United Sum
EnviFonnwtittl Protection
Agency
Off ice of
Drinking Water (WH-550)
Washington, DC 20460
EPA 570/9-85004
May 1985
Economic Impact Analysis
of Proposed Regulations to
Control Volatile Synthetic
Organic Chemicals (VOCs)
in Drinking Water
-------
ECONOMIC IMPACT ANALYSIS
OF
PROPOSED REGULATIONS TO CONTROL VOLATILE
SYNTHETIC ORGANIC CHEMICALS (VOCs)
IN DRINKING WATER
OCTOBER 1985
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF DRINKING WATER (WH-550)
WASHINGTON, D.C. 20460
EPA 570/9-85-004
-------
ERRATA
Shortly before completion of this report, new information
on the health effects of tetrachioroethylene (PERC) was provided
to the Agency. This information requires the Agency to repropose
the Recommended Maximum Contaminant Level for PERCI This economic
analysis describes impacts associated with PERC as estimated
before receipt of the new information. While PERC is a commonly
found contaminant, it was not found to contribute significantly
to the cost of the proposed regulations. Based on the new informa-
tion, this finding will be reevaluated during preparation of
the PERC MCL rule.
-------
TABLE OF CONTENTS
Pa2e
1. SUMMARY. OF RESULTS
1.1 Introduction I—I
1.2 Problem Definition 1—2
1.3 Alternatives for EPA Action I— 4
1.L Benefit Assessment 1—5
1.5 Cost Assessment 1—6
1.6 Regulatory Flexibility & Paperwork Analysis 1—7
1.7 Summary of Costs & Benefits of VOC Removal 1—8
2. PROBLEM DEFINITION
2.1 Introduction 11.-i
2.2 Health Effects I l_i
2.3 Occurrence Il—S
2.U Control. Technologies . 11—11
2.5 Market Charécteristics, Imperfections,
and the Need for Regulation
2.6 State Actions to Control VOCs in the
Absence of Federal Regulations 11—23
2.7 Actions Taken by Water Utilities in
Response to VOC Contamination 11—26
2.8 Control of VOCs Via Other Federal
Legislation 11—28
3. REGULATORY AND NON—REGULATORY ALTERNATIVES
3.1 Alternative No. 1: No Federal Regulations 111—1
3.2 Alternative No. 2: Set Federal Monitoring
Regulations and Provide Health Advisories
as Appropriate 111—3
3.3 Alternative No. 3: Set Primary Drinking
Water Regulations for Certain of the VOCs 111— U
3.L1 Other Authorities to Protect Drinking Water 111—5
1$. ASSESSMENT OF BENEFITS
L .1 Introduction IV—1
L1.2 Baseline Exposure and Cancer Risk IV—1
L4.3 Methodology for Evaluation of Multiple
Occurrence IV—4
L1.L Analysis of Regulatory Alternatives IV—7
14.5 Uncertainties in Benefit Assessment IV—13
5. ASSESSMENT OF COSTS
5.1 Intrcduct on V—i
5.2 Capital and O&M Costs V—i
5.3 Uncertainties in Cost Assessment V_1L
5.14 Cost to State Governments of Federal
‘ ICC Regulations V—16
-------
TABLE OF CONTENTS (continued)
Page
6. SYSTEM LEVEL NET BENEFITS ANALYSIS
6.1 Introduction VI—1
6.2 Methodology VI—3
6.3 Results and Discussion VI—9
7. REGULATORY FLEXIBILITY ANALYSIS AND PAPERWORK ANALYSIS
7.1 Regulatory Flexibility Analysis Vu—i
7.2 Paperwork Analysis VII—6
8. SUMMARY OF COSTS AND BENEFITS
8.1 Introduction VI II—1
8.2 Probable Actions, Health Benefits and
Costs - VIII—i
APPENDIX A -— HEALTH EFFECTS OF INDIVIDUAL CHEMICALS
APPENDIX B -— ASSUMPTIONS REGARDING COMPLIANCE CHOICES
-------
LIST OF EXHIBITS
EXHIBIT
NUMBER PAGE
1—1 Summary of Health Effects of Volatile
Organic Compounds 1—3
2—1 Estimated Number of Water Systems
Exceeding Adjusted Acceptable Daily
Intake for VOCs 11—3
2—2 Strength of Evidence on Carcinogenicity
of VOCs 11—6
2—3 CAG Risk Estimates for VOCs and Rounded
Values Determining Equivalent MCLs 11—7
2— 4 - Summary of Single and Multiple Occurrence
of VOCs as a Class 11—9
2—5 Cumulative Occurrence for Water Supplies
in the Ground Water Supply Survey (GWSS)
with TVOC Concentrations Above Indicated
Levels 11—10
2—6 Estimated Number of Ground Water •Systems
Having Concentrations Greater Than or
Equal to 0.5 ug/]. for the Indicated VOCs 11—12
2—7 Estimated Number of Surface Water Systems
Having Concentrations Greater Than or
Equal to 0.5 ug/l for the Indicated VOCs 11—13
2—8 Confidence Intervals for Number of Water
Systems Having VOC Contamination II 1U
2—9 Ir,fluent Concentrations of VOCs in
Groundwater Systems 11—15
2—10 Influent Concentrations of VOCs in
Surface Water Systems 11—16
2—11 Summary of VOC Program Actions in
Thirteen States 11—25
Example Cumulative Distribution of the
Number of People Exposed to VOCs IV—3
4—2 Proportions of Single (vs. Multiple)
Occurrence for Individual VOCs IV —6
-------
LIST OF EXHIBITS (continued)
EXHIBIT
NUMBER PAGE
Annual Risk of Cancer Factors for
Individual VOCs and the Multiple
Contaminant Case IV—8
4_Li Total Number of Cancer Cases Avoided
Per Year Under Alternative MCLs IV—lO
4—5 Total Number of Cancer Cases Avoided Per
Year by System Size IV—12
4—6 Selected Factors Contributing to
Uncertainty in Risk Estimation IV—15
5—1 - Total Annual Market Costs of VOC Removal V—7
5—2 Total Annual Social Costs of VOC Removal V —9
5—3 Number of Systems Required to Remove VOCs V—lO
5_LI Annual Market Costs Per Affected Water
System V—li
5—5 Annual Market Costs Per Unit of Production V—12
5—6 Annual Market Cost of VOC Removal Per
Eousehold V—13
Summary of Current and Additional Costs
of VOC Regulation by State Governments V—18
5—8 Estimated Costs for VOC Regulation for
Sample States V—19
5—9 Estimated National Costs of State
Government Implementation of EPA
Regulations for VOCs V—19
6—1 Present Value N -et Benefits as a Function
of MCL and the Assumed Value Per Cancer
Case Avoided VI—5
6—2 Maximum Present Value Net Benefits as a
Function of MCL and Relation to Marginal
and Average Cost VI—7
6—3 Summary of Results of Present Value Net
Benefits Analysis V 111
-------
LIST OF EXHIBITS (continued)
EXHIBIT
NUMBER PAGE
7—1 Number of Community Water Systems by
Population Served VII—3
7—2 Systems Required to Treat as a Function
of MCL VII—5
7—3 Water Systems with Volatile Organic
Compounds V117
7...1 Proposed Monitoring for Unregulated VOCs VII—15
7—5 Unregulated VOCs Considered for
iMonitoring VII—17
7—6 Costs for Monitoring Unregulated VOCs VII—18
8—i 68 Confidence P1 t of Cancer Cases
Avoided as a Function of Annualized Cost VIII—2
8—2 95% Confidence Plot of Cancer Cases
Avoided as a Function of Annualized Cost VIII—3
8—3 Summary of Impacts of the Regulatory
Options VIII—5
-------
1. SUMMARY OF RESULTS
1.1 Introduction
This document has been prepared to provide information
on the economic costs and benefits associated with regulatory
alternatives, as required by Executive Order 12291. It considers
alternatives required to be examined by the Executive Order.
However, some of the alternatives arguably may not be allowed
under the Safe Drinking Water Act.
The resultant analysis is not the basis for decision—making
on drinking water regulations, but it does provide useful information
to assist the Administrator in understanding the impacts of
the alternatives. The basis for the regulatory proposal is
discussed in the preamble, and comes directly from the Safe
Drinking Water Act. Specifically, the MCL. is to be as close
to the Recommended MCL (health goal) as is feasible. Feasibility
involyes determination of what technologies are available to
provide for compliance with the regulation, as well as examination
of the ability to routinely determine the level of contaminants
in water. Costs are considered when examining feasibility.
These costs were found to be reasonable for treatment technologies
and analytical methods used for the volatile organic contaminants
proposed for regulation.
The proposed maximum contaminant levels (MCLs) are:
Tr chloroethylene 5 ugh Benzene 5 ugh
Carbon Tetrachioride 5 ugh 1,1_Dichloroethylefle 7 ugh
1,2—Dichloroethane 5 ugh 1,1,1—Trichloroethane 200 ug/l
Vinyl Chloride 1 ugh p—Dichloroberizene 750 ugh
Of the regulatory alternatives we examined, the one which is
most similar to the proposal above would control each constituent
to 5 ugh. The costs the nation would pay to meet that alternative
are $21 million/year and the benefits of the alternative are
32 cases of cancer avoided each year.
The report which follows describes the manner in which
these and other estimates of regulatory impact were reached.
It depends heavily on estimates of contamination occurrence,
removal technology cost and health effects which are discussed
each in their own background documents.
The data on contamination occurrence are contained in reports
prepared by EPA in 1983 and 198 4 under the general title,
Occurrence of VOCs in Drinking Water. Food. and Air . There
is a separate report for each contaminant. Information on
removal technology is contained in EPA, TechnologieS ar
Costs for the Removal of Volatile Organic Che ica1s fro
Potable t !ater Su iies . Information on health effects are
contained in Drinking Water Criteria Documents prepared by
EPA. Th re is a separate document for each contaminant.
I—i
-------
1.2 Problem Definition
As many as 29 volatile organic chemicals (VOCs) have been
evaluated through various EPA studies. Some of these have been
dropped from consideration for regulation due to the inadequacy
of available data on health effects. In a number of instances,
the risks posed by the chemical were determined to be non—existent
because the chemical was not actually found In drinking water.
Nine VOCs currently are being considered for regulation. These
nine chemicals are trichioroethylene, tetrachioroethylefle, carbon
tetrachioride, 1,1,1—trichioroethane, 1,1—dichiorOethYlefle,
1,2—dichioroethane, benzene, vinyl chloride, and para—dichioro—
benzene. One or more of these compounds occur in 12 percent
of the community water systems of the United States, typically
at low concentrations.
Some VOCs are thought to be carcinogenic, some are known
carcinogens, and all are toxic (see Exhibit 1—1). They enter
the body from drinking water through one of three route.s, either
by ixigestion, inhalation, or dermal absorption. Ingestion and
inhalation are estimated to be responsible for approximately
50 cases of cancer per year in the United States (see page IV—2).
Dermal absorption might add to this, although the magnitude
of cancer cases from VOCs absorbed through the skin is unknown.
Both surface and groundwater supplies are vulnerable to
VOC contamination. The source of the contamination is likely
to be improper disposal of industrial waste materials.
The role of government in VOC contamination problems can
be viewed in terms of the inability of the market to adequately
account for and respond to contamination problems in drinking
water supply. These “imperfections tt in the market are due to:
a) a poorly developed econcmic demand for health risk reduction,
and b) monopoly ccnditions in the supply of water. The demand
f r removing VOCs is poorly developed because knowledge of the
occurrence of VOCs is far from complete. In addition, health
risks ai-e poorly understood. Although the public is intensely
concerned about involuntary exposure to carcinogens, the trade—offs
between costs of VOC removal and reduced health risk due to
VOC removal are unclear.
Some states such as Florida and New Jersey have already
moved to promulgate standards on certain VOCs. Others, such
as California, New York, Pennsylvania, and Missouri have established
action levels or guidelines for removal of VOCs. In the abseOce
of federal regulations, states are approaching VOC control On
a case—by—case basis. Not all states have VOC programs, however.
This unevenness in state programs and the resulting unevenness
to the protection of the public health highlights the need for
federal standards on VOCs. Uncer the Safe Drinking Water Act
(SDWA), EPA is mandated to develop National Primary Drinking
Water Regulations.
1—2
-------
EXHIBIT 1—1
SUMMARY OF HEALTH EFFECTS OF VOLATILE ORGANIC COMPOUNDS
Cancer
o quality of evidence
Toxic Effects
o liver
o kidney
o central nervous system
o heart
o cardiovascular system
o gastrointestinal system
o adrenal system
o pulmonary system
o circulatory system/blood
o immunity
o other
0)
C
0 1 C O
C
4) 4, . 1.3
C — 4,
0) 0
p - I
0
0) —
0
0) 0
o 0
‘-4 1.
o 4 J
p-C 0 I
C., -
I —
I.. 0)
.1.1 1-
I
I I
I ‘ . I
I I
I I
41
C
4,
N
C
4,
0
I-
0-
0) ‘ -I
0)
N
C
41 I
C.
4)
4) 4)
C ‘-I
Co
C -
01 .3 .1.3
01 4)
— 0 0
1.. L.
o o 0
-4 -4 -4
.C . :
C) C) Q
.— _4
— .c •
I I
C C”) •
—
> I—
ol 0
III L I. LI
-‘ • ‘ .‘ I
— I i•..
P I I
I
I, I
I I , I
* I=known or probably arc1nogen; II:possible; III non—carcinogen or insufficier.
data
o Indicates that the VOC in the column is associated with. the health effec
in the row
HEALTH EFFECT
a,
0
‘-4
C)
Co
4)
4,
.1.)
C
0
. 0
I.
CO
C,
0
I
III
0
I
‘ I
I •- I .1
I “I
I #sI
I i’
I I
p lo I°
F 10 1 1:
1—3
-------
In general, the states prefer EPA to set standards for
VOCs. The health advisories developed by EPA over the past
five years are depended upon heavily by the states as is the
guidance provided on treatment techniques. Many states believe,
however, that setting standards on VOCs is something EPA is
better equipped to do and is a proper role for EPA in the federal—
state partnership.
At the local utility level, public pressure or the potential
for public pressure has led many utilities to take immediate
actions to address incidents of VOC contamination. The public
is sensitized to the threat posed by chemicals in drinking water
as a result of media coverage of hazardous waste disposal stories.
The treatment technologies that have been demonstrated
to be effective for VOC removal are some form of activated carbon
adsorption or aeration. Packed tower aeration, slat tray aeration,
and other variations have been used by some systems to remove
VOCs.
Many communities would opt for nontreatment alternatives
if faced with VOC contamination. Nontreatment alternatives
often considered by water utilities include the following:
o Weilfield management —— shutting down of contaminated
wells and increasing production from other existing
wells.
o Source protection —— monitoring of surface sources of
water and identifying polluters if contamination is
detected, thereby pressuring diachargers of VOCs to
cease pollution of water supplies.
o Regionalization —— interconnection with a nearby system’s
uncontaminated supply.
o Alternative source —— developing a new source to replace
• the contaminated source.
