United States
Environmental Protection Office of Water EPA 811-R-92-005
Agency (WH-550) August 1992
&EPA FRAMEWORK FOR DECISION
MAKING: AN EPA PERSPECTIVE
-------
Framework for Decision Making: EPA Perspective
Stig Regli1, John E. Cromwell2, xin Zhang2, Allen B. Gelderloos3,
William D. Grubbs4,
Frank Letkiewicz3, Bruce A. Macler1
The United States Environmental Protection Agency (EPA) is
developing national regulations to control for disinfectants and
disinfection by-products (D/DBPs) in public drinking water
supplies. EPA intends to propose regulations that will apply to
all public water systems using disinfection and serving non-
transient populations — approximately 220 million people. This
regulation would be unlike the current maximum contaminant level
(MCL) for total trihalomethanes (TTHMs) which only pertains to
systems serving more than 10,000 people.
A major issue is to ensure that the limits set for D/DBPs do
not cause changes in treatment that result in significant
increases in risk from waterborne pathogens or other DBPs not
regulated at this time. A fundamental goal in setting any new
regulation is to be able to conclude with confidence that the
resultant changes in treatment would lead to a significant
overall decrease in risk to public health. This paper will
describe the analysis that EPA is undertaking in pursuit of
meeting this goal.
Background
The Safe Drinking Water Act (SDWA) passed by the Congress of
the United States requires the EPA to specify a maximum
contaminant level goal (MCLG) for each regulated contaminant.
MCLGs are set at levels at which no adverse health effects are
expected to occur.
The SDWA also requires the EPA to set MCLs as close to the
MCLG as is technically and economically feasible to achieve; the
rule must specify such best available technology (BAT). If it is
not technically or economically feasible to measure a contaminant
1 - U.S. Environmental Protection Agency
2 - Wade Miller Associates
3 - Malcolm Pirnie, Inc.
4 - Science Applications International Corporation
$ - Abt Associates Inc.
1
-------
at the level of concern, then the EPA must set a treatment
technique requirement in lieu of an MCL.
Systems are not required to use BAT in order to comply with
the MCL but can use other technologies as long as they meet all
drinking water standards and are approved by the State. Systems
unable to meet the MCL after application of the BAT can get a
variance, extending the time for achieving compliance. Systems
that obtain variances must meet a schedule approved by State
health officials.
MCLGs are set at zero for contaminants that have
carcinogenic potential from a drinking water exposure. In such
cases the EPA attempts to set MCLs at levels which assure that
lifetime risk from cancer through exposure in drinking water is
between less than one in ten thousand to one in one million (10"*
to 10"6) , using conservative margins of safety in calculating such
risk. Risk assessments usually entail: extrapolation from toxic
effects in animal studies using high doses of exposure to predict
the effects of low doses of exposure in humans; use of the
linearized multistage model which is more conservative than most
other models; and use of a 95% upper confidence interval for
defining the risk factor.
Based on available occurrence, exposure, and health effects
data, twelve chemicals have been identified for possible
regulation (USEPA 199la; USEPA 1992a; USEPA 1992b). Table I lists
the candidate compounds for regulation, possible MCLGs, the
health effect of concern, and the tentative cancer risk
classification.
Table II indicates the concentration level for DBPs
classified as 87 compounds in drinking water at which excess
lifetime cancer risk of 1 in 10,000 and 1 in 100,000 would occur
as predicted by the upper 95% confidence interval, assuming 2
liters are consumed per individual per day. Table II also
indicates the range at which the candidate compounds have been
reported to occur in US drinking water supplies (for a more
detailed characterization of D/DBP occurrence see USEPA, 1992a).
The basis for-the MCLGs, cancer risk classification, and the
indicated cancer risks are described elsewhere (USEPA 1992b).
The EPA intends to set MCLs for each compound listed in
Table I. However, three regulatory options are being considered
for trihalomethanes (THMs) and haloacetic (HAAs) acids. For THMs
the EPA is considering: 1) MCLs for each of the four THMs, 2) an
MCL for total trihalomethanes (TTHMs), and 3) an MCL for each of
the THMs and an MCL for TTHMs (which may 'be different from the
sum of the individual THMs). Individual THMs are being considered
as their health risks differ significantly; and the technical
feasibility for limiting their formation can vary greatly,
-------
depending upon source water quality. An MCL for TTHMs is being
considered to act as a surrogate regulation to limit other DBFs
and because of the precedent already established with the current
TTHM standard.
For HAAs the EPA is considering: 1) MCLs for
trichloroacetic acid (TCA) and dichloroacetic acid (OCA) as
health effects data are available for each of these compounds;
2) an MCL for total HAAs (THAAs) including mono-, di-, and tri-
chloroacetic acid; monobromo and dibromo acetic acid; and
possibly bromochloroacetic acid; and 3) the combination of
options 1 and 2. A limit for THAAs is being considered because
all the compounds can be analyzed simultaneously using the same
analytical method at no additional cost. A limit for THAAs would
act as a surrogate to limit production of other HAAs and DBFs for
which health risks are not yet determined. Future MCLs, when
more data become available, may pertain to specific HAAs other
than DCA and TCA, and/or an expanded list of HAAs including
chloro-bromo acetic acids.
i'.
Issues With Defining BAT for DBFs
BAT must include disinfection but now much?
The easiest means for avoiding risk from D/DBPs is not to
disinfect water. Obviously this is not a viable option, at least
not for surface water supplies, since without disinfection
waterborne disease would be rampant. Although not specified in
the SDWA, the EPA must define BAT for controlling DBFs in
conjunction with criteria to control pathogens.
The feasibility of complying with any D/DBP standard with a
given BAT is directly affected by the need to also comply with
microbial standards. Currently only two national standards have
been promulgated in the U.S. to control for microbial concerns:
the total coliform rule, and the surface water treatment rule
(SWTR). Ground water disinfection requirements are under
development. _
The SWTR (USEPA, 1989a) requires that systems using surface
water, or ground water under the direct influence of surface
water, provide at least 99.9% (3 logs) and 99.99% (4 logs)
removal or inactivation of Giardia cysts and viruses,
respectively. These performance levels are not determined by
measurement of Giardia or viruses but by assuming removal
efficiencies by filtration and. inactivation efficiencies by
disinfection based upon design and operating conditions.
Well operated filter systems that meet recommended design
and operating, and turbidity performance, criteria are assumed to
-------
achieve 2 to 2.5 log removal of Giardia cysts, and a 1 to 2 log
removal of viruses by filtration alone. Inactivation efficiencies
for Giardia and viruses are determined based upon measurement of
the product of "C" (disinfectant concentration in mg/1) and WT"
(disinfectant contact time in minutes); the "CT" for a system
equates to an inactivation efficiency indicated in tables in the
SWTR.
The SWTR also requires systems to maintain a) a disinfectant
residual of at least 0.2 mg/1 at all times in the water entering
the distribution system and, b) a detectable disinfectant
residual or a heterotrophic plate count of less than 500/ml in at
least 95% of the measurements throughout distribution system.
The total coliform rule (USEPA, 1989b) requires all public
water systems (surface and ground water supplies) to measure for
total coliforms in the distribution system and to ensure that no
more than 5% of the measurements are total coliform positive.
For any positive total coliform measurement, follow-up
measurement of fecal coliforms or E. coli is required. No more
than one positive fecal coliform or E. coli measurement is
allowed in any given month.
The GWDR will establish minimal disinfection requirements
for all ground water systems unless they are able to demonstrate
they are not vulnerable to fecal viral and bacterial
contamination. The vulnerability of wells for each system is to
be determined based upon the location of fecal contamination
sources, such as septic tanks and sewage lines, with respect to
each well, and hydrogeological features.
Major issues are whether more microbial standards, and the
disinfection required to meet these standards, are needed to
limit microbial risk or to prevent significant increases in
microbial risk under new D/DBP regulations. To what extent, if
any, should the definition of BAT for the D/DBP regulations
include additional microbial treatment requirements to address
this concern?
Which technologies should be considered and how should treatment
performance be defined?
How the EPA defines BAT will directly affect the level at
which the MCLs are set and the potential for downside risks.
Because health risks from by-products of alternate disinfectants
are not as well characterized as they are for by-products of
chlorination, it may be appropriate to define BAT as certain
precursor removal technologies with chlorination for primary
disinfection under this rulemaking (since MCLs do not have to be
met using BAT this would not preclude the use of alternate
disinfectants). Examples of precursor removal technologies
include: a) conventional treatment optimized for DBP precursor
-------
removal; b) granular activated carbon (GAG) as a filter media
replacement; c) GAC following filtration; and d) membrane
filtration.
Other technologies for BAT consideration, depending upon the
source water quality, include conventional treatment or precursor
removal technologies with use of alternate disinfectants, such as
ozone for primary disinfection, and chloramines for residual
disinfection.
The cost and performance of the above technologies will vary
greatly depending upon design and operation. The technical and
economic feasibility of using these technologies to achieve
different MCLs can be greatly influenced by source water quality
and system size. One approach to this dilemma is to define MCLs
based on what some reasonable percentage (e.g., 90% or 95%) of
systems could at least achieve if they were to apply BAT to their
source water.
If the MCL were based on what all systems (or a very high
percentage such as 99%) could achieve using the BAT, then the MCL
would be higher and potentially much less protective for the
population at large. On the other hand, if the MCL were set
based on what a lower percentage of systems (e.g., 75%) could
achieve, then treatment costs to comply, or the transactional
costs to enforce compliance, might be unreasonable.
For logical decision making, a methodology is needed to
predict performance of different BATs in the U.S. given the
diversity of source waters and the effect that such diversity has
on the treatability for OBPs and microbial concerns. Such a
-methodology should allow for prediction of BAT performance for
controlling DBFs while also achieving different objectives for
pathogen control.
Implications of Systems not Using BAT to achieve MCLs
Systems are not obligated to use BAT to achieve the MCL. If
BAT were defined as one of the OBP precursor removal strategies
as described a~bove, systems could choose less expensive
technologies such as adopting choramines or ozone and chloramines
to achieve the MCL. In fact, use of such technologies might lead
to lower levels of risk from chlorinated OBPs (e.g., levels well
below the MCL) than if the BAT were used. However, depending
upon the source water quality, such treatment changes may
introduce new risks from other DBFs that are not yet well
defined. Also, since chloramines are a much weaker disinfectant
than chlorine, switching to chloramines may significantly
increase microbial risk. Switching to ozone — a stronger
disinfectant than chlorine — followed by chloramines to maintain
a residual in the distribution system may lead to greater
nutrient availability for bacteria to grow in the distribution
-------
system. Such effects will largely depend upon the source water
quality, and the treatment following ozonation.
If the MCLs are defined based on what., conventional
filtration followed by deep bed GAG and chlorine, or what
membrane filtration and chlorine, could achieve, many systems
would be expected to attempt meeting such MCLs using chloramines
or ozone and chloramines due to cost. If chloramines, rather
than chlorine, were included as a residual disinfectant in the
definition of BAT, then the MCLs for chlorinated DBFs would be
lower and even greater shifts to use of alternate disinfectants
to achieve compliance could be expected. The potential health
risk implications that might result from different shifts in
technology needs to be considered when defining BAT.
Downside Risks with DBP Precursor Removal Technologies
Precursor removal technologies can remove substantial
organic material and disinfectant demand. As more DBP precursors
are removed, and the water is cleaner, the efficiency of
disinfection is expected to increase per dose of disinfectant
applied. A decrease in disinfectant demand allows for lower
disinfectant doses to be applied and yet still maintain a
residual in the distribution system. However, a significant
decrease in the level of disinfection between the point of
application and early points in the distribution system could
increase microbial risk unless precautions are taken (e.g.,
increasing the disinfectant contact time prior to the first
customer). The significance of such changes in risk depend on
the pathogen concentrations in the source water, the change in
effect that the water quality matrix has on interference with
disinfection, the net change in the value of the CT product with
respect to different populations receiving the finished water in
the distribution system, and the net change in bacterial
pathogens that might grow in the distribution system.
Precursor removal technologies can also change the ratio of
disinfectant residual with respect to organic (e.g., humic and
fulvic acids) -and inorganic (e.g., bromide) precursors. This can
lead to a change in the relative amounts of different types of
DBPs that are formed. For example, depending upon the bromide
concentration in the source water it becomes possible, with
increased precursor removal, to decrease the amount of TTHMs or
THAAs but to increase the concentration of specific brominated
compounds (e.g., bromoform or brominated HAAs). Such changes in
the relative amounts of the different DBPs could result in no net
gain in risk reduction and possibly even an increase in risk. No
health effects data are yet available on the brominated HAAs.
-------
Strategy for Defining BAT and Appropriate MCLs
The EPA is taking the following steps to address the above
issues and to determine appropriate MCLs for OBPs: 1)
identification of candidate BATs for controlling both DBFs and
pathogens; 2) definition of BAT in the context of also achieving
microbial treatment objectives (the SWTR, coliform rule, or other
criteria); 3) evaluation of candidate BAT performance with
respect to pathogen and DBF control for different source water
qualities; 4) consideration of candidate MCLs, based on what
level is achievable by candidate BATs for a "reasonable"
percentage of systems; and 5) consideration of potential net
changes in risk from exposure to DBFs and pathogens (known and
unknown), and the costs to implement such changes, to determine
the optimum MCLs. The EPA is attempting to implement this
strategy through use of a disinfection by-product risk assessment
model. (DBPRAM).
Overview of DBPRAM
Figure 1 illustrates the DBPRAM modeling framework. OBPRAM
simulates raw water qualities for treatment plants throughout the
U.S. and predicts: 1) the most likely treatment technologies that
utilities would adopt to meet different regulatory targets (MCLs
for OBPs and microbial treatment objectives); 2) the occurrence
of trihalomethanes and haloacetic acids and the associated cancer
risk (i.e., that which is currently quantifiable) after treatment
changes are implemented; 3) the occurrence of pathogens — using
Giardia as the target organism — in the finished water after
treatment changes are implemented, and the associated risk from
infection; and 4) the costs for meeting the treatment objectives.
A detailed description of the model, assumptions, and results
determined to date are described in a series of papers (USEPA,
1992c, Letkiewicz et al., 1992; Gelderloos et al., 1992; Grubbs.
et al., 1992; Cromwell et al., 1992).
Focus of Analysis
To date the modeling analysis has focused on systems which
currently filter and disinfect surface water. Since these
systems include those with the highest levels of DBF precursors
and pathogens in the source water, they represent those with the
highest possible changes in risk that could result from treatment
changes to comply with D/DBP regulations. If regulatory
scenarios can be described that assure a significant decrease in
overall risk in these systems, then scenarios can also be
described that ensure there is at least a positive net benefit in
overall risk for systems with better source water quality (e.g.,
most ground water systems). Preferably, such regulatory scenarios
-------
would also ensure there is no significant increase in risk for
any particular system, no matter what its source water quality.
