are available, and they have
to be estimated. One of these two parameters is the
constant kd, which describes the rate of bacterial mor-
tality. It is expected to vary both seasonally and from
plant to plant; therefore, the use of this approach
would require considerable data collection in the fil-
ters being modeled. An examination of CHABROL
model predictions shows that BDOC removal is essen-
tially directly proportional to influent BDOC at a spe-
-------
cific EBCT. particularly if the existence of Smin is
assumed and model predictions for BDOC influent
concentrations close to zero are ignored. This sup-
ports the empirical approach of Huck et al,64 which
showed that BOM removal within a biofilter can be
approximated by a first-order process.
Other researchers have modeled biofiltration by
dividing BOM into easily and slowly biodegradable
fractions and allowing biomass to decrease with filter
depth.65-70 This model provides predictions of bio-
mass and DOC concentration versus filter depth. Mea-
sured and predicted biomass profiles indicated a major
stratification, with a difference in biomass of one
order of magnitude between the top of a sand filter
and at a depth of 5 cm or more. This result appears to
conflict with full-scale data and model predictions
from others.75-87
The previously discussed models provide important
process insights for drinking water biofilters. Never-
theless, they are relatively complex and cannot be
directly used by utilities because they require para-
meters that are not readily available. An empirical
modeling approach useful for predicting approximate
performance for practical purposes has been devel-
oped.6472 This approach showed that the amount of
BOM removed in a given biofilter was directly pro-
portional to the influent concentration. Thus removals
can be approximated as a first-order process, and a
biofilter at apparent steady state will essentially
achieve a constant percentage removal at a given
EBCT and temperature. The relationship has been
shown to hold for AOC, BDOC, THMFP, chlorine
demand, and carboxylic acids.64120
All of the models discussed previously address
steady-state conditions. In practice, single-stage biofil-
ters may not reach steady state for BOM removal
because of the loss of biomass during backwashing and
the gradual replenishment of biomass during the sub-
sequent filter cycle. The extent of this non-steady-
state behavior has not yet been demonstrated exper-
imentally. However, simulations performed with a
recently developed non-steady-state model predict
that the removal of an easily biodegradable compo-
nent (acetate) declines sharply immediately follow-
ing backwashing but that the biofilter rapidly recov-
ers and nearly complete acetate removal occurs for the
majority of the filter cycle.121 Removal predictions
for more slowly biodegradable material (e.g., NOM)
exhibit a more pronounced sawtooth pattern through-
out the filter cycle, implying that the amount of bio-
mass is more critical.
Conventional performance
of biological filters
Turbidity removal. Several investigations have
generally indicated that biological filters provided
good removal of turbidity and reliably met current
guidelines. One pilot study showed no difference in
effluent turbidities among prechlorinated. backwash-
chlorinated. and biological (nonchlorinated) anthra-
cite-sand filters.122 Nevertheless, major differences
in particle counts were observed among the filters. 1 _
another study, the quality of a laboratory-scale biofil-
ter effluent at the end of a 48-h filter run indicated
that biofiltration did not impair turbidity, which was
always <0.4 ntu.85 Other authors observed no dif-
ferences in turbidity removal between biological
GAC-sand and anthracite-sand filters at pilot scale,
and both types of filters averaged effluent turbidities
<0.2 ntu.15-41 At full scale, biological anthracite-sand
and GAC-sand filters on average produced an efflu-
ent turbidity <0.1 ntu.42-44
Particle removal. The available literature on
panicle removal, particle detachment, and panicle
size distribution indicates some differences between
biological and nonbiological filters. The current
understanding of these phenomena is limited, how-
ever, especially in biofilters, in which the number of
bacterial cells is considerably higher than in con-
ventional filters. Several authors have shown that
the detachment of particles from nonbiological fil-
ters plays an important role in changing the particle
size distribution through a filter (influent versus
effluent).123-124
During the study conducted by Goldgrabe et al.122
the prechlorinated filter consistently outperformed
the biological filters in all panicle size ranges exam-
ined (1-150 |im), yielding a 0.4-0.5-log better total
particle removal, although no differences in turbid-
ity were discerned. Goldgrabe and co-workers attribj
uted the better performance of the prechlorinated
filter to either improved panicle destabilization by Cl2
in this filter or panicle generation by biological activ-
ity in the biofilters. Coffey et al44 reponed average
particle removals of 2.2-2.3 logs in biofilters and
effluent levels of 3-26 particles/mL. These results
are similar to data obtained in conventional full-
scale filters in which particle removals were in the
range of 1.3-2.5 logs.125
In GAC filters, Servais et al69 observed that total
bacterial counts in the effluent were higher during the
period of bacterial colonization than under steady-
state conditions. These authors suggested that colo-
nization was accompanied by the delayed develop-
ment of a protozoan population, which at steady state
efficiently removed some of the produced bacterial
biomass. Based on this, they noted that GAC filters
designed for adsorption, i.e.. with regular regenera-
tion of the GAC, cannot escape the process of bacte-
rial colonization and therefore possibly release more
bacteria than biologically equilibrated filters.
Head loss buildup and filter run length. A bio-
logical GAC-sand filter backwashed without air scour-
ing (Table 2) yielded a 40 percent increase in clean-
bed head loss between startup (essentially no biomass)
and three months later (steady-state biofilm).44 The
same trend was observed for the head loss accumu-
lation rate, which increased 50 percent over the same
three-month period. Given that both turbidity and
temperature did not vary to a major extent during the
study period, the data suggest that biofilm accumu-
lation may have resulted in increased head loss.44
-------
Goldgrabe et al1-2 found inat head loss buildup was
higher in two biological filters (nonchlorinated and
backwash-chlorinated) compared with a conven-
tional prechlorinated filter After a 14-week period of
acclimation, terminal head loss was reached in the
nonchlorinated. backwash-chlorinated, and prechlo-
rinated filters after about 90. 100, and 144 h, respec-
tively. Although the nonchlorinated filter contained
about seven times more biomass at the top than the
backwash-chlorinated filter did, the head loss did not
differ substantially.
Ahmad and Amirtharajah55 indicated that water
wash at a bed expansion of 60 percent is an insuffi-
cient backwashing procedure for biofilters because
head loss increased in successive filter runs How-
ever. when the filters were backwashed using col-
lapse pulsing followed by water wash with 20 percent
bed expansion, head loss increase in subsequent fil-
ter runs was negligible. Coffey et al42 confirmed that
clean-bed head loss increases over time when biofil-
ters are backwashed without air scouring. Other
authors have reported a major increase in head loss
accumulation in a GAC-sand filter backwashed with
water containing 2 mg/L Cl2. compared with an iden-
tical filter backwashed without Clj.94 The researchers
gave no explanation for this unexpected observation.
Restarting biological filters. Pilot-scale results
have shown no substantial difference between bio-
logical and nonbiological filters in terms of turbidity
removal except that the effluent turbidities at the
beginning of filter ripening varied.9 The initial tur-
bidity peak was much higher for biofilters (GAC-sand
and anthracite-sand) compared with a conventional
prechlorinated anthracite-sand filter. In laboratory
studies, endotoxin levels of biofilter effluents were
greatly increased for a brief period following restart
of a filter that had been shut down for 2 h.9 These
authors noted that the health significance of the ele-
vated endotoxin levels was unknown and that fur-
ther investigation was required. Possible remedies
for full-scale plants may be to operate all biofilters,
regardless of plant flow rate, or to backwash filters
before returning them to service. Additional research
is needed to assess the necessity of such operational
strategies. Other authors reported that filter shut-
downs of between 4 and 6 h. followed by back-
washing, represented an effective strategy to con-
trol the proliferation of higher organisms, i.e.,
annelids, in biological GAC filters.126
Summary and research needs
Most of the European literature on biological treat-
ment of drinking water focuses to a great extent on
second-stage biofiltration. In contrast. North Ameri-
can literature primarily emphasizes filtration that
combines microbial activity with an existing phvsic-
pchemical unit operation, referred to here as single-
stage biofiltration Single-stage filtration appears to be
the most economical approach to implementing a
biological process in drinking water treatment and
was the focus of this article
The widely reported increase of BOM upon ozona-
tion may lead to regrowth problems in the distribution
system if ozonation is not followed by biofiltration.
Thus ozonation and biofiltration should be considered
as a coupled process, rather than two independent
process steps. Anthracite-sand and GAC- sand filters
appear to provide similar average BOM removals, and
different studies have shown that biofiltration can be
successfully implemented in anthracite-sand filters.
Nevertheless. GAC-sand filters seem to provide better
aldehyde removals at colder temperatures, establish a
BOM-removing biofilm more rapidly, provide increased
protection against oxidant residuals in the filter influ-
ent, and permit faster reestablishment of BOM removal
after periods out of service. In addition. GAC-sand
showed better DOC and TOC removals compared with
anthracite-sand, possibly because of slow adsorption
processes or continuous bioregeneration. This indicates
that optimized filter media selection for biofilters depends