1.3 Alternatives for EPA Actions
There are three basic alternatives that EPA can choose
to limit human exposure to VOCs:
o Take no further action (other than dissimanating existing
health advisories) and let states and utilities continue
to deal with VOC problems.
o Require monitoring and report±ng of VOC concentrations
(coupled with existing health advisories), on the prem ise
that improved information will lead to efficient state
and utility actions.
1—4
-------
o Set maximum contaminant levels (MCLs) on the premise
that the difference in actions taken from state to state
under the other alternatives will not adequately protect
the public health. This will result from the fact that
some communities and states will not act without EPA
pressure and because some communities and states may
overreact and invest excessively in YOC removal. MCLs
are preceded by setting of recommended MCLs and are
accompanied by monitoring and reporting requirements.
Under the Safe Drinking Water Act, EPA must establish recommended
maximum contaminant levels for each contaminant which may have
any adverse effect on the health of persons. These recommended
MCLs must be set at a level at which no known or anticipated
adverse effects on the health of’ persons occur and which allows
an adequate margin of safety. EPA is required to issue maximum
contaminant levels as close to the recommended maximum contaminant
level as is feasible using the best technology, treatment techniques,
and other means, taking costs into account. EPA published proposed
recommended MCLs in the Federal Register on June 12, 198!!. -
1.!! Benefit Assessment
To assess the degree of health benefits from regulation
of VOCs, it is necessary to estimate the number of cancer cases
avoided per year at various MCLs. This analysis has estimated
the number of cancer cases avoided for v rious al ernative MCLs
including MCLs that are equivalent to 10 and 10 risk levels..
(Under the Safe Drinking Water Act, a risk—based approach to
setting MCLs is not permitted. However, Executive Order 12291
requires consideration of regulatory alternatives that are outsid
of curr%nt legislative authority. Thus MCLs equivalent to 10
and 10 risk levels were studied as a part of the development
of this Econimic Impact Analysis.)
Results show that a total of L19 cases of cancer would be
avoided per year if MCLs for all of the VOCs studied were set
equal to the analytical limit of detection achievable in the
best research labs (0.5 ugh for most of the VOCs, but 1.0 ugh
for vinyl chloride). For MCLs above the analytical limit of
detection, high and low estimates of cancer cases avoided were
developed. MCLs of 1.0 ugh for all VOCs produce estimates
of the number of cancer cases avoided annually ranging from
38 142. MCLs of 5.0 ugh for all VOCs produce estimates in a
range of 27—32. MCLs higher than these result in progressively
fewer cancer cases avoided per year.
MCLs equivalent to 10 and io 6 individual lifetime risk
lev p1s also were evaluated under Executive Order 12291. Thg
10 MCLs are approximate .ly equivalent to 1.0 ug/l and the 10
MCLs are roughly equivalent to 5.0 ug/l (See Exhibit 2—3).
Thus, the total number of cancer cases avoided by these two
alternatives are similar to those achieved by MCLs of 1.0 and
5.0 ugh, respectively. The actual estimates of the number
1—5
-------
of cancer cases avo tded per year are 38—42 for a risk level
and 26—32 for a 10 risk level.
These cancer risks would not necessarily be evenly spread
throughout the population. Within any population, some groups
of people are more susceptible to cancer from VOCs than are
others.
Although economists have tried to infer dollar values for
health risk avoided, it is difficult to put a dollar value on
a cancer case avoided. Economists have suggested values of
life ranging from $300,000 to $7,000,000. Such values may facilitate
policy analysis, but it should be recognized that most people
do not consciously put any dollar values on their own lives.
1.5 Cost Assessment I
Executive Order 12291 requires that “major rules” proposed
by federal government agencies be reviewed by the Pffice of
Management and Budget- COMB). There are three tests for a rule
being considered “major”: annual national cost, major cost
increases, and significant adverse effects on competition and
other aspects of the economy.
The “major rule” threshold for annual national cost is
$100 million per year. A substantial part of the costs of removing
VOCs can be traced to the capital investment required and the
operation and maintenance expense of treatment oj- .n.ontrea.trLent
measures. These costs to society (in 1983 dollars) are $157.L
million (annualized cost) if MCL.s are set at 0.5 micrograms
per liter for all VOCs; $101.7 million (annualized cost) for
MCLs of 1.0 ug/l; $21.2 million (annualized cost) for MCLs of
5.0 ugh; $11.1 million (annualized cost) for MCLs of 10 micrograms
per liter; $7.0 million (annualized cost) for MCLs of 20 ugh;
$6.0 million for MCLs of 25 ugh; $3.8 million (annualized cost)
for MCLs of 50 ug/l; and S2.L million for MCLs of 100 ugh.
Costs also were estimated for the MCL alternatives equivalent
to 10 and 10 ’ individual lifetime risk levels. The e annualized
costs were estimate at $98.7 million for the 10 ’ risk level
and $27.0 for the 1O risk level.
A major increase in costs and prices indicates, a major
rule. Cost increases in water production are more or less passed
along to customers. For the 5.0 ugh MCL . alternatives, the
additional costs of water production for the average system
in each of several size categories would be about as follows:
o For systems serving 25—500 people,- $0.54 per thousand
gallons, a nineteen percent increase over current costs.
o For systems serving 501—3300 people, $0.29 per thousand
gallons, a sixteen percent increase over current costs.
1—6
-------
o For systems serving 3301—50,000 people, $0.07 per thousand
gallons, a six percent increase over current costs.
o For systems serving more than 50,000 people, $0.02 per
thousand gallons, a two percent increase over current
costs.
Thus in terms of the Executive Order 12291 cost thresholds,
the 5.0 ugh MCL alternatives would not result in annual costs
over 100 million dollars and would not produce major cost increases
in most water systems.
1.6 Reffulatorv Flexibility and Paperwork Analysis
Under the Regulatory Flexibility Act, EPA must analyze
the impacts of proposed regulations on small entities. In the
case of water systems, small entities are likely to be water
systems serving fewer than 50,000 people. However, less than
10 percent of these small water systems would be affected by
a VOC regulation (other than by any monitoring requirements
applicable to all water systems). EPA guidelines on compliance
with the Regulatory Flexibility Act define the threshold for
a ttsignificant impact on a substantial number of small entities”
as an impact affecting 20 percent of the total population of
such entities. Thus, there is unlikely to be a “significant
impact on a substantial number of small entities” in this case.
Under the Paperwork Reduction Act, EPA must consider the
impacts of proposed regulations on the response burden of utilities
and states providing information on water quality. The monitoring
of drinking water for VOCs and reporting of violations of the
MCL are likely to be the largest component of reporting require—
ments.
Several approaches to monitoring requirements are being
considered by EPA. Three specific options have been developed.
The primary differences between the options relate to the extent
of specific sampling requirements and the provision of state
discretion. In each option, monitoring requirements are propos ed
to be phased in depending upon the size of the systems. Systems
that are most vulnerable to VOC contamination should be sampled
first. While EPA, studies have not shown a cleat distinction
between potential source of contamination and actual VOC contami-
nation, the ground water supply survey (GWSS) found that the
best correlation was between the size of systems and VOC contami-
nation. Therefore, proposed monitoring requirements will require
that the largest systems sample first. -
In addition to monitoring for VOCs, EPA is proposing to
establish monitoring requirements for unregulated contaminants
under Section 1L 145(a) of the Act. The rationale for proposing
monitoring regulations is that similar analytical techniques
to those used to measure the nine VOCs also can be used to measure
other VOCs of concern at relatively small additional cost.
.1—7
-------
EPA has developed monitoring cost estimates for each option
and these are summarized below. EPA is proposing that Option
2 be selected as the minimum enforceable monitoring requirement.
All costs shown are expressed in millions of dollars annually.
Initial Round Option 1 O tion 2 Option
Compliance $25.0 $9.3 $3.8
Unregulated 2.7 2.3 0.5
Reoeat Monitoring
Compliance 63.7 17.L 2.9
Unregulated 2.7 0 0
1.7 Summary of Costs and Benefits of VOC Removal
The total national benefits and costs of the alternative
MCLs were computed by multiplying the total number of systems
affected by an MCL times either the average cost of cdmpliance
or the number of cases of cancer avoided annually. This was
computed separately for different size classes of water systems
and the results were then summed to produce national totals.
Comparison of the total national benefits and costs of
alternative MCLs reveals that the options fall into two groups
or clusters; a relatively high cost group consisting of MCLs
of 0.5 and 1.0 ugh and a relatively low cost group consisting
of MCLs of 5.0, 10.0, and 20.0 ugh. The analysis shows that
there is as much as a four or five fold difference in cost between
the two groups compared to a difference of only a factor of
two on the benefits side. The degree of difference in these
results is due primarily to the fact that more stringent MCLs
would affect a larger number of water systems.
Analysis of net benefits at the level of individual water
systems indicates that the more stringentMCLS in the vicinity
of 0.5 and 1.0 ugh would only produce positive net benefits
at the system level in the case of certain of the more highly
carcinogenic VOCs. The net benefits analysis also shows that
regulation of all VOCs will produce the greatest positive net
benefits due to the fact that VOCs commonly occur together.
Removal of the less carcinogenic compounds will often result
in removal of some of the more harmful compounds as well.
1—8
-------
2. PROBLEM DEFINITION
2.1 Introduction
This chapter provides answers to the obvious questions
that are relevant and contribute to an understanding of the
problem of drinking water contamination with volatile organic
chemicals (VOCs). The chapter is organized into four major
sections and covers the following topics: 1) health effects
of VOCs; 2) nature and extent of VOC occurrence in drinking
water; 3) available technologies for the removal of VOCs from
drinking water; 14) the structure of the water industry; and,
5) the need and the available mechanisms for regulation of VOC
contaminants.
Some of the data presented in this chapter deals with the
general problem of VOC contamination, encompassing some 29 organic
chemical compounds. Only nine of these are the subject of -egulatory
actions being proposad. The others were studied in the process
of arriving at a decision to take action on the nine. NOTE:
All nine were originally suspected to be either animal or human
carcinogens. Two were later determined to be non—carcinogens
during the course of preparation of this RIA. As •a result,
much of the background data presented in Chapter 2 covers nine
chemicals and the analyses presented later in this document
cover eight suspected carcinogens. This has no significant
effect on the interpretation of the results of these analyses,
however, since the last compound dropped from consideration
was assumed from the start to pose much less carcinogenic risk
than the others.
2.2 Health Effects
The health effects attributable to exposure to VOCs may
be classified into two groups: 1) acute and chronic toxic effects
(non—carcinogenic effects); and, 2). carcinogenic effects. The
specific effects associated with each individual chemical are
described in Appendix A.
2.2.1 Acute and Chronic Toxic Effects
Exposure at very high levels to VOCs has been shown to
result in a variety of acute and chronic toxic effects. These
levels are usually much higher than those found in public drinking
water supplies. Damage to the liver and kidneys is a common
effect demonstrated in animals from high exposure to several
VOCs, as well as central nervous system effects and cardiovascular
changes.
The Acceptable Daily Intake (ADI) of a contaminant is the
level of intake from air, food, and water that is experimentally
determined to be the tT 0 effect” level; meaning no acute and
chronic toxic effects. The ADI is expressed in mg/kg body weight
11—1
-------
per day. For the nine VOCs, the ADIs have been converted to
“Adjusted” Acceptable Daily Intakes (AADIs) to represent the
equivalent no effect level divided by a safety factor specific
to each chemical, where the intake is totally due to drinking
water. Accordingly, AADIs are given in terms of mg/i; assuming
consumption of 2 liters of water per day by a 70 kg adult.
AADIs for VOCs are presented in Exhibit 2—1. This exhibit
also presents data on the estimated number of water systems
believed to have concentrations of VOCs in excess of the indicated
AADIs. As shown, there are believed to be very few water systems
that display VOC concentrations exceedIng the AADIs. Thus,
acute and chronic toxic effects due to VOC contamination are
probably not extensive.
2.2.2 Carcinogenic Effects
Carcinogenic effects have also been demonstrated from exposure
to certain VOCs. Two compounds are demonstrated human carcinogens,
while others have exhib-ited carcinogenic effects in animal studies.
The evidence of carcinogenicity for the nine compounds ranges
from sufficient evidence in humans to very limited or no evidence
in animals.
In the VOC RMCL proposal, EPA raised the question of the
strength of evidence on the evidentiary threshold required to
conclude that a substance should be considered to be a “carcinogen”
for the purposes of regulation. Subsequent to the June 12,
i98Lr, RMCL proposal, EPA proposed an approach for classifying
chemicals’ based upon the strength of evidence of carcinogenicity
(Proposed Guidelines for Carcinogen Risk Assessment 9 FR 4629 4,
November 198L1). EPA proposed a categorization scheme based
upon the International Agency for Research on Cancer (IARC)
criteria. The categorization consists of a five category approach,
as shown below. In contrast, the IARC classification consists
of three categories with the primary difference being that IARC
does not distinguish between those chemicals with inadequate
animal evidence •of carcinogenicity and those chemicals with
no evidence for carcinogenicity, while the EPA scheme makes
that distinction.
EPA Prooosed Categorization for Carcinogens
Group A — Human carcinogen (sufficient evidence from epidem—
iolôgical studies).
Group B — Probable human carcinogen.
Group Bi — At least limited evidence of carcinogenicity to
humans.
Group B2 — Usually a combination of sufficient evidence in
animals and inadequate data in humans.
ii—;
-------
EXHIBIT 2—1
ESTIMATED NUMBER OF WATER SYSTEMS EXCEEDING ADJUSTED
ACCEPTABLE DAILY INTAKE FOR VOCs
System Size CPooulation Served )
3301—
Comoound AADI* 25— 00 S01—1 00 co.ooo co.ooi+
Trichloro—
ethylene
0.26 mg/i 0 (a) 0 (a) 0 (a) 0
Tetrachloro—
ethylene 0.68 mg/i 0 0 0 0
1,1,1—Tn—
chloroethane 1.00 mg/i - 0 0 - 0
Carbon Tetra—
chloride 0.025 mg/i 0 0 7( 2 (c)
Vinyl Chloride 0.0146 mg/i 0 0 3 0
Benzene 0.025 mg/i 12 (d) 2 (d) 0 0
1 ,2—Dichloro—
ethane 0.26 mg/i 0 0 0 0
1 ,1—Dichloro—
ethylene 0.35 mg/i 0 0 0 0
p—Dichloro—
benzene 3.75 mg/i 0 0 0 0
* Adjusted Acceptable Daily Intake (149 FR 214338), excluding contributions
from air and food.