Due to limited data, the analysis has focused on a subset of
filtered surface water supplies: those which do not use lime
softening and which each serve more than 10,000 people. About 103
million people are currently served by systems meeting this
description. Figure 2 illustrates key input and output
parameters in the DBPRAM and the conceptual approach for
estimating finished water quality from different treatment
configurations. Probability distributions for raw water quality
parameters affecting THM and HA formation were generated from
different data sets (Letkiewicz et al 1992). Raw water Giardia
data reflecting variation among cities (LeChevalier et al., 1991)
and within cities (Hibler et al., 1987) were used to simulate
national raw water occurrence (Grubbs et al 1992). A water
treatment plant model was used to predict Giardia cyst
concentrations and THMs and HAs in the distribution system as a
function of different filtration and disinfection conditions
(USEPAc, 1992, Gelderloos et al 1992).
THMs and HAAs have been targeted in the DBPRAM because they
are among the DBPs identified to date that occur at highest
concentrations in most drinking waters. They are also likely to
be of the greatest health concern, and models are available for
predicting their formation. Giardia cysts are targeted in the
DBPRAM because more data are available on their occurrence in
source waters than other pathogens; Giardia are much more
resistant to disinfection than most other waterborne pathogens;
and changes in treatment to control for DBPs, assuming that
systems must still meet the minimal requirements of the SWTR, are
likely to result in more significant changes in exposure from
Giardia than from most other waterborne pathogens identified to
date. Also, a dose response curve for Giardia is available for
estimating risk from exposure (Rose et al 1991).
Treatment Constraints
The DBPRAM was run to predict the ability for systems to
comply with different MCL targets for TTHMS and THAAs under two
sets of treatment constraints. The first set of constraints,
termed the "SWTR scenario", requires systems to:
- achieve at least a 3 log reduction in Giardia cysts by
filtration and disinfection prior to the first customer;
8
-------
- maintain a residual of at least 0.2 mg/l free chlorine (or
1 mg/l total chlorine if chloramines are used) at the end of
the distribution system1;
- maintain the free chlorine residual below 2 mg/l at the
first customer because of concern from taste;
- maintain the chloramine residual below 3 mg/l at the first
customer to remain below the possible MCLG (see Table II)2
and;
- maintain a pH of at least 7.6 in the disinfected water
entering the distribution system if the hardness is greater
than 100 mg/l; maintain a pH of at least 8.0 if the hardness
is less than 100 mg/l.3
The second set of constraints apply to an "Enhanced SWTR"
(ESWTR) scenario. This scenario assures that systems with poorer
source waters disinfect sufficiently to maintain a low pathogen
risk (using Giardia as the target organism) while still complying
with the D/DBP rule. The constraints in this scenario are
identical to those of the "SWTR scenario11 except that systems are
required to achieve higher levels of treatment, depending upon
the concentration of Giardia cyst in the source water/ as
follows:
1 The disinfectant residual constraint in the distribution
system is assumed necessary to meet the total coliform rule and
the requirement to maintain a "detectable" residual in at least
95% of the samples taken in the distribution system under the
SWTR.
2 Since the above analysis was conducted, EPA has begun
considering an MCLG for chloramines of 4 mg/l measured as total
chlorine (USEPAb, 1992) to account for the health effects of
monochloramine which cannot be practically measured by itself.
In future model runs EPA will consider a constraint of 4 mg/l
rather than 3 mg/l measured as total chlorine.
3 These constraints are to simulate corrosion control and
compliance with the lead rule.
-------
Giardig
cyst concentration/100 L Giardia
Removal/Inactivation
S 1 3-log
1-10 4-log
10-100 5-log
~~ * 6-log
7-log
10-100
100-1,000
1,000-10,00
The modeling approach assumes that under both the SWTR and
ESWTR treatment constraints, filtration plants are operating in
accordance with the SWTR and are achieving 2.5 log removal of
Giardia cysts. The model assumes that any additional treatment
which is required is achieved through disinfection. The modeling
approach assumes that plants will operate disinfection slightly
beyond the minimum requirements to ensure continuous compliance
during abnormal conditions. For disinfection through the
treatment plant with chlorine, the chlorine dose is calculated to
achieve at least 20% greater inactivation than required in the
SWTR or ESWTR.
The only difference between the modeling approach for the
ESWTR and the SWTR scenario is the amount of disinfection
required in the plant depending on the raw water Giardia
concentrations.
Initial Conditions and Compliance Sorting Routine
One hundred hypothetical treatment plants were individually
evaluated through the treatment model to predict plant effluent
and distributed water quality and to determine whether a system
meets the treatment constraints and DBF MCLs.
Due to the large number of treatment configurations, and
because systems are currently making changes in treatment to
comply with the SWTR, total coliform, and lead rules, it was not
possible to simulate a baseline of treatment practice. Rather, it
was assumed that all treatment plants initially use chlorine as
the only disinfectant, though plants use alternate disinfectants
often because of source water quality with substantial THM
precursors and the need to meet the current TTHM rule. The
validity of the model for predicting treatment changes that
4 EPA guidelines in the Guidance Manual to the SWTR (USEPA,
1991b) are based on a geometric mean but in this modeling
analysis the arithmetic mean is used for purposes of conducting
statistical analysis.
10
-------
systems would make to comply with new DBFs standards was in part
confirmed by comparing the percentage of systems predicted to
modify their treatment to achieve a TTHM MCL of 100 yg/1 with the
percentage of plants that actually made such modifications to
meet a target of 100 ng/l MCL (Gelderloos et al 1992, Cromwell et
al 1992).
The predicted water quality of the 100 plants is evaluated
through a "compliance sorting routine1*, sorting the treatment
model output according to alternative DBF MCLs. If a system does
not meet the constraints established by the SWTR (or the ESWTR)
and the MCL of interest, the system implements the lowest cost
modification necessary to meet the constraints. If the new
treatment method is not sufficient to achieve compliance with
either the treatment constraints or the DBF MCL of interest, the
next lowest cost modification is added. This process is repeated
for each system until it complies with the treatment constraints
and the DBF MCL of interest.
The compliance sorting routine assumes that systems strive
to comply with the MCL with a 20% margin of safety. In other
words, for an MCL of 40 ug/1, it is assumed that systems would
add sufficient treatment to consistently meet a level of 32 ug/1
(40 - 0.2 x 40 = 32). Similarly, if a system needs 0.5 log of
Glardia inactivation to minimally meet the SWTR treatment
constraints (2.5 log removal is assumed by filtration) then a
system would strive to attempt to achieve this level with a 20%
margin of safety, i.e., 0.6 log inactivation (0.5 + 0.5 x 0.2 =
0.6). If higher levels of inactivation are needed to meet an
ESWTR, then these levels are adjusted upward by 20% (e.g.,
achieving a 3 log level of inactivation would require 3.6 log of
inactivation).
For coagulation/filtration plants using chlorine as the only
disinfectant, the approach assumes the following order of
alternatives, from the lowest cost alternative to the highest
cost alternative:
• Elimination of pre-chlorination, if practiced;
• Installation of ammonia feed system to provide chloramines
as secondary disinfectant;
• Increasing coagulant dose to improve DBF precursor removal;
• Switching to ozone/chloramines for disinfection; and
• Adding GAC (30 minute empty bed contact time with 6 month
carbon regeneration frequency) or membranes.
The implementation of a chlorine/chloramine or an
ozone/chloramine disinfection strategy may be effective for
11
-------
complying with both the SWTR and ESWTR, and TTHM or THA limits.
However, these alternative disinfection strategies may not be
effective for other DBFs. For example, if a system has high
bromide concentrations in the source water the use of ozone may
lead to high bromate concentrations. Therefore, under some
possible regulatory scenarios, the implementation of certain
disinfection alternatives to chlorine may not be sufficient to
comply with all MCLs. Therefore, the impacts of alternative DBF
MCLs were evaluated under two different compliance approaches
(see Figure 3). One compliance approach assumes that alternative
disinfection strategies are available, with the other assuming
that alternative disinfection strategies are not available, i.e.,
systems can only chlorinate.
The compliance approach assuming no availability of alternate
disinfectants allows for prediction of the percentage of systems
that would achieve different MCL targets through only use of
precursor removal technologies. Estimation of the feasible
performance of candidate BATs, such as optimized filtration for
DBP precursor removal, or optimized filtration and GAC, if
applied to source waters throughout the U.S., could form the
basis for defining the MCL, e.g., the level of DBP(s) that at
least 90 percent of the systems could achieve using the candidate
BAT. The same approach can be applied for defining the
performance of candidate BATs using alternate disinfectants.
Analysis of these two compliance approaches also makes possible
national cost comparisons between allowed use of alternate
disinfectants versus sole use of precursor removal strategies to
achieve different regulatory targets. Such cost analysis is
discussed elsewhere (Cromwell et al.. 1992).
Risk Analysis
DBFs
The mean ~OBP concentrations predicted in the model described
above were used to estimate cancer risk in each system. Figure 4
illustrates the methodology and the risk factors used to estimate
the cancer cases per year from the predicted DBP drinking water
concentrations. Risk factors for both the upper 95% confidence
interval and the maximum likelihood (MLE) estimate were used in
the analysis.
Though chlorodibromomethane has been tentatively classified
as a group C carcinogen, a cancer risk factor equivalent to that
for bromodichloromethane is assumed. A cancer risk factor for
trichloroacetic acid was used based on animal data (USEPAb,
1992), even though it also is tentatively classified as a group C
12
-------
carcinogen.5 No risk numbers are available for the other HAAs
(monochloroacetic, and mono and dibromoacetic acid) included in
the predicted THAA occurrence levels and thus a zero risk factor
is assumed for them.
Not included in the analysis are occurrence or risk
estimates for chloral hydrate or bromate, or the chloro-bromo
HAAs.
Giardia
Figure 5 illustrates the methodology for estimating Giardia
occurrence and risk at the first customer. The analysis assumed
a reduction in treatment performance of 1 log 5% of the time, all
viable cysts as infective in humans, infections equivalent to
illness for the purpose of identifying outbreaks, and outbreaks
as identified if the illness rate at the first customer exceeded
1% within a 30 day period. Other assumptions are also used in
different analyses (Grubbs et al 1992) but are not described
here. A dose response curve (Rose et al., 1991) was used to
estimate infections from the predicted occurrence levels.
Not indicated in Figure 5 is the assumption that the
exposure predicted at the first customer represents 10% of the
exposure for the total population:
number of infections per system * infection rate at first
customer x total system population x 0.10.
This equations assumes die off of Giardia cysts from exposure to
a disinfectant residual as they move through the distribution
system. People receiving water in the middle of the distribution
system, because of the large CT values achieved through such
points, would be expected to be exposed to much lower Giardia
cyst concentrations than people near the first customer. The 10%
factor was assumed to account for an approximate weighted average
for all systems for such effects.
Not included in this analysis are risks from other
pathogens. This exclusion may be a significant limitation
regarding pathogens more resistant to disinfection than are
Giardia cysts (e.g.., Cryptosporidium, possibly some viruses) and
bacterial pathogens which might grow in the distribution system
(e.g., Legionella and mycobacteria). Except for such concerns,
it appears reasonable to assume that if risks from Giardia cysts
5 Risk factors were used for these compounds because,
conceivably, the cancer classification for these compounds could
change to 83 at a later date based on additional review of the
existing data or new evidence from new data.
13
-------
are shown to be negligible then the risks from most other
pathogens would also be negligible. If the risks from Giardia
cysts are shown to be significant, then the risks from exposure
to pathogens less resistant to disinfection than Giardia may also
be significant, but to a lesser degree.
Modeling Results
Aggregate Analysis
Figures 6-13 illustrate the predicted ability, based on
the DBPRAM, for public drinking water supplies in the U.S. to
meet different MCL targets for THAAs and TTHMs by different
treatment technologies given treatment constraints to control for
pathogenic organisms. The analyses in these Figures represent
systems serving more than 10,000 people each and 103,000,000
collectively. These systems currently filter but do not soften.
For the purpose of estimating exposure and associated risk, each
of the 100 hypothetical water treatment plants was assumed to
Represent the same population (1,030,000 people). A detailed
description of the information in Figure 6 is provided as an
example of how to interpret the information in Figures 7-13 which
are similarly structured.
Figure 6 - SffTR ft/Alternative Disinfection. This Figure
illustrates the analysis under which systems were allowed to use
alternate disinfectants and DBF precursor technologies to meet
MCL targets for THAAs ranging from 60 to 10 ug/1.
Column one indicates the MCL target level for THAAs (i.e., mono,
di, and tri-chloroacetic acid, and mono and dibromo-acetic acid).
Column two indicates the degree of treatment applied (see
treatment code description in upper left). Treatment code #1
indicates no treatment has been applied other than that necessary
to meet the SWTR scenario treatment constraints.
Column thr«« indicates the percentage of systems predicted to use
different levels of treatment (from column two) to meet the
target MCL based on lowest cost. For example, at an MCL of 30
the model predicts that 48% of the systems would meet this MCL
without any treatment change; 12% would eliminate
prechlorination; 22% would eliminate prechlorination and add
chloramines; 14% would eliminate prechlorination, add
chloramines, and increase alum dose to improve precursor removal;
etc. The numbers in column three sum to 100% for each MCL
alternative.
Column four indicates the cumulative percent of systems meeting
the target MCL through the specified level of treatment
(graphically represented in the small figure on the right hand
14
-------
side of the page). For example, at an MCL of 20 ug/1, 83% of the
systems would be able to meet this standard (a level of 16 ug/1
with the 20% margin of safety) without using GAC (treatment code
#6) . It is important to note that all of the treatment
constraints would have to be met through each level of specified
treatment. Thus, if chloramines are used, the system must still
provide enough CT with either chlorine or ozone (if chlorine can
not be feasibly used, i.e., treatment code /5) prior to the
addition of ammonia, to achieve the 0.6 log inactivation of
Giardia cysts necessary to minimally meet the SWTR.
If an MCL for THAAs were set at 10 ug/1, the model predicts that
only 89% (last line in column) would be able to meet this MCL
even with installation of GAC following filtration. As a result,
11% of systems would qualify for a variance if they installed GAC
(see small figure in lower left).
Columns five and six indicate the predicted mean concentration of
TTHMs and THAAs for the systems through the indicated level of
treatment to meet the target THAA MCL. Note the effect of
decreasing TTHMs as a function of lowering the MCL for HAAs
without having any MCL for TTHMs. The fourth small figure from
the bottom left side of the page graphically represents this
relationship.
The first lines in each MCL target grouping indicate the
predicted occurrence of TTHMs (55 ug/1) and THAAs (24 ug/1) with
no treatment beyond the requirement to minimally meet the SWTR
scenario treatment constraints using chlorine as the disinfectant
and not having any MCL for either TTHMs or THAAs. However, this
mean value does not include systems which are unable to meet the
-treatment constraints. Lines 2 through 6 under each MCL grouping
indicate the cumulative mean concentration through each treatment
tier, i.e., the average of those meeting and not meeting the MCL
through the indicated level of treatment. Thus, the mean values
in columns 5 and 6 through line 5 under-represent the expected
mean value for all systems making no treatment changes to comply
with an MCL.