on site-specific characteristics such as water quality
(i.e., BOM composition and water temperature) and
the plant's specific operational issues. In general, wood-
based GACs and coal-based GACs differed only slightly
in terms of BOM removal and the amount of biomass.
Contact time, usually expressed as EBCT, is a key
vanable for biological BOM removal. Ln general, BOM
removals increase with increasing EBCT in a less than
proportional way. Therefore, EBCTs above a certain
value may not be economically justifiable. Different
investigations have shown that for a given EBCT,
removal of BOM is essentially independent of
hydraulic loading. Consequently, external mass trans-
fer appears to play a minor role in BOM removal in
drinking water biofilters. Some studies have shown
acceptable removals of ozonation by-products at
EBCTs as short as 2-4 min, whereas the removals of
BDOC, chlorine demand, and chlorination by-prod-
uct precursors appear to require considerably longer
EBCTs In general, full-scale biofilters are designed
for EBCTs below -30 min, and single-stage filters
usually have lower EBCTs (I to -15 min) than sec-
ond-stage filters (10 to -30 min)
Available literature suggests that backwashing bio-
logical filters with or without air scouring may not lead
to major losses of attached biomass. This empirical ob-
servation seems to be Ln agreement with fundamental
considerations, because bactena appear to attach more
strongly to the media than nonbiological particles do.
CKin the backwash water substantially decreases the
amount of biomass in anthracite-sand filters, although
BOM removal capability is much less affected, espe-
cially removal of the rapidly biodegradable fraction
(e.g.. formaldehyde and AOC). This suggests that no
linear relationship exists between attached biomass and
the removal of different BOM components and that
the amount of biomass is not the limiting factor for
BOM removal above some minimum level of biomass.
Thus, the acceptability of Cli in the backwash
water of biofilters may depend on treatment objec-
tives If the objective is the removal of relatively eas-
ily biodegradable components (e g . ozonation by-
-------
products or AOC). Cl2 in the backwash water might
be acceptable. This may not be true tf biofiltration is
intended to remove chlorine demand, chlorination by-
product precursors, or both Recent results have
shown a decrease in BOM removal performance
tov\ard the end of a filter cycle and an improvement
immediately following backwashing The reason for
this inhibition is currently not fully understood, al-
though one proposed explanation is that the accu-
mulation of particles and floes on the filter medium
leads to a decrease in the actual contact time.
Several studies have indicated that BOM removal
increases with increasing temperature, an expected
finding given that both mass transfer and biodegra-
dation kinetics are favored at higher water tempera-
tures. Inhibition of biological activity in anthracite-
sand filters has been observed at 03 concentrations of
-0.1-0.2 mg/L in filter influents, whereas H202 resid-
uals of -1 mg/L in the influent of an anthracite-sand
filter did not impair the removal of several easily
biodegradable components. The periodic and con-
tinuous presence of Cl2 and NH2Cl residuals in biofil-
ters inhibited the biological activity, especially in
anthracite-sand filters.
Available BOM removal models include mecha-
nistic and empirical models and generally assume
steady-state conditions. Because of their complexity,
however, mechanistic models cannot be readily used
by water utilities for filter design or optimization.
Recent modeling work addresses the possible non-
steady-state (dynamic) conditions of single-stage
biofilters resulting from the periodic loss of biomass
during backwashing.
Several investigations have shown good turbid-
ity removal m biofilters, indicating that such filters can
reliably meet current turbidity guidelines. Biological
and nonbiological filters do not seem to differ in terms
of effluent turbidity, and approximately equal tur-
bidity removals have been observed in biologically
operated anthracite-sand and GAC-sand filters. Nev-
ertheless. some literature data indicated that biofil-
tration led to higher total particle counts, compared
with conventional filtration. Compared with con-
ventional filters, biofilters appear to exhibit increased
head loss buildup, possibly caused by the attached
biomass Compared with conventional filters, biofil-
ters also showed higher turbidity peaks during filter
ripening and increased concentrations of endotoxin
following a short period out of service. In addition,
backwashing immediately following an out-of-ser-
vice period helped to control the excessive buildup of
higher organisms, e.g., annelids, in biofilters.
The authors have identified several areas war-
ranting further research.
•	Well-established criteria have not been estab-
lished for maximum BOM concentrations entering
the distribution system, even though this issue rep-
resents one of the most important factors affecting
future design and operation of biofilters
•	The removal of BOM components and chlori-
nation b\ -product precursors does not linearly depend
on the amount of attached biomass Therefore, a bet-
ter understanding is needed of the relationships
among the amount of biomass, its metabolic func-
tion, and the degradation of different BOM compo-
nents. In addition, further research must be under-
taken to define the role of suspended biomass within
biological filters, as well as the practical significance
of microbial community structure with biofilter depth.
•	Optimum backwashing criteria must be estab-
lished for biological single-stage filters with the dual
goal of particle and BOM removal. Such criteria may
vary with temperature and the amount and nature of
BOM present Their establishment requires a deeper
understanding of the significance of biomass loss dur-
ing backwashing in order to assess the extent to which
biofiltration for BOM removal should be considered
a non-steady-state process. Other areas requiring fur-
ther study are the effects of C12-NH2C1 in the back-
wash water and the control of head loss increase.
•	Biofiltration for BOM removal requires the as-
sessment of biological performance on a short time
scale, i.e., within a single filter cycle. Research needs
to uncover the reasons for observed decreases in BOM
removal performance toward the end of a filter cycle.
•	The effects of the continuous and periodic pres-
ence of oxidant residuals, i.e., 03, H202, and Cl2, in
biofilters must be better understood.
•	In the case of some BOM components (e.g.,
aldehydes, carboxylic acids), microbial degradation
and formation may occur in parallel under certain
conditions. The way in which these compounds are
metabolized should be further investigated.
•	No mechanistic BOM removal models exist that
are readily available to water utilities for filter design
or process optimization and control. Such BOM
removal models based on the available approaches
need to be developed and should consider the non-
steady-state behavior of biofilters.
•	The removal of particles in biofilters requires
funher research because the literature indicates some
differences between biological and nonbiological fil-
ters, possibly attributable to the different nature of par-
ticles, i.e., bacterial cells versus inorganic particles.
Phenomena such as particle detachment and particle
size distributions in biofilter effluents are not well
understood and should be investigated.
•	Full-scale operation of biofilters necessitates the
establishment of guidelines regarding the response
of these filters to events such as filter shutdown and
subsequent restarting.
Researchers pursuing these or other biofiltration
investigations may want to consider the recommen-
dations in the sidebar on page 87. These recommen-
dations derive, in part, from issues that arose during
the authors' attempt to interpret results of certain
studies included in this review article.
Acknowledgment
This research was partially funded by the AWWA
Research Foundation under project RFP-2'52, with
Project Officers Joe Roccaro and Ann Scarritt The au-
-------
thors appreciate the contributions of Appiah
Amirtharajah and Edward J. Bouwcr. who served as
co-investigators in the project. Financial support was
also provided by the National Sciences and Engi-
neering Research Council of Canada (NSERC) Chair
in Drinking Water Treatment, Department of Civil
Engineering, University of Waterloo. Waterloo, Ont.
Partners in the chair are NSERC. the City of Brantford
(formerly the Brantford Public Utilities Commission),
the Regional Municipality of Ottawa-Carleton. the
Regional Municipality of Waterloo, the Windsor Util-
ities Commission. Water Technology International,
Conestoga-Rovers and Associates Ltd.. Hewlett-
Packard Canada Inc., and the University of Waterloo.
Partial funding for the first author was provided
through a scholarship from the Swiss National Sci-
ence Foundation and the Swiss Academy of Engi-
neering Sciences.
References
1. Rittmann, B.E. & Huck, P.M. Biological Treat-
ment of Public Water Supplies. CRC Crii. Reviews
in Environ Control. 19:119(1989).
2 Lundgren, B.V.; Grimvall, A.; & Savenhed, R.
Formation and Removal of Off-flavour Com-
pounds During Ozonation and Filtration Through
Biologically Active Sand Filters. Water Sci. d> Tech-
no!., 20:8/9:245 (1988).
3.	Manem, J.A. Er Rittmaxx. B.E. Removing Trace-
level Organic Pollutants in a Biological Filter.
Jour. AWWA, 84.4:152 (Apr. 1992).
4.	Rjttmann, B.E. Transformation of Organic Micro-
pollutants by Biological Processes. Quality and
Treatment of Drinking Water. (J. Hrubec, editor).
The Handbook of Environmental Chemistry,
5B:31 (1995)
5.	Rrn.MANN. B.E.: Gantzer. C.J.: & Montiel. A. Bio-
logical Treatment 10 Control Taste and Odor
Compounds in Drinking Water Treatment
Advances in the Control of Tastes and Odors in Drink-
ing Water (M. Suffet and J. Mallevialle, editors).
. AWWA. Denver (1 995).
6 Boi'wer. E.J &• Crowe, PB Biological Processes
in Drinking Water Treatment. Jour. AWWA,
SO 9.S2 (Sept 1988).
7. Boiaver, E.J.: Hozalski, R.M.; &- Goel. S. RemovaJ
of Natural Organic Matter in Biofilters (90670).
AWWARF, Denver (1995).
8	Rittmaxx, B.E Fundamentals and Application
of Biofilm Processes in Drinking Water Treat-
ment. Quality and Treatment of Drinking Water (J.
Hrubec. editor). The Handbook of Environmen-
tal Chemistry. 5B:61 (1995).
9	Huck. PM et al Design of Biological Processes
for Organics Control (90722) AWWaRF. Denver
M997).
|l 0 SovTHEtMER, H. et al. The Mtilheim Process. Jour