(a) data for systems with concentrations above 0.100 mg/i: 147(214—500
persons), 114 (501—3300 people), 3(3301—50,000 people)
(b) five systems estimated over 0.1 mg/i
(c) number of systems with concentrations between 0.020 and 0.030 mg/i
(d) systems with concentrations over 0.020 mg/i
SOURCE: EPA, Occurrence of VOCs in Drinking Water. Food. and Air , 1983
and i98 i. There is a separate volume for each chemical.
11—3
-------
Group C — Possible human carcinogen (limited evidence of carcino—
genicity in animals in the absence of human data).
Group D — Not classified (inadequate animal evidence of carcinogeni—
city).
Group E — No evidence of carcinogenicity for humans (no evidence
for carcinogenicity in at least two adequate animal
tests in different species or in both epidemiological
and animal studies).
IARC Criteria
Group 1 — Chemical is carcinogenic to humans (sufficient evidence
from epidemiological studies).
Group 2 — Chemical is probably carcinogenic to humans.
Group 2A — At least limited evidence of carcino.genicity
to humans.
Group 2B — Usually a combination of sufficient evidence
in animals and inadequate data in humans.
Group 3 — Chemical cannot be classified as to its carcinogenicity
to humans.
Both of these classification schemes are based upon a qua1i—
tative review of all availabLe evidence. Information considered.
in each assessment includes short—term tests, long—term animal
studies, human studies, pharmacokinetic studies, comparativ&
metabolism studies, structure — activity relationships and other
relevant toxicological studies.
Other groups have supported the concept of assessing carcinogens
by degree of evidence. The National Academy of Sciences (NAS)
in Drinking Water , Health , 1977. Vol. I, classified chemicals
into four groups: human carcinogens, suspected human carcinogens,
animal carcinogens and suspected animal carcinogens.
The Office of Science and Technology Policy’s recent review
of the science and associated principles of chemical carcinogenic
risk (50 FR 10372) generated a series of principles to be used
to establish specific guidelines for assessing carcinogenic
risk. The review discussed the type of tests (short— and long—term)
used to assess potential carcinogens and the interpretation
of data in light of numerous uncertainties. A classification
..system for carcinogens was not provided in the report; however,
the assessment of chemicals based upon weight of evidence was
supported.
The June 12, 198L RMCL proposal consisted of a two—category
approach to setting RMCLs; chemicals were classified as carcinogens
or non—carcinogens. All chemicals with evidence of carcinogenicitY
11—4
-------
ranging from limited to sufficient were classified as “carcinogens”
and their RMCLs were proposed at zero. A number of commenters
pointed out that this approach did not adequately take into
account the varying degrees of evidence of the chemicals classified
as carcinogens. Since the proposal, the EPA guidelines for
carcinogen risk assessment have been proposed which contain
a classification scheme for chemicals based on strength of evidence.
In response to comment and consistent with the EPA guidelines,
a three—category approach has been developed which considers
all of the available scientific data, as shown below:
Three Cate orv Anoroach
Category I — Known or probable human carcinogens: Strong
evidence of carcinogeriicity.
o EPA Group A or Group B
o IARC Group 1, 2A or 2B
Category II — Eq ivocal evidence of carcinogenicity.
o EPA Group C
o IARC Group 3
Category III — Non—carcinogens: Inadequate or no evidence
of carcinogenicity in animals.
o EPA Group D or E
o IARC Group 3 -
Category I includes those chemicals which, in the judgment
of EPA, have sufficient human or animal evidence of carcinogenicity
to warrant their regulation as known or probable human carcinogens.
Category II includes those chemicals for which some limited
but insufficient evidence of carcinogenicity exists from animal
data. Category III includes those substances with inadequate
or no evidence of carcinogenicity.
Exhibit 2—2 presents a summary classification of the nine
VOCs in terms of all of the three different schemes for categorizing
chemicals on the basis of strength of evidence.
Exhibit 2—3 presents the estimated level of carcinogenic
risk (lifetime) associated with the nine VOCs based on the analysis
of’ the Carcinogen Assessment Group (CAG).
2.3 Occurrence
Since 1975, EPA has conducted six nationwide surveys to
assess the occurrence of VOCs in drinking water supplies. The
most recent of these surveys are the most useful due to advances
h—S
-------
EXHIBIT 2-2
STRENGTH OF EVIDENCE ON CARCINOGENICITY OF VOCs
EPA IA or Thr
Evide e of Cuid qtd.valøit Category Proposed
C aT OLnd CrCirK enia .tY Lines* a s ficati Class fieatiou* _______
‘1 ich1Q r oethyl e limited andnel B2 3 I
evidexe
Ir deqt te
evidexe
Te ach].oroethy1ene limited an 1 C 3 II
evidexe
Inaden te Fn n
evidexe
Carbou Tetrachioride Carcinogenic in 3 2B I zero
species by the oral.
route
1,2—Dichioroethene Carcinogenic in 2 2B I
sp bytheoral
route
Vinyl Qiloride Carcinogenic in A 1 I
anin lc by the
ii 1aciou and oral.
route and cerdm-
genicinh nansby
in 1ation
Carcinogenicinaninnis A 1 I
and htnnns by inhelation
and in ax i l-c by gavage
1 ,1—Dichloroethylene Conflicthig evidence C 3
of carcazx genicity in
pii ids, riutagenic
p-Dichlorobeizene No evidexe E 3 m O.75 r#J
1,],1-Trichloroethene Prelin nary aninn]. D 3 Q.2 J1
evidexe being a itnd
Inedeqtnte luzan
evidexe
* pj tion in text.
11—6
-------
EXHIBIT 2—3
CAG RISK ESTIMATES FOR VOCs
AND ROUNDED VALUES
DETERMINING EQUIVALENT MCLs
ROUNDED VALUES 3
CAG RISK ESTIMATES DETERMINING EOUIVALENT MCL
io 6 MCLs MCLs
1O Risk for Equivalent Equivalent
Risk for Ingestion tp t
Inges9.on and 10° 1O
Chemical Only Inhalation 2 Risk Risk
Trichioroethylene 1.8 ugh .9 ugh 1 ugh 9 ugh
Tetrachioroethylene 1.0 .5 1 5
Carbon Tetrachioride 0.27 .135 1 2
1,2—Dichloroethane 0.5- .25 1 3
Vinyl Chloride 0.015 .0075 1 1
1,1—Dichioroethylene 0.24 .12 1 2
Benzene 0.67 .335 1 4
1,1,1—Trichloroet ane 4 21.7 10.85 11 109
p—Dichlorobenzene’
NOTES TO TABLE:
1 . The CAG risk estimates given represent the concentration of each individual
con aminant in drinking water that would be required to produce a
10— lifetime individual cancer risk. CAG risk estimates are based
on ingestion only.
2. The risk posed by inhalation is assumed to be equivalent to that posed
by ingestion. Thus, CAG risk estimates are halved to represent the
combined effects of ingestion and inhalation.
3. MCLs equivalent to the CÁO risk estimates are determined by rounding-
the CÁO estimates to the nearest integer. Rounding is necessary
because the analytical limit of detection for these chemicals is in
a range of 0.5 to 1.0 ug/l il_n the best research labs . Equivalent
MCLs provide a useful point of comparison only if they are in a range
that can be detected in the labs used by public water systems. A
conclusion of this table is that many of the rounded values are Still
not in such a range.
U. Determined not to be a carcinogen during the course of this regulatory
analysis; included in analysis of chapters 4 and 5.
5. Determined not to be a carcinogen; not included in analysis of chapters
4 and 5.
11—7
-------
in the state—of—the—art of analytical chemistry that have made
it possible to detect organic chemical contaminants at increasingly
lower concentrations. The two most recent surveys are the Ground
Water Supply Survey (GWSS) and the Community Water System Supply
Survey (CWSS).
As shown in Exhibit 2— 4, in the GWSS, 99 of ‘466 (21.2%)
randomly selected groundwater supplies had at least one of the
29 VOCs identified in that survey. In the CWSS, 50 of the 330
(15.2%) groundwater supplies had at least one of 10 VOCs identified
in that survey; and, iLl of 106 (13.2%) surface water supplies
were found to have one or more of the 10 VOCs present.
Exhibit 2_LI also provides data on multiple occurrences
of VOCs; ULI of 99 (I4LLLI%) sites in the GWSS where VOCs were
present had two or more of them, while 19 of 50 (38.0%) of the
groundwater supplies in the CWSS that contained VOCs had two
or more present and 5 of iLl (35.7%) of surface water supplies
that contained VOCs had two or more present.
An indication of the level of contamination possible when
multiple contaminants are present is illustrated by the figures
shown in Exhibit 2—5. These data, taken from the GWSS, show
that in 97 percent of the cases where VOCs are present, the
sum of the concentrations of’ the VOCs present is below 10 ugh.
Concentrations of VOCs totalling more than 50 ug/l are shown
to occur less than one percent of the time.
The above facts about the occurrence of VOCs are plausible
in view of the suspected pattern of’ causation behind VOC contam-
ination. VOCs are man—made organic compounds that are associated
with industrial waste disposal practices. VOC contamination
is characteristic of many hazardous waste disposal facilities
that have been identified by EPA as problem sites requiring
clean—up action. Thus, it is not surprising: 1) that VOCs are
found to affect only a portion (13—21%) of the total number
of water systems and not the entire water industry because proximity
to such wastes is a prerequisite; and, 2) that the presence
of’ multiple VOCs is fairly common (35—L4LI%) when contamination
exists since numerous chemicals are involved in industrial
operations.
Because VOCs are “volatile,” their presence in groundwater
supplies is both more probable and more likely to occur in higher
concentrations than in surface water supplies that are naturally
aerated. The CWSS data cited above show the detection frequencies
to be roughly the same in both ground and surface water supplies
(15.2% vs. 13.2%), but there are more than four times as many
community water systems using groundwater sources than surface
water sources.
Evaluation of’ the impacts of’ VOC regulatory alternatives
requires calculation of’ the total VOC exposure. Total exposure
is a function of the number of people exposed and the concentrations
11—8
-------
EXHIBIT 2_14
SUMMARY OF SINGLE AND MULTIPLE OCCURRENCE OF VOCs AS A CLASS
Ground water 14
280(814.9%)
50(15.2%)
19 (5.8%)
6 (1.8%)
14 (1.2%)
2 (0.6%)
0
0
0
1 Based on analyses for 29 VOCs. (Ground Water Supply Survey)
21466 supplies studied. (Community Water System Survey)
3 Based on analyses for 10 VOCs.
33O supplies studied.
io6 supplies studied.
Cw s s 2
No. of Contaminants
0
>1
>2
>3
> 14
>5
>6
>7
>8
GWSS 1
Random 3
367(78.8%)
99(21 .2%)
1414 (9.14%)
25 (5.6%)
1141(3.0%)
8 (1.7%)
Ii n
, v.,ø
2 (0.14%)
0
Surface water 5
92(86.8%)
114(13.2%)
5 (14 ,7%)
1 (0.9%)
0
0
0
0
0
11—9
-------
EXHIBIT 2—5
CUMULATIVE OCCURRENCE FOR WATER SUPPLIES IN THE GROUND WATER
SUPPLY SURVEY (GWSS) WITH TVOC CONCENTRATIONS ABOVE INDICATED LEVELS
(Total Number of Water Supplies Sampled = 166)
Number of Supplies Percent
Concentration Levels Above Indicated Level of Samol
detection limit’ 99 21.2
5.0 ugh 20 i4.3
10.0 12 2.6
50.0 2’
100.0 0 0.0
* Detection limits varied by chemical from 0.5 to 1.0 ugh.
11—10
-------
to which they are exposed over time. EPA has prepared estimates
to serve this purpose, drawing on the best information from
all recent survey work. Exhibits 2—6 and 2—7 present estimates
of the number of water systems by size and source which have
VOC contaminat•ion above the analytical limit of detection of
0.5 ugh for each of the individual VOCs. Confidence intervals
associated with these estimates are presented in Exhibit 2—8.
Exhibits 2—9 and 2—10 present estimates of the proportion
of ground and surface water systems that are above various levels
of influent concentration for each of the individual VOCs.
2.L Control Technologies
The range of technologies for controlling VOCs in drinking
water includes five treatments. The following list includes
the treatments considered and the basic assumptions used in
this economic impact assessment. The complete basis for defining
the available treatments is given in Technoloiies and Costs
for Removal of Volatile Organic Chemicals from Potable Wate
Supplies . A more detailed summary of this document is also
included in the Federal Register notice proposing these regulations.
Treatments are assumed to be designed for various influent concentra-
tions, ranging from 5 ugh to 100 ugh and various effluent
concentrations ranging from 0.5 ugh to 50 ugh.
o Granular Activated Carbon (GAC) : a new installation
if selected.
— it is assumed that 100 percent oT the water is contam-
inated and therefore treated.
— systems with flows less than 2 mgd use factory—assembled
steel pressure vessels.
— carbon usage rates depend on the VOCs present and
their influent concentration.
o Packed Tower Aeration : new installation if selected.
— it is assumed that 100 percent of the water is contam-
inated and therefore treated.
— a maximum column diameter o 10 feet, a maximum liquid
loading rate of 3 gpm/ft’, and maximum blower size
of 6L 00 standard cubic feet per minute were assumed.
— in some states, granular activated carbon may also
have to be used to prevent air pollution from air
stripped VOCs; these costs are not considered here.
o Diffused Air Aeration : retrofit of existing treatment
equipment if selected.
II—”
-------
EXHIBIT 2—6
ESTIMATED NUMBER OF GROUND WATER SYSTEMS HAVING CONCENTRATIONS
GREATER THAN OR EQUAL TO 0.5 ugh FOR THE INDICATED VOCs
System Size (Population Served)
Contaminant 2c— oO c01 — 0O O1— 0.00O co,ooi+
Trichioro—
ethylene 1128 330 153 20
Tetrachioro— I
ethylene 996 290 259 7
1,1,1—Tn—
chioroethane 961 283 138 8
Carbon Tetra—
chloride 82 196 53 5
Benzene 1453 132 142 a
p—Dichloro—
benzene 367 106 35 2
Vinyl Chloride 4 0 0 27 14
1 ,2—Dichloro—
ethane 9 14 27 37 3.
1, 1—Dichloro—
ethylene** 593 173 86 6
1 or more
VOCs 36140 1160 528 3’l
Number of Sys-
tems in U.S. 314799 10168 3358 133
* Number of Systems with concentrations greater than or equal to 1
ugh.
4* Number of Systems with concentrations greater than 0.2 ugh.
SOURCE: EPA, Occurrence of VOCs in Drinking Water. Food. and Air , 1983’
and 19814. There is a separate volume for each chemical.
11—12
-------
EXHIBIT 2—7
ESTIMATED NUMBER OF SURFACE WATER SYSTEMS HAVING CONCENTRATIONS
GREATER THAN OR EQUAL TO 0.5 ug/]. FOR THE INDICATED VOCs
System Sf e (Population Served)
Contaminant 2 — OO O1— OO O1 _ O .OOO O.OO1.i .