The mean value- in line 6 for each MCL grouping represents the
mean value for all 100 hypothetical systems having made treatment
changes to comply with the target MCL.
Column 7 shows the predicted cancer cases per year, at different
MCLs for THAAs, caused by exposure from each of the THMs,
assuming the risk factors at the 95% upper confidence interval.
For example, a THAA standard of 20 ug/1 (with no standard for
TTHMs) is predicted to reduce the incidence of cancer attributed
to THMs from at least 33.8 cases per year to 16.1 cases per year.
A more precise way to interpret the data is to consider only line
6 under each MCL grouping. For example, a standard of 60 ug/1
vs. a standard of 20 ug/1, assuming no TTHM standard were in
15
-------
place, is predicted to reduce the incidence from 27.9 to 16.i
cases per year from exposure to THMs.
Column 8 indicates predicted cancer cases per year, at different
THAA MCLs, attributed to THMs using the risk factors based on
the maximum likelihood estimate. Note the much fever predicted
incidents of cancer versus those for the upper 95% confidence
interval estimates.
Column 9 indicates the predicted cancer cases per year caused by
exposure from THAAs assuming the risk factors are based on the
95% upper confidence interval.
Column 10 indicates the predicted cases of cancer attributed to
THAAs (i.e., only di and trichloroacetic acid since no risk
factors were available for the other HAAs), assuming the risk
factors are based on the maximum likelihood estimate. The second
small figure on the bottom left represents the combined annual
cancer cases caused from exposure to both THAAs and TTHMs using
the MLE risk factors.
Column 11 indicates the predicted .incidence of infection per
year. As the target MCL for THAAs decreases, the predicted
incidence of infections increases (see also the small third
figure from the bottom left). There are several factors
contributing to this phenomena.
The first is the presence of more DBF precursors preceding
disinfection resulting in higher disinfectant demand. As more
precursors are removed at lover MCLs because of the adoption of
precursor removal technologies, a lover disinfectant demand
exists in the vater and less disinfectant can be used to meet the
constraint of providing a disinfectant residual at the end of the
distribution system. In other words, in high oxidant demand
waters the constraint of maintaining a residual in the
distribution system is much more significant in affecting a
higher level of inactivation than the constraint of minimally
having to achieve the 0.5 log Giardia. inactivation criterion.
A second is that many of the systems that switch to chloramines
in lieu of chlorine as a residual disinfectant to lower THM or
HAA formation have high disinfectant demand water. When using
chlorine, such systems have to maintain a high dosage, and
subsequently a high CT prior to the first customer, in order to
maintain a free chlorine residual in the distribution system.
When these systems switched to a chloramine residual in the
distribution system, a much weaker disinfectant than chlorine,
these systems no longer provided the high level of inactivation
prior to the first customer. Rather, their level of inactivation
was reduced to only meeting the 0.5 log inactivation criterion.
16
-------
The difference in predicted Giardia incidence at high versus low
THAAs may be understated because the SWTR scenario assumes that
systems are designed and operated to minimally meet the
disinfection constraints, in reality, many systems may provide
higher levels of inactivation than would be minimally required
and still meet the high THAA MCLs. However, recent water
treatment plant data (LeChevalier et al 1991) appear to support
the validity of the predicted Giardia infections in the high MCL
range (Cromwell et al 1992).
Column 12 shows the predicted incidence of Giardia outbreaks
at the 90th percentile for different THAA MCLs. For example, ten
out of every one hundred systems, those with the highest Giardia
cyst concentrations, are predicted to have an outbreak within
every 3,089,000 years.at an MCL of 60 ug/1, and within every
0.658 years at an MCL of 40 ug/1.
The large deviations in the predicted outbreak intervals for
different target MCLs are explained by the large deviations
around the mean among the systems with high concentrations of
Giardia in their source water and the concurrent significant
decreases in inactivation that theoretically could occur with
treatment modifications. The model predictions suggest
potentially very significant increases in the occurrence of
waterborne outbreaks as the MCL for THAAs is lowered, i.e.,
without a parallel requirement for increased stringency of
regulation to also control for Giardia (e.g., an ESWTR).
It is important to note that the assumptions in this
analysis, such as the definition of what constitutes an outbreak
or a symptomatic response (i.e., infection versus illness) have
large uncertainties and will substantially affect the predicted
output. The assumed criteria that define an outbreak, for the
purpose of this analysis, appear to be overly conservative and
the predicted outbreak intervals cannot be considered reliable
estimates. Probably the most important indication of this
analysis (rather than the specific predicted outbreak intervals)
are the potentially large relative differences in likelihood for
waterborne outbreaks to occur as a function of meeting
increasingly stringent MCLs for DBFs.
Figure 7 - SWTR Without Alternative Disinfection. Under this
modeling scenario systems can only use precursor removal
technologies to meet the MCL target. The result of this
constraint is that more systems would require use of GAC to
achieve the target MCLs. If an MCL were set at 10 ug/1 the model
predicts that, even with use of GAC, 17% of the systems would not
be able to meet this limit.
17
-------
At an MCL of 60 ug/1, this scenario predicts a higher number
of Giardia infections (246,531) than those predicted when "
alternate disinfectants could be used (213,045, Figure 6). This
higher number is attributed to more systems reducing disinfectant
demand than in Figure 6, because of greater use of precursor
removal technologies, and subsequently lowering their
disinfection dose while still being able to meet the minimal
disinfection treatment requirements.
Figure 8 - Enhanced SffTR filth Alternative Disinfection. This
analysis differs from Figure 6 in that systems must achieve
proportionally higher levels of Giardia inactivation as the
source water concentration of Giardia cysts increases. Predicted
Giardia infection incidence is reduced to below the 1/10,000
infection per person per year goal of the SWTR through
implementation of the recommended guidance. The probability for
outbreaks to occur is eliminated.
The predicted level of treatment necessary to meet the different
MCL targets (columns 2, 3, and 4) is not much different than
that predicted in Figure 6 without the ESWTR. The additional
contact time within the plant required by the ESWTR (which is
assumed can be added, if needed) to meet the inactivation target,
does not significantly change the predicted concentration of
THAAs at the average customer. As average residence time in the*
distribution system is about 1.6 days, the additional 1 to 3
hours in the plant, if chlorine is used, does not significantly
increase THAA concentration to the average customer.
These results imply that requiring greater inactivation of
Qiardia for more highly contaminated source waters will not have
a significant impact on the national costs of compliance. The
cost for increased contact basin size following
coagulation/filtration or GAG to meet the inactivation
requirements of the ESWTR does not significantly influence the
total cost. However, if systems do not have space to increase
contact basin size, other options might then be required (e.g., a
stronger disinfectant such as ozone which could significantly
increase treatment costs).
Figure 9 - Enhanced SWTR Without Alternative Disinfection. This
scenario is the same as Figure 8 except systems can only use
precursor removal technologies to achieve the different THAA
MCLs. The results in Figure 9 are similar to those in Figure 8
except higher levels of precursor removal technology are needed
to achieve the MCLs. The technologies predicted to achieve the
MCLs are almost identical to those in Figure 7 where no alternate
disinfectants were allowed but less stringent disinfection
requirements were required. This result is explained by the same
logic as described under the description of Figure 8.
18
-------
Figures 10 - 13 - These Figures are analogous to Figures 6 -9
except analysis targets for TTHMs in lieu of THAAs. Note how the
concentrations of predicted THAAs are affected by systems
targeting to achieve different TTHM MCLs, indicated in the
fourth figures from the bottom left.
The results at the target MCL of 100 ug/1 in Figures 10 are of
particular interest. The constraints in the model that led to
these results most closely represent the current situation today
under the existing TTHM standard of 100 ug/1. The main difference
is that some plants have not yet made treatment modifications to
comply with the requirements of the SWTR, total coliform, and
lead rules and that this may influence existing levels of TTHMs.
Figure 10 indicates that 61% of the systems using only chlorine
as their disinfectant would be able to meet a TTHM MCL of 100
ug/1 without making any treatment modifications. It also
indicates that 21% would eliminate prechlorination, and another
18% would switch to chloramination in order to meet a TTHM MCL of
100. These results appear to approximate the changes that large
filtered systems actually made to comply with the TTHM rule since
it was promulgated in 1979 (Gelderloos 1992; Cromwell et al
X992).
Figure 10 also indicates predicted mean of 40 ug/1 for TTHMs and
20 ug/1 for THAAs in the distribution systems for all systems
required to comply with a TTHM MCL of 100 ug/1. These results
also appear to approximate the existing mean occurrence level for
these groups of compounds in large filtered supplies today
(Cromwell et al 1992).
Disaggregate Analysis (Changes in Risk to Individual Systems)
The modeling results discussed so far, which pertain to 103
million people, indicate that the risk from Giardia infections
may increase by several hundred thousand per year under
significantly more stringent DBF regulations than currently
exist, unless-revisions are made to the SWTR. What may be even
more significant are the potential changes in microbial risk that
might occur in populations of individual systems under new DBF
regulations.
In order to understand the potential for changes in risk at
the system level it becomes essential to examine the source water
quality and level of treatment for individual systems. Figure 14
presents a plot of influent Giardia concentrations (Yw axis)
versus estimated total log reductions achieved ("X" axis) at the
plant effluent (first customer) based upon survey results from 46
different plants in the U.S. (LeChevalier et al 1991). A
19
-------
critical underlying feature of these data, which can be
considered a pre-sWTR baseline, made clear by the random scatter
pattern in this plot, is that there appears to be no correlation
between the influent Giardia concentration and the total log
reduction achieved.
The vertical line at 3 log reduction indicates the minimal
level of treatment required through the first customer by the
SWTR, although higher levels of treatment may be needed to
maintain the disinfectant residual in the distribution system.
The diagonal line through the points represents the formula of
providing the level of treatment, prescribed in the SWTR Guidance
Manual, for systems to achieve a level of risk of less than one
infection per 10,000 people per year (1/10,000). The points to
the right of the diagonal line represent plants achieving more
total log reduction than needed to meet the 1/10,000 risk target;
points to the left represent plants achieving less total log
reduction than needed to meet the 1/10,000 risk target.
The lack of correlation between the influent Giardia concen-
tration and the total log reduction achieved could be attribut-
able to numerous factors: l) the lack of monitoring capability
for Giardia and the knowledge of when to treat more or less; 2)
attempts to comply with the current TTHM MCL (significantly, many
of the plants to the left of the line are using chloramines.); 3)
attempts to comply with current standards for coliforms and to
maintain a chlorine residual in the distribution system (e.g.,
Ten State Standards); or 4) other treatment objectives served by
pre-disinfection.
Figure 14 illustrates the relevance of considering changes
in risk at the plant level rather than the average levels across
all plants. The clear implication of the data in Figure 14 is
that the minimum 3-log reduction requirement of the SWTR does not
appear sufficient to achieve the target 1/10,000 risk level in
most systems. The higher levels of total reduction specified in
the SWTR Guidance Document appear to be necessary in order to
assure meeting the target risk level. Another relevant issue is
how the distribution of points indicated in Figure 14 might
change as a function of different DBF regulatory targets.
The Monte Carlo simulation framework established in the
DBPRAM is extremely well-suited to analyzing the relationship
between influent Giardia levels and total log-reductions achieved
at each individual plant. Since the influent Giardia
distribution used to drive the simulation was developed from the
above-described data, some direct similarities to the findings
are to be expected in the model results. Unfortunately, a
complete calibration check of the model against the survey
findings (LeChevallier et al 1991) is not possible because the
data were collected in a pre-SWTR compliance environment whereas
20
-------
the model is constructed to include the SWTR requirements as
initial constraints. Some similarities are nonetheless evident.
As discussed above, there are many possible reasons why the
data in Figure 14 indicate a lack of correlation between the
influent Giardia level and the total log reduction achieved. In
the over-simplified reality of the DBPRAM, a few assumptions
dominate this relationship. Figure 15 presents a plot similar to
that of Figure 14 showing simulation results for a scenario that
assumes implementation of the SWTR (with alternate disinfectants)
and a TTHM standard of 100 ug/1. Of 100 simulated plants, 39
were determined by the DBPRAM algorithm to require treatment
adjustments in order to meet a TTHM MCL of 100 ug/1. Only these
39 points are plotted in Figure 15. The plotted points reflect
the relationship between influent Giardia and total log reduction
after compliance with these modeling constraints.
Many of the 39 points in Figure 15 appear in a vertical line
reflecting the SWTR requirement for a minimum of 3-logs of total
reduction. By contrast to the pre-SWTR baseline reflected in the
data in Figure 14, some of the points along the 3-log line in
Figure 15 would represent a higher level of total reduction than
in the pre-SWTR condition while others would represent a lower
level. In the simplified model logic, those plants that were not
previously achieving a 3-log level would have increased their
level of total reduction to equal 3-logs, while some of those
achieving more than a 3-log reduction would have dropped their
level of total reduction to the 3-log minimum. Notably, all the
points aligned vertically along the 3-log minimum are to the left
of the diagonal line that defines the 1/10,000 risk threshold
that the SWTR sought to achieve, indicating that they are
incurring risks greater than this target level.
The remaining points plotted in Figure 15 show more than 3-
logs of total reduction. Some of them are above the 1/10,000
risk threshold — to the right of the diagonal — while others
are below it — to the left of the diagonal. In tracing the
source of these results through the model logic, it appears that
they exhibit more than the 3-log minimum due to the additional
SWTR constraint of maintaining a disinfectant residual in the
distribution system. The simulated plants achieving more than a
3-log total reduction at the first customer (see Figure 15)
represent conditions where enough total organic carbon (TOC) is
present to require elevated chlorine dosages to overcome the
chlorine demand of the TOC in the distribution system, but yet
where the resulting contribution to TTHM formation is not
significant enough to require additional TOC removal.
Figure 16 illustrates the effect of a tighter by-product MCL
of 25 ug/1. Here 83 out of 103 plants would be required to
change treatment in order to comply. As shown in the graph, the
21
-------
result is that all the points are aligned along the 3-log minimum
line. Those plants with more than 3-logs of total reduction in
Figure 15 have been guided by the treatment algorithm into TOC
removal strategies which in turn reduced the chlorine demands
that had to be overcome in order to meet the SWTR residual
requirement in the distribution system. As a result, the model
predicts that all plants required to change treatment to meet a
TTHM standard of 25 ug/1 would reduce their total log reduction
to the 3-log minimum.