.4U1\',4, 70-7-393 (July 1978)
11 Sontheimer. H. Applying Oxidation and Adsorp-
tion Techniques. A Summary of Progress Jour
JUir.4, 71:1 1.612 (Nov. 1979)
12 Sontheimer, H. Design Criteria and Process
Schemes tor GAC Filters Jour	71 11 6IS
(Nov 1979).
13.	Rittmaxx, B.E. &¦ Snoeyixk, V.L Achieving Bio-
logically Stable Drinking Water Jour AWWA,
76.10'106 (Oct. 1984).
14.	Huck, PM.; Fedorak. P.M.; & Anderson. W.B.
Formation and Removal of Assimilable Organic
Carbon During Biological Treatment Jour AWWA.
83:12:69 (Dec. 1991).
1 5. LeChevallier, M.W. et al. Evaluating the Per-
formance of Biologically Active Rapid Sand Fil-
ters. Jour AWWA, 84:4:136 (Apr. 1992).
16.	Huck, P.M. Use of Biological Processes in Drink-
ing Water Treatment—Review of European Tech-
nology. Report to the Biotechnology Research
Institute, Montreal (1988).
17.	Huck, P.M. Measurement of Biodegradable Or-
ganic Matter and Bacterial Growth Potential in
Drinking Water. Jour AWWA, 82:7:78 (July
1990).
18.	Woolschlager, J. & Rittmann, B.E. Evaluating
What is Measured by BDOC and AOC Tests. Rev.
Sci. Eau. 8:371 (1995).
1	9. Booth S.D.J, et al. A Mechanistic Approach for
Modeling the Removal of Ozonation Byprod-
ucts in Biologically Active Filters. Proc. 1995
AWWA WQTC. New Orleans.
20. Amirtharajah, A. Some Theoretical and Con-
ceptual Views of Filtration. Jour AWWA. 80:12:36
(Dec. 1988).
2	1. Cleasby, J.L. Filtration. Water Quality and Treatment
(F.W. Pontius, editor). McGraw-Hill, New York
(4th ed., 1990).
22.	vax der Koou, D , Visser. A.; & Hunen, W.A.M.
Determining the Concentration of Easily Assim-
ilable Organic Carbon in Drinking Water. Jour
AWWA. 74:10:540 (Oct. 1982).
23.	Werner, P. & Hambsch. B. Investigation of the
Growth of Bacteria in Drinking Water. Water Sup-
ply. 4:3:227 (1986).
24 Servais, P.; Bille.n, G.; & HascoEt, M.C. Deter-
mination of the Biodegradable Fraction of Dis-
solved Organic Matter in Waters. Water Res.,
21:4.445 (1987).
25.	Lienhard, H. & Sontheimer. H. Influence of
Process Conditions on the Effect of Ozone Treat-
ment of Organic Substances in Water. Ozone Sci
CEngrg., 1:61 (1979)
26.	Amy. G.L., Kuo. C.J , & SrERKA. R.A. Ozonation of
Humic Substances: Effects on Molecular Weight
Distributions of Organic Carbon and Tri-
halomethane Formation Potential. Ozone Sci &
Engrg.. 10.39 (1988)
27 Glaze, W.H & Weinberg. H.S. Identification and
Occurrence of Ozonation By-products in Drink-
ing Water (90625). AWWARF, Denver (1993).
28. Goel, S., Hozalski. R.M : & Bouwer. E.J.
Biodegradation of Natural Organic Matter: Effect
of NOM Source and Ozone Dose. Jour AWWA.
871:90 (Jan. 1995).
-------
29. van der Koou. D . Hijn'en. VV.A.M : & Kruithof,
J.C Effects of Ozonation. Biological Filtration,
and Distribution on the Concentration of Easily
Assimilable Organic Carbon in Drinking Water
Ozone Sci & Engrg, 1 1:297 11989).
30 Price, M L. Ozone and Biological Treatment for
DBP Control and Biological Stability (90649).
AVVWARF, Denver (1994).
31.	LeChevallier, M.W.; Welch, N.W.; & Smith, D.B,
Full-Scale Studies of Factors Related to Coliform
Regrovvth in Drinking Water. Appl & Envir Micro-
biol., 62:7:2201 (1996).
32.	DeWaters, J.E. & DiGiano, F.A. Influence of
Ozonated Natural Organic Matter on the Bio-
degradation of a Micropollutant in a GAC Bed.
Jour. AWWA, 82:8:69 (Aug. 1990).
33.	Reckhow, D.A. et al. Control of Disinfection By-
products and AOC by Preozonation and Biolog-
ically Active In-Line Direct Filtration. Proc. 1992
AWWA Ann. Conf., Vancouver, B.C. (1992).
34.	Harward, C. & Croll. B. AOC Control in the
Treatment of Eutrophic Surface Waters Using
Ozone and GAC. Proc. 11th Ozone World Con-
gress, San Francisco (1993)
35.	Swertfeger, J.W. et al. Control of Ozonation
By-products by Biological Filtration. Proc. 1993
AWWA Ann. Conf., San Antonio, Texas.
36 Hacker, P.A. et al. Production and Removal of
Assimilable Organic Carbon Under Pilot-Plant
Conditions Through the Use of Ozone and PER-
OXONE. Ozone Sci OEngrg. 16:197 (1994).
37.	Xie, Y. 6- Reckhow, D.A. A New Class of Ozona-
tion By-products: Ketoacids. Proc. 1992 AWWA
Ann. Conf., Vancouver, B.C.
38.	Andrews, S.A. Organic By-product Formation
From the Ozonation and Chlorination of Aquatic
Natural Organic Matter. Doctoral thesis. Uni-
versity of Alberta. Edmonton, Alta. (1993).
39.	Andrews, S.A. & Huck, P.M. Using Fractionated Nat-
ural Organic Matter to Quantitate Organic By-prod-
ucts of Ozonation. Ozone Sa. & Engrg, 16:1 (1994).
40.	Najm, I.N. & Krasner, S.W. Effects of Bromide
and NOM on By-product Formation. Jour
AWWA, 87:1:106 (Jan. 1995).
41 Krasner, S.W.; Sclimenti, M.J.; &- Coffey, B.M.
Testing Biologically Active Filters for Removing
Aldehydes Formed During Ozonation. Jour
AWWA. 85:5.62 (May 1993K
42. Coffey B.M. et al. Comparison of Backwashing
Techniques for Biological Filters: Effects of Air
Scour. Proc. 1996 AWWA WQTC, Boston.
43 Wang, J. & Summers, R.S. Biodegradation Behav-
ior of Ozonated Natural Organic Matter in Sand
Filters. Rev Sci Eau. 1:3 (1996).
44.	Coffey, B.M. et al. Comparison of Biologically
Active Filters for the Removal of Ozone By-prod-
ucts, Turbidity, and Particles. Proc. 1995 AWWA
WQTC, New Orleans.
45.	Wang, J.Z.; Summers, R.S.: & Miltner, R.J. Biofil-
tration Performance1 Pan 1, Relationship to Bio-
mass Jour AWWA, S7 12 5 5 (Dec. 1995).
46	AWWA Research and Technical Practice Com-
mittee on Organic Contaminants An Assess-
ment of Microbial Activity on GAC. Jour .-Uv'lv;-!,
738.447 (Aug 1 9S 1)
47	Werver. P. Mikrobiologische Untersuchungen
der AJaivkohiefilter zur Trinkwasseraufbereitung.
Publication No. 19, Department of Water Chem-
istry, Engler-Bunte Institute, University of Karl-
sruhe, Karlsruhe, Germany (19S2).
48.	Nayar, S.C & Sylvester. N.D. Control of Phenol
in Biological Reactors by Addition of Powdered
Activated Carbon. Water Res., 13:201 (1979).
49.	Chang, H.T. &- Rnr.MA.VN, B.E. Mathematical Mod-
eling of Biofilm on Activated Carbon. Envir Sci.
& Techno!., 21:3:273 (1987).
50.	Li, A.Y.L. & DiGiano, F.A. Availability of Sorbed
Substrate for Microbial Degradation on Granular
Activated Carbon. Jour. W?CF, 55:4:392 (1983).
51.	van der Koou, D. Effect of Treatment on Assim-
ilable Organic Carbon in Drinking Water. Treat-
ment of Drinking Water for Organic Contaminants
(P.M. Huck and P. Toft, editors). Pergamon Press,
New York (1987).
52.	Maloney, S.W. et al. Bacterial TOC Removal on
Sand and GAC. ASCE Jour Envir. Engrg .
1 10:3:519 (1984).
53 Pr£vost, M. et al. Removal of Vanous Biodegrad-
able Organic Compounds by First- and Second-
Stage Filtration. Proc. 12th Ozone World Con-
gress, Lille, France (1995).
54. Weinberg. H.S. et al. Formation and Removal
of Aldehydes in Plants That Use Ozonation. Jour
AWWA. 85:5:72 (May 1993).
55 Sonthei.mer, H.; Crittenden, J.C.; & Summers,
R.S. Activated Carbon for Water Treatment DVGW-
Forschungsstelle of the Engler-Bunte Institute,
Universitv of Karlsruhe, Karlsruhe. Germany
(1988)
56.	Merlet. N et al. Enlevement de la Matiere
Organique dans les Filtres CAB. Rev. Sci. Eau.
(No special) 5.143 (1992).
57.	Findlay, R.H.; King, G.M.; & Watllng, L. Efficacy"
of Phospholipid Analysis in Determining Micro-
bial Biomass in Sediments. Appl & Emir Micro-
biol., 55:1 1.2888 (1989).
58.	Miltner. R.J. et \l. Response of Biological Fil-
ters to Backwashing. Proc. 1992 AWWA WQTC,
Toronto.
59 Miltner, R.J et al. Comparative Evaluation of
Biological Filters. Proc. 1996 AWWA WQTC,
Boston.
60.	Namkung, E. & Rittmann, B.E. Soluble Micro-
bial Products (SMP) Formation Kinetics by
Biofilms. Water Res , 20:6:795 (1986).
61.	Carlson, K.H. et al. Importance of Soluble
Microbial Products in Drinking Water Biofiltra-
tion. Proc 1996 AWWA WQTC, Boston.
62.	Moll, D.M.: Wang, J.Z ; & Summers, R.S. NOM
Removal by Distinct Microbial Populations in
Biofiliration Processes. Proc 1995 AWWA Ann.
Conf., Anaheim, Calif.
-------
63	Servais. P.. Avzil. A . &- Ventresque. C Simple
Method for Determination of Biodegradable Dis-
solved Organic Carbon in Water. Appl e1 Envir
Microbiol ."55 10:2732 119S9).
64	Huck, P.M., Zhang. S.; & Price, M.L. BOM
*
Removal During Biological Treatment: A First-
order Model. Jour.	S6:6 61 (June 1994).
65	Wang, J & Summers R.S. A Heterogeneous Biofil-
tration Model for Natural Organic Matter Uti-
lization. Proc. 1995 AWWA Ann. Conf.. Ana-
heim, Calif
66.	Zhang, S. &• Huck, P.M. Biological Water Treat-
ment. A Kinetic Modeling Approach Water Res..
30:5 1195 (1996).
67.	Zhang, S. Modeling of Biological Drinking Water
Treatment. Doctoral thesis. University of Alberta,
Edmonton. Alta. (1996).
68.	Sonthelmer, H. &- Hubele. C. Use of Ozone and
Granular Activated Carbon in Drinking Water
Treatment. Treatment of Drinking Water for Organic
Contaminants (P.M. Huck and P. Toft, editors).
Pergamon Press, New York (1987).
69.	Servais P.; Billen. G.; & Bouillot, P. Biological
Colonization of Granular Activated Carbon Fil-
ters in Drinking Water Treatment. ASCE Jour.
Envir Engrg., 20:4:888 (1994).
70.	Wang, J. Assessment of Biodegradation and
Biodegradation Kinetics of Natural Organic Mat-
ter in Drinking Water Biofilters. Doctoral thesis.
University of Cincinnati, Cincinnati (1995).
I?l. Carlson, K.H. & Amy, G.L. Relative Importance
of EBCT and HLR on the Removal of BOM Dur-
ing Biofiltration. Proc. AWWA 1995 WQTC, New
Orleans.
72.	Huck, P.M. & Anderson. W.B. Quantitative Rela-
tionships Between the Removal of NVOC, Chlo-
rine Demand, and AOX Formation Potential in
Biological Water Treatment. Vom Wasser, 78.281
(1992).
73.	Prevost, M. et al. Chlorine Demand Removal
by Biological Activated Carbon Filtration in Cold
Water. Proc. 1990 AWWA WQTC, San Diego.
74 Servais P. et al. Pilot Study of Biological GAC
Filtration in Drinking Water Treatment. Jour
Water SRT-Aqua. 41:3 163 (1992).
75.	Billen, G. et al. Functioning of Biological Fil-
ters Used in Drinking Water Treatment: The
CHABROL Model. Jour Water SRT-Aqua 41:4:231
(1992)
76.	Price, M.L. et al. Evaluation of Ozone/Biologi-
cal Treatment for Disinfection By-produns Con-
trol and Biologically Stable Water. Ozone Set. &
Engrg.. 1 5:95 (1993)
77.	Hozalski, R.M.; Goel. S.; & Bouwer, E.J. TOC
Removal in Biologically Active Sand Filters: Effect
of NOM Source and EBCT. Jour AWWA. 87:12.40.
(Dec 1995)
S Bablov G.P.; Ventresque. C.; & Ben Aim, R.
Developing a Sand-GAC Filter to Achieve High-
rale Biological Filtration Jour AWWA, 80:12'47
(Dec 1 988).
79.	Camper. A.K. et al.' Operational Variables and
the Release of Colonized GAC Panicles in Drink-
ing Water. Jour AWWA. 79:5,70 (May 1987).
80.	Graese. S.L . Snoeyt.nk. VL: & Lee. R.G. Granu-
lar Activated Carbon Filter-Adsorber Systems.