Trichioro—
ethylene 101 98 234 65
Tetrachioro—
ethylene 34 33 - 87 26
1,1, 1—Tn—
chioroethane 0 0 62 21
Carbon Tetra—
chloride 96 94 298 89
Benzene 106 102 81 12
Para—Dichioro—
benzene 0 0 9 3_
Vinyl Chloride* 0 0 9 3.
1 ,2—Dichloro—
ethane 68 66 143 39
1, 1—Dichloro—
ethyaene** 0 0 26 9
1 or more
VOCs 427 414 616 150
Number of Sys—
terns in U.S. 3937 3817 2994 454W
* Number of systems with concentrations greater than or equal to 1 ugh.
** Number of systems with concentrations greater than 0.2 ugh.
SOURCE: EPA, Occurrence of VOCs in Drinking Water. Food, and Air . 1983
and 1984. There is a separate volume for each chemical.
11—13
-------
EXHIBIT 2—8
CONFIDENCE INTERVALS FOR NUMBER OF WATER SYSTEMS
HAVING VOC CONTAMINATION
1Q
trichioroethylene
tetrachioroethylefle
1 , 1 , 1 —trichloroethane
carbon tetrachioride
benzene
p—d ichlorobenzene
vinyl chloride**
1 ,2 —dichloroethafle
1 ,1_dich].oroethylene***
95% Confidence Interval for Number
of Systems Having VOCS*
Groundwater Systems Surface Water Systems
1031—2228 216—783
10147 2066 13_3 146
799—198k 29—136
12 —545 299—850
330—939 614 537
197—822 0_3L1.
13— 8 0—3w
0—337 82—55.1
336—13’49 0—81
* Number of systems with VOC concentration greater than or equal to
0.5 micrograms per liter.
Number of systems with concentration greater than or equal to 1 ugh.
Number of systems with concentration greater than or equal to 0.2
U g/l.
SOURCE: EPA, Occurrence of VOCs in Drinking Water. Food. and Air , 1983
and 198L1. There is a separate volume for each chemical.
**
“4
11—14
-------
EXHIBIT 2—9
INFLUENT CONCENTRATIONS OF VOCs IN GROUNDWATER SYSTEMS
Percentage of Systems With Influent Concentration
Greater than Concentration Indicated
Contaminant O. ugh c.o u I1 20.0 u L1 co _ p ughi
trichioroethylene 3.37 0.87 0.U3 0.27
tetrachioroethylene 3.20 0.66 0.1 4 0.01
1,l,].—trichloroethane 2.87 0.76 0.16 0.01
carbon tetrachioride 0.69 0.23 0.00 0.00
benzene 1.30 0.28 0.03 0.00
p—dichlorobenzene : 1.05 0.00 0.00 0.00
vinyl chloride* 0.06 0.0L 0.01 0.01
1,2—dichioroethane 0.33 0.00 0.00 0.00
1,1_dichloroethylene** 1.77 0.17 0.00 0.00
* Lowest influent concentration is 1 ugh
** Lowest influent concentration is 0.2 ugh
Source: EPA, 0c urrence of VOCs in Drinking Water. Food, and Air . 1983
and 198L1.. There is a separate volume for each chemical.
11—15
-------
EXHIBIT 2—10
INFLUENT CONCENTRATIONS OF VOCs IN SURFACE WATER SYSTEMS
Percentage of Systems With Influent Concentration
Greater than Concentration Indicated
Contaminant 0.5 u /l 5.0 u Il 20.0 uç/1 50.0 u Jl
trichioroethylene Ll.Z5 0.08 0.08 0.00
tetrachioroethylene 1.61 0.00 0.00 0.00
1,1,1 —trichioroethane 0.7 14 0.00 0.00 0,00
carbon tetrachloride 5.15 0.25 0.08 0.00
benzene - 2.69 0.00 0.00 0.00
p—dichlorobenzene - 0.11 0.00 0.00 0.00
vinyl chloride 4 0.11 0.00 0.00 0.00
1,2—dichloroethane 2.82 0.73 0.00 0.00
1,1 —dichloroethylene 44 0.31 0.00 0.00 0.00
* Lowest influent concentration is 1 ugh
** Lowest influent concentration is 0.2 ugh
Source: EPA, 0ec urrence of VOCs in Drinking Water. Food, and Air , 1983
and 198L . There is a separate volume for each chemical.
II—i6
-------
— all water is treated.
— an existing basin with a detention time of 20 minutes
is assumed to be available for retrofitting with
diffusers.
— fine bubble diffusers are assumed.
— blowers are assumed to provide an air—to—water ratio
of 8:1,.
— in general, a removal efficiency of 90 percent or
less is achieved.
o Tray Tower Aeration
— it is assumed that 100 percent of the water is contam-
inated and therefore treated.
— aerators are assumed to be 16 feet in height.
— aerato area is based on a hydraulic loading of 50
gpm/ft ; maximum surface area of a single aerator
is 200 square feet.
— in general, a removal efficiency of 90% or less- is
achieved.
o Powdered Activated Carbon (PAC) : additional PAC used
by systems already using PAC.
— all water is treated.
— PAC use is increased by conventionally designed water
plants which treat surface water and which already
add PAC for taste and odor control.
— no capital costs are required.
— an additional 20 mg/l above that normally used at
the plant for taste and odor control is assumed.
— on average 25 to 50 percent of VOCs are removed with
this method.
A number of alternatives to central treatment solutions were
evaluated. These techniques are listed below with the respective
assumptions used in this analysis. Bottled water and point—of—use
treatments were not thcluded in the list, however. Point—of—use
devices were considered, but they were not included in the cost
analysis beca-use they were judged to be too expensive; most
water systems would probably select another alternative. Bottled
water does not address the problems of inhalation and dermal
absorption.
11—17
-------
o Source Protection : Source protection consists of surface
water systems monitoring their water supply to identify
VOCs. This has a short_termeffectofdiminishingdischargeS
of contaminants into the water source, since producers
of the contaminants (typically industries) know that
they are being monitored. Small systems are often “free
riders,” since larger utilities drawing upon the same
source will engage in monitoring and then report results
to smaller systems at no cost to the smaller systems.
o Re iona1ization : Regionalization consists of inter-
connection with an existing system with sufficient capacity
to supply all of the affected community’s water. This
technique could be used by small water systems. It
is assumed that the host utility has an adequate supply
of water and could meet any VOC standard. Costs include
new pumping stations and transmission pipelines. Costs
exclude the procurement and treatment of’ water because
the net cost of this water depends on the characteristics
of the host system and the contaminated systems; net
costs (the difference between the host system and dis-
continued and contaminated system) may be very small.
A weighted average of pipeline lengths of’ 3, 7.5, and
10 miles is assumed.
o We11fi 1d Management : We-ilfield Management cons ists
of pumping to waste of contaminated wells and increasing
production from remaining wells. Twenty percent of
the wells in a welifield are assumed to be contaminated.
• This alternative would be selected only by groundwater
systems with sufficient well capacity remaining to supply
their needs without drilling new wells. Thus, only
larger groundwater systems would adopt weilfield management
(greater that 1 mgd).
o Alternative Source : Alternative source consists of’
drilling of new wells and installation of appropriate
transmission pipelines to replace- a contaminated ground
or surface source. For larger systems not all the water
is assumed to be contaminated. The distance to the
new wells is a function of system size, but new wells
are assumed to be less than five miles away except for
systems serving over 500,000 people. These new wells
would supply all or some of the contaminated utility’s
water, depending on system size. It is assumed that
alternative groundwater sources are of such quality
that existing treatment is adequate. The exising distri-
bution system is utilized.
11— 18
-------
2.5 Market Characteristics. Imperfections. and the Need for
Regulation
2.5.1 Structure of the Water IndustrY
The American public is served by approximately 216,800
water systems. These are the water systems defined by the Safe
Drinking Water Act as public water systems and which are under
federal jurisdiction in terms of regulation. These systems
are divided into two categories: community water systems (CWS),
and non—community water systems (NCWS). CWS constitute 27 percent
of the systems and serve primarily residential areas, while
NCWS make up the other 73 percent of the water systems and serve
mainly transient or non—residential areas. There is a total
of 158,100 non—community water systems which serve approximatelY
36 million persons.
Community water systems are defined as those serving 25
or more persons, or, having at least 15 service connections.
There are approximately 58,718 community water systems in the
country. Of these, approximately. 37,813 (611.k%) can be categorized
as “very small” —— serving 25—500 people; 13,915 (23.7%) can
be categorized as “small” —— serving 500—3300 people; 3,9 3
(6.7%) are “medium” —— serving 3,301—10,000 people; 2,770 (U.7%)
are “large” —— serving 10,001—100,000 people; and only 277 (0.5%)
are• classified as “very large” —— serving more than 100,000
persons. More than 6 4 percent of the systems serve less than
2.6 percent of the population, whereas about 0.5 percent of
the systems serve more than U4 percent of the popuLatiofl.
Urban water systems (a subset of CWS) are defined as those
systems which serve 50,000 or more persons. There are approximately
750 urban water systems in the U.S.; these large systems serve
56 percent of the community water system population.
Surface water is the primary source for 18.8 percent of
community water systems and is the primary source for 65.6 percent
of the population served by community systems. Ground water
is the source for 81.2 percent of the systems, serving 3i4.
percent of the CWS population. In general, the CWS falling
into the very small, small, and medium population categories
use ground water as their primary source, while the larger size
categories use surface water to a greater extent. Conversely,
96 percent of the non commUnitY water systems are served by
ground water sources. EPA data show an increase in the use
of ground water sources between 1975 and 1980 in the smaller
categories, and a decrease in the larger population categories.
2.5.2 Ownershi D
Public systems are predominantly owned by municipal governments,
although a sizeable number of systems also are owned by the
federal government. Large wholesalers of water, such as the
Metropolitan Water District of Southern California, are one
11—19
-------
of’ the major owners of very large systems. Publicly owned systems
serve approximately 85 percent of the total population which
use community water supplies. Approximately 73 percent (553)
of the urban water systems are publicly owned.
According to the Survey of 0Deratin and Financial Character-
istics of Community Water Systems (U.S. EPA Office of Drinking
Water, 1982), there are about 15,7 4O private water systems serving
some 31,000,000 people. Private systems are usually investor
owned in the larger population size categories. In the small
and medium size categories, however, they tend to be. owned by
homeowners associations. In addition, there are about 16,907
“ancillary” systems serving another 1.7 million people who live
in trailer parks and other small developments. As evident by
the comparatively small population served, these are typically
very small systems and are generally not thought of’ as private
water systems in the conventional sense of a regulated utility.
The largest number of investor owned syste rs are in
Pennsylvania, Florida, Texas, and California. In addition,
two states, Connecticut and New Jersey, have the majority of
their populations served by private water companies.
The total number of investor owned systems is increasing.
This is primarily the result of growth in rural, underdeveloped
areas in Florida, Texas, Arizona, California, and other Sunbelt
states. In suburban or rural areas, housing developers have
little choice but to “go into the water business.”
The trend in larger private systems is in the other direction.
Suburban systems are being taken over by cities either through
condemnation suits, or because the water system owner cannot
obtain rate increases (from the state public service commission)
large enough to yield sufficient profits.
2.5.3 Characteristics of the Municiosi Water Suoolv Industry
The water supply industry is both mature and conservative.
Because it is mature, the rate of innovation is low. Consequently,
the conventional process by which drinking water is treated
has not appreciably changed in the past few decades. Surface
waters typically are treated by a combination of unit processes
that includes chemical mixing, coagulation and flocculation,
sedimentation (or clarification), filtration (usually through
sand or a dual media), and disinfection (usually chlorination).
Special treatment processes often are needed to remove iron
and manganese, color, hardness, or organic contaminants such
as THMs or VOCs.
Innovative treatment techniques-are periodically introduced
by U.S. equipment manufacturers, but acceptanc&of either new
technologies or those proven to be effective in Europe (e.g.,
ozone or granular activated carbon) has been slow.
11—20
-------
Water supply is a highly capital intensive, yet not highly
profitable business. Water has the second highest asset/revenue
ratio of any utility. Water utilities exhibit tremendous economies
of scale and can be considered natural monopolies since no two
water utilities serve the same geographic area.
2.5.11 Regulation of the Water Su olv Industry
Primary responsibility for the provision of water has tra—
ditionally been with the local governments. The role of the
federal government has been to support local and state govern—
inents in meeting water supply needs (primarily through construction
of large water resource development projects) and to address
problems beyond the scope of lower levels of government. The
other federal role has been regulatory in nature. In 19711,
Congress enacted the Safe Drinking Water Act which required
EPA to set national drinking water standards and also provided
support to state programs in the form of grants. Under the
provisions of this act, EPA promulgated National Interi.m Primary
Drinking Water Regulations (NIPDWR) which went into effect on
June 211, 1977.
Both public and private systems are regulated by state
health departments. Fifty—three of 57 states and territories
have accepted primary enforcement responsibility under the Safe
Drinking Wate.r Act. All 57 must approve plans and specifications
for either new water facilities or additions to water facilities.
Investor owned systems also are regulated by state public
utility commissions (PUC’s). State PUC’s have the authority
to approve rate increases, regulate quality and quantity of
service, approve areas in which a utility can operate, and approve
methods and levels of financing. In reality, however, most
PUC’s regulate only rate increases.
2.5.5 Rate Levels
The price charged for water is a function of the average
cost of producing the commodity. The average cost is then marked
up in order to attain a profit level. Pricing is also affected
by “unaccounted—for” water —— total production of water minus
total deliveries of water. If water is “lost,” rates are set
higher, so in fact metered customers pay for more water than
tney actually receive. This occurs when utilities cover the
total cost of production. Cost includes operating and maintenance
costs, depreciation, taxes or payments in lieu of taxes (however,
publicly—owned systems do not generally pay taxes, and as a
rule, record rio depreciation expense), and interest expense.
Operating and maintenance costs are the direct costs of
producing water arid maintaining the water system — labor, fuel,
electricity, chemicals, repairs, and the i.ike. Treatment costs
can affect operating and maintenance costs. Treatment costs
vary depending on the source of the water, contaminants in the
II—2
-------
water, and compliance with drinking water standards. The proportion
of delIveries (only in the larger systems) going to residential
versus industrial users also has an impact on cost.
Water rates vary throughout the country. In many small
systems, the consumer may pay as little as $2—5/month flat rate
for water. In some larger investor owned systems and in arid
areas of the west and southwest, water rates can be as high
as $2—3 or more per 1000 gallons. The trend in water rates
is- toward more frequent increases. Also, because water utilities
exhibit tremendous ec’onomies of scale, declining block structures
typically have been the rate structure preferred by utilities.
This too is changing as the public perception of the value of
water, which always has been low, increases.