As shown in Figure 16, the convergence to the 3-log minimum
results in a risk level higher than the 1/10,000 target for all
of the 83 plants compelled to modify treatment in meeting a TTHM
MCL of 25 ug/1. The resulting change in endemic incidence of
Giardia infections will be larger for some plants than for
others. In particular, there may be some plants for which the
co-occurrence of high influent TOC and high influent Giardia
result in a very large increase in incidence due to significant
drops in chlorine dosages for compliance with SWTR residual
requirement after TOC removal has been triggered by the need to
meet a TTHM MCL. The increase in incidence of Giardia infections
has been computed based on the change in total log-reduction
between one MCL and another, as computed by the OBPRAM. Results
of this analysis are presented in Figures 17 and 18.
Figure 17 plots the cumulative distribution of the 100
simulated plants by their level of increase in annual Giardia
infections per 1 million persons when moving from a TTHM MCL of
100 ug/1 to an MCL of 75 ug/1. The plot confirms that some
plants would face extreme conditions, resulting in significant
increases in Giardia infections. Nearly 10 percent of the plants
could experience increases of between 1,000 and 10,000 infections
per 1 million population. The top end of this range approaches
outbreak proportions. Figure 18 shows that the impact of a shift
from a TTHM MCL of 100 ug/1 to a TTHM MCL of 25 ug/1 would
produce increases in Giardia incidence on this same order of
magnitude for over 30 percent of the plants. Figures 19 and 20
present the same results that are plotted in Figures 17 and 18 as
bar charts, emphasizing the impacts at the extreme ends of the
range. -
The ultimate analysis that must be made is a comparison
between the increase in incidence of Giardia infections and the
corresponding decrease in cancer incidence that results from
alternative MCLs. Combining the above analysis of Giardia
infections with the by-product concentrations simultaneously
projected by the DBPRAM produces such an analysis. Results are
summarized graphically in the log-log plots presented in Figures
21 and 22. Figure 21 plots increased incidence of Giardia
infections per 1 million persons versus decreased cancer
incidence per 1 million persons for the .plants having to change
22
-------
treatment when the TTHM MCL is changed from loo ug/1 to 75 ug/1
in a SWTR scenario (with alternate disinfectants).
Figure 22 presents the same picture for the plants that
would have to change treatment if the TTHM MCL is changed from
100 ug/1 to 25 ug/1. The pattern of the points plotted in these
diagrams indicates that the degree of change in cancer incidence
spans only one order of magnitude (less than one case per 1
million persons per year) while the degree of change in incidence
of Giardia infections spans several orders of magnitude (from one
to more than 10,000 infections per 1 million persons per year).
This outcome indicates that the microbial risk is more sensitive
to treatment changes needed to meet TTHM MCLs.
Figures 23 through 30 present the same series of graphs as
Figures 15 through 22 except that the entire analysis is
structured in terms of alternative MCLs for Total Haloacetic
acids. The interpretation is identical to that described above.
One obvious direction indicated by these results would be to
strengthen the SWTR before proceeding with requirements for
additional by-product controls. In the ESWTR scenario, it is
assumed that the portion of the SWTR Guidance Document which
recommends an additional log of inactivation for each additional
log of influent Giardia. is converted to a mandatory requirement.
In terms of the graphs presented in Figures 15 and 16, this
requirement would mean that the model logic would result in the
plants being lined up along or to the right of the diagonal line
representing the 1/10,000 target risk level rather than along the
vertical line representing the 3-log minimum required by the
present SWTR. By definition of the 1/10,000 risk level as the
target, there would therefore be no microbial risk trade-offs.
The evaluation of this option with the DBPRAM showed that
adherence to the 1/10,000 risk level resulted in very little
additional cancer incidence and in very little modification in
treatment, and therefore very little additional cost. This
outcome results from the model assumption that additional
inactivation requirements at the first customer would be met with
additional contact time at the plant effluent. This additional
contact time allows inactivation requirements to be met without
increased chlorine dosages. While the additional contact time
contributes to inactivation, it does not contribute much in the
way of additional by-product formation because the additional
increment of contact time is trivial in proportion to the total
residence time in the distribution system.
23
-------
Model Shortcomings and Next Steps
This discussion will only pertain to model analysis or
modifications anticipated to occur in support of the development
of a proposed rule for D/DBPs. Much more research will be needed
over a longer time frame to address the unknowns identified in
the shaded areas of Figure 1.
Develop Predictive capability for Bromate
Under the modeling scenarios with alternative disinfectants
allowed, the DBPRAM currently selects ozone, with chloramines as
a residual disinfectant, as an alternative disinfectant to
chlorine to achieve lower levels of THMs and HAAs. This
selection may lead to high levels of bromate formation in systems
with high bromide concentration in the source water (AWWA, 1992).
Since exposure from bromate may result in greater risk than
exposure from THMs or HAAs (Table II), depending upon the bromide
concentration in the source water and the subsequent formation of
bromate upon ozonation, the model may be inappropriately
selecting ozone in the compliance sorting routine. The model may
also be underestimating the total cancer risk from the selected
distribution of technologies to meet a target MCL for THMs or
HAAs. EPA is now developing formation equations for bromate and
intends to include these equations in the modeling analysis in
developing the proposed rule*
Extend Modeling Analysis to Smaller Systems
The current analysis only pertains to systems serving more
than 10,000 people. Since EPA intends for the new DBP
regulations to pertain to smaller systems, it becomes important
to evaluate the potential risk and cost impacts in such systems.
EPA is now expanding the application of DBPRAM to systems
serving less than 10,000 people. This analysis will consider the
following factors as being different from larger systems: a)
possible smaller distribution system residence times, b) the
potential for higher treatment failure rates, c) the effects of
not having to already meet an existing TTHM MCL on net changes in
total risk and d) economies of scale in installing and operating
different technologies (e.g., using membrane filtration in lieu
of GAC as a precursor removal technology).
Extend Modeling Analysis to Systems Using Lime Softening
Systems that use surface water and lime softening may
represent those systems having the greatest difficulty in meeting
microbial and DBP standards. Since these systems operate at high
pH, it is more difficult to achieve effective inactivation with
24
-------
most disinfectants. High pH conditions may also make it
substantially more difficult to remove DBP precursors.
Therefore, systems using lime softening may be the best indicator
for characterizing potential changes in risk from exposure to
DBFs and pathogens that might result under new regulations.
EPA is developing formation equations to predict THMs and
HAAs for systems using lime softening. Also, new data are being
developed to evaluate the inactivation efficiency of disinfection
at high pHs (Logsdon et al, 1992). EPA anticipates using this new
information to expand DBPRAM to predict potential changes in risk
in lime softening systems under new OBP regulations.
Model Validation and Refinement
Some model validation analysis has already been conducted
(Cromwell et al 1992, USEPAc, 1992) but much more is needed.
DBP occurrence data being collected at water treatment plants
will be compared with model predictions to evaluate the accuracy
of model predictions. Sensitivity analyses are being conducted to
determine which of the parameters and model assumptions have the
greatest impact on affecting model predictions. EPA is also
considering the use of different TKM and HAA formation equations
than are now being used in the DBPRAM (AWWA 1992).
The DBPRAM currently assumes that the data collected from
one survey (LeChevalier et. al., 1991) is representative of the
amplitude and variation among communities using filtered surface
supplies in the US. The DBPRAM also assumes that the data
collected from another source (Hibler et. al., 1987) is
representative of the variation within a community. EPA will
jreevaluate and possibly modify these assumptions as more data
become available. Also, several "what if" scenarios pertaining
to the probabilities that infections may lead to illness, and the
rate of illnesses that would likely cause an outbreak to be
identified, will be considered in future model runs.
Conclusions
1. EPA has developed a model that predicts the occurrence of
trihalomethanes (THMs), haloacetic acids (mono, di, and tri-
chloroacetic acid; and mono, and di-bromoacetic acid,
collectively cited as THAAs), and Giardia cysts in finished
drinking waters in the U.S. as a function of source water quality
and treatment practice. The model predicts risk of cancer from
THMs, and di- and trichloroacetic acid, and risk of infection
from Giardia cysts. The model is being used to predict the
potential changes in treatment and risk that might result from
different regulatory targets.
25
-------
2. Model predictions currently only pertain to filtered systems
that do not use lime softening, serve greater than 10,000 people,
and collectively distribute water to 103 million people in the
U.S. The model is currently being expanded to include analysis
for systems'serving less than 10,000 people and those which use
lime softening.
3. More data are needed to evaluate the validity of the model
predictions. Refinements in the model assumptions and predictive
equations are anticipated and, therefore, the model predictions
should be considered tentative. Two risk estimates for cancer
are given: numbers without parenthesis indicate estimates based
upon maximum likelihood risk factors, and numbers within
parenthesis indicate estimates based upon the upper 95%
confidence interval. Risk estimates for Giardia infections are
based upon a maximum likelihood analysis.
4. Model predictions indicate that at the existing TTKM MCL of
100 ug/1, following compliance with the SWTR, total coliform
rule, and lead rules, 45 (173) cases of cancer per year will
occur based upon predicted exposure from THMs [1, (24) cases] and
di- and tri-chloroacetic acid [44, (139) cases].
5. Model predictions indicate that if an MCL of 25 ug/1 were set
for TTHMs, and alternative disinfectants to chlorine were not
allowed, 41% of the systems would be able to meet this standard
.without the use of GAC. 5% of the systems could not meet this
standard with the use of deep bed GAC following optimized
filtration for precursor removal, with allowed use of alternate
disinfectants, 97% of the systems could meet a 25 ug/1 standard
without the use of GAC.
6. Model predictions indicate that if an MCL of 10 ug/1 were set
for THAAs, and alternative disinfectants to chlorine were not
allowed, 43% of the systems would be able to meet this standard
without the use of GAC. 17% of the systems would not be able to
meet this standard with the use of deep bed GAC following
optimized filtration for precursor removal. With allowed use of
alternate disinfectants, 50% of the systems could meet the
standard without the use of GAC. 11% of the systems could not
meet the standard with the use of GAC.
7. Model predictions indicate that at a TTHM MCL of 25 ug/1, with
no other DBF standard and allowed use of alternate disinfectants,
27 (102) cancer cases per year would occur based upon predicted
exposure from THMs 0.4 (10 cases) and di- and tri- chloroacetic
acid 26.2 (82 cases). At a THAA MCL of 10 ug/1 (with no other
DBF standard), 12 (45) cancer cases per year would occur based
upon predicted exposure from THMs [0.4 (10.4) cases] and di and
tri-chloroacetic acid [11.3 (35.3) cases].
26
-------
8. Model predictions indicate that if systems were to only
minimally comply with SWTR, i.e., not follow guidance of
providing higher levels of inactivation for poorer source water
quality, 340,000 Giardia infections per year would result under
the TTHM MCL of 100 ug/1. A new standard of 25 ug/1 for TTHMs
alone would result in 170,000 additional Giardia infections per
year. A new standard of 10 ug/1 for THAAs alone would lead to
226,000 additional Giardia infections per year.
9. Under any new DBF MCL, e.g., 75 ug/1 for TTHMs or 50 ug/1 for
THAAs, some systems would experience increases of 1000 to 10,000
Giardia infections per million people per year with decreases in
cancer risk of 0.1 to 1 case per million people per year (based
upon maximum likelihood estimates. If MCLs were set at, e.g., 25
ug/1 for TTHMs or 10 ug/1 for THAAs, more than 25% of the systems
are predicted to experience this range of respective net changes
in risk.
10. The modeling results indicate that it is not possible to
conclude with certainty that the total risk will decrease as a
result of any specific regulatory action.
11. If an enhanced SWTR could be implemented to achieve an
infection rate of less than 1/10,000 at the first customer, where
higher levels of treatment for Giardia were required depending
upon the level of Giardia in the source water, the predicted
number of Giardia infections per year would be less than 500.
12. The model predicts that the universe of systems that use
surface water and which practice coagulation and filtration could
use almost the same decision tree of treatment technologies (but
with differences in the level of disinfection) to meet the
existing SWTR and stringent DBF MCLs as they could to meet an
enhanced SWTR and stringent DBF MCLs. If this conclusion is
correct, the costs for complying with these two different
regulatory scenarios would not be significantly different even
though much greater microbial protection could be provided under
an enhanced SWTR, if such a rule could be implemented.
13. If an enhanced SWTR could be implemented, any MCL for
controlling THMs or HAAs would lead to greater use of alternate
disinfectants, unless prevented through regulation, since these
technologies are more cost effective for lowering THMs or HAAs
than use of 6AC or membrane filtration. Since the health risks
from DBFs of alternate disinfectants to chlorine are largely
unknown, it remains possible that greater use of alternate
disinfectants could lead to greater health risks than those from
chlorinated DBFs such as THMs and HAAs.
27
-------
14. Precluding the use of alternate disinfectants to achieve
lower levels of THMs and HAAs would lead to substantially greater
treatment costs than if they were allowed. Also, use of
alternate disinfectants allows for substantially lower THM and
HAA concentrations to be achieved than if only precursor removal
technologies were allowed.
References
Cromwell, J.E.; Zhang, X.; Letkiewicz, F. J.; Regli, S. and B.
Macler. "Analysis of Potential Tradeoffs In Regulation of
Disinfection By-Products." Office of Ground Water and Drinking
Water Resource Center. Washington D.C. USEPA in press, 1992.
Gelderloos, A.B.; Harrington,G.W.; Owen, D.W.; Regli, S.:
Schaefer, J.K.; Cromwell, J.E. and X. Zhang. "Simulation of
Compliance Choices For Regulatory Impact Analysis". Office of
Ground Water and Drinking Water Resource Center. Washington D.C.
US EPA in press. 1992.
Grubbs, W. D., Macler, B. and S. Regli. "Modelling Giardia
Occurrence and Risk." Office of Ground Water and Drinking Water
Resource Center. Washington D.C. USEPA. EPA-811-B-92-005. 1992.
Hibler, C.P. Analysis of Municipal Water Samples for Cysts of
Giardia. Report prepared for Office of Drinking Water, U.S. EPA,
1987.
LeChevallier, M. W., Norton, D.N. and R. G. Lee. Occurrence of
-Giardia and Cryptosporidium in Surface Water Supplies. Applied
and Environmental Microbiology 57:2610-2616. 1991.
Letkiewicz, F.J.; Grubbs, W.G.; Lustiik, M.; Mosher, J.; Zhang,
X. and S. Regli. "Simulation of Raw Water Quality and Water
Treatment Characteristics in Public Water Supplies in Support of
the Disinfection By-Products Regulatory Impact Analysis". Office
of Ground Water and Drinking Water Resource Center. Washington
D.C. USEPA. EPA-811-R-92-001. 1992.
Logsdon, G, et al. The Removal and Disinfection Efficiency of
Lime Softening Process for Giardia and Viruses. Poster
presentation, AWWA Annual Conference, Van Couver, Canada. 1992.
Rose, J.B., Haas, C.N., and Regli, S. Risk Assessment and
Control of Waterborne Giardiasis. Amer. Jour. Public Health
81:709-713. 1991.