Jour AWWA. 79 12:64 (Dec 19S7).
81 Chipps. M.J.; Bauer, M.J.; & Bayley. R.G. Achiev-
ing Enhanced Filter Backwashing With Com-
bined Air Scour and Subfluidizing Water at Pilot
and Operational Scale. Filtration t? Separation,
1-55 (1995).
82.	Amirtharajah, A. Optimum Backwashing of Sand
Filters. Jour Envir Engrg—ASCE, 104:5:917 (1978).
83.	Amirtharajah, A. et al. Optimum Backwash of
Dual Media Filters and GAC Filter-Adsorbers
With Air Scour (90584). AWWARF, Denver 1991.
84.	Amirtharajah, A. Optimum Backwashing of Fil-
ters With Air Scour: A Review, Water Sci & Tech-
no!.. 27:195 (1993).
85.	Ahmad, R. & Amirtharajah, A. Detachment of
Biological and Nonbiological Panicles From Bio-
logical Filters During Backwashing. Proc. 1995
AWWA Ann. Conf., Anaheim, Calif.
86 Ahmad, R. et al. Optimum Backwashing Strate-
gies for Biological Filters. Proc. 1994 AWWA
WQTC, San Francisco.
87.	Servais, P. et al. Microbial Activity in GAC Filters
at the Choisy-le-Roi Treatment Plant. Jour.
AWWA, 83:2.62 (Feb. 1991).
88.	Lu. P. & Huck, P.M. Evaluation of Methods for
Measuring Biomass and Biofilm Thickness in
Biological Drinking Water Treatment. Proc. 1993
AWWA WQTC, Miami.
89 Mlltner. R.J.; Summers. R.S.; & Wang, J.Z. Biofil-
tration Performance: Part 2, Effect of Back-
washing Jour AWWA, 87:12:64 (Dec. 1995).
90.	Carlson, K.H. et al. Ozone-induced Biodegra-
dation and Removal of NOM and Ozonation By-
products in Biological Filters. Advances in Slow
Sand and Alternative Biological Filtration (N. Gra-
ham and R. Collins, editors). John Wiley &¦ Sons,
Chichester, Great Britain (1996).
91.	Daniel, P. & Teefy, S. Biological Filtration: Media,
Quality, Operations, and Cost. Proc. 1995 AWWA
Ann. Conf, Anaheim. Calif
92.	Bouillot, P. et al. Elimination du Carbone
Organique Dissous Biodegradable Durant la Fil-
tration Biologique sur Charbon Actif en Grains.
Rev Sci Eau, (No special) 5:33 (1992).
93.	Boere, J.A. Reduction of Oxidants by Granular
Activated Carbon Filtration Proc. 10th Ozone
World Congress, Monaco (1991).
94.	DiGiano, F.A. et al. Microbial Activity on Filter-
Adsorbers (90606). AWWARF, Denver (1992).
95.	Snoeyink, V.L. et al. Organic Compounds Pro-
duced by the Aqueous Free-Chlorine-Activated
Carbon Reaction Envir. Sci. & Techno!., 15:188
(1981).
96.	McCreary. J.J ; Snoeyink. V.L.: & Larson, R.A.
Comparison of the Reaction of Aqueous Free
Chlorine With Phenolic Acids in Solution and
December 1997
-------
Adsorbed on Granular Activated Carbon Envir.
Sci. ^ Technol. 16:339 (1982)
97 Voudrias. E.A.; Larson, R.A . & Snoeyink, V.L
Effects of Activated Carbon on the Reactions of
Free Chlorine With Phenols. Envir Sci & Tech-
nol. 19:441 (19S5).
98.	Hwang, S.-C.; Larson, R.A.: &- Snoeyink, V.L.
Reactions of Free Chlorine With Substituted Ani-
lines in Aqueous Solution and on Granular Acti-
vated Carbon. Water Res., 24:4:427 (1990).
99.	Kac.
-------
Chapter
15
Water Fluoridation
Thomas G. Reeves, P.E.
National Fluoridation Engineer
U.S. Public Health Service
Centers for Disease Control
Atlanta, Georgia
Fluoridation of public water supplies has been practiced since 1945.
Few public health measures have been accorded greater clinical and
laboratory research, epidemiologic study, clinical trials, and public at-
tention than has water fluoridation. This chapter will present an over-
view of the history and public health and engineering aspects of fluo-
ridation.
Fluoridation is the deliberate adjustment of the fluoride concentra-
tion of a public water supply in accordance with scientific and medical
guidelines. Fluoride, a natural trace element, is present in small but
widely varying amounts in practically all soils, water supplies, plants,
and animals, and is a normal constituent of all diets. The highest
concentrations in mammals are found in bones and teeth. Virtually all
public water supplies in the United States contain at least trace
amounts of fluoride from natural sources.
History
The discovery of the relationship between fluoride in drinking water
and dental health has an interesting and intriguing history. The se-
ries of studies that led to a demonstration that fluoridated water had
caries-inhibitory properties was one of the most extensive programs
carried out in the epidemiology of chronic disease. It began in 1901,
when" a U.S. Public Health Service (USPHS) physician stationed in
Naples, Italy, wrote that black teeth observed in emigrants from a
nearby region were popularly believed to have been caused by using
933
-------
934 Chapter Fifteen
water charged with volcanic fumes. It was later determined that the
water supply contained an extremely high amount of fluoride and that
everyone drinking it was afflicted with discolored (or "mottled") teeth,
a condition referred to as dental fluorosis.
In its mildest form, dental fluorosis is characterized by very slight,
opaque, whitish areas on some posterior teeth. As the defect becomes
more severe, discoloring is more widespread, and changes in color
range from shades of gray to black. In the most severe cases, gross cal-
cification defects occur, resulting in pitting of the enamel. In some of
the latter cases, teeth are subject to such severe attrition that they
wear down to the gum line, and complete dentures must be obtained.
In 1916 Dr. Frederick S. McKay, a practicing dentist, reported that
many of his patients in Colorado Springs, Colo., had this defect.2 After
further study, McKay later concluded that the condition was caused
by an undetermined substance in the drinking water. McKay recom-
mended that the water supply of Oakley, Idaho, be changed because of
the high incidence of such dental defects among the children there.
The supply was changed in 1925 to a nearby spring that had been
used by a few other children whose teeth were not discolored.
Several years elapsed before the cause of dental fluorosis was dis-
covered. Almost simultaneously, two different groups of scientists
working independently with different tools and methods identified the
causal agent. A. W. Petrey, a chemist who was head of the testing di-
vision of the laboratory at the Aluminum Company of America at
Pittsburgh, noticed the calcium fluoride band in a spectroscopic exam-
ination for aluminum in a water sample from Bauxite, Ark. The chief
chemist of these laboratories, H. V. Churchill, reported in 1931 that
similar examinations of water samples from areas where dental
fluorosis was endemic invariably showed the presence of fluoride.3
At almost the same time, Drs. H. V. Smith, M. C. Smith, and E. M.
Lantz at the University of Arizona reported the cause of mottling by
duplicating the condition in rats by feeding concentrated naturally
fluoridated water and comparing the results with the mottling ob-
served when a diet high in fluorides was used.3 The strikingly similar
appearance of the mottling uncovered a long-standing mystery.
On April 10,1931, an abstract of Churchill's report appeared in Indus-
trial and Engineering Chemistry. This was the first printed notice of the
possible relationship between fluoride and dental fluorosis. The Septem-
ber issue of the Journal of the American Water Works Association
(AWWA) carried Churchill's complete paper.4 Churchill found that en-
demic regions of mottling had waters containing 2 mg/L or more fluoride,
while those areas without mottling had water supplies with less than 1*0
mg/L. This division of fluoride waters was confirmed by the Smiths u
Arizona, who reported that water sources from nonendemic mottling &
eas contained less than 0.72 mg/L fluoride.8
Soon ;
signed t
fluorosis
investigj
containii
containii
from Noi
linois, In
British \
South Ai
Throug
varying t
that less
nottling.
cities witl
of this stu
able relat
ies incidei
1.	When t
ther ini
and fill
and sev
2.	At a fl.
curs—n
cant mc
percent
3- At fluor
caries rt
¦ decrease
provemf
Irinki^^w
Rapids, Mi
Jrantford,
ared to a r.
-------
Water Fluoridation 935
Soon after this, Dr. Clinton T. Messner, head of the USPHS, as-
signed the dentist Dr. H. Trendley Dean to do research on dental
fluorosis. Dean began carefully prepared and executed epidemiologic
investigations. He confirmed that many localities have water supplies
containing fluoride. Areas with the largest number of such supplies
containing the highest levels of fluorides include those states running
from North Dakota to Texas, those along the Mexican border, and Il-
linois, Indiana, Ohio, and Virginia. Similar supplies are found in the
British West Indies, China, Holland, Italy, Mexico, North Africa,
South America, Spain, and India.
Through observation of thousands of children in communities with
varying fluoride levels, Dean established what he termed a mottled
enamel index—a numerical method for measuring the severity of
fluorosis.6 Using this index, he established the fluoride level below
which the use of such water contributed no significant discoloration.
This"level in the latitude of Chicago was about 1.0 mg/L.
Many investigators, including McKay, observed during the 1920s
that less decay occurred in children whose teeth were afflicted with
mottling. In order to confirm this, Dean examined 7257 children in 21
cities with water supplies containing varying fluoride levels. Results
of this study, some of which are shown in Fig. 15,1, revealed a remark-
able-relationship between waterborne fluorides and fluorosis and car-
ies incidence.7 Three conclusions were drawn from Dean's study:
1.	When the fluoride concentration exceeds about 1.5 mg/L, any fur-
ther increase does not significantly decrease the decayed, missing,
and filled (DMF) tooth incidence, but does increase the occurrence
and severity of mottling.
2.	At a fluoride concentration of about 1.0 mg/L, the optimum oc-
curs—maximum reduction in caries with no aesthetically signifi-
cant, mottling. At this level DMF tooth rates were reduced by 60
percent among the 12- to 14-year-old children.
3.	At fluoride concentrations below 1.0 mg/L, some benefits occur, but
caries reduction is not as great and decreases as the fluoride level
decreases until, as zero fluoride is approached, no observable im-
provement occurs.
Studies on fluoride were interrupted by World War II, but in 1945
and 1947, four classic studies were initiated with the intent to dem-
onstrate conclusively the benefits of adding fluoride to community
drinking water.8-11 Fluoridation began in January 1945 in Grand
Rapids, Mich.; in May 1945 in Newburgh, N.Y.; in June 1945 in
Bradford, Ont.; and in February 1947 in Evanston, 111. When com-
pared to a nonfluoridated "control city" a 50 to 65 percent reduction in
-------
936 Chapter Fifteen
T
T3
Of
c