2.5.6 Market Imperfections and the Need for Rejulation
Some economists believe that market forces by themselves
are an efficient mechanism for allocating resources. ..EconomiC
theory holds that in à perfect market, the natural market forces
allocate resources efficiently and that the common good is repre-
sented by economic efficiency, a state in which society has
maximized its net benefits. Under the conditions of perfect
competition with no externalities, efficiency can be reached
through the process of individual actions in the marketplace.
In such a perfect market, there would be no need. for government
intervention.
The marketplace is characterized by ttimperfections,?’ however,
and experience has shown that government intervention often
is justified. The question to be examined in this section is
whether the market can determine the best course of action for
limiting human exposure to VOCs from drinking water.
There are several reasons why-the market, acting alone,
cannot -produce the most efficient set of action3 with regard
to VOC control:
1) the general public is generally uninformed about the
presence of VOCs in drinking water;
2) the public is generally unaware of the health risks
posed by exposure to VOCs;
3) consumers typically cannot place a dollar value on
health risks avoided; and
L ) water is supplied to the public by regulated monopolies,
not by competing utilities.
Because these conditions exist, it is difficult to register
the demand for drinking water quality as a function of the amount
consumers are willing to pay. The same conditions make it difficult
to represent competitive costs of supply in the marketplace.
11—22
-------
A brief illustration of conditions on both the demand and
supply sides of the market for drinking water quality shows
that imperfections exist and that government intervention is
needed. On the demand side, several conditions must be met
before a perfectly operating market can exist. These conditions
deal with the preferences of water consumers. One of the demand
side conditions says that money and drinking water quality must
be comparable; that is, each individual must be able to state
preferences between money and drinking water quality. In the
absence of a centralized, authoritative source of market information
on the subject (such as the EPA), this condition clearly cannot
be met.
On the supply side of the market, it is clear that water
utilities do not represent a competitive market. Each community
served by a water utility is served by only one water supplier.
Therefore, monopoly conditions exist. Water utilities are natural
monopolies in that they exhibit large economies of scale and
competitive services from other suppliers would notbe cost—
effective.
In conclusion, neither the demand nor supply sides of the
market for drinking water meets the conditions for a perfectly
competitive market. As a result, economic efficiency cannot
necessarily be realized from interacting market forces alone.
Market imperfections are so strong that government intervention -
is required to achieve efficiency.
In the specific case of VOC regulations,, the following•
reasons for imperfection apply:
1) the economic demand for VOC removal is not quantifiable;
2) the public is not generally knowledgeable about health
risks;
3) information 2bout VOC occurrence is incomplete;
14) suppliers of drinking water are natural- monopolies;
and
5) drinking water cannot be provided at different quality
levels in the same.service area.
Thus, the market does not provide a mechanism for registering
the costs and benefits of VOC removal and for maximizing the
net benefits of VOC removal.
2.6 State Actions to Control VOCs in the Absence of Federal
Regulations
In 19811, EPA conducted an informal survey of 13 state water
supply directors to determine what actions states were taking
to control VOCs in the absence of federal regulations. These
11—23
-------
states were distributed across all ten federal regions. The
matrix shown in Exhibit 2—11 summarizes the results of the survey.
There is substantial variation among state programs, as
could be expected. There are many comrnonalities in approaches,
however. The principal characteristics of state VOC programs
are as follows:
o Virtually all states react to incidents of VOC contamination
on a case—by—case basis. The level of clean—up required
is decided on a judgmental basis after the water supply
professionals consult with state toxicologists and epi-
demiologists.
o All states interviewed depend heavily upon health advisories,
reports, and other toxicological and health data developed
by EPA.
o Six of 13 states contacted had conducted extensive.,statewide
surveys of VOC contamination in water systems. Some
of the others cited lack of resources, few occurrences,
or dependency on EPA as reasons for not having conducted
monitoring surveys.
o Several states have established guideline or “action
levels” for certain VOCs. The action level is the concen-
tration at which a utility must notify the state; substantial.
exceedence of an action level requires mitigative- acti.on -
by the utility. California has set action levels for
the same nine VOCs being considered for regulation by
EPA; Pennsylvania, New York, and Missouri have established
guideline levels for certain VOCs.
o The State of Florida has established maximum contaminant
levels for eight VOCs; these MCLs went into effect in
May, i98 4. Several other states have indicated that
they will move to establish regulations if EPA does
not do so. These states include Wisconsin, Washington,
Maryland, and California. New Jersey, although not
interviewed, is moving quickly to establish their own
regulations. These standards are not uniform from state
to state.
o Some states indicate that they will not establish MCL.s
for VOCs even in the absence of EPA regulations. Some
states, such as Colorado, are prohibited from establishing
MCLs unless EPA first promulgates regulations.
o Some states would prefer that EPA not promulgate regulations
for VOCs. This is a minority opinion and is based on
the concern that federal standards will be established
at too high a level. However, most existing state standards
or action leve -ls are well below the proposed MCLs.
11—24 -
-------
EXhIBIT 2—Il
SUMMARY OF VOC PROGRAM ACTIONS IN ThIRTEEN STATES
the nine VOCs in RiA ____
the nine VOCs in RIA ____ ____
the nine VOCs in RIA ____
N
N
N
N
N
I. Type of State Re uiat ion or Action
• NCL
- Total VOCs
- One or more of
- Other VOC
• Action Level
— Tolai VOCs
- One or more of
— Other VOC
• ieaIth Advisory
- Total VOCs
-. One or more of
- Other VOC
•. Case by Case Consideration
2. MonItoring Requirements
• Total VOCs
• One or more of the nine VOCs in RIA
• Other VOC
3. Survey for VOC Contamination
• Total VOCs
• One or more of the nine VOCs in RIA
• Other VOC
4• Provision of Technical Assistance
• islaraat ion coiicerning removal measures
• Conduct IoiIcoiogicai studies
• Review ane evaluate public health data
• Conduct epldemlOlOgiCai siudies
• Uses U .S. (PA lteailh Assessmeni Data
N
N
N
N
N
N
N
CA
cO
FL
LA
HA
Hi
HO
NY
PA
WA
WI
I——_
—_———a———
i
N
N
N
N
N
N
N
N
N
N — — N
N
N
N
— N
I
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
U
N
N
-------
o The state water supply directors feel that the amount
of scientific evidence on the health effects of these
chemicals is sketchy. Therefore, any concentration
of VOCs in drinking water is a matter of public concern.
As a result, most state directors feel that national
standards based on good health data are needed.
o States would like EPA to adopt flexible monitoring and
reporting requirements. The state directors believe
they are better positioned to decide which systems to
monitor than is the EPA.
As noted above, all states approach occurrences of organic
contamination on a case—by—case basis. After confirming the
levels of contamination by specific chemicals, state public
health personnel review EPA’s Health Advisories and other available
public health information and develop recommendations for action.
For carcinogenic (actual or suspected) contaminants, recommendeg
target action levels have, in some cases, been based 0-n a 10
risk level.
Predominant actions taken in response to occurrences of
VOC contamination appear to be closing down the contaminated
wells and drilling new wells in the same well field.
Although most states want EPA to set MCLs for VOCs, and,
quickly, five state water supply directors contacted would prefer
to see no EPA regulation of VOCs or other ontaminants . As
they view proposed regulations, the primary problem centers
on lack of sufficient credible scientific data upon which to
fix specific maximum exposure levels.
In the absence of EPA MCLs, these state officials are able
to convince affected water utilities that “less contamination
is better than more,” and these affected systems are taking
actions appropriate to the contaminants and their concentrations.
In the absence of regulations, a water system having a 10 ppb
level of TCE, for example, will most likely take action to minimize
this level of contamination. On the other hand, if EPA sets
a TCE MCL at a higher level, the supplier with 10 ppb may be
able to justify not taking any action to reduce contamination
levels. -
2.7 Actions Taken by Water Utilities in Resoonse to VOC
Contamination
Informal -interviews were conducted with nine individual
water supply managers to determine the actions taken by local
water utilities in response to VOC contamination of water supplies.
Pertinent information also was obtained from informal interviews
with the state water supply directors.
Questions specifically asked managers of water supply systems
were directed at the nature of their VOC problems, how they
11—26
-------
learned of the problem, what actions were taken or are planned,
and how action decisions were made. In those cases involving
shutdown of a contaminated water supply, information was solicited
as to how the utility proceeded to find alternate sources of
supply.
2.7.1 Nature Of The VOC Problems Encountered -
All VOC incidents cited occurred in groundwaters, and nearly
all are caused by chlorinated organics. A few cases involve
benzene, toluene, and xy].enes, which are thought to be present
as a result of leaking underground gasoline storage tanks.
Many instances of VOC contamination stem from leakage of waste
disposal sites; others from discharges of contaminating chemicals
into rivers which feed underground welifields.
In instances for which the contaminating source has been
found, levels of contamination have been reduced significantly
after cessation of discharge or cleanup of the polluting site.
However, in many other instances, the sources of contamination
have not yet been located, and interim measures, including treatment,
have been instituted.
In many instances of VOC contamination, other organic chemicals
also are present which are not listed by EPA as VOCs. These
include pesticides, PAHs, and similar organics.
2.7.2 Aetions Taken To Date
The types of actions taken in response to ‘VOC occurrences
vary widely and are always site—specific. As a result, no clearcut
trends could be established to. predict specific actions based
upon the small sample of systems contacted.
It was hypothesized that large, medium and small water
systems would respond differently to VOC contamination problems.
as a result of differences in availability of resources, analytical
equipment, and trained personnel.
Similarly, it had been anticipated that the smaller systems
would not be capable of taking much responsive action, nor would
they have the resources to take action. The larger utilities
were expected to be able to take fast action, and the medium
sized utilities were expected to respond it an intermediate
fashion, depending upon the extent of their resources.
In fact, all small public water systems contacted which
have VOC contamination have taken significant actions on their
own initiatives. It can be argued that such actions were taken
in- response to public pressure, or to avoid public pressure,
or through lack of balanced information of the risks that are
associated with the contamination found.
11—27
-------
The most common response to VOC contamination problems
is to close the contaminated well. If other wells are available
in the field, these are used to replace the contaminated source.
If another well is not available, the typical action is to drill
a new well.
When the source of contamination is an uncontrolled hazardous
waste site on the CERCLA National Priority List, this has caused
mixed reactions. When a Superfund site is involved, indications
are that more testing and surveying of the extent of contamination
is conducted, and more extensive treatment is likely to be installed
to correct the problem, not only at the hazardous waste site
itself, but also to treat the contaminated water supply.
2.7.3 Factors Which Influenced How Action Decisions Were
Made
The overriding factor affecting actions taken was public
response to notification of VOC contamination. One state official
advised that the public can cause faster responses from local
utilities than can state agencies. However, several local water
supply managers have found the key to harnessing the power of
public pressure to work to their benefit.
In several cases local water officials have approached
the public with presentation of all data available, a discussion
of what is known about the health risks, alternative action
plans, costs, and timing. Unanimously, these officials have
found that .when they have been forthright with the publicS, the
first benefit is lack of pressure to move too rapidly without
proper planning.
Most local water supply utilities take direction and seek
guidance from the state water supply agency and from local health
departments. They are not generally concerned about EPA actions,
although they do appreciate that state water agencies respond
toEPA regulations, and adjust and adapt state actions accordingly.
2.8 Control of VOCs Via Other Federal LegislatiOn
There are eight major pieces of legislation that control
exposure to toxic substances such as the nine VOCs. Al). eight
Acts control one or more of the VOCs in this analysis to some
extent. Each -Act uses a variety of control measures to form
a complex pattern of regulation. For example, while the Food
and Drug Administration (FDA) has authority to control levels
of contaminants in a wide range of foods, the Department of
Agriculture’s Food Safety and Quality Service (FSQS) controls
the levels of contaminants in meats through the Federal Meat
Inspection Act and the Poultry Products Inspection Act. Howe.ver,
the FSQS usually incorporates FDA’s tolerances for food additives.
In addition to overlapping authorities, each Act usually empowers
a number of regulatory programs. These regulatory programs
do not control all of the nine VOCs in the same manner or to
11—28
-------
the same degree. Under the Clean Air Actts new source performance
standards (NSPS), for example, only a few of the VOCs are regulated.
The following discussion provides a brief summary of the
regulatory programs authorized by the above—mentioned legislation
concerning control of exposure to one or more of nine VOCs as
well as to other toxic substances.
o Clean Water Act : This act provides protection to surface
waters, i.e., water bodies including those not specifically
designated as drinking water supplies through regulatory
programs such as Toxic Pollutant Effluent Standards,
Water Quality Criteria, Effluent Limitations, New Source
Performance Standards, Pretreatment Standards for New
and Existing Sources, and Designation of Hazardous Substances
and Reportable Quantities. It indirectly controls human
exposure through limiting emission of toxic substances
into surface waters.
o Clean Air Act : This act protects air quality. Programs
include the National Ambient Air Quality Standards,
Emission Standards for Hazardous Air Pollutants, Prevention
of Significant Deterioration, and New Source Performance
Standards. It controls exposure through limiting emission
of toxic substances into the ambient air. -
o Occupational Safety and Health Act : The Occupational
Safety and Health Administration develops recommended
practices and prescribes permissible exposure limits
in the workplace. This Act has juria-diction over au-
workplace settings. It includes all routes of exposure
including inhalation, ingestion, and dermal absorption.
Controls range from prescriptive correction measures
to allowable limits of exposure.
o Food DruL and Cosmetic Act : This legislation controls
contamination of food through treatment, packaging and
preparation processes applied to food. VOCs are controlled
mainly as contaminants and packaging ingredients.
o Consumer Product Safety Act : This Act controls consumer
products and consumer formulation, but not production.
Its jurisdiction includes physical as well as chemical
safety of consumer products.
o Resource Conservation and Recovery Act. Comnrehensive
Environmental Resoonse Compensation and Liability Act
and the To,cic Substances Control Act : All three of
these Acts primarily concern tlie prevention of human
exposure through control of disposal, spills, and pro-
duction. RCRA’s mandate -is to control hazardous wastes
from their generation through ultimate disposal. CERCLA’s
mission is to clean up abandoned hazardous waste sites
that pose acute and chronic health risks. TSCA can
11—29
-------
be used to contrbl toxic substances in any context;
TSCA’S primary focus, however, is to prevent toxic substances
from entering the marketplace.
Of these eight pieces of legislation, the two most important
ones from the standpoint of controlling VOCs in drinking water
are RCRA and CERCLA. Drinking water standards and health advisories
may serve as the basis for clean—up goals under RCRA and CERCLA.
As noted in an earlier section, some communities are pursuing
Superfund monies as a funding source for cleaning up contaminated
drinking water sources.