28
-------
USEPA. National Primary Drinking Water Regulations: Filtration;
Disinfection; Turbidity, Giardia Iambiia, viruses, Legionella,
and Heterotrophic Bacteria. Final Rule. Fed.Reg. 54:27486 (June
29,1989). 1989a.
USEPA. National Primary Drinking Water Regulations: Total
Coliforms. Fed. Reg. 54:27544 (June 29, 1989). 1989b.
USEPA. Status Report on Development of D/DBP Regulations. Office
of Ground Water and Drinking Water. USEPA. Wash. D.C.
199la.
USEPA. Guidance Manual for Compliance with the Filtration and
Disinfection Requirements for Public Water Systems Using Surface
Water Supplies; USEPA, Wash., D.C. 1991b.
USEPA. Occurrence Assessment for Disinfectants and Disinfection
By-Products (Phase 6a) in Public Drinking Water. Office of Ground
Water and Drinking Water Resource Center. Washington D.C. USEPA.
EPA-811-R-92-003. 1992a*
USEPA. Status Report on Development of MCLGs for Disinfectants
and Disinfection By-products. Office of Science and Technology.
1992b.
USEPA o Water Treatment Plant User's Manual. Office of Ground
Water and Drinking Water Resource Center. Washington D.C. USEPA.
EPA-811-8-B-92-001. 1992C
29
-------
Table I
Candidate Compounds For Regulation
trihalomethanes
possible health effect
MCLG and cancer status
*(RFD not "(tentative - not yet
yet approved) approved by CRAVE)
chloroform 0
bromodichloromethane 0
dibromochloromethane 60
bromoform 0
total THMs 0
haloacetic acids
trichloroacetic acid 100 /*g/l*
dichloroacetic acid 0
cancer, B2
cancer, B2
liver, C
cancer, B2
cancer, B2
liver, C*
cancer, B2*
total HAs 0 cancer, B2*
(mono, di, and tri chloroacetic acid; mono and dibromoacetic acid)
Other
chloral hydrate
bromate
chlorine
chloramines
chlorine dioxide
chlorite
chlorate
5/tg/l
0
4 mg/1*
3 mg/1**
0.8 mg/1*
0.3 mg/1 *
health advisory
liver, C*
cancer, B2*
blood, D*
blood, D*
blood, neurological, D*
blood, D*
** 4 mg/1 if measured as total chlorine
B2 compounds include possible human carcinogens, C compounds
include possible human carcinogens, and D compounds include those
with inadequate or no human and animal evidence of
carcinogenicity.
-------
Table II
Drinking Water Concentrations G*g/l) for
Lifetime Cancer Risks (upper 95% confidence interval)
DBF
Bromodichloromethane
Bromoform
Chloroform
Dichloroacetic acid**
Bromate**
10-
-5
known
range of
60
400
600
10
5
6
40
60
1
0.5
«-*
occur
0-100
0-50
0-340
0-80
?
•
** tentative risk estimate
-------
Table I
Candidate Compounds For Regulation
trihalomethanes possible health effect
MCLG and cancer status
(tentative)**
chloroform 0 cancer, B2
bromodichloromethane 0 cancer, B2
dibromochloromethane 60 jig/1 liver, C
bromoform 0 cancer, B2
total THMs 0 cancer, B2
haloacetic acids
trichloroacetic acid 100 /ig/1 liver, C
dichloroacetic acid 0 cancer, B2
total HAs 0 cancer, B2
(mono, di, and tri chloroacetic acid; mono and dibromoacetic acid)
Other
chloral hydrate 5 /zg/1 liver, C
bromate 0 cancer, B2
chlorine 4 mg/1 blood, D
chloramines 3 mg/1* blood, D
chlorine dioxide 0.8 mg/1 blood, neurological, D
chlorite 0.3 mg/1 blood, D
* 4 mg/1 if measured as total chlorine
** B2 compounds include probable human carcinogens, C compounds
include possible human carcinogens, and D compounds include those
with inadequate or no human and animal evidence of
carcinogenicity.
-------
Table II
Drinking Water Concentrations 0*g/l) for
Lifetime Cancer Risks (upper 95% confidence interval)
DBF
IQ"4
known
10'! range of
occurrence
Bromodichloromethane
Bromoform
Chloroform
Dichloroacetic acid**
Bromate**
100
400
600
10
5
10
40
60
1
0.5
0-100
0-50
0-340
0-80
9
•
** tentative risk estimate
-------
DBPRAM Modelling Framework
Monte Carlo Simulation
of Influent Water Quality
1
Batch Mode
Treatment and Distribution Model
, for Given Combinations of Treatments
Compliance Sorting Routine
for Alternative DBP MCLs
and MCL Combinations
ASSUMPTIONS:
Model Plant and
Model Distribution System
Characteristics
CONSTRAINTS:
• Microbial Treatment Objectives
(SWTR vs ESWTR)
• Corrosion Control Objectives
• Taste and Other Objectives
Compliance Vectors:
Percentage of Plants
Requiring Given Treatment
Combinations to Meet Given MCLs
EXPOSURE ASSESSMENT
Source
Water
Pathogens
G
i
a
r
d
i
a
O
t
h
*
r
; S
Distrib.
Sys.
Pathogens
THMs
Halo
Acids
C
h
1
o
r
0
B
r
0
m
o
Other
8
t
o
m
a
t
9
O
t
f)
£
r
$
I
Unit Cost Vectors:
Estimated Unit Costs for
Given Treatment
Combinations to
Meet Given MCLs
I
''FEASIBILITY-
ASSESSMENT
Definition of BAT
for Given MCL
Combinations
National ]
Cost
Estimate J
i
: Regulatory
Impact Analysis
Under E012291
RISK ASSESSMENT
Source
Water
Pathogens
G
i
a
r
d
i
a
0
t
h
6-
f
8
Distrib.
Sys.
Pathogens
THMs
Halo
Acids
C
h
I
0
r
0
B
r
o
m
o
Other
B
r
0
m
a
t
e
o
t
h
e
r
3
RISK
MANAGEMENT
Estimates
^Uncertainty
•••Unknowns
Decision
Estimables
Unknowns
-------
Figure 2
Major Parameters
Mfcrobial
Occurrence
Giardia
Raw Water Quality Parameters
IA
Hardness
IA
Turbidity
I/X
Alkalinity
|/\i
TOG
\^\
PH
\/\
UV
IA
Temp.
I/V
Bromide
Treatment
Process
Model
-------
Compliance Decision Tree
Switch to
i
Increase
Alum Dose
Switch to
03/NH2CI
1
Add
GAC
Eliminate Pre-CL
(if applicable)
I
1
Add
Nanofiltration
Add
QAC
Increase
Alum Dose
i
1
Add
Nanofiltration
-------
Figure 4
DBF Cancer Risk Calculation Data and Analysis
\
Population Exposed (persons) x OBP Concentration (jig/I) x Annual DW Risk Factor
/ cases/person/year \
ufl/l
- cases/year
. Cl Lifetime Risk Factor:
MLE or Upper 95% a
Lifetime Risk Factor
(cases/person/lifeti
mg/kg/day
iime\
Conversion Factors:
1
70kg x
x 10»ug/rng
Lifetime DW
Risk Factor
/ cases/person/lifetime'
UQ*
Conversion Factor
Annual DW
Risk Factor
/ cases/person/year \
/ 70 years/lifetime \
Divide Lifetime DW Risk Factor
into10MO»,10«to
obtain DW concentrations i
for these lifetime risk levels
MLE and Upper 95% Cl Lifetime
Risk Factors Used:
DBP
Chloroform
Bromodichlormethane (BDCM)
DibronKKhtoromethane (DBCM)
Bromotorm
Dichloroacettc acid
Tichkxoacetic acid
MLE
3.5xio*
8.2x10*
8.2x10*
4.4X104
88x10J
2.1xia»
Upper 95%
Cont. Int.
61x1&3
2.5xiaa
2.5x1 a2
7.9x1 a»
2.8x1 a*
6.3x10*
DW Concentrations (ufl/l)
for Lifetime Risks (for Upper 95% Cl)
10"
600
140
140
400
12.5
55
10*
60
14
14
40
1.25
5.5
10*
6
1.4
1.4
4
0.125
0.55
Comments
MLE from R. Bull, 95% Cl from draft HA
MLE from R. Bull, 95% Cl from draft HA
Assumed to be the same as BDCM
MLE from R. Bull. 95% Cl from draft HA
Provided by R. Cantilli (OST)
Provided by R. Cantilli (OST)
-------
Overview ofGiardia Modelling
Hibler Data
73 plants
variation between
plants: lognormal
LeChevalier Data
46 plants
variation between
plants: lognormal
t
temporal variation
within plants:
delta negative binomial
1
distribution used
for influent simulation
1
Assuming:
• 48% recovery
• 13% viability
• deletion of
values estimated at
limit of detection
100
simulated influents
1
Treatment Model
100
plant effluents (1 st customer)
t
summary statistics
distribution used
for simulation
of within plant
temporal variation
mean 90th%
1 I
compute expected
value of Rose dose
response function
compute expected value
and variance of Rose
dose response function
estimate
of endemic
incidence
rate
estimate
of outbreak
risk
Assuming:
• 5% failure rate
(foss of 1 log)
• 25% secondary
infection rate
•1% per 30 days
outbreak threshold
• viable cyst-
infective in humans
• infection-illness
-------
Ontt3O-Jtn-K
Figure 6
MODEL OUm/T(«Ml4M«*>«*4nlr>Q):
4 - pn-ahkM * •nmaila +
6 - pn-cMof « •Mnanta * •kin * aura
0 - pn-ahtoi •» ••onjila * ttom • arm* * QMS
MCt
(THMU)
00
00
00
oo
00
00
60
SO
60
60
60
•V
40
40
40
40
40
40
JO
10
JU
JU
JO
•
JU
30
20
JO
20
*>
to
10
10
10
10
«J
Tit.
Cafe
1
2
3
4
s
•
1
2
S
4
6
•
1
2
3
4
S
•
l
2
J
4
S
•
1
2
J
4
S
•
1
2
3
4
5
•
«at8v*
70
12
10
0
0
•
04
16
17
4
0
•
SO
10
22
6
1
O
44
12
22
14
3
1
3'
0
0
23
7
IT
10
3
1
22
6
80
OmiMW
*<*•*•.
tiMUnwu tiw (nckKta* ZO% CM
-------
Oretl 30-Jan-ez
Figure 7
MODEL OUTPUT (witaw w/o nRonkio): 8WTH W/O ALTERNATIVE D»»KCTON
pm-ehtof » moOtf •kun doM
pra-ohkw » Aim doM + OAC
MCL
(TMM*)
00
00
00
OP
so
so
so
e»
40
40
40
40
30
30
30
ao
20
JO
20
to
10
10
10
10
TM.
Cod*
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
.3
4
*f*8y».
Endk*
70
12
11
7
04
IS
14
7
SO
18
to
12
4S
12
10
«i
37
a
20
96
IB
3
21
ft
CunuMlM
*<48y».
24
10
14
•
24
10
13
•
0M
33.0
30.2
202
m*
33.6
20.0
26.6
W.7
33.6
26.0
26.0
M.0
33.0
27.0
24.6
mi
330
250
22.2
W.I
33.0
25.2
20.3
14.0
*» - - - 1 "-• -
._WiHUfMll
It
»A£
1.3
1.1
1.1
1.1
1.3
1.1
1.1
1.0
1.3
1.1
1.0
08
1.3
1.0
00
a0
1.3
10
0.0
tt7
1.3
0.0
0.7
««
""W
06«
1S33
1MO
134.0
1»7
1S33
137.3
1204
IMS
1543
1315
1160
tas.o
1533
120.0
1061
00-0
1S33
1W1
030
M.O
153,3
124.2
04.2
•0.0
56
IAE
40.4
4S.O
431
41.7
404
44.1
41 3
40.0
404
42.3
375
89.7
40.4
414
336
as-0
404
40.5
301
17*
40.4
300
27.1
M
-tzssA
kttaolkm
(Mint
aw^0»
277.142
•MOM
40HM6
H0.430
•atmMmsUan
^^^^BB5nP~^^^
(0M^ *.Vm. to OuHin^i
219M0
1.0*1
aoe0
0.174
0.094
aoo0
IMntOAC
10 tO » M
fHAAMCL (HfA4
It W 10 40 00
tHU MCL (y«AJ
•p«Mnl d *yM«n» kwWIttig Mch l/Mlmtnl tlw (InckjdM 20K ovw-d««tgn tictcK)
•cumuMlM pwcwit a« (yolwra ibl«lo moot MCL * Mch IfMlnMnI llw (toludM 20% mw-daolgn Itdor)
' m**n canccntoallan «l Mch Iradrrant tlw c* oil syMwm; twM mMIng MCL (ndttwwnal mMlhg MCL
1O% o( papuMkxi «npoo«J M lit* cuMomw
* lo> Ban p«c«ntll*
-------
Doit: ao-Jtn-02
Figure 8
MOGELOUmfTa): ENHANCED8«nRWVM.1BVkTnEIMMirECnON
Tl - Ml ra«iMig tfl*M» tntfMnl modi
2 - (IMMMipiv-iMafMkn
a - dtatoai Bt»-cMo« * add MMnunli
1
iwcvil ol •ytura,
"»«
ion.
THMte)
00
80
80
00
80
00
60
60
60
60
60
00
40
40
40
40
40
40
30
30
30
ao
30
00
ao
ao
ao
ao
ao
to
10
to
10
10
w
tfummmf —
W.
Cote
1
2
3
4
6
•
1
2
3
4
6
•
1
2
«<*8y*
Endktf
70
11
17
2
0
0
02
10
10
3
1
a
68
10
10
10
2
P
47
12
12
16
10
1
35
10
4
16
10
1?
17
3
0
16
16
•
MC
1.3
1.2
l.t
1.1
1.1
M
1.3
1.2
1.1
1.1
1.1
M
1.3
1.1
1.0
1.0
10
1.0
1.3
1.0
0.0
0.6
O.6
ao
1.3
1.0
0.6
0.7
0.6
ao
1.3
1.0
0.6
0.0
0.4
*«
IW
OS*
1S83
1432
1440
143.4
1434
HM
1683
130.6
138.3
1381
136.3
1«M
1S6.3
, 13&1
131.6
1232
121.7
»»».»