•





£
1,000
-





0







Qt
C
41
900





_
3







£
o
800
L*




_
O

•





o

» •






700
-• •




-
CI

•





V
600
-




_


¦ •





c
4i
c
500
-




-
I

•





CU
Q.
400
•




-
£


•
A



£
300
—

!


-
a.



•
• *
•

QJ
kA
200
-













o
u
100
_





o







c



_



o
0
'
1

'
1

00 0 5	10	1.5	2.0 2 5 3.0
fluoride (f ) content of the public water supply, mg/L
Figure 1S.1 Relation between the amount of dental caries (perma-
nent teeth) observed in 7257 selected 12- to 14-year-old white school
children of 21 cities of four states and the fluoride content of public
water supply. (Source: J. F. McClure, "Ingestion of Fluoride and
Dental Caries—Quantitative Relations Based on Food and Water Re-
quirements of Children 1-12 Years Old," Am. J. Diseases Children,
vol. 66, 1943, p. 362.)
dental caries was found in the fluoridated cities without evidence of
any adverse effects. These initial studies established fluoridation as a
practical and effective public health measure that would prevent den-
tal caries.
Once the safety and effectiveness of fluoridation had been estab-
lished, engineering aspects needed to be developed before community
water fluoridation could be implemented. In the 1950s and 1960s,
Franz J. Maier, a sanitary engineer, and Ervin Bellack, a chemist,
both with the USPHS, made m^jor contributions to the engineering
aspects of water fluoridation. Maier and Bellack helped determine
which chemicals were the most practical to use in water fluoridation,
the best mechanical equipment to use, and the best process controls.
Bellack also contributed to major advances in fluoride testing. In
1963, Maier published the first comprehensive book on the technical
aspects of fluoridation,12 and, in 1972, Bellack, then with the U.S. En-
vironmental Protection Agency (USEPA), published an engineering
manual13 that has only recently been replaced.
Over the past 40 years, fluoride and fluoridation have been the sub-
ject of numerous studies undertaken by the USPHS, state health de-
-------
Water Fluoridation 937
partments, and nongovernmental research organizations. Since 1970,
over 3700 such studies have been conducted.14 These studies have
overwhelmingly supported the beneficial effect of water fluoridation.
Recently, studies in the United Kingdom, and in several states by
the Robert Wood Johnson Foundation, have reaffirmed the safety and
benefits of fluoridation.16,16 In a 1983 United Kingdom study, dental
caries experience of 5-, 12-, and 15-year-olds from Anglesby (fluori-
dated) was compared with nonfluoridated Afron. The study found an
approximate 50 percent reduction in DMF teeth in Anglesby verses
Afron.15 The Robert Wood Johnson Foundation conducted a 4-year
study (1977-1981) to assess whether dental caries in children could be
reduced or eliminated by various combinations of preventive mea-
sures, including water fluoridation. The study concluded that fluori-
dation is an important, extremely effective, yet inexpensive preven-
tive measure.16
Concerns have recently been expressed that increases in the preva-
lence of fluorosis are occurring in communities with negligible and op-
timal water-fluoride concentrations, because of increased total fluo-
ride consumption from various sources.17 Studies from the University
of Minnesota and the National Institute of Dental Research indicate
that no increase in fluoride consumption from foods and beverages
over the last 40 years has occurred and preliminary investigations in-
dicate that no increase in the prevalence of fluorosis in either fluori-
dated or nonfluoridated communities can be demonstrated.18,19
Present Status of Fluoridation
The Centers for Disease Control (CDC) estimated that approximately
130 million Americans, or about 60.5 percent of those served by public
water supplies, consumed fluoridated water daily as of January 1,
1988.20 Some 9.0 million of these people are served by naturally fluo-
ridated supplies. Approximately 70 percent of all cities with popula-
tions of 100,000 or more have fluoridated water. More than 22 states,
the District of Columbia, and Puerto Rico provide fluoridated water to
over one-half of their population. As of 1985, eight states require flu-
oridation, at least for cities above a minimum population. The in-
crease in the U.S. population served by fluoridated drinking water
systems is shown in Fig. 15.2.21
In 1981, approximately 38 countries reported community water flu-
oridation benefiting approximately 208 million people. The United
States, Ireland, Canada, Brazil, Australia, Venezuela, and the United
Soviet Socialist Republic have large populations consuming fluori-
dated water. The city-states of Hong Kong and Singapore are totally
fluoridated.
Considerable progress has been made toward achieving community
-------
938 Chapter Fifteen