11—30
-------
3. REGULATORY AND NON—REGULATORY ALTERNATIVES
OVERVIEW
The major alternatives to be considered for limiting human
exposure to VOCs in drinking water were discussed briefly in
the Proposed Rulemaking notice of June 12, i98 (Z49 FR 243 4U.
These alternatives are listed below:
1. No Federal Regulations. Provision of health advisories
for State action as appropriate.
2. Set federal monitoring regulations and provide health
advisories for State action as appropriate.
3. Set Primary Drinking Water Regulations for certain
of the VOCs.
There are several variations that can be incorporated into
each of these alternatives. In establishing primary drinking
water regulations for instance, EPA could choose to set a technology
based standard or develop MCLs based on feasibility and costs.
The authority for establishment of recommended rnaximum
contaminant levels (RMCLs) and maximum contaminant levels (MCLs)
is the Safe Drinking Water Act (SDWA). The SDWA (42 U.S.C. 300f
sea.) requires the EPA to establish RMCLs for “each contaminant
which, in (the Administrator’s] judgment may have any adverse
effect on the health of persons.” Section 1412(b)(2) requires
the establishment of an MCL for each contaminant for which an
RMCL is established. Further, the MCLs must be as close to
the RMCLs as feasible, taking cost into consideration.
.In addition to the regulatory mandates, the SDWA provides
authorities for ensuring the safety of the •nation’s drinking
water in a non—regulatory context. Section 114L 2(a)(2)(B) authorizes
EPA to provide technical assistance to states and publicly owned.
water systems in response to and alleviation of any emergency
situation which the Administrator determines to be a substantial
danger to public health. In the absence of appropriate State
or local action, Section 1431 authorizes EPA to take such actions
as the administrator deems necessary to protect public health
form a contaminant that may present an imminent and substantial
endangerment to the health of persons.
Each of the three major alternatives for controlling VOCs
.in drinking water is discussed in the following sections.
3.1 Alternative No. 1: No Federal Regulations
States and local utilities have been addressing site—specific
VOC contamination problems since the late 1970s in the absence
of federal regulations. EPA has assisted the states in addressing
hI—i
-------
this problem by issuing health advisories and providing advice
on treatment techniques and analytical methods.
EPA has issued health advisories* on 20 different chemical
contaminants to date. Anecdotal evidence indicates that state
officials value the health advisories and depend heavily on
them when making decisions regarding VOC control. When the
first health advisories on certain VOCs were issued in the form
of SNARLS (Suggested No Adverse Response Levels), states interpreted
and applied them in different ways. Some states applied the
health advisories as if they were standards while some considered
adopting them as state standards.
More recent experience has shown that several states are
moving to establish their own standards in the absence of EPA
regulations. Florida already has standards in effect while
New Jersey, California, Washington, and Wisconsin are taking
actions to set standards. Maine has a monitoring regulation.
In the absence àf federal regulations, states will either
design control, strategies to address incidents of contamination
on a case—by—case basis or establish statewide standards for
VOCs that occur frequently. Either of these approaches is likely
to be inefficient for four reasons. First, state drinking water
programs tend to be uneven in quality; therefore, some states
may tend to overregulate while others may not regulate at all.
Second, many state programs are faced with increasing constraints
on resources; this lack of resources could lead to no action
in some states. Third, if each state moves to adopt regulations,
it will be a lengthy process. State agencies will have to be
educated and convinced, a process that could take years. Fourth,
some states cannot establish drinking water regulations for
contaminants not regulated by the federal government (e.g.,
Colorado).
* Health advisories are developed for substances not regulated
under the SDWA to provide scientific guidance to federal,
state, and local officials concerning the health effects
of substances that are detected in drinking water supplies.
These advisories may be used to assist in determining appropriate
“immediate actions.” Immediate actions can be defined as
actions in the interim between discovery of contamination
and a complete regulatory decision (‘48 FR ‘45507). Health
advisories for rioncarcinogens suggest levels of contamination
at which no adverse health effects would occur, including
a safety margin for sensitive populations, such as pregnant
women. Health advisories for carcinogens provide risk estimates
but do not recommend levels. Since most of the nine VOCs
considered here are suspected carcinogens, health advisories
have reported each substance’s cancer risk.
111—2
-------
These factors, all of which contribute inefficiency to
the process of limiting human exposure to VOCs, provide strong
justification for federal standards for these chemicals.
3.2 Alt rnativ No. 2 Set Federal Monitoring Refulations
and Provide Health Advisories as Aooropriate
This alternative would result in all public water systems
determining if they have VOCs in their drinking water. Section
1445 of the Safe Drinking Water Act (SDWA) authorizes EPA to
require recordkeeping, reporting, monitoring, and any other
information to 1) assist in compliance with the SDWA, 2) evaluate
health risks of unregulated contaminants, and 3) advise the
public of such healthrisks. This alternative is based on the
premise that states and utilities could more efficiently deal
with VOC contamination incidents if better information on occurrence
and concentrations of contaminants were available to them and
to the public. Since health advisories have been issued, the
information generated by monitoring would enhance public knowledge.
The results of no EPA action other than health advisories
would thus be modified by more systematic monitoring for VOCs
and possibly by greater public concern in response to VOC contam-
ination. States would still be free to set their own standards,
but theoretically they would be working with better information.
There is evidence, gained from EPA studies, which indicates
that several states already have completed, or are in the process
of completing, monitoring surveys of their c.ommunity.-water systems.
There are several disadvantages to the “monitoring only”
alternative:
1. It provides no guidance regarding safe levels of contam-
ination. The states would have to depend upon the
- EPA health advisories and apply their own judgment
on a case—by—case basis.
2. A monitoring requirement would not have the impact
of an MCL; utilities would not be forced to remove
a contaminant once it was determined to exist.
3. Some states may elect to take no action to establish
regulations.
4. Better information, the major benefit of a monitoring
only alternative, is only one of several “market imper-
fections” that exist. Government intervention still
would be justified by other imperfections.
Under this alternative, the public would be better informed,
thus placing them in a position to exert pressure on utilities
to remove VOCs from drinking water supplies. Anecdotal evidence
acquired by EPA indicates that public action has indeed resulted
in utility actions to remove VOCs in a number of communities.
111—3
-------
3.3 Alternative No. : Set Primary Drinking Water Re gulations
for Certain of the VOCs
This alternative would involve establishment of RMCLs,
MCLs, and monitoring and reporting requirements for a selected
number of VOCs. Section 11112 of the SDWA authorizes EPA to
set maximum contaminant levels (MCLs) for those contaminants
in drinking water that may have “any adverse effect on the health
of persons.” Recommended Maximum Contaminant Levels (RMCLs)
are set at a level at which “rio known or anticipated adverse
effects on the health of persons occur nd which allows an adequate
margin of safety.” MCLs are to be set as close to RMCLs as
is feasible, taking costs into consideration (Section 11112 (b)
(3) of the SDWA].
RMCLs are health goals and are non—enforceable; MCLs, by
contrast, are enforceable standards that also require regular
monitoring and reporting by affected utilities. Under the statute,
the RMCLs and MCLs are to be established in separate and consecutive
rulemaking actions Ci.e., RMCLS are proposed; RMCLs are then
promulgated at the same time the MCLs are proposed; MCLs are
then promulgated). If an RMCL is promulgated for a particular
contaminant, an MCL also must be established.
If “it is not economically or technologically feasible
to ascertain the level of a contaminant in drinking water,”
a treatment technique requirement is to be established (in lieu
of MCLs) (Section 11101). The nine VOCs being considered for
regulation are detectable using conventional analytical techniques.
Thus, EPA is not considering a treatment—based requirement for
these chemicals.
Proposed RMCL.s were published in the Federal Register on
June 12, 19811 (149 FR 2L3L 8). RMCLS of zero were proposed for
seven chemicals: trichloroéthylene, tetrachioroethylene, carbon
tetrachioride, 1,2 —dichloroethane, vinyl chloride, benzene,
and 1, 1 —dichloroethylefle. RMCLS also were established for 1,1, 1 —tn—
chioroethylene (0.2 mg/i) and p —dichlorobenZene (0.75 mg/i)..
The approach used by EPA to establish these proposed RMCLs was
to select zero levels for potential or known carcinogens. Proposed
RMCLs for the two non—carcinogens were derived by calculating
an AADI (Adjusted Acceptable Daily Intake) level and assuming
a proportional exposure contribution from drinking water.
The basis for proposing and setting MCLs for drinking water
contaminants traditionally has been feasibility of removal,
taking costs into consideration. Application of certain treatment
technologies (such as GAC), without consideration of costs,
can generally remove chemicals to or below the limit of detection.
Such low levels may result in extremely low risks to the general
population and high cost per theoretical benefit. These risks
may be lower than those on which EPA has acted in the past.
Thus, a generally acceptable risk approach and the traditional
feasibility and cost approaches represent alternative regulatory
111—4
-------
mechanisms. Adoption of a generally acceptable risk approach
would require changes in the Safe Drinking Water Act and is
not within the limits of current authority. Such an approach
is nonetheless considered in various sections of this economic
impact analysis because Executive Order 12291 instructs that
options outside the limits of current authority should not be
excluded from analysis.
Benefits and costs for MCLs equivalent to 1C and
individual lifetime risk levels have been calculated. Benefit
calculations and results are presented in Chapter 14 and costs
are discussed in detail in Chapter 5. Results indicate that
these MCL alternatives are essentially equivalent to MCL alternatives
that were being considered by EPA on the basis of feasibility
and cost.
A primary advantage of setting MCLs for VOCs is that it
would provide consistent, national controls for these contaminants.
In the absence of federal regulations, the control of-volatile
organic chemicals will provide uneven protection of the public
health due to the inefficiency of the state—by—state, case—by—case
approach.
Another direct benefit of federal regulations of VOCs is
that these MCLs can be used as guidance by state and federal
officials grappling with the question of “how clean is clean.”
Officials responsible for cleanup of waste sites under CERCLA
(i.e., Superfund) commonly rely on drinking waber standards
as the basis for aquifer restoratian and protection.
The majority of state officials want federal standards
for VOCs. The kinds of activities undertaken by the EPA in
setting standards —— extensive evaluation of analytical methods,
exposure assessments, research and demonstration of .feasibility
of treatment technologies, and cost assessments —— is clearly
beyond the capabilities possessed by many states.
State officials view this set of activities as the proper
role for the EPA in the federal—state partnership.
3 .L Other Authorities to Protect Drinkinu Water
Section H 142 of SDWA authorizes EPA to provide technical
assistance to states and publicly owned water systems in response
to and alleviation of any emergency situation which the Administrator
determines to be a substantial danger to public health. In
the absence of appropriate state or local action, Section 1L 31
authorizes EPA to take such actions as the Administrator deems
r ecessary to protect public health from a contaminant that may
present an imminent and substantial endangerment to the health
of persons.
The court system also provides an institutionalized means
to achieve removal of VOCs from drinking water. Some water
“I—s
-------
systems, such as the one serving Acton, Massachusetts, have
filed lawsuits against firms believed to be responsible for
contamination of drinking water supplies as a result of improper
disposal of waste materials. Citizen suits against polluters
and public water systems appear to be a developing trend as
public concern about chemicals in drinking water increases.
111—6
-------
14 ASSESSMENT OF BENEFITS
4.1 Introduction
Assessment of the benefits of regulating exposure to potentially
carcinogenic substances may be undertaken via a straightforward
calculation of the number of cases of cancer that are likely
to be produced in the U.S. population at varying levels of exposure.
This calculation requires knowledge of the following: 1) the
extent of exposure to the substance in the absence of regulation;
2) the degree of reduction in exposure that will be produced
by different regulatory strategies; and 3) the probability,
or risk, of getting cancer from a given amount of exposure.
This chapter presents such an analysis for proposed primary
drinking water standaras for volatile organic chemicals (VOCs).
4.2 Baseline Exoosure and Cancer Risk
As discussed in Chapter 2, the Office of Drinking Water
has conducted a number of nationwwide studies to determine the
nature and extent of drinking water contamination with VOCs.
Various summaries of this data are presented in Chapter 2.
As mentioned in the NOTE on page 2—1, there were originally
nine chemicals that were the subject of this document. The-
background data in chapter 2 covers nine chemicals. Two were
subsequently determined to be non—carcinogens. One of these
detern inations was made too late for the chemical to be deleted
from the analysis supporting chapters 4 and 5. Eight chemicals
are therefore covered in these chapters. This does not have
a major effect on the results (see Exhibit 4—3).
The exposure data for all of the chemicals under examination
display a similar pattern. Generally speaking, there are many
water systems having low concentrations of these contamInants
and relatively few water systems having high concentrations.
Accordingly, the cumulative distribution of the number of con—
tarninated water systems with respect to concentration, takes
the form of a log—log curve. The data were fit to log—log equations
to facilitate analysis of regulatory options across the entire
range of concentrations. Equations were developed for each
of twelve size classes of both ground and surface water systems.
These equations were used to estimate the total number
of oeoole exposed above a given concentration. If, for example,
a maximum contaminant level (MCL) is set at 20 ugh, the number
of water systems in each size/source category having concentrations
equal to or greater than 20 ugh can be determined from the
family of equations. Then, using the average population of
the water systems within each category it is possible to convert
from the number of water systems having VOC contamination above
a given level to the number of oeoole exposed to VOC concen-
trations above a given level. This produces a cumulative distri—
‘v—i
-------
bution of the number of people exposed, as illustrated in Exhibit
Li—i. These people are currently incurring a certain risk of
cancer as a result of this exposure. An MCL. established at
this level would remove or reduce this exposure, thus avoiding
an associated number of cancer cases per year. This is the
appropriate measure of the benefit of the regulatory alternatives.
The exposure data may be used to develop an estimate of
the baseline number of eases of cancer occurring annually from
exposure to VOC’s by multiplying the number of people exposed
times the average concentration at which they are exposed times
the annual risk of cancer per person per ugh of exposure to
each chemical. The formula is given in equation 1.
Eqi: #CASES #EXPOSED(i,j,k) x AVECONC(k) x RISK(i)
Where:
#EXPOSED(1,J,k) number of people in size/source category
(j) exposed to chemical (i) in concentration
range (k)
AVECONC(k) = the average concentration at which those
people are exposed to Cthe mid—point
of concentration range ki
RISK(i) ; the annual risk of cancer per person
per ugh of exposure to chemical (1) -
The risk factors, represented as RISK(i) in the above
expression, are those developed by the Carcinogen Assessment
Group (CAG). The CAG risk numbers for all of the chemicals
are presented in Exhibit 2—3. The numbers given show the risks
from both. ingestion and inhalation exposure routes. As these
risk numbers indicate, vinyl chloride is by far the most potent
of the VOCs.
Using the above equation, it is estimated that there are
a total of approximately fifty cancer cases. induced annually
by the combined effects of the eight VOCs. Of these fifty eases,
thirty—seven are attributable to vinyl chloride.