IS83
1326
1249
107.2
101.6
MM
1683
1280
1200
96.9
82.0
OM
1683
127.0
1220
88.6
87.8
18,4
Jto
HUE
60.3
48.1
40.6
48.1
40.1
40.1
60.3
44.8
44.8
43.6
43.6
*.V«. to OMbmfc
• WM»
MMy
WhBy
hOjttf
MMy
WMy
100%
00
I"
I
I
A UCt <««M
IrMlminl Ito (incMM iO% orar-dMOn IKIOC)
M>l* lo im« MCt. M Mch U*«m«nl Uw (iickxjM 2O% avw -dMlgn ticttn)
ton* mMttna MCt «nd tx>w na tnMtttg MCt
-------
Drdt 3O-Jtm-IU>
Figure 9
MODEL OUTPUT (witac* w/o «*mlnfl): ENHANCED 8WIH ¥«D ALIBWAT1VE DBIIFECIION
1 - not rapMifl *"•>•• IntfnnnlinodMMlkin
3 - «IMMte pm-cMofkuMan
• - rfMxl* pm-flhtof » MM«y Oun M
*• iftMtdaw * O*C
MCL
(IHAto)
00
80
00
•0
60
60
60
M
40
4O
40
40
30
30
30
•0
AJ
*'
J1J
»
IU
10
10
to
lit.
Goto
1
2
3
4
1
2
3
4
1
2
3
• 4
1
2
3
4
i
J
>
4
i
/
)
4
KalSr*
Endhg-
70
It
12
7
02
10
16
f
SO
to
ts
1»
47
12
21
3D
V,
to
JO
»
i;
3
22
M
«<48ya.
67
44
40
»
57
42
. 30
91
S7
4O
32
M
57
30
28
I/
i^LOn.?
26
21
20
10
26
21
10
1*
26
20
10
M
25
10
18
13
25
10
14
' 0
26
10
13
0
Cm«iki|
FTW fe
06«
340
. 31.2
20.0
*M
34.0
30.7
20.2
•M
34.0
20.0
27.3
tt.0
340
200
263
as.*
340
200
227
f*«
34.0
20.1
200
M4
MLE
1.3
1.2
1.1
».1
1.3
1.2
1.1
1.1
1.3
1.1
10
OL*
13
to
00
ao
13
1.O
00
0.7
1.3
1.O
0.0
at
*ta««,
HMto
06%
150.3
1432
1353
ttt.0.
160.3
1306
120.4
1MB
150.3
1361
1102
MM.4
15O3
1325
100.2
02.0
150.3
120.0
05.1
6»-4
1603
127.0
040
MM)
ULE
60.3
401
43.5
42.1
603
44.0
41.3
400
603
434
30.3
•&•
503
42.0
341
ao.4
50.3
41 4
30.5
W.»
60.3
41.0
27.2
M
-G&SA
htaBUoM
8M
aw
270
•10
MO
MO
•nfctnhiHton*
nJilm^r ~B*
(00») ».Yi» to aulbiwk
brfbkkM
•<^wi
fcigoj _H_
VHPHiy
MhRy
hIMy
WM»
MMy
100%
90%
O 8°*
3
0 70%
tu 00%
£
(0
§40%
ra 30%
° 20%
10%
0%
10 20 30 40 50
THAA MCL (ufl/L)
60
90M
00)1
•pwovil at lydwra kwUIUng Met) IfMlmtnt Itof (kiclxlM 2O% ovw-dMlgn tactoi)
• oumuWIm p*fcanl a« tyttwra ri>l« to fmrt MCL « Mcfi IrMlrranl llw (hdudM 20% «w - dMl^i tcdor)
n omonhrilan • Met) Iculmanl llw of M syMxra. >IOM mMlIng MCI tndtmMnol mMthg MCI.
10% of pcpuMion upoad * IIM cudamw
' MlimM* lu eon p*c«nlil*
-------
Onrt.
MODEL OUIPUT(«iilm«*»Mta*IQ): 8W1RWS ALTERNATIVE UBMECnON
Figure 10
HDL
J1TOM*
100
100
100
100
100
MP
76
76
76
76
76
n
so
so
60
SO
so
•
»
»
M
»
2S
*
Til.
Onto
1
2
9
4
6
•
1
2
3
4
6
•
1
2
3
4
J
•
1
2
3
4
i
•
• otSy*.
En**'
01
21
18
O
0
0
4»
26
23
3
0
0
33
23
34
e
i
a
i;
10
30
27
13
•
OMWMMB
«<*•*•.
.. .M
1.9
1.0
0.6
0.0
0.0
*•
13
00
0.7
0.0
0.0
»•
1.9
0.0
0.6
0.4
0.4
0.4
H to
am
1S3.3
1381
190.7
130.7
190,7
WM
1633
1921
12810
127.6
127.6
«7.i
1S33
1200
110.0
11D4
100.0
WM
1693
1240
107.7
006
04.0
*1.7
»AE
404
497
440
44.0
44.0
«M
49.4
42.6
41.7
410
41.0
«J>
40.4
40.7
97.6
96.4
96.9
M*
49.4
40.1
34.6
204
27.2
ma
EflMfMfllD*
•Hvmivvv
••MM
«0«M
474.906
•10.146
tilS'faL^
. M«
QiU
ao4
aoo»
20 30 40 SO 00 70 80 90 100
TTHM MCL (ug/L)
ft
hg MohliMimM ter (InoUtn 20%ow-dMl(n I«lo0
41 Mch IfMtoMnl U*c at •* (yitani*: *KM« mMthg MCX "Hi ttim* not tiMttng MCt
la OOti pMMndto
-------
'Oiftt
Figure 11
MODEL OUTPUT (urlM K/O KjUnkifl}. SWTR MOD AtTEHNATTME OmNFECTION
TratfiMnl Code:
I -
9 - •*•*>«• pm-cMot * modNy *ktm
4 - pra-ohkx •» iftMiKtaM + CMC
fTTHM.)
too
too
too
wo
7S
75
76
1C
60
SO
SO
to
25
25
2S
T*
EndkV
17
10
14
SO.
-------
DnH.3O-Jtn-aa
Figure 12
MODEL aUIPUT(«urtB»«*>**tntoa): ENHMKXO8WmW«LietlMTIMEO«l»FECnaN
MCL'
(I1HMJ)
100
100
too
too
100
WP
7S
7$
76
74
76
1*
60
60
60
60
60
00
26
26
26
2S
86
• 86
Til.
Onto
1
2
3
4
6
•
1
2
3
4
6
•
1
2
3
4
S
' •
1
2
3
4
s
• •
%***•.
KatddhMa'
UMBUI
•i
20
19
3
0
• «
40
24
. 32
4
1
«
32
22
26
17
4
o
14
14
17
27
26
a
OttHuMH»
%
-------
Figure 13
MOOEL OUTPUT (iuitaM«/OMlUnktg): EMMNC^ BOTH MW ALTERNATIVE OBUFECI1ON
TmdfMT* Opd»:
1 - IK* raqpMig tirthv Inrtmnl mod
2 - •*«*»*• pm-cMai»MUan
a - dMutepn-oMof + modNy akim
4 - fm-chkx » *im.
0.7
0.0
1.3
0.0
07
a4
HMte
06*
1503
1380
134A
1M»
15O3
136.0
1230
lit*
1SO.3
13Q4
1042
M.i
1SO3
120.0
aoo
SB.»
MLE
60.3
44.0
43.3
410
603
43.7
30.7
M.I
60.3
41.0
336
37.1
503
41 1
28*
I8.S
htaothm
ata
mi
•12
aw
EpJ*MBk)'
(00«) » Vm. to Oulbr«k
MMT
MM*
***»
MMy
100%
00%
80%
70%
00%
50%
4O%
30%
20%
10%
0%
Til #4
2030405000 70 8000 100
TTHM MCL (ug/L)
I-
OOM
ooa
0«B
00>
{::
f 0*1
TTHU MCL (1^.1
'pwoml a tyttam Iratallhg Mch lt«Mlm«nl llw (IncUlw 20% cvw-dMlp) laclor)
'ounuMlw p«c«nl ol lyttemi Alt) to IDM! MCL * MOh UMlnwnl tlw (InoludM 20% avw-dwlp) Itctoc)
•iDMn oanMntraUan « Mch ItMtiranl Uw al m lyilwra; tiOM mMlhg MCL Btd (KM*nd mMlhg MCL
4 (MurnM 10% al paiwMlan MpcMd 41 lim cuptomw
'••Urn** lor 00h p«ctnlll*
-------
Figure
Influent Glardia vs. Total Log Reduction (Winter) From LeChevailier
10,000
1,000
8
I
,0,
£
O
100
10
Total Log Reduction
-------
SWTR Scenario: With Alternate Disinfectants
TTHM MCLs 100 (ug/l) #. of Plants = 39
10,000
1,000
§
I
&
3
3
too
10
Total Log Reduction Achieved
10
-------
Figure 16
10,000
1,000
3
I
I
3
100
10
SWTR Scenario: With Alternate Disinfectants
TTHM MCL s 25 (ug/l) #. of Plants Treating = 83
7
Total Log Reduction Achieved
-10
-------
SWTR Scenario: With Alternate Disinfectants
TTHM MCL = 100 ug/l to 75 ug/l
o
£
§
I
100%
90%
80%
70%
60%
50%
40% -i-
30% -i-
-20% -:-
10% 1-
0% -*-
1
10 100 1,000 10.000
Increase in Annual Glardia Infections Per 1 Million People
100.000
-------
figure J.a
SWTR Scenario: With Alternate Disinfectants
TTHM MCL = 100 ug/l to 25 ug/l
0%
10 100 1,000 10,000
InerMtt in Annual Glardta Infections Pw 1 Million People
100,000
-------
100% —
90% -
80% -
70% -
(ft
1 60% -
I
Percentage of
Si 5
i i
30% -
20% -
10% -
0% -
TTHM MCL = 100 ug/l to 75 ug/l
••
85%
-
5- 7%
2% 1V
1-10 10-100 100-1,000 1,000-10,000 10.000-100,000
Increase in Annual Giardte Infections Per 1 Million People
-------
Figure 20
100%
90%
1-10
SWTR Scenarios: With Alternate Disinfectants
TTHM MCL = 100 ug/l to 25 ug/l
10-100
100-1,000 1,000-10.000 10.000-100,000
in Annual Glardta Infections Per 1 MUlton Peopto
-------
SWTR Scenario: With Alternate Disinfectants
From TTHM MCL = 100 ug/l to 75 ug/l
Change in Risks Per 1 Million People
10
I '
o.ot
10 100 1.000
Increase in Annual Giardia Infections
10,000
100.000
-------
Figure 22
SWTR Scenario: With Alternate Disinfectants
From TTHM MCL = 100 ug/l to 25 ug/l
Change in Risks Per 1 Million People
I
3
10
1
1
0.1]
f
<
1
1
1
"
jT
.01*
1
=a
<
«
b
g>
• •> I
d
B
i_
O
<
n
1
1
4
• «
0
/%>
7
•
—
<
^
...
R?
g
<
i
«
n
10 100 1,000 10.000 100,
inerMM in Annual Glardia Intacttons
-------
SWTR Scenario: With Alternate Disinfectants
THAA MCL s 60 (ug/l) #. of Plants = 29
10,000
1,000
8
u,
3
9
Total Log Reduction Achieved
10
-------
Figure 24
10,000
1,000
8
,0,
a
(9
100
10
SWTR Scenario: With Alternate Disinfectants
THAA MCL = 10 (ug/l) #. of Plants Treating = 80
10
Total Log Reduction Achtewd
-------
SWTR Scenario: With Alternate Disinfectants
THAA MCL a 60 ug/l to 50 ug/l
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
10 100 1,000 10,000
Increase in Annual Giardta Infections Per 1 Million People
100,000
-------
figure 26
SWTR Scenario: With Alternate Disinfectants
THAA MCL = 60 ug/l to 10 ug/l
o% -t
1
10 100 1,000 10,000
Incniie in Annual Gfardta Infections Per 1 Million People
100,000
-------
1 WW /O
90% -
80% -
70% -
| 60% -
CO
1 50% -
Q|
S
I 40%;-
30% -
20% -
10% -
0% -
95%
^M
_
SWTR Scenarios: With Alternate Disinfectants
THAA MCL = 60 ug/l to 50 ug/l
40/ ««/ 3%
I • ° • i . 1% - i ,J^*m o* -
1-10 10-100 100-1,000 1.000-10,000 10,000-100.000
IncreaM in Annual Glardia Infection* Per 1 Million People
-------
ngure
100% —
90% --
80% --
70% --
60% «
50%
40% —
1-10
SWTR Scenario: With Alternate Disinfectants
THAA MCL = 60 ug/l to 10 ug/l
20% -
10% -
0% -
M»
•
H
12%'
24%
•
• : ; : '• '. • : • : • : • : • : • : | : : : • ; - ; : ; •
-_:'•_: ''.I. •_:'•' •'.}:'•:'•:'•:':•;'•'.
: : : ; :':•':•: ; : - : : :••-"'-•'••'••
4%
— I — fcPife1*^^"-1* — i
Inc
10 '^00 100-1.000 1.000-10,000 10,000-100.000
in Annual Glardia Infections Pw 1 Million Peopto
-------
SWTR Scenario: With Alternate Disinfectants
From THAA MCL = 60 ug/l to 50 ug/l
Change in Risks Per 1 Million People
10-
1
1 ; -
n Annual Cancer Incit
>
A
Decrease 1
c
'
<
0.01 T
1
i
t
i
i
>
4
-
4
1
10 100 1,000 10,000 100,
Increase in Annual Glardla Infections
-------
Figure 30
SWTR Scenario: With Alternate Disinfectants
THAA MCL = 60 ug/l to 10 ug/l
Chang* in Risks Per 1 Million People
10
1
mm
1CI..I
0.1
0.01
10
100 1,000
in Annual Outfit Infection*
10,000
100,000
-------
STATUS REPORT ON ANALYTICAL METHODS
TO SUPPORT THE DISINFECTANT/DISINFECTION
BY-PRODUCTS REGULATION
August 1992
Background
The United States Environmental Protection Agency (EPA) is
developing national regulations to cc..trol disinfectants and
disinfection by-products (D/DBPs) in public drinking water
supplies. Twelve D/DBPs have been identified for possible
regulation under this rule12-3, based, on available occurrence,
exposure, and health effects data. EPA intends to set maximum
contaminant levels (MCLs) for these D/DBPs, and analytical
methods will be specified for use in demonstrating compliance
with each MCL. This document provides a summary of the
analytical methods that EPA intends to propose as compliance
monitoring methods. A discussion of surrogate measurements that
are being considered for inclusion in the regulation is also
provided.
There are several technical issues that EPA is trying to
resolve before the methods discussed in this document are
proposed/promulgated. The public is encouraged to contact EPA,"
if they have information that will help EPA in this process.
Information should be addressed to:
Patricia Snyder Fair, Chair
D/DBP Analytical Methods Task Force
U.S. Environmental Protection Agency
Office of Ground Water & Drinking water
26 W. Martin Luther King Drive
Cincinnati, OH 45268
Disinfectants:
Chlorine and Chloraaines. The measurement of chlorine
residuals will be required as part of the D/DBP Rule in order to
demonstrate compliance with MCLs for chlorine and monochloramine:
The utility will have the option of either measuring free or
total chlorine residuals for the chlorine MCL. Compliance with
the monochloramine MCL will be based on a total chlorine residual
measurement,' because monochloramine cannot be practically
measured by itself.