300 i
J 200
1
e
i
0
1	100
o
a
Total
U.S.
population
Population
«rved br
Public
watw
supply
Population
Mrv«dby
total
fluoridation
Population
ttrvtd by
natural
fluoridation
1945 SO 55 60 65 70 75 80 85 1990
Year
Figure 15.2 Fluoridation growth in the United States,
1945-1984. (Source: Centers for Disease Control, "Dental Caries
and Community Water Fluoridation Trends," MMWR, vol. 34, no.
6, 1985p. 77.)
fluoridation in Central and South America, especially in Brazil. Brazil
requires fluoridation for all communities with populations over
50,000. The Pern American Health Organization (a branch of the
World Health Organization) has been very active in the promotion of
fluoridation in Latin America.
Although community water fluoridation has been shown to be safe
and the most cost-effective method to prevent dental caries, a small
percentage of the population continues to oppose its introduction into
community water systems. When fluoridation is being considered for
adoption by a community, persons opposed to fluoridation often at-
tempt to refute the benefits, safety, and efficacy of this effective public
health measure. Charges against fluoridation and the corresponding
truths have been discussed elsewhere.22 Assistance in responding to
false charges against fluoridation may be obtained from the Dental
Disease Prevention Activity of the CDC, Atlanta, Ga. The National
Institute of Dental Research, Bethesda, Md., a branch of the National
Institutes of Health, is another source of information concerning flu-
oridation.
Causes of Dental Caries
Tooth decay is a complex process, and all factors involved are not en-
tirely understood. It is usually characterized by loss of tooth structure
(enamel, dentin, and cementum) as a result of destruction of these tis-
sues by acids. Evidence indicates that acids are produced by the action
-------
Water Fluoridation 939
0f oral bacteria and enzymes on sugars and carbohydrates entering
the mouth. This takes place beneath the plaque, an invisible film com-
posed of gummy masses of microorganisms that adhere to the teeth.
Oral bacteria are capable of converting some of the simpler sugars
into acids, and the bacteria and enzymes acting in combination are ca-
pable of converting carbohydrates and more complex sugars into ac-
id£-The production of acids is a result of the natural existence of bac-
teria and enzymes in the mouth.
Nearly everyone is attacked by dental caries, the most prevalent
chronic disease of humans.23 It is truly universal. Until water fluori-
dation became widespread, almost 98 out of 100 Americans experi-
enced some tooth decay by the time they reached adulthood. The high-
est tooth decay activity is found in school children. Tooth decay begins
in early childhood, reaches a peak in adolescence, and diminishes dur-
ing adulthood.24
D&olal Benefits of Fluoride in Drtnkiflty
When water containing fluoride is consumed, some fluoride (about 50
percent) is retained by fluids in the mouth and is incorporated onto
tfeetfr by surface uptake (topical effect). The rest (about 50 percent) en-
ters the stomach where it is rapidly adsorbed by diffusion through the
stomach walls and intestine. Fluoride enters the blood plasma and is
rapidly distributed throughout the body, including teeth {systemic ef-
fect). Because of the systemic effect, the fluoride ion is able to pass
freely through all cell walls and is available to all organs and tissues
of the body. "Distributed in this fashion, the fluoride ion is available to
all skeletal structures of the body in which it may be retained and
stored in proportions that generally increase with age and intake.
Bones, teeth, and other parts of the skeleton tend to attract and re-
tain fluoride. Soft tissues do not retain fluoride. Fluoride is a "bone
seeker," with about 96 percent of the fluoride found in the body depos-
ited in the skeleton. Because teeth are part of the skeletal system, in-
corporation of fluoride in teeth is basically similar to that in other
bones. It is most rapid during the time of the child's formation and
growth. Erupted teeth differ from other parts of the skeleton in that
once they are formed, with the exception of the dentin (inner part of
the tooth) and the root, cellular activity virtually ceases. As a result,
very little change occurs in the fluoride level in teeth after they are
formed. Children must drink the proper amount of fluoridated water
during early development of permanent teeth, preferably before they
start school, in order to realize full benefits.
High levels of fluoride in drinking water have been found to cause
-------
940 Chapter Fifteen
adverse health effects. As a result, the USEPA has established regu-
latory limits on the fluoride content of drinking water. Based on a de-
tailed review of health effects studies on fluoride,26,26 the USEPA set
a maximum contaminant level of 4 mg/L in water systems to prevent
crippling skeletal fluorosis.27 A secondary level of 2 mg/L was estab-
lished by USEPA to protect against objectionable dental fluorosis.27
These limits, to be reviewed by USEPA every three years, or as new
health data become available, primarily impact systems that have
naturally high fluoride levels.
The relationship between dental caries, dental fluorosis, and fluo-
ride level is shown in Fig. 15.3.28 The beneficial effect of optimally flu-
oridated water ingested during the years of tooth development has
been amply demonstrated. At the optimal concentration in potable
water, fluoride will reduce dental caries from 20 to 40 percent among
children who ingest this water from birth.29 Evidence that water flu-
oridation is effective in preventing caries has been repeatedly demon-
strated, starting with the the initial-community trials in the United
States and Canada in the 1940s. In recent years, however, the relative
impact of water fluoridation appears to have diminished as other
sources of fluoride supplementation (toothpastes, food, etc.) have in-
creased. Continuation of benefits into adult life is inevitable. Stronger,
teeth result in fewer caries, which require fewer and less extensive"
fillings, fewer extractions, and fewer artificial teeth. Fluoridated wa-
ter helps prevent cavities on exposed roots as a result of receding
gums in adults who develop periodontal disease. Early evidence indi-
cated that higher levels of fluoride would strengthen bones of older
10.0-
100 _
00 1.0 2.0 3.0 4.0 5.0 6.0
Fluoride (F~) contvnt, mg/L
Figure 15.3 Dental caries and dental fluorosis in relation
to fluoride in public water supplies. (Source: J. M. Dun-
ning, Principles of Dental Public Health, Harvard Univer-
sity Press, Cambridge, Mass , 1962.)
-------
Water Fluoridation "941
jeople, thereby reducing the incidence of bone fractures.JU,<*i This now*
ippears not to be true.
Various studies in fluoridated communities over the last 30 years
iave shown a dramatic increase in the number of teenagers that are
jompletely caries-free. Teenagers without lifetime exposure realize
jenefits from fluoridation, and benefits increase for those with life-
ime exposure. Conservative estimates indicate that 20 percent of the
leenagers in a fluoridated community will be caries-free.32 This is
about six times as many as are caries-free in a fluoride-deficient com-
munity. Fluoride makes teeth more resistant to bacterial acids and in-
hibits the growth of certain kinds of bacteria that produce acids.33 In
addition, fluoride appears to actually aid in the remineralization of
teeth.34,35
As a result of the positive effects of fluoride on teeth, fluoridation
can substantially reduce costs associated with restorative dentistry.
For every dollar spent (1980) on water fluoridation, a potential $50 in
dental bills may be saved.36 In 1985, the CDC estimated the cost of
fluoridation to be about the cost of a candy bar per person per year.
Optimal Fluoride Levels
The optimal fluoride level in water is the level that produces the
greatest protection against caries with the least risk of fluorosis. Ini-
tially, this figure was obtained after examining the teeth of thousands
of children living in various places with differing fluoride levels. It
was not based on any direct or accurate knowledge of how much water
children drank at various times and at different places. Early in such
investigations, variation in the optimal figure was observed, depend-
ing on the local air temperature, which had a direct bearing on the
amount of water children at different ages consumed. Studies in Cal-
ifornia and Arizona, where temperatures are considerably above the
average of other parts of the United States, showed a definitely lower
optimal fluoride level. This was demonstrated by observing dental
fluorosis prevalence in places with various natural fluoride levels in
their public water supplies and by estimating the actual quantities of
water ingested by children of various age groups and weights.
The optimal fluoride level for a water system is usually established
by the appropriate state regulatory agency. Optimal fluoride concen-
trations and control ranges recommended by the USPHS and CDC
Jnay be used as guidelines if state limits have not been established.
These levels, shown in Table 15.1,37 are based on the annual average
the maximum daily air temperature in the particular school or com-
munity.
Many water supplies contain fluorides naturally. For these systems,
-------
942 Chapter Fifteen
TABLE 15.1 Optimal Fluoride Levels Recommended by the U.S. Public Health
Service, Centers for Disease Control
Annual	Recommended
average of	Recommended fluoride	control community	Range school
maximum	concentration systems	systems, mg/L
daily air				
temper-
Community,
School,
0.1
0.5
20%
20%
atures;"?
mg/L
mg/L
Below
Above
Low
High
40.0-53.7
1.2
5.4
1.1
1.7
4.3
6.5
53.8-68.3
1.1
5.0
1.0
1.6
4.0
6.0
58.4-63.8
1.0
4.5
0.9
1.5
3.6
5.4
63.9-70.6
0.9
4.1
0.8
1.4
3.3
4.9
70.7-79.2
0.8
3.6
0.7
1.3
2.9
" 4.3
79.3-90.5
0.7
3.2
0.6
1.2
2.6
3.8
'Based on temperature data obtained for a minimum of 5 years.
tBased on 4.5 times the optimal fluoride level for communities.
Source: From T. G. Reeves, Ref. 37.
the question of the practicability of supplementing natural fluoride
with enough fluoride to bring the concentration up to the optimal
level must be addressed. Usually, addition of the small amount of flu-
oride needed to reach the optimal level can be shown to be economi-
cally justified based on the resulting benefits to the community.
Fluoride Chemicals and Chemistry
Fluorine, a gaseous halogen, is the thirteenth most abundant element
found in the earth's crust. It is a pale yellow noxious gas that is highly
reactive. It is the most electronegative of all elements and cannot be
oxidized to a positive state. Fluorine is not found in a free state in na-
ture; it is always found in combination with chemical radicals or other
elements as fluoride compounds.
Fluoride can be found in a solid form in fluoride-containing miner-
als such as fluorspar, cryolite, and apatite. Fluorspar is a mineral con-
taining from 30 to 98 percent calcium fluoride (CaF2). Cryolite
(Na3AlF6) is a compound of aluminum, sodium, and fluoride. Apatite
[Ca10(PO4, C03)6(F, CI, OH)2] is a deposit of a mixture of calcium com-
pounds that include calcium phosphates, calcium fluorides, and cal-
cium carbonates. Trace amounts of sulfates are usually present as im-
purities. Apatite contains from 3 to 7 percent fluoride and is the main
source of fluoride used in water fluoridation at the present time.
Apatite is also the raw material for phosphate fertilizers. Cryolite is
not a major source of fluoride in the United States.
Fluoride is widely distributed in the lithosphere and hydrosphere.
Because of the dissolving power of water and movement of water in
-------
Water Fluoridation 943
the hydrologic cycle, fluoride is found naturally in all waters. High
concentrations of fluoride are not common in surface water, but may
occur in groundwater, hot springs, and geothermal fluids.
Fluoride forms compounds with every element except helium, neon,
and argon. Polyvalent cations such as aluminum, iron, silicon, and
magnesium form stable complexes with the fluoride ion. The extent to
which complex formation takes place depends on several factors, in-
cluding the complex stability constant, pH, and the concentrations of
fluoride and the complexing species.38
Sodium fluoride, sodium silicofluoride, and hydrofluosilicic acid are
the three most commonly used fluoride chemicals in the United
States. Standards for these chemicals are published by AWWA for use
by the water industry.39^11 All chemicals used for fluoridation should
Be comparable in quality to the requirements of these standards.
Prom time to time, shortages of fluoride chemicals have occurred.
Generally, most "shortages" are not shortages at the manufacturer's
plant, but a temporary shortage at the local distributor level. Local
shortages are usually eliminated quickly. In the past, shortages at the
manufacturing level, especially of hydrofluosilicic acid and sodium
silicofluoride, have occurred.
Hydrofluosilicic acid and sodium silicofluoride (and most sodium
fluoride) are by-products of phosphoric acid manufacture, the main in-
gredient of phosphate fertilizer. Sales of fertilizer will have a direct
effect on the volume of fluoride chemicals produced. In the past, slow
sales of fertilizer have resulted in a temporary shortage of these two
chemicals. Shortages have been relatively mild because the number of
fluoridated communities was smaller andlower volumes of sodium silico-