The baseline estimate of fifty cases per year may represent
the low end of the range due to analytical limitations in the
measurement of VOCs. The analytical limit of detection in the
surveys performed by the Office of Drinking Water was generally
0.5 ug/l. Given the shape of the cumulative distribution of
occurrence for VOCs, exemplified by the curve for carbon
tetrachioride in Exhibit Zi...i, it is conceivable that there could
be a significant amount of occurrence at concentrations below
the analytical limit of detection (0.5 ugh for most VOCs, but
1.0 ug/l for vinyl chloride). The estimate of fifty cases per
year is based only on the occurrence above the analytical limit
of detection.
IV—2
-------
EXHIBIT 4—1
EXAMPLE CUMULATIVE DISTRIBUTION OF
THE NUMBER OF PEOPLE EXPOSED TO VOCs
c:.ARE;OiJ TETRACHL.O RIDE. EXPOSURE..
G UNDWAT S? TEMS 0. — 1.0 MIL POP.
1 7
1 . -
I . -
J’] 1.4-
II
z
c 1
± 1.1-
I-
1-
U i i .j .9
:::i - III
:.
! ‘:. -
_1
0.4
a
0.2
Ci. I
i:-
2 4 6
CONCT T ON ( /l)
IV—3
-------
1 1.3 Methodology for Evaluation of Multiple Occurrence
Calculation of the baseline number of cancer cases by the
procedure described in the preceding section is dependent upon
the underlying assumption that the risks posed when two or more
chemicals are present are additive. As indicated in the equation
shown above, the number of cases Is calculated separately for
each chemical Ci) and then the totals are added. This assumes
that there are no synergistic or antagonistic effects. As explained
in the preamble to the final RMCL regulation, EPA is unable
to set a total or multiple contaminant RMCL. These contaminants
may be synergistic, antagonistic, or additive in their effects
and data on multiple VOC effects are not available. However,
EPA believes that absent conclusive scientific data, it is necessary
and protective to apply the individual MCLs to multiple exposure
situations.
Given the assumption of additive risk, the above—described
procedure provides a good estimate of the baseline .number of
cases of cancer prodpced in the aggregate. The procedure must
be modified for purposes of estimating the benefits of alternative
regulatory strategies, however, because it cannot provide information
about the benefits of alternatives such as regulating the eight
chemicals individually or of regulating on the basis. of risk
(not permitted under SDWA, but evaluated under E.0. 12291).
For example, 87 percent of the time, trichioroethylene (TCa)
is found in the presence of other VOCs. Because the treatment
technologies are the same for all VOCs, a regulation affecting
TCE will produce benefits greater than those indicated by the
removal of TCE; some amount of the other VOCs present in the
multiple occurrence cases also will be removed.
The multiple occurrence case is important since almost
50 percent of all VOC occurrence is joint occurrence. Moreover,
vinyl chloride, the most potent VOC, appears to occur only in
the presence of other VOCs. To permit analysis of the benefits
of regulating individual VOCs and of the effect of regulatory
alternatives on the multiple occurrence case, it is. necessary
to evaluate the benefits of single and multiple occurrences
separately from one another.
Data on the occurrence of VOCs is inadequate to fully specify
the joint probability distributions of the eight chemicals.
It is not possible to know with certainty which chemicals are
most likely to occur together and in what concentrations they
are likely to be present. All that is known from the available
data is the percent of the time that a given chemical Is likely
to occur singly and the percent of the time It is likely to
occur jointly with others. Given this fundamental piece of
information, it is possible to restructure the formula presented
earlier as follows:
IV-4
-------
Eq2: #CASES
#EXPOSED(i,j,k) x %SINGLE(i) x AVECONC(k) x RISK(i)
ij k
+ #EXPOSED(m,j,k) x AVECONC(k) x RISK(m)
ij k
Where:
%SINGLE(i) the percentage of the time that chemical
Ci) occurs by itself, with no other VOCs,
present
#EXPOSED(m,j,k) = the number of people in size/source category
(j) exposed to “typical mixture” (m)
in concentration range (k)
RISK(m) = the annual risk of cancer per person
per ugh of “typical mixture” Cm)
The above formula is based on the concept of defining a
number of “typical mixtures,” denoted by the index (m), to represent
the multiple occurrence case. This approach requires two sets
of assumptions:
(1) The extent of the occurrence of these “typical mixtures”
must be defined in terms of the number of people. exposed. -
and the concentrations to which they are exposed .
This is reflected in the variable, OEXPOSED(m,j,k).
(2) The degree of cancer risk represented by the constituents
of each of the “typical mixtures” must be defined,
reflected by the variable RISK(m).
As noted earlier, trichloroethYlefle (TCE) occurs 87 percent
of the time in the presence of other VOCs. As shown in the•
table in Exhibit Z _2, this is the largest proportion of multiple
occurrence of a!]. the VOCs. In addition, TCE is the most commonly
occurring VOC. Thus it is appropriate to use the occurrence
distribution of TCE as a stand—in for the occurrence of the
multiple contaminant case. More formally, this assumption is
as follows:
0.87 x #EXPOSED(TCE,j,k) #EXPOSED(m, ,k)
This yields an occurrence distribution which is given in terms
of ugh of TCE instead of ugh of Total VOCs (TVOC) This dis-
crepancy is compensated for by defin.ing the RISK(m) variable
as the cancer risk of the “typical mixture” per ugh of TC E
instead of per u /1 of TVOC , simply a change in units.
There are two conceivable methods for assigning a value
to the variable RISK(m):
IV—5
-------
EXHIBIT 4—2
PROPORTIONS OF SINGLE (VS. MULTIPLE) OCCURRENCE
FOR INDIVIDUAL VOCs
Chemi. cal Percent Single
Tr-ichlcr ethylene
Tetr chl r ethvlene
Carbon Tetrachicride S7
1,1, 1—Trichioroethane
1. ,2—Dichlcroethane 20
Ben:ene 60
1., .—Dichloroethylene 0
Vinyl Chloride 0
IV—6
-------
(1) Bottom—up Approach —— Using the available occurrence
data, “typical mixtures” can be constructed on the
basis of judgment; specifying the constituents and
concentrations which appear typical.
(2) Top—down Approach —— Using the total number of cancer
cases per year estimated by the aggregate approach
in Equation 1 (50 cases) as a control total, an amount
of risk can be assigned to RISK(m) which will be just
sufficient to cause Equation 2 to produce the same
result; this is the implied amount of risk that should
c.haracterize the “average typical mixture.”
The top—down approach is clearly the preferred choice of
the two because it is possible to “calibrate” to the result
produced by the aggregate method. The re.sults of this calibration
step are summarized in Exhibit 4—3 which compares the risk factors
for the individual VOCs to the risk factor derived for the “average”
“typical mixture” by this procedure.
The data presented in Exhibit 4—3 allows a check on the
realism of the top—down results. The “typical mixture” displays
more risk than many of the individual constituents. The calib—
ration result might be interpreted as representing the mid—point
between two extremes; low risk mixtures featuring chemicals
such as trichloroethylene, tetrachioroethylene, and 1,1,1—tn—
chioroethane; and, high risk mixtures featuring chemicals such
as vinyl chloride and carbon tetrachloride. On the other hand,
there is growing evidence that vinyl chloride may be a by—product
of biological degradation of the other VOCs. If this hypothesis
is correct, small concentrations of vinyl chloride may be expected
in many cases of VOC occurrence (when likely parent compounds
are present), which would also account for the calibration result.
By contrast, a bottom—up approach to assessing the multiple-
occurrence case was selected as a basis for treatment cost estimates
used in Chapter 5. On the cost side, however, the problem is
somewhat different since it is assumed that the same treatment
techniques provide some level of removal for all VOCs. Thus,
all that was necessary was a set of assumptions representative
of the type of equipment that would have to be installed. Cost
estimates for the multiple contaminant case were based on the
assumption that carbon usage rates are additive for granular
activated carbon and on the most difficult constituent to remove
for aeration treatments.
14•4 Analysis of Re ulatorv Alternatives
Regulation of contaminants by the Office of Drinking Water
takes the form of Maximum Contaminant Levels (MCLs) which stipulate
the level to which contaminants must be removed. The methodology
described in the preceding two sections applies to calculation
of the baseline number of cases of cancer that will result in
the absence of regulation. This methodology must be modified
IV—7
-------
EXHIBIT 4—3
ANNUAL RISK OF CANCER FACTORS FOR INDIVIDUAL VOCs
AND THE MULTIPLE CONTAMINANT CASE
Tn chl oroethy 1 ene
Tetrach 1 oroethyl ene
Carbon Tetrachioride
1,1., l —Trichloroethane**
1 , —Dicrt1oroethane
Ben:ene
1, 1—Dichroroethylene
Vinyl Chloride
Multiple Contaminant Case
* Annual risk: of cancer per person per ugh
(All figures are rounded.)
*- lO 8
* 1.O —B
* lO ’—7
* lO -9
* jO —8
* 1’) —8
* 1.O —7•
* j •’—
7.7 * tO ’—7
** Determined to be a non—carcinogen during the course of
preparation of this document. It was included in the analysis
of the total number of cases of cancer avoided. As evident from
the risk. factor, it does not have a major effect on the results.
Chemi . cal
Risk Factor*
l.a
—S
1.1.
!. 7
4.:
I
S —
1.9
IV-8
-------
to calculate the number of cases that will be avoided as a result
of alternative regulatory actions. This is accomplished by
replacing the variable AVECONC(k) with the expression:
(AVECONC(k)—MCL(i))
This substitution causes Equation 2 to produce an estimate
of the cases avoided per year as a result of alternative MCLs.
The result produced by this procedure reflects a strict inter-
pretation of the effect an MCI.. will have on the water Industry;
that is, it assumes that water systems will only install the
minimum amount of treatment that is required of them. In reality,
many water systems may opt to treat to levels below the MCI..
in an attempt to completely remove organic contaminants. This
pattern has been followed in a number of cases of VOC contamination.
(It is noted however that MCLs could not be effectively monitored
and enforced at these levels due to the limitations of analytical
methods of detection.) Non—treatment alternatives such as develop-
ment of alternative water sources also are popular solutions.
Public attitudes regarding this form of contamination can influence
the treatment decision in the direction of more thorough removal.
To accommodate this uncertainty in treatment selection,
the results of the analysis are presented as the high and Low
ends of a range. The high estimate is the result produced when
the- above—described substitution in Equation 2 is not made,
reflect±ng the assumption that all water systems affected by
the MCLs would attempt to treat to levels below the limit of
detection in the best available research labs. The low estimates
is the result produced when the above—described substitution
j made in Equation 2, implying an assumption of minimum compliance.
where all water systems treat only to the level of the MCL.
Exhibit U—4 presents estimates of the total number of cancer
cases that would be avoided at a national level as a result
of alternative MCLs. MCLs evaluated range from 0.5 ugh (the
limit of detection) to 100 ugh. For simplicity, the totals
represent the case in which the MCLs are set at the same level
for all eight VOCs. This need not always be the case, however,
and results also are presented for “risk—based” MCLs where the
MCLs differ for each contaminant but each is controlled to the
same level of risk. “Risk—based” MCLs are evaluated t represen
regulatory strategies designed to obtain-uniform 1O and 10
risk levels for all contaminants. (It is noted that, under the
Safe Drinking Water Act, a risk—based approach to setting MCL.s
is not permitted. However, Executive Order 12291 requires considera-
tion of regulatory alternatives that are out ide curre t legislative
authority. Thus, MCLs equivalent to 1O and 10 ’ risk levels
were studied as part of the development of this Economic Impact
Analysis.)
The results in Exhibit Z_Z show a total of 49 cases of
cancer per year that would be avoided under an MCL of 0.5 ugh.
This differs slightly from the control total of 50 cases obtained
with Equation 1 due to rounding errors inherent in the calibration
procedures described above.
IV—9
-------
EXHIBIT 4—4
TOTAL NUMBER OF CANCER CASES AVOIDED PER YEAR
UNDER ALTERNATIVE MCLs
MCL No. Cancer Cases Avoided Per Year
High Estimate Low Estimate
42
2. 2á
O u /1 49 49
42 :8
1 -‘ • i
t.
29 to.
2 .O 29 t3
to 4 .-
to(: . t: 1
‘v—b
-------
Since 0.5 ugh is the analytical limit of detection (for
most VOCs), the answer of 9 cases is the same for both the
high and low estimates. This results from the engineering reality
that designing for the limit of detection is the technical equivalent
of designing for zero. The practical effect of detection limits
will carry over to other very low MCLs as well. At an MCL of
1.0 ugh, for example, it is likely that many treatment systems
would be designed the same way if the MCL were zero. Hence,
the low estimate of cancer cases avoided at an MCL of 1.0 ug/l
should probably be ignored in favor of the high estimate. For
MCLs of 5.0 ugh and ‘below, there will be a tendency to design
for zero to compensate for variability in influerit concentration.
As a result, the high estimates of the number of cancer cases
avoided are closer to the correct figure.
Exhibit L _ 4 also p esents r sults for “risk—based” MCLs
intended to achieve 10 and 1O risk level . The io6 MCLs
are in the vicinity of 1.0 ugh and the 10’ MCLs are in the
vicinity of 5.0 ug/l. It is therefore no surprise, that the
total number of cancer cases avoided by these two alternatives
are very similar to those achieved by MCLs of 1.0 and 5.0 ugh.
It may be concluded from this analysis that MCLs of 1.0 and
5.0 u%/l imply roughly the same benefits as “risk—based” MCLs
of 10 and 10 ’ , respectively.
Exhibit LI. 5 presents a breakdown of the results by four
different system’ size categories. The table shows that most
of the benefits reside in the larger. system size categor.ies.
This is a product of’ the simple fact that there are many more
people exposed to cancer risk as a result of’ VOC contamination
in larger size systems.
Similar analysis of costs by system size categories presented
in Chapter 5 shows that the smaller’ system size categories bear
a disproportionately larger share of the total costs. There ’ore,
an alternative regulatory strategy that would provide a varian ” e
to systems under 10,000 population also was studied. (Such variances
are not permitted on the basis of cost or system size under
the SDWA. However, the alternative was studied in complying
with Executive Order 12291.) The results show that the effect
of such a variance would be to reduce the total national benefit
from L19 to Lt2 cancer cases avoided per year at an MCL of 0.5
ugh. The corresponding reduction in total annual. costs would
be approximately one—third.
A final alternative approach considered was that of having
a separate TVOC (Total VOCs) MCL to assure adequate treatment
in the multiple contaminant cases. It is conceivable that a
mixture of VOCs could be treated to the point where the MCLs
of the individual constituents are satisfied but the additive
effect of the residuals still amounts to a significant exposure.
A separate MCL for TVOC concentration would be one approach
to dealing with this problem (assuming the implied monitoring
requirement would be manageable).