There are many analytical methods available for measuring
chlorine residuals, and each has certain limitations. All
methods are subject to interferences, so the accuracy and
-------
precision of the analysis will vary with matrix. LOW level free
chlorine measurements are subject to interference errors fron
chloramines and other oxidants.
Several methods for measuring chlorine residuals were
promulgated with the Surface Water Treatment Rule (SWTR)*: l)
amperometric titration; 2) DPD ferrous titrimetric method; 3) DPD
colorimecric method; and 4) the leuco crystal violet (LCV)
method. These were referenced to the 16th Edition of Standard
Methods for the Examination of Water and Wastevater3. EPA used
these methods as the starting point for determining methods that
are applicable to the D/DBP Rule. Methods listed in the 17th
Edition of Standard Methods4 were also considered. Table l
summarizes the methods most likely to be proposed with the D/DBP
Rule. The methods required to demonstrate compliance with the
SWTR will also be updated to be consistent with the D/DBP Rule.
EPA intends to propose thre« of the four chlorine residual
methods from the SWTR, when the D/DBP Rule is proposed. The LCV
method will not be proposed. It was not included in the 17th
Edition of standard Methods, due to its relative difficulty and
lack of comparative advantages. EPA thinks it is unlikely that
this method is being used, so dropping it should have little
effect on the regulated community.
EPA is considering approval of two additional methods for
compliance monitoring: 1) the syringaldazine (FACTS) method for
free chlorine and 2) the iodometric electrode method for total
chlorine. Both methods are included in the 17th Edition of
Standard Methods.
Method 4500-C1 E for measuring low levels of total chlorine
residual is also available in the 17th Edition of Standard
Methods. Although it is not useful for measuring total chlorine
residuals at concentrations near the MCL, it is useful for
determining compliance with the SWTR requirements for a
detectable residual in the distribution system. Therefore, EPA
is considering approval of this method when it updates methods
for the SWTR. Data obtained using this method can also be used
to demonstrate compliance with the total chlorine residual MCL in
the D/DBP Rule.
The 18th Edition of Standard Methods7 will be available for
purchase in August, 1992, so the D/DBP Rule will cite it for
descriptions of the chlorine residuals methods. The method
numbers did not change from the 17th to-the.18th edition, so the
numbers given in Table 1 can be used with either edition. Since
EPA is required to review its regulations every 3 years, EPA
anticipates incorporating the latest versions of the methods into
the D/DBP Rule each time it is reviewed.
-------
Historically, EPA has required all analyses that are done to
demonstrate compliance with an MCL, be performed by a certified
laboratory. Because chlorine residuals are not stable, these
samples must be analyzed immediately efri cannot be transported to
an off-site laboratory. Utility per-ainnel have been performing
chlorine residual analyses either in the field or in the
treatment plant. EPA feels these measurements should continue to
be made at the water system by any person acceptable to the
State. It is not EPA's intent to require the certification of
each water system for chlorine residual measurements.
EPA will continue to allow the use of DPD colorimetric test
kits for field measurements of chlorine residuals, if they are
approved by the State.
Chlorine Dioxide. The SWTR promulgated two methods for
measuring chlorine dioxide residuals: 1) amperometric titration
and 2) the DPD method. The 16th Edition of Standard Methods was
cited as the reference. As discussed above for chlorine residual
measurements, the latest versions of methods for measuring
chlorine dioxide residuals (see Table 1) will be cited in the
D/DBP Rule.
The 17th Edition of Standard Methods proposes a second
amperometric titration method (4500-C102 E) for measuring
chlorine dioxide residuals. EPA is considering adding it as an
acceptable compliance monitoring method.
The methods for chlorine dioxide residuals are indirect
methods, because the concentrations are determined by difference.
Amperometric titration methods are preferred over the DPD method,
if they are used by highly trained personnel.
It is EPA's intent to continue to allow the measurement of
chlorine dioxide residuals at the water system by any person
acceptable to the State.
Due to the limitations of the current methodology, EPA is
seeking information on new methodology that may be applicable for
compliance monitoring. New methods must provide demonstrated
advantages over the current methods and have the potential for
being distributed in a standard format to interested public in
the timeframe of the D/DBP regulation. New methods must be
useable in the field or by utility personnel.
Disinfection By-products:
Trihaloaethanes. There are two methods currently approved
for THM compliance monitoring: EPA Methods 501.1 and 501.2.
These packed column, gas chromatography (GC) methods were
-------
promulgated with the 1979 THM Rule*. EPA intends to propose two
capillary column GC methods by the end of 1992: EPA Methods
502.2" and 524.2iJ, After a 45 day comment period, the methods
will be promulgated, and they can be used for THM compliance
monitoring effective 30 days after promulgation. EPA Methods
502.2 and 524.2 are in general use in many laboratories, because
they are used to measure the concentrations of volatile organic
compounds (VOCs) in drinking water.
EPA Methods 502.1 and 524.1* are not being proposed in 1992
and they are not being considered for the D/DBP Rule, because
they use packed column chromatography. The GC technology has
progressed to the point that packed columns are becoming
obsolete. EPA is considering eliminating packed column GC
methods (EPA Methods 501.1 and 501.2) from the list of approved
THM compliance monitoring methods when the D/DBP Rule is
proposed. This would not be implemented until the monitoring
requirements of the D/DBP Rule become effective.
EPA Method 551" is the only new method that is likely to be
addeu for THM compliance monitoring when the D/DBP Rule is
promulgated. It involves adjusting the ionic strength of the
sample, extracting the analytes into methyl-tertiary-butyl ether
(MTBE), and analyzing the extract by capillary column GC with
electron capture detection (ECD). This method can also be used
to measure the concentrations of haloacetonitriles, chloropicrin,
1,1-dichloropropanone, 1,1,1-trichloropropanone, and chloral
hydrate, if the appropriate dechlorinating agents are used.
Several laboratories around the country have modified EPA
Method 551 by using pentane, instead of MTBE, as the extraction
solvent. When this is done, chloral hydrate (CH) is not included
in the *--lysis, because CH is too polar to be extracted by
pentane. ?A Method 551 permits the analyst to modify GC
columns, .C conditions, detectors, extraction techniques,
concentration techniques, internal standard or surrogate
compounds, as long as the analyst demonstrates the modified
method still meets the performance criteria established in the
nethod. Therefor«, if Method 551 is approved for THM compliance
•mitoring, approval to use pentane will not be considered
-=>ssary as lonq is the method performance criteria are met.
. ...» will eliminate some of the need for approval of alternative
test procedures (ATPs).
Many commercial sources of MTBE are contaminated with
chloroform, necessitating cleanup prior to use in THM analyses.
MTBE can be purified by distillation using appropriate safety
precautions, but it should not be stored for long periods of
time, in order to prevent the formation of peroxides. At least
some lots of OmniSolv (EMScience) MTBE labeled "Suitable for
Spectr©photometry, Liquid Chromatography, Gas Chromatography,
-------
Residue Analysis, Assay by GC: 99.9% pure" have not contained
measurable levels of chloroform or chlorinated solvents,
indicating suitable purity MTBE can be produced commercially.
This should encourage laboratories to make the need for high
purity MTBE known to their local solvent suppliers. • '
Manufacturers may be willing to prepare and market special lots,
when they are aware of a market for their product.
The EPA methods that are likely to be approved for
compliance monitoring of THMs are summarized in Table 2. All of
these methods include the option to use ascorbic acid as a
dechlorinating agent. This practice is under review, because
ascorbic acid may cause the loss of brominated THMs under some
conditions. If laboratories are aware of alternatives to the HC1
and ascorbic acid required by EPA Method 524, they are encouraged
to share that information with EPA. (Sodium thiosulfate and HC1
were originally included in 524, but they cause interference
problems with some of the early eluting analytes included in the
method. Thiosulfate and HC1 can still be used, if the sample
does not have to be analyzed for the early eluting compounds.)
Chloral Hydrate. EPA Method 551" will be proposed as the
compliance monitoring method for chloral hydrate (CH), but
additional work must be done before the method is ready for
general use. Chloral hydrate is subject to base-catalyzed
hydrolysis, and the current version of the method does not
provide a mechanism for preventing hydrolysis. This method will
require the addition of a preservative, probably in the form of
acidification. Work is underway to evaluate preservatives for
CH. EPA solicits data demonstrating the stability of CH in
drinking water samples. The data must indicate the
dechlorinating agent used, preservation procedure, sample pH,
storage conditions and holding time.
The regular dechlorinating agent (NH4C1) recommended in
Method 551 cannot be used in some drinking water matrices,
because it interferes with the CH analysis. Either ascorbic acid
or sodium sulfite can be used under those circumstances. Since
ascorbic acid causes problems with THM analyses in some matrices,
EPA prefers the use of sodium sulfite. As part of the effort to
evaluate preservation techniques for CH, EPA will look at how
THMs are affected. It is anticipated that a technique can be
developed in which THMs and CH could be measured in the same
sample using EPA Method 551. If successful, this would reduce
the monitoring costs associated with these analytes.
Some laboratories have expressed concern about the safety of
using MTBE. EPA recommends that ether extracts should be stored
in an explosion-proof refrigerator/freezer.
-------
The availability of standards for CH has been a problem
laboratories. However, dilute solutions of CH in acetone are -
available from NSI Environmental Solutions, Inc. (Research
Triangle Park, NC) ^Catalogue # 001179-01-01, 332.00] This is a«
EPA certified standard. A neat standard is available from
Supelco (Bellefonte, PA) [Catalogue # 4-8048]. Other commercial
sources may become available during the next year.
Very few laboratories are doing CH analyses. EPA is not
aware of any commercial laboratories, but this should change over
the next few years as utilities include it as a by-product of
interest in their bench and pilot scale treatment studies.
Chloral hydrate will be included in the EPA Performance
Evaluation (PE) Studies by the end of 1992. This will provide
laboratories an opportunity to evaluate how well the method is
performing for them. It will also provide EPA with an estimate .
of how many laboratories are doing CH analyses and how well they
are doing.
Haloacetic Acids. The EPA method that will be proposed for
haloacetic acid (HAA) compliance monitoring will give the analyst
several options. The initial (standard) version of Method 552"
specified the following steps in the procedure: 1) extraction
with MTBE after sample pH adjustment to >11.S; 2) discard MTBE
fraction and adjust pH of sample to <0.5; 3) extraction with
MTBE; 4) concentration and drying of the extract; 5) conversion
of the HAAs to their methyl esters using diazomethane; and 6)
analysis by capillary column GC/ECD.
A microextraction option was later added to EPA Method
55212. This option eliminated the cleanup extraction, extract
concentration, and extract drying steps. The 18th Edition of
Standard Methods will include a microextraction method for
measuring HAAs. EPA is considering it as a compliance monitoring
method, because it is equivalent to the option described in
552.
EPA is also developing a liquid/solid extraction technique .
using ion exchange resins and an acidic methane 1 derivatization
procedure13. A written procedure should be available for public
distribution by the end of 1992, and it will be designated EPA
Method 552. l'°.
All of the above procedures will be proposed as compliance
monitoring techniques; all will be covered under EPA Method 552.
Giving the analyst the flexibility to choose extraction technique
and derivatization method should make it easier to begin this
analysis in the laboratory. The acidic methanol derivatization
procedure will also eliminate the concern expressed by some
states and other entities over the use of diazomethane due tc
-------
safety issues.
The quantisation procedures specifed in the current version
of EPA Method 552" must be clarified prior to proposal of the
D/DBP Rule. The initial procedure (described-above in the first
paragraph under HAAs) does not require that standards be
extracted, and no correction is made for analytes which are not
fully extracted from the aqueous sample (e.g., dibromoacetic acid
[DBAA]). The microextraction option recommends the preparation
of aqueous standards that are carried through the same procedure
as the samples. Using this option, the analytical results are
automatically corrected for less than 100% extraction
efficiencies. Either calibration method will be acceptable for
compliance monitoring, if EPA only sets MCLs for dichloroacetic
acid (DCAA) and trichloroacetic acid (TCAA), because both acids
are extracted from water at > 90% efficiency. However, EPA is
also considering a total HAA (THAA) MCL. Samples from utilities
with high levels of bromide ion in their source water may contain
high levels of the brominated and mixed broraochloro-acetic acids.
Since these acids are not extracted from water as efficiently as
the chlorinated acids, use of the initial HAA analytical
procedure (described above in the first paragraph under HAAs)
would give a lower THAA result than the microextraction
procedure. For this reason, future versions of Method 552 will
specify that aqueous standards be used for calibration purposes.
Analyses performed to meet compliance monitoring requirements for
a THAA MCL must use aqueous standards for calibration purposes.
Five of the HAAs (monochloroacetic acid [MCAA], DCAA, TCAA,
monobromoacetic acid [MBAA], and DBAA) have been included in EPA
PE studies, since 'study WS026 (spring, 1990). A calibration
procedure for the HAAs was not specified with the PE studies, so
laboratories reported data from both procedures. Since the two
procedures are not equivalent for several of the HAAs, EPA will
specify that HAA PE samples be analyzed using aqueous standards
in future PE studies.
EPA Method 552 specifies the HAA standards be prepared in
MTBE, but many laboratories are using methanol instead of MTBE.
'Recent work by Yuefeng Xie14 at the University of Massachusetts,
indicates that over time the acids will undergo conversion to
their methyl ester analogs when stored in methanol. The
conversion rate varies with analyte and storage conditions.
Unless the analyst specifically checks for this conversion by
analyzing a non-derivatized standard, it is unlikely the problem
would be detected. The use of ester or mixed free acid and ester
standards will provide inaccurate results, due to differences in
extraction efficiencies between acids and esters. EPA has not
studied this problem, but recommends that if laboratories must
continue to use methanol, they monitor their methanol standards
for this conversion, and prepare fresh standards when esters are
-------
detected. As an alternative, laboratories should prepare HAA
standards in MTBE.
There are 9 haloacetic.acids'(HAAs; that could potentially
be included in a THAA MCL. However'-~r.e EPA Method 552 only
includes 6 of the HAAs (MCAA, DCAA, TCAA, MBAA, DBAA, and
bromochloroacetic acid [BCAA]). Tribromoacetic acid (TBAA) is
not included, because it is not reliably measured using the
current techniques. The remaining 2 HAAs (bromodichloroacetic
acid [BDCAA] and dibromochloroacetic acid [DBCAA]) have not been
tested with Method 552, because standards are not commercially
available. It is unlikely that TBAA, BDCAA, or DBCAA will be
included in Method 552 in the timeframe of this regulation.
EPA is aware of three suppliers that are selling or plan to
sell HAA standards:
1) Supelco, Inc. (Bellefonte, PA) is distributing a
calibration standard that contains MCAA, DCAA, TCAA, MBAA,
DBAA, and BCAA (catalogue #4-8047). They.will also market
a BCAA standard in solution form as a custom chemical.