fluoride and hydrofluosilicic acid were needed than at the present
time. Shortages occurred in 1955 to 1956, in the summer of 1969, in
the spring and summer of 1974, in the summer of 1982, and in the
early part of 1986, this last one being the most severe.
During production of fluoride chemicals, trace amounts of impuri-
tfes may be introduced into the chemical, especially arsenic, lead,
and/or zinc. Normally, impurities are at levels far below that which
would necessitate the establishment of maximum impurity limits.42
Sodium fluoride
Sodium fluoride (NaF) is a white, odorless material available either as
a powder or in the form of crystals of various sizes. It has a molecular
weight of 42.00, a specific gravity of 2.79, and a practically constant
solubility of 4.0 g/100 mL (4 percent) in water at temperatures gener-
ally encountered in water treatment practice. When added to water,
sodium fluoride dissociates into sodium and fluoride ions:
-------
944 Chapter Fifteen
NaF ^ Na+ + F"	(15.1)
The pH of a sodium fluoride solution varies with the type and
amount of impurities present. Solutions prepared from common
grades of sodium fluoride have a pH near neutrality (approximately.
7.6). Sodium fluoride is available in purities ranging from 97 to over
98 percent, with impurities consisting of water, free acid or alkali, so-
dium silicofluoride, sulfites, and iron, plus traces of other substances.
Approximately 8.6 kg (19 lb) of sodium fluoride will add 1 mg/L of flu-
oride to 1.0 mil gal (3.8 ML) of water.
Sodium sillcofluoride
Sodium silicofluoride (Na2SiF6) is a white, odorless crystalline mate-
rial with a molecular weight of 188.06 and a specific gravity of 2.679
Its solubility varies from 0.44 g/100 mL of water at 0°C to 2.45 g/100
mL at 100°C.
When sodium silicofluoride is dissolved in water, virtually 100 per-
cent dissociation occurs rapidly:
NaaSiFg si 2Na* + SiFg"	(15.2)
Silicofluoride ions (SiF|~) may react in two ways. The most common
is hydrolysis of SiFfreleasing fluoride ions and silica (Si02):
SiFg" + 2H20 ^ 4H+ + 6F" + Si02	(15.3)
Silica, the main ingredient in glass, is very insoluble in water. Al-
ternatively, SiF§~ dissociates very slowly, releasing fluoride ions and
silicon tetrafluoride (SiF4):
SiFj" ss 2F" + SiF4	(15.4)
Silicon tetrafluoride is a gas that will easily volatilize out of water
when present in high concentrations. It also reacts quickly with water
to form silicic acid or silica:
SiF« + 3H20 4HF + H2Si03	(15.5)
SiF< + 2H20 ?£ 4HF + Si02	(15.6)
Solutions are acidic, with saturated solutions usually exhibiting *
pH of between 3 and 4 (approximately 3.6). Sodium silicofluoride tf
available in purities of 98 percent or higher. Principal impurities are
water, chlorides, and silica. Approximately 6.3 kg (14 lb) of sodium
silicofluoride will add 1 mg/L of fluoride to 1.0 mil gad (3.8 ML).
-------
Water Fluoridation 945
Hydrofluosilicic acid
Hydrofluosilicic acid, also known as hexafluosilicic, silicofluoric, or
fluosilicic acid (H2SiF6), has a molecular weight of 144.08 and is avail-
able commercially as a 20 to 35 percent aqueous solution. It is a straw-
colored, transparent, fuming, corrosive liquid having a pungent odor
and an irritating action on the skin. Solutions of 20 to 35 percent
hydrofluosilicic acid have a low pH (1.2), and, at a concentration of 1
mg/L in poorly buffered potable waters, a slight depression of pH can
occur.
Hydrofluosilicic acid dissociates in solution virtually 100 percent.
Its chemistry is very similar to Na2SiF6:
Hydrofluosilicic acid should be handled with great care because of its
low pH and the fact that it will cause a "delayed burn" on skin tissue.
Hydrofluosilicic acid (23 percent) will freeze at approximately 4°F
(-15.5°C). Approximately 20.8 kg (46 lb) of 23 percent acid are re-
quired to add 1 mg/L of fluoride to 1.0 mil gal (3.8 ML).
Hydrofluoric acid and silicon tetrafluoride are common impurities
in hydrofluosilicic acid that result from production processes.
Hydrofluoric acid is an extremely corrosive material. Its presence in
hydrofluosilicic acid, whether from intentional addition (i.e., "forti-
fied" acid) or from normal production processes, demands careful han-
dling. Hydrofluosilicic acid fumes are lighter than air and will rise.
Other fluoride chemicals
Ammonium silicofluoride, magnesium silicofluoride, potassium fluo-
ride, hydrofluoric acid, and calcium fluoride (fluorspar) are being, or
have been, used for water fluoridation. Each material has properties
that make it desirable in a specific application, but each also has un-
desirable characteristics. None of these chemicals has widespread ap-
plication in the United States. Calcium fluoride, however, is widely
used in South America.
Fluoride Feed Systems
Three methods of feeding fluoride are common in community water
supply systems:
t
!• Dry chemical feeder with a dry fluoride compound
H2SiF6 2HF + SiF4
SiF< + 2H20 4HF + SiOj
SiF« + 3H20 ^ 4HF + H2Si03
(15.7)
(15.8)
(15.9)
-------
946 Chapter Fifteen
2.	Chemical solution feeder with a liquid fluoride compound or with a
prepared solution of a dry chemical
3.	Fluoride saturator
The first two methods are also commonly used to feed other water treat-
ment chemicals. The saturator is a unique method for feeding fluoride.
Selection of the best fluoridation system for a situation must be
based on several factors, including popul?f
-------
Water Fluoridation 947
je a channel where other water treatment chemicals are added, a
uain coming from the filters, or the clearwell. If a combination of fa-
aJities exist, such as a treatment plant for surface water plus supple-
mental wells, a point where all water from all sources passes must be
selected. If no common point exists, a separate fluoride feeding instal-
fltion is needed for each facility.
Another consideration in selecting the fluoride injection point is the
possibility of fluoride losses in filters. Whenever possible, fluoride
should be added after filtration to avoid substantial losses that can oc-
nir, particularly with heavy alum doses or when magnesium is
jresent and the lime-soda ash softening process is being used. A flu-
iride loss of up to 30 percent can result if the alum dosage rate is 100
ng/L.13 In some situations, addition of fluoride before filtration may
je necessary, such as when the clearwell is inaccessible.
When other chemicals are being fed, the question of chemical com-
jatibility must be considered. The flunnde injection point should be as
;ar away as possible from the injection point of chemicals that contain
ialcium, in order to minimize loss of fluoride by local precipitation.
For example, if lime is being added to the main leading from the fil-
ters for pH control, fluoride can be added to the same main at another
joint or at the clearwell. If lime is added to the clearwell, fluoride
should be added to the opposite side. If injection point separation is not
[jossible, an in-line mixer must be used to prevent local precipitation
3f calcium fluoride and to ensure that the added fluoride dissolves.
In a single-well system, the well pump discharge can be used as the
fluoride injection point. If more than one well pump is used, the line
leading to the distribution system can be used as the injection point.
In a surface water treatment plant or softening plant, the ideal loca-
tion of the fluoride injection point is in the line from the filters to the
:learwell. This location provides for maximum mixing. Sometimes the
clearwell is located directly below the filters, and discharging chemi-
:als directly to the clearwell is difficult. In this situation the fluoride
injection point must be at another location, such as in the main line to
the distribution system or before the filters.
Safety considerations
Fluoride levels to which a water plant operator may be exposed can be
much higher than the optimal level. Proper handling of fluoride chem-
icals is necessary to prevent overexposure. Dusts are a particular
problem when sodium fluoride and sodium silicofluoride are used. Op-
erators must be aware of the hazards involved in the feeding of fluo-
ride chemicals and should always follow accepted safety practices.
Manuals describing operational hazards and safety practices for fluo-
ride chemical feed systems are available.22,37,43
-------
948 Chapter Fifteen
Waste disposal
An important consideration in the design and operation of a fluorida-
tion system is the disposal of empty fluoride chemical containers. The
temptation to reuse fiber drums is difficult to overcome because the
drums are convenient and sturdy. Paper bags are dusty and could
cause a hazard if they are burned, and empty acid drums could contain
enough acid to cause contamination. The best practice is to rinse all
empty containers, including paper bags, with water. After all traces of
fluoride are removed, bags and drums should be disposed of according
to the requirements of the state's environmental protection program
solid waste division.
School Fluoridation
Children living in areas not served by community water supplies are
often unable to benefit from drinking water that has been optimally
fluoridated. In 1985, about 17 million people (7 percent of the U.S.
population) lived in areas that lacked central water systems. Because
community fluoridation is not feasible for these areas, other ways of
preventing dental caries must be developed if the natural fluoride
level in the water supply is inadequate. Fluoridation of water supplies
of rural schools has been implemented in many areas and is particu-
larly appealing because it reaches sizable numbers of children with
minimal demands on personnel, equipment, and funds. As of June
1987, 459 schools in 12 states provide fluoridated water to approxi-
mately 150,700 students.44 School fluoridation is another way to pro-
vide fluoridated water, but it is not considered an alternative to com-
munity water fluoridation because of its limitations.
The most obvious limitation of school water fluoridation is that chil-
dren are approximately 6 years old before they begin attending school,
whereas maximum dental benefits occur when fluoridated water is
consumed from birth. Data obtained from communities with con-
trolled fluoridation, however, indicate that children who are 6 years of
age or older at the time fluoridation is initiated do derive dental ben-
efits. These findings are not surprising, considering that, at age 6, a
significant amount of calcification will occur in later-erupting perma-
nent teeth. In addition, considerable fluoride uptake occurs between
the completion of permanent tooth calcification and eruption. Evi-
dence also indicates that the topical action of fluoridated water will
confer some caries inhibition to erupted teeth.
A second factor limiting the effectiveness of school water fluorida-
tion is that exposure to fluoridated water in a school is intermittent
Children attend school only 5 days a week for only part of the day and
only part of the year. One concern is the generally short water lines
found in a school system. The impact of an overfeed would be more
-------
Water Fluoridation 949
serious than in a community system, because a slug of fluoride only
needs to travel several feet rather than many yards or even milesnbe-
fore it might be consumed.
Several studies have been conducted to determine the optimal fluo-
ride level for school fluoridation systems. Studies at Pike County, Ky.,
and Elk Lake, Pa., found a 35 to 40 percent reduction in dental cavi-
ties in school children consuming water at 4.5 times the recommended
optimal level for community fluoridation.45,48 Results of a 12-year
study at a school near Seagrove, N.C., showed that little additional
benefits resulted when the fluoride level was increased to 7 times
the recommended optimal level for communities.47 As a result, 4.5
times the optimal level recommended for community systems has
been established as the recommended level for school fluoridation sys-
tems.
When school fluoridation is considered, the issue of safety must al-
ways be addressed. Full-time exposure to fluoride levels as low as
twice the optimum can cause some degree of dental fluorosis. Yet,
early epidemiologic studies have found that children consuming
fluoride-free water at home were uniformly free of any objectionable
signs of dental fluorosis when water with a natural fluoride at a level
of 6 mg/L was consumed at school. Because fluorosis is a development
disturbance that Is produced only at the initial stage of enamel forma-
tion, teeth of school-age children may be too advanced to be adversely
affected by higher levels of fluorides. Other epidemiologic findings
support this evidence.48"48
Alternatives to Water Fluoridation
Alternative means of providing the benefits of fluoride besides the flu-
oridation of municipal water supply systems are available. Municipal
water fluoridation, however, is the most cost-effective means available
for reducing the incidence of caries in a community. This conclusion is
based.on the mass of evidence demonstrating the efficacy of the mea-
sure and on the most current information on costs of implementing
fluoridation. Alternative methods should only be considered in situa-
tions where municipal water fluoridation is not possible.
In general, five alternatives to water fluoridation exist that use ei-
ther the topical or systemic method:
Topical fluoride methods
Systemic fluoride methods
l.Gela
1. Tablets
2. Mouth rinses
2. Drops
3. Dentifrices

-------
Water Fluoridation 951
table 15.3 Effectiveness of Topical Fluoride Applications


Cavity reduc-
Application
Frequency
tion, %
1. Professionally applied

35
2. Self-applied


a. 0.2% NaF rinse
Weekly
25
b. Supervised brushing with 9.0% stannous
2/year
25
fluoride

c. Toothbrushing at home with 0.1% fluoride
Daily
20
dentifrice

Source: From S. B. Heifeti, Ref. 50.


table 15.4 Dally Supplemental Ruorlde Dosage Schedule

Ciacentralion of fluoride in water, mg F/day
Less than 0.3 0.3 mg/L to 0.7
Greater than 0.7
Age, years mg/L
mg/L
mg/L
lirth to 2 0.25
0.0
0.0
r3 0.50
0.25
0.0
-13 1.00
• 0.50
0.0
'2.2 mg of sodium fluoride contains 1 mg of fluoride (F).
jounce: From American Dental Association, Ref. 5.
funded by CDC with the state of Minnesota to evaluate the perfor-
mance characteristics of these units.
Recent shortages of fluoride chemicals have stimulated research to
Sad an alternative chemical that is readily available and not subject
to shortages. Calcium fluoride is readily available in fluorspar rock
and is prevalent throughout the United States, especially in southern
Illinois. Calcium fluoride is also used as the feed chemical for many
water systems in Brazil and other South American countries. A three-
year demonstration project funded by CDC to determine the feasibil-
ity of using calcium fluoride to fluoridate community water supply
systems is being conducted by the Ohio State University Water Re-
sources Foundation. Results have not looked promising.
References
1.	H.C. Hodges and FA. Smith, in J.H. Simons (ed.), Fluorine Chemistry, Vol. 4, Ac-
ademic Press, New York, 1965.
2.	F.S. McClure, Water Fluoridation, the Search and the Victory, National Institutes of
Health, Bethesda, Md., 1970.
3.	D.R. McNeil, The Fight for Fluoridation, Oxford University Press, New York, New
York, 1957.
4.	H.V. Churchill, "The Occurrence of Fluorides on Some Waters of the United
States," J. AWWA Vol. 23, No. 9,1931, p. 1399.
-------
952 Chapter Fifteen
5.	M.C. Smith, E.M. Lantz, and H.V. Smith, 'The Cause of Mottled Enamel, a Defe*
of Human Teeth," University of Arizona Agricultural Experiment Station Bulletin
No. 32, 1931.	^
6.	H.T. Dean, "Chronic Endemic Dental Fluorosis (Mottled Enamel)," JAMA, VoL
107, 1936, p. 1269.
7.	F.J. McClure, "Ingestion of Fluoride and Dental Caries—Quantitative Relations
Based on Food and Water Requirements of Children 1-12 Years Old," Am. J. Du-
eases Children, Vol. 66, 1943, p. 362.
8.	H.T. Dean F.A. Arnold, J. Phillip, find J.W. Knutson, Studies on Mass Control of
Dental Caries through Fluoridation of Public Water Supply, Public Health Reports
65, Grand Rapids-Muskegon, Mich., 1950.
9.	D.B. Ast, D.J. Smith, B. Wachs, H.C. Hodges, H.E. Hilleboe, E.R. Schesinger, H.C.
Chase, K.T. Cantwell, and D.E. Overton, "Newburgh-Kingston Caries-Fluorine
Study: Final Report," J. Am. Dental Assoc., Vol. 52, 1956, p. 290.
10.	H.K. Brown and M. Poplove, "Brantford-Samia-Statford, Fluoridation Caries
Study, Final Survey, 1963," J. Can. Dental Assoc.. Vol. 31, No. 8, 1965, p. 505.
11.	I.N. Hill, J.R. Blayney, and W. Wolf, "Evanston Fluoridation Study—Twelve Yean
Later," Dental Progr., vol. 1, 1961, p. 95.
12.	Franz J. Maier, Manual of Water Fluoridation Practice, McGraw-Hill, New York.
1963.
13.	E. Bellack, Fluoridation Engineering Manual, U.S. Environmental Protection.
Agency, Washington, D.C., 1972; reprinted September 1984.
14.	Michigan Department of Public Health Policy Statement on Fluoridation of Commu-
nity Water Supplies and Synopsis of Fundamentals of Relation of Fluorides and Flu-
oridation to Public Health, Michigan Department of Public Health, 1979.
15.	D. Jackson, P.M.C. James, and F.D. Thomas, "Fluoridation in Anglesby 1983: A
Clinical Study of Dental Caries," Br. Dental J., voL 158, 1985, p. 45.
16.	FL 130 National Prevention Dentistry Demonstration Program Reaffirms Benefits of
Community Water Fluoridation, United States Department of Health and Human
Services, Public Health Service, Centers for Disease Control, Atlanta, Ga., 1985.
17.	D. Leverett, "Prevalence of Dental Fluorosis in Fluoridated and Nonfluoridated
Communities—A Preliminary Investigation," J. Pub. Health Dentist., voL 46,1988,
P- 4.
18.	L. Singer, and R.H. Ophaug, "Fluoride Intake of Humana" in J.L. Shupe, H.B.
Peterson, and N.C. Leone (eds.), Fluorides, Effects on Vegetation, Animals and Hu-
mans. Paragon Press, Salt Lake City, Utah, 1983.
19.	W.S. Driscoll, H.S. Horowitz, R.J. Meyer, S.B., Heifex, A. Kingman, and E.R.
Zimmerman, "Prevalence of Dental Caries and Dental Fluorosis in Areas with Neg-
ligible, Optimal, and Above-Optimal Fluoride Concentrations in Drinking Water,1*
J. Am. Dental Assoc., vol. 113, 1986, p. 29.
20.	Fluoridation Census 1985, United States Department of Health and Human Ser
vices, Public Health Service, Centers for Disease Control, Atlanta, Ga., 1988.
21.	"Dental Caries and Community Water Fluoridation Trends—U.S.," Morbidity and
Mortality Weekly Report, vol. 34, no. 6, 1985, p. 77.
22.	Water Fluoridation Principles and Practices, 2d ed., AWWA Manual M4, AWWA
Denver, Colo., 1984.
23.	D.F. Striffler, W.O. Young, and BA. Burt, Dentistry, Dental Practice and the Com-
munity, 3d ed., Saunders, Philadelphia, 1983.
24.	The Prevalence of Dental Caries in United States Children, 1979-1980. The No-
tional Dental Caries Prevalence Survey, National Institute of Dental Research, Na-
tional Caries Program, National Institute of Health, Bethesda, Md., Decembtf
198L
25.	United States Environmental Protection Agency, Final Draft for the Drinking Wo*
ter Criteria Document on Fluoride, Criteria and Standards Division, Office of Drink*
ing Water, Washington, D.C., 1985.	__j
26.	"National Primary Drinking Water Regulations; Fluoride; Final Rule and Propow
Rule," Federal Register, vol. 50, 1985, p. 47142.	,
27.	"National Primary and Secondary Drinking Water Regulations; Fluoride; F"1*
Rule." Federal Register, vol. 51, April 2, 1986, p. 11396.
-------
Water Fluoridation 953
j3. J M. Dunning, Principles of Dental Public Health, Harvard University Press, Cam-
bridge, Mass., 1962.
29.	E. Newbrun, "Effectiveness of Water Fluoridation,"./. Public Health Dentistry, vol.
49, no. 5, special issue, 1989.
30.	J. Jowsey, L.B. Riggs, P.J. Kelly, and D.L. Hoffman, "Effect of Combined Therapy
with Sodium Fluoride, Vitamin D, and Calcium in Osteoporosis," Am. J. Med., vol.
53, 1972, p. 43.
jl. L.B. Riggs, E. Seeman, S.F. Hodgson, D.R. Taves, and W.M. O'Fallon, "Effects of
the Fluoride/Calcium Regimen of Vertebral Fracture Occurrence in
Postmenopausal Osteoporosis," N. Engl. J. Med., vol. 306, No. 8, 1982, p. 446.
32. FL-98 Caries-Free Teenagers Increase with Fluoridation, United States Department
of Health and Human Services, Public Health Service, Centers For Disease.Control,
Atlanta, Ga., 1978.
J3. Paul H. Keyes, "Present and Future Measures for Dental Caries Control," J. Am.
Dental Assoc., vol 79, 1969, p. 1395.
14.	Leon M. Silverstone, "The Significance of Remineralization in Caries Prevention,"
J. Can. Dental Assoc., vol. 50, No 2, 1984, p. 157.
15.	J R. Mellberg and D.E. Mallon, "Acceleration of Remineralization, in vitro, by So-
dium Monofluorophosphate and Sodium Fluoride," J. Dental Res., vol. 63, No. 9,
1984, p. 1130.
16.	C.W. Gish, "The Dollar and Cents of Prevention," J. Indiana Dental Assoc., vol. 58,
1979.
7. Thomas G. Reeves (ed.). Water Fluoridation—A Training Course Manual for Engi-
neers and Technicians, United States Department of Health and Human Services,
Public Health Service, Centers for Disease Control, Centers for Disease Control, At-_
lanta, Ga., 1986.
38.	Bert A. Eichenberger and Kenneth Y. Chen, "Origin and Nature of Selected Inor-
ganic Constituents in Natural Waters," in Roger A. Minear and Lawrence H. Keith
(ed.). Water Analysis, Vol. 1: Inorganic Species, Part 1, Academic Press, New York,
1982.
39.	AWWA B701, Standard for Sodium Fluoride, AWWA, Denver, Colo.
40.	AWWA B702, Standard for Sodium Silicofluoride, AWWA, Denver, Colo.
41.	AWWA B703, Standard for Hydrofluosilicic Acid, AWWA, Denver, Colo.
(2. National Academy of Sciences, Committee on Water Treatment Chemicals, Water
Chemicals Codex, National Academy Press, Washington, D.C., 1982.
13. Safety Practice for Water Utilities', AWWA Manual M3 AWWA, Denver, Colo., 1983.
M. Center for Disease Control, Dental Disease Prevention Activity Files, July 1987.
15.	H.S. Horowitz, S.B. Heifetz, F.E. Law, and W.S. Driscoll, School Fluoridation Stud-
ies in Elk Lake, Pennsylvania and Pike County, Kentucky—Results after 8 Years,"
Am. J. Pub. Health,.vol. 58, No. 12, 1968, p. 2240.
16.	H.S. Horowitz, S.B. Heifetz, and F.E. Law, "Effect of School Water Fluoridation on
Dental Caries: Final Results in Elk. Lake, PA, after 12 Years," J. Am. Dental
Assoc., vol. 84, 1972, p. 832.
17.	H.S. Horowitz, S.B. Heifetz, and J A. Brunelle, "Effect of School Water Fluoridation
on Dental Caries: Results in Seagrove, NC, after 12 Years," J. Am. Dental Assoc.,
vol. 106, 1983, p. 334.
W. GA. Kempf and F.S. McKay, "Mottled Enamel in a Segregated Population," Pub.
Health Rep., vol. 45, 1930, p. 2923.
19. Charles Gish, "Relative Efficiency of Methods of Caries Prevention in Dental Public
Health," Proc. Workshop on Preventive Methods in Dental Public Health, University
of Michigan, Ann Arbor, Mich., June 1978.
& S.B. Heifetz, "Cost-Effectiveness of Topically Applied Fluorides," Proc. Workshop on
Preventative Methods in Dental Public Health, University of Michigan, Aim Arbor,
Mich., June 1978.
'1- American Dental Association Council on Dental Therapeutics, Accepted Therapeu-
tics, 39th ed., American Dental Association, Chicago, 111., 1982.
'2. J. Leo, The Venturi Fluoridator, Indian Health Service, United States Public
Heflth Service, August 1981.
FPA Lihrarv Recuon 4
lU3bb/b
-------