2) Absolute Standards, Inc. (New Haven, CT) is selling a
HAA QC sample, containing MCAA, DCAA, TCAA, MBAA, DBAA, 2,4-
. dichlorophenol (24DCPh), and 2,4,6-trichlorophenol
(246TCPh). A standard containing the methyl derivatives of
the same compounds is also available, thus, providing
laboratories a mechanism for checking the derivatization
efficiency of their method. BCAA will soon be added to the
free acid standard mixture, and should be available for
release in August, 1992 (catalogue # 30054).
3) ULTRA Scientific (North Kingstown, RI) plans to release
a standard containing MCAA, DCAA, TCAA, MBAA, DBAA, BCAA,
24DCPh, and 246TCPh (catalogue # PHM-552A). The standard is
tentatively scheduled for release in early August, 1992,
assuming successful results from a 60 day time storage
study.
EPA believes there will be adequate laboratory capability
available by the time compliance monitoring for HAAs is required.
Sixteen laboratories participated in the WS029 PE study by
analyzing the HAA sample. This number is expected to increase
significantly over the next few years, as laboratories expand
their analyses to address the new regulations.
chlorite, (Chlorate) and Broaate. An ion chromatography
method will be proposed as the compliance monitoring method for
these anions. EPA does not intend to propose an MCL for the
chlorate ion (CIO,*), but a description of the analytical method
is presented here since it can be determined with the chlorite
ion (C102') . The current version of EPA Method 300.0 Part B15 can
be used for measuring C10:~ and CIO,' in drinking waters. In some
cases, a weaker carbonate eluent may be required16, because Clo:
-------
elutes in the -void volume in high ionic strength samples. A
change in the eluent strength is permitted as part of Method
300.0 Part B, as long as the analyst demonstrates that the
quality control requirements outlined in the method are met (see
Sections 10.1.1 and 11.1 of the method).
Chlorite ion is unstable in many waters1", so a preservation
technique will be required if the samples cannot be analyzed
within 15 minutes of collection. Recent studies'*" indicate
ethylenediamine (EDA) is a suitable preservative for C10: , and it
does not adversely affect analysis of the other anions at the
concentrations typically found in drinking water. Therefore, EPA
is considering requiring the addition of EDA to samples analyzed
for C102~ compliance monitoring.
Measurement of CIO,' in samples containing a free chlorine
residual will also require the use of EDA. Free chlorine reacts
with CIO," to form CIO, and chloride.
Samples containing a chlorine dioxide residual must be
sparged with an inert gas (e.g., He or Ar) at the time of
collection to eliminate the chlorine dioxide. Otherwise, C10:
will continue to form C10:' and CIO,' in the samples. Prior to
sparging, the samples must be protected from light to prevent
photodecomposition of the chlorine dioxide to form C102' and CIO,'
19
•
EPA is aware of other techniques (e.g., flow injection
analysis [FIA]) that are being used to measure chlorite and
chlorate. FIA and/or any other applicable procedure will be
considered as a potential compliance monitoring method, if it is
demonstrated to be as accurate and precise as ion chromatography.
The technique would have to be available in a standard format for
public distribution.
Analyses for bromate ion in drinking water samples will
require a modification to the EPA ion chromatography method, due
to the interference from chloride present in these matrices, one
approach is to pretreat the sample with a silver media in order
to remove the chloride20. When this technique is used, traces of
silver are deposited on the 1C analytical column, necessitating
the removal of silver before using the column for bromide
analyses. Another alternative is to change the eluent. Recent
studies16-11 demonstrate that substituting a borate eluent for the
carbonate eluent specified in the method not only provides the
resolution required to quantitate bromate ion in drinking water,
but also gives a more stable baseline. All the DBF anions can be
measured using the same chromat©graphic conditions with the
borate eluent. Sections 10.1.1 and 11.1 of EPA Method 300.0 Part
B permit the use of the borate eluent as long as the analyst
demonstrates that the quality control requirements outlined in
-------
the method are met.
EPA is concerned about the ability to reliably measure
bromate at the levels necessary for.tttis regulation. Studies'are
underway in several laboratories, including EPA, to examine ways
to lower the detection limit for this anion. Concentration
techniques are the most promising avenues of research. EPA
requests information on any techniques that are demonstrated to
be effective in drinking water matrices.
Chlorite, chlorate and bromate ions are now included in EPA
PE studies. Thirteen laboratories participated in study WS029.
Surrogate Measurements!
Simulated Distribution System (80S) Test. EPA may consider
the use of a Simulated Distribution System (SOS) test as a
surrogate measurement of DBP concentrations in the distribution
system. EPA will recommend a modified version of the procedure
described in Method 5710 E in the 17th edition of Standard
Methods (reference will be updated to the 18th edition when
proposed). The method is written for use in determining THM
concentrations, but it can be extended to other DBFs as long as
appropriate dechlorinating agents are used after the storage
period. Results of the test are not valid, if there is no
detectable chlorine residual at the end of the storage period.
If the test is used to determine a "worst case11 for consumer
exposure, then the sample storage period should be representative
of the maximum distribution system temperature, pH, chlorine
concentration and the longest detention time in the distribution
system. EPA will accept data comparing results from the SOS test
to actual distribution system samples. The data must include a
full description of the SOS test conditions, including
dechlorination procedures and analytical methods used to
quantitate the DBFs. Information concerning the data from
distribution system samples should include temperature, chlorine
residuals, pH, and approximate detention time of the water in the
system.
Total organic Carbon. There are 3 methods for measuring
organic carbon listed in the 17th Edition (18th Edition) of
standard Methods, and 2 of them are applicable to requirements of
the D/DBP rule. The persulfate-ultraviolet oxidation method
(5310 C) and the wet-oxidation method (5310 0) provide the
sensitivity necessary for low level organic carbon measurements,
so they will be recommended if organic carbon is proposed in the
D/OBP rule as a surrogate measurement for DBP precursors.
Depending upon the sample pretreatment, several different organic
carbon fractions are measured by these methods. If the sample is
filtered through a 0.45pm pore size filter before analysis,
Dissolved Organic Carbon (DOC) is measured. Purging the sample
10
-------
prior to analysis results in the measurement of Nonpurgeable
Organic Carbon (NPOC). Both of these parameters have been used
as indicators of raw^water quality. EPA will accept information
concerning the use of DOC or NPOC as a surrogate for DBP
precursors in source water.
Total Organic Halida. Method 5320 B in the 17th Edition
(18th Edition) of Standard Methods will be recommended for
determining Total Organic Halide (TOX). The sample must be
dechlorinated and acidified at the time of collection. Sodium
sulfite crystals or a FRESHLY prepared sulfite solution should be
used for dechlorinatipn. Following dechlorination, the sample is
acidified to a pH < 2 using nitric acid in order to preserve the
sample. If the bottles must be shipped to the sampling site with
the reagents already present in the bottles, then sulfuric acid
should be substituted for nitric acid. Department of
Transportation (DOT) regulations must be followed when shipping
bottles containing sulfite and sulfuric acid.
UV Absorbance. There is not a standardized method for this
parameter. The technique originally used to establish a
relationship between raw water UV absorbance and THM formation
involved filtering the sample through a 0.45jim pore size filter
and then measuring UV absorbance at 254 nm. The filter must be
prewashed to remove water-soluble organics. EPA will accept
information concerning the applicability of alternative
procedures for defining this measurement.
EPA is evaluating its data to determine the relationship
between TOC and UV absorbance. It may be necessary to measure
both parameters in order to better characterize raw water
quality.
EPA will accept data demonstrating interferences to the UV
measurement at 254 nm that would prevent its use in specific raw
waters.
L1 Consents t
EPA knows the laboratory capacity for doing DBP compliance
monitoring samples is not yet in place. However, EPA believes
the capacity can be developed by the time monitoring requirements
take effect. PE samples will be available for all the DBPs
during 1992, so laboratories can take advantage of those
opportunities to demonstrate competency. Part of the laboratory
certification process will involve successful performance in PE
studies. Laboratories can request PE samples through their state
Certification Officer and participation is free.
EPA does not have interlaboratory method performance data
for the non-THM DBP methods. EPA anticipates that good
11
-------
laboratories will be able to achieve quantitative results on
analyses of PE samples that are within at least ± 40 % of the
sample's true value for the non-THM DBFs. However, this has yet
to be demonstrated. Some performance data will be generated
using the results from PE studies. EPA will also examine
alternatives ways to obtain the data, such as round robin, studies
among several laboratories currently analyzing drinking water
samples for DBPs.
The most recent PE studies that included the THMs indicated
that the majority of the laboratories participating in the
studies were able to achieve results within ± 20 % of the THM
sample's true value. Therefore, EPA will continue to require
that level of performance for certification for THM analyses.
References: .
1. USEPA. Status Report on Development of D/DBP Regulations.
Office of Ground Water and Drinking Water. Washington, D.c.
1991.
2. USEPA. Occurrence Document in Support of the Development of
the D/DBP Regulations. USEPA Publications. Washington, D.C.
in press. 1992.
3. USEPA. Status Report on Development of MCLGs for
Disinfectants and Disinfection By-products. Office of
Science and Technology. Washington, D.C., 1992.
4. USEPA. Drinking .Water; National Primary Drinking Water
Regulations; Filtration, Disinfection; Turbidity, Giardia
- lamblia, Viruses, Legionella, and Heterotrophic Bacteria;
Final Rule. Fed. Rea.. 54:124:27486 (June 29, 1989).
5. Standard Method's for the Examination of Mater and
Wastewater, 16th Edition, American Public Health.
Association, American Water Works Association, and Water
Pollution Control Federation, 1985.
6. Standard Methods for the Examination of Water and
Wastewater, 17th Edition, American Public Health
Association, American Water Works Association, and Water
Pollution Control Federation, 1989.
7. Standard Methods for the Examination of Water and
Wastewater, 18th Edition, American Public Health
Association, American Water Works Association, and Water
Pollution Control Federation, 1992.
8. ' 40 CFR 141, Subpart C, Appendix C.
12
-------
9. USEPA. Methods for the Determination of Organic Compounds in
Drinking Water, EPA/600/4-83/039, PB91-231480, National
Technical Information Service (NTIS), December 1988 (revised
July 1991). r *
10. USEPA. Methods for the Determination of Organic Compounds in
Drinking Mater - Supplement II, Environmental Monitoring
Systems Laboratory, Cincinnati, OH, August 1992.
11. USEPA. Methods for the Determination of Organic Compounds in
Drinking Water - Supplement I, EPA/600/4-90-020, PB91-
146027, NTIS, July 1990.
12. Barth, R.C. & Fair, P.S. "Disinfection By-Prodcuts:
Analysis of Haloacetic Acids and Chlorophenols;
Microextraction Procedure vs. USEPA Method 552." AWWA Water
Quality Technology Conference (WQTC) Proceedings, November
1990. •
13. Hodgeson, J.W., Collins, J., & Becker, D. "Advanced
Techniques for the Measurement of Acidic Herbicides and
Disinfection Byproducts in Aqueous Samples." Proceedings of
the 14th Annual EPA Conference on Analysis of Pollutants in
the Environment. May 1991.
14. Xie, Yuefeng, University of Massachusetts, Amherst, MA,
personal communication, May, 1992.
15. USEPA. The Determination of Inorganic Anions in Water by Ion
Chromatography Method 300.0. Environmental Monitoring
Systems Laboratory, Cincinnati, OH, August 1991.
16. Hautman, p.p. & Bolyard, M. "Analysis of Oxyhalide
Disinfection By-Products and Other Anions of Interest in
Drinking Water by Ion Chromatography/" International Ion
Chromatography Symposium Proceedings. October 1991.
17. Pfaff, J.D. & Brockhoff, C.A. "Determination of Inorganic
Disinfection By-Products by Ion Chromatography," Jour.
82:4:195 (April 1990).
13. Hautman, D.P. & Bolyard, M. "Analysis of Inorganic
Disinfection By-Products Using Ion Chromatography," AWWA
WQTC Proceedings. November 1991.
19. Zika, R.G. ET AL. "Sunlight-Induced Photodecomposition of
Chlorine Dioxide," Water Chlorination Chemistry:
Environmental Impact and Health Effects. Vol. 5, Lewis
Publ., Inc., Chelsea, Mich. (1985).
13
-------
20. Kuo, C. and Weinberg, H.S. "Analysis of Inorganic
ocprocH-u IOn Chromatography," AWWA
WQTC Proceedings. November 1990.
14
-------
Table 1. Methods Being Considered for Use in Compliance Monitoring of
Disinfectant Residuals Under the D/DBP Rule
Residual
Free Chlorine
Methodology
Amperoraetric Titration
DPD Ferrous Titrimetric
DPD Colorimetric
Syringaldazine (FACTS)
Std Meth*
4500-C1 D
4500-C1 F
4500-C1 G
4500-C1 H
Working
Range (mg/L)
> 0. 1
> 0.1
> 0.1
> 0. 1
Total Chlorine
Chlorine Dioxide
Araperoraetric Titration
Amperonetric Titration
DPD Ferrous Titrimetric
DPD Colorimetric
lodometric Electrode
Amperometric Titration
DPD Method
Amperometric Titration
(proposed)
4500-C1 D
4500-C1 E
4500-C1 F
4500-C1 G
4500-C1 I
4500-ClOj C
4500-ClOj D
4500-C1O, E
0.1
0.2
>
<
>
>
> 0.1
0
0
1
1
> 0. 1
> 0. 1
> 0.1
Method Number Used in the 17th and 18th Editions of Standard Methods.
15
-------
Table 2. Methods Being Considered for Use in Compliance Monitoring of Disinfection
By-Products Under the DBF Rule
Working
Residual Methodology* EPA Method (Ref)"* Range4
Trihalomethanes PfcT/GC/ElCD & PID 502.2 (9) > 2.0
P4T/GC/MS 524.2 (10) > 0.5
LLE/GC/ECD 551 (11) > 0.5
Chloral Hydrate LLE/GC/ECD 551 (11) > 0.5
Haloacetic Acids LLE or SPE /GC/ECD 552 (104,11) > 5.0
Chlorite & Chlorate 1C , 300.0 Part B (15) >10**
Broroate 1C 300.0 Part B (15) >1044
P&T = purge and trap; GC = g3S chromatography; E1CD = electrolytic conductivity
detector; PID = photoionization detector; MS = mass spectrometer; LLE = liquid/ liquid
extraction; BCD = electron capture detector; SPE = solid phase extraction; 1C = ion
chromatography
Reference for method is given in ().
* The concentrations listed in this table are estimates of the lowest levels
laboratories can routinely measure with confidence. For methods involving
multianalytes (e.g., THM methods), the range is given for the analyte with the
highest detection/quantitat ion level.
*f Based on experience in EPA's laboratory using a borate eluent. These levels may
be optimistic for many drinking water laboratories.
16
------- |