‘v—li
-------
EXHIBIT 4—S
TOTAL NUMBER OF CANCER CASES
AVOIDED PER YEAR BY SYSTEM SIZE
25-100 50t—3301 01—5OK SOK+
High Low High Low High Low High Low
MCL Est. Est. Est. Est. Est. Est. Est Est.
1 1 2 20 16 to
1 o 1 14 11 16
0.5 ugh 1 1 2 22. 22:
1.0 1 1 2 20 19. 19 tOi
5.0 1 1 2 .1 14 12. 16
10.0 1 0 2 1 1 10 15 LZ
20.0 1 0 1 1 12 7 15
25.0 C’ 1 1 12 6 15 7
50. 0 0 0 1 0 1 L
1 00. o o 0 0 0 0 0 0 0
IV—12
-------
This concept was analyzed; it was found to not have a sig-
nificant effect on the total benefits. The reason stems. from
the technical nature of the treatment technologies employed
to remove VOCs. For the most part, both aeration and activated
carbon approaches will remove all VOCs present in a mixture
to low enough levels that the residual does not represent a
significant exposure level. -
1.5 Uncertainties in Benefits Assessment
There are many sources of uncertainty in judging the health
effects of specific chemicals and in estimating the risks due
to these chemicals’ presence in environmental media. Ideally,
one would like to separate the various sources of uncertainty
and then estimate the magnitude of each using statistical methods.
Typically, however, this requires knowledge of the frequency
distribution of the relevant variables, and this knowledge is
usually sparse. It is generally easier to estimate uncertainties
in risk using results based on experimental experience, or obtained
by consulting experts —— the so—called “Delphi” method. While
each of these approach&s has its drawbacks, the lack of statistical
data often gives scientists and policy—makers few other alterna—
tives. The discussion below attempts to identify the factors
that contribute to the overall uncertainty and to describe what
is known about the likely magnitude of each factor.
A widely used technique for estimating the risk corresponding
to a particular dose level involves the formulation of a dose—
response model. Models such as the probit, one—hit and multistage-
models have been investigated in detail over the past decade,
and these and other models have been commonly postulated as
useful approximations for estimating risk due to a particular
dose level. A problem arises, however, in using these models
to extrapolate from the results obtained with laboratory animals
to the low doses typical of humar , exposures (i.e., those producing
lifetime risks in the range of 10’ to 10 per person). Statistical.
procedures such as those described above are only valid within
the range of the original experimental data; when the procedures
are used outside this range, extrapolation errors result. Further-
more, the size and direction of these errors vary with the choice
of model used; therefore, the estimation of human risk can vary
widely with the choice of extrapolation model. At low doses,
the differences among predictions range over as much as five
orders of magnitude (100,000 times).
In addition to the problems posed by the choice of extrapolation
model, other factors introduce uncertainty into the estimation
of’ risk and health impacts. These include:
—. Experimental error . Risk factors for animals are obtained
from bioassay experiments. Inappropriate choices of
test speciesor protocols, or improperlaboratory procedures,
can lead to inaccurate estimates of risk factors. Even
where these problems are absent, the application of
IV—13
-------
statistical analysis to a sample of animals yields finite
confidence intervals around the point—estimate result.
— Unit exDosure . Calculations of human exposure generally
assume that 2 liters of water are consumed per day.
In the case of VOCs, it has been further assumed that
an amount of the contaminant equivalent to that found
in two liters of water is inhaled daily by each person,
as a result of aeration from shower water, etc. The
bases for these assumptions have not been documented
as fully as possible, and it is to be expected that
they are another source of possible error.
o Numbers of peoDle e oosed . The distribution of population
by concentration of VOC is derived from surveys of limited
samples of water systems. The 95 percent confidence
interval in the number of people exposed to a given
concentration level typically ranges from j 25 percent
of the mean at low concentrations (where sample sizes
are relatively large) to greater than ± 100 percent
of the mean at high contaminant levels (for which only
small sample sizes are available; see EPA, “Occurrence
[ of VOCs] in Drinking Water, Food, and Air,” 1983).
In addition, the estimates of the proportionof people
exposed to a given VOC alone versus those exposed. to
the same VOC in corijunctthn with other VOCs is known
only from limited surveys of groundwater systems.
Exhibit L _6 lists some of the major contributors to uncertainty
in estimating risks and health impacts, and gives estimates
of the magnitude of the resulting uncertainty to the extent
it is possible to do so.
Overall, it has been estimated that, x 1udin the five—
order—of—magnitude uncertainty in the extrapolation model, the
remaining uncertainty in any particular risk estimate may be
as high as two or three orders of magnitude (see tineertaintv
in the Re ulatorv Dec ision—Makin Process , U.S. EPA Office of
Drinking Water, September 28, 198 4).
In order to account for these sources of uncertainty in
the estimates of benefits accruing to VOC regulations, the fo1l 4ing
assumptions were made:
o Uncertainty due to the choice of a risk extrapolation
model were ignored. The Agency has standardized on
a linealized multi—stage model. Therefore, the risk
estimates for VOCs will at least be consistent with
those for other substances and other media.
o The remaining uncertainty in the unit risk estimates
derived by the Carcinogen Assessment Group was assumed
to have a standard error of either 20 percent, 50 percent,
or 80 percent of the value given. (The CAG figures
IV—14
-------
EXHIBIT 4—6
SELECTED FACTORS CONTRIBUTING TO UNCERTAINTY IN RISK ESTIMATION
(Adapted from Uncertainty in the Re ulatorv Decision —Mak1n
Process , U.S. EPA Office of Drinking Water September 28, 198’s)
Factor
Possible Contribution to Uncertainty
1. Diet of test animals
2. Laboratory procedure
3. Decision criteria re
carcinogenicity
factor of 2
— 2 orders of magnitude
+ 2 orders of magnitude
‘4. Synergism/antagonism
among substances
Unknown
5. Number of animals used
and distribution by
dosage
6. Selection of experimental
dose levels
7. Choice of extrapolation
model
8. Statistical noise
9. Other:
Ability of experimental
personnel
Choice of species, sex,
age and strain of test
animals
Diseases in test animals
Lack of corresponding
human tissues
Inappropriate statistical
methodology
Choice of significance
level
± 2 orders of magnitude
± 2 orders of magnitude
Possibly j 5 orders of magnitude;
see text
factor of 2
Errors in any of these
factors can cause a given
experiment to be invalid
in whole or in part
A more detailed breakdown is given in Techniques for the Assessment
of Carcino enie Risk to the U.S. Population Due to Exposure
from Selected Volatile Organic Compounds from Drinking Water
Via the Ingestion. Inhalation and Dermal Routes ; Cothern, Coniglio,
and Marcus; EPA 570/9—85—001.
IV—15
-------
themselves are conservative, representing the upper
95 percent confidence limit of the experimental- results
extrapolated to humans. The size of the confidence
limit itself was not available.)
o The standard errors in the occurrencedata were derived
by fitting log—log regression curves to the distributions
of population by concentration, as obtained from the
reference cited above. Separate regressions were estimated
for each VOC and for each water source (surface and
ground). The standard errors of the regressions were
a few percent of the point estimates at low concentrations,
ranging up to one or two hundred percent at high concen—
trations.
o For each VOC, the standard error in the proportion of
exposures represented by single occurrences was estimated
directly from the Ground Water System Survey data.
The standard error was always less than one percent
of the pr.oportion.
For any VOC occurring singly, the expected number of cancer
cases in the population within a given concentration interval
is derived by multiplying the exposed population by the unit
risk factor and again by the mean concentration in the interval
and the proportion of single occurrences. (For the multiple
occurrence case, the last term is replaced by one minus the
single—occurrence proportion for TCE.) Because the terms are.
multiplicative, the standard errors described above can be combined
using the following expression:
(S 2 + ? 2 ) (Sr 2 + R 2 )’(Sc’ + C 2 )’(Sf 2 + F 2 )
—
where:
S = standard error of the estimate of cancer cases in
the exposure interval i
P mean population exposed in the interval
R mean point estimate of the unit risk
C average concentration in the interval
F proportion of all exposures that are single exposures.
The resulting standard errors were combined (by taking the square
root di the sum of their squares) over all exposure intervals
and VOCs to obtain the standard error in the total number of
cancer cases estimated to occur from exposure to the VOCs in
IV— 16
-------
question. This total standard error was approximately 10 percent
of the point estimate of the number of cancer cases. This is
equivalent to the standard error in the number of cases avoided
if the MCL were to be set to zero for all VOCs.
The standard errors in cases avoided for specific MCLs
other than zero were not explicitly computed. For any MCL below
20 ugh, however, these standard errors would be about the same
proportion of the mean as in the case of a zero MCL.. This is
because the vast majority of the total uncertainty is attrib-
utable to exposures higher than 20 ugh, which would be eliminated
in either case. Total uncertainty was found to be dominated
by the errors in the exposure data. The standard errors in
the various estimates of unit cancer risk, which were allowed
to vary between 20 percent and 80 percent of the mean, contributed
negligibly to the standard error in the total number of cases
avoided.
IV—17
-------
Influent Levels Effluent Levels
(ugh) (ug/].)
0.5 0.5
5.0 5.0
20.0 20.0
50.0 50.0
100.0
The model was run separately for each VOC and effluent level,
generating, for each VOC—MCL combination, capital and 0&M costs
of compliance for each of the 24 industry segments. Each pair
of capital and 0&M costs was reduced to an equivalent annualized
cost, using an appropriate rate of interest (interest rates
are discussed further below). The annualization expression
used was:
A 1 F 1 x Ni—(1+i) )/i] + V 1
where:
A 1 = the annualized cost
j = an appropriate interest rate, and
n the useful life of the capital equipment.
These annualized costs were then summed for surface and groundwater
systems within each size category. Finally, a regression curve
was fitted to the annualized PTm results for each VOC, with
annualized cost as the dependent variable and MCL as the independent
variable. The resulting expressions enabled the analyst to
obtain an annualized cost of compliance for any arbitrary MCL.-
For each VOC, separate regressions were run for each of the
12 system size categories.
In all, 84 cost regressions were estimated (six individual
VOCs plus one multiple occurrence case, times 12 system size
categories). For half of these, a log—log curve was fitted
by the method of ordinary least squares. For the remainder,
a semi—log curve was used because it provided a better fit to
the PTm results. The algebraic forms of the two equations are
as follows:
Log—log: ln(Y) = a + b x ln(MCL)
Semi—Log: Y = a + b x ln(MCL).
In the vast majority of cases, the value of is in excess
of 0.98.
Note that this interpoiation procedure lumps capital and
0&M costs together into an annualized cost. Although PTm itself
produces separate estimates for capital and 0&M costs, it was
not possible to obtain statistically valid regression results
‘/—3
-------
for O&M costs alone from the PTm output. Therefore, the method
described above was chosen as the best way of incorporating
information about both types of costs in the national cost
estimates.
In addition to compliance costs, PTm generates estimates
of the numbers of systems requiring treatment to achieve a given
MCL for a particular VOC. Like the cost calculations, these
estimates are computed for a discrete set of influent and effluent
levels. Therefore, to generalize the results to any arbitrary
MCL, regression curves were fitted to the PTm number—of—systems
outputs in a fashion similar to that described above, with number
of systems as the dependent variable and MCL as the independent
variable. In order to obtain a good fit, however, it was necessary
to aggregate the PTm outputs from 12 size categories down to
four.
5.2.2 Inouts and Assumptions
The PTm model w s run for seven contaminant conditions.
Six of these represent specific VOCs as sin Ie occurrences;
the last represents all multiple occurrences. Single and multiple
occurrences had to be treated separately to avoid double—counting
the costs of removing VOCs that occur in association with others
that are also being removed. The single—occurrence cases were
as follows:
Trichioroethyl ene
Tetrachloroethylene
Carbon Tetrachioride
1,1, 1—Trichioroethane
Benz ene
1 ,2—Dichloroethane.
Vinyl chloride and 1,1—dichioroethylene, two of the VOCs being
regulated by this action, do not occur singly. Therefore, their
costs of removal were not specifically computed by PTm; instead,
they were subsumed into the “multiple occurrence” case. The
relative frequencies of occurrence of the eight VOCs of interest
as single and multiple occurrences were obtained for groundwater
systems from EPA ’s Ground Water Supply Survey; the same proportions
were used to extrapolate these occurrences to surface’ water
systems. The frequencies of occurrence have been summarized
in Exhibit 14 2 above.
For each of’ the single—occurrence VOCs, a distribution
of’ drinking water systems by raw water concentration was obtained
from the series of’ EPA reports entitled, Occurrence fof VOCsI
in Drinking Water. Food. and Air . Separate distributions were
obtained for each water source (surface and ground) and system
size category; they are summarized in Chapter 2. No data exist,
however, on the composition of’ multiple occurrences or their
frequency distribution by size category and influent level.
Therefore, for the multiple occurrence case, the distribution
V—4
-------
of contaminated systems was assumed to have the same shape as
the distribution of (single—occurrence) trichloroethylene (TCE).
TCE is one of the two most prevalent VOCs in drinking water,
and is found in association with other VOCs 87 percent of the
time. (The other substances in the multiple occurrence case
were assumed, on the basis of engineering judgment to be 1,1,1—
trichloroethane and tetrachioroethylene.) The distributions
thus derived were used to generate the frequency—of—occurrence
portion of the decision tree matrix, one of the inputs to PTm
(see Section 5.2.1 above).
The probabilities that systems of a given size and water
source would choose a particular treatment, given a specified
influent level and MCI.., were generated by convening a panel
of water supply engineers, water chemists, and economists familiar
with the decision making processes of the water supply industry.
The panel considered, for each VOC, system size, influent level
and MCL, the relative unit cost and effectiveness of each available
control measure. The panel participants also applied their
practical experience in estimating the frequency with which
each treatment would be chosen. Their probability estimates
were entered in the remaining portion of the PTm decision tree
matrix.
The costs of treatment per unit of water treated were estimated
for each treatment method. Treatment methods considered included.
the following:
Granular activated carbon
Packed tower aeration
Diffu.sed air aeration
Slat tray aeration
Powdered activated carbon
Welifield management (for groundwater
systems only)
Source protection (for surface water
systems only)
Regionalization
Use of an alternative source
Point of use treatment devices
These methods of treatment are discussed further in Chapter 2.
Also, complete technical analyses of these options may be found
in Technologies and Costs for the Removal of Volatile Organic
Chemicals for Potable Water Su olies , U.S. EPA, Office of’ Drinking
Water, December 13, 198’4.
Capital costs per unit of capacity and O&M costs per unit
of production were generated separately for each combination
of VOC, treatment method, influent level, and effluent level.
For the multiple occurrence case, the following contaminant
levels were assumed for the purpose of generating unit costs:
V—5
-------
Influent Concentration
Trichioroethylene Tetrachioroethylene 1,1,1—Trichioroethane
100 ugh 50 ugh 50 ugh
50 20 20
20 5 5
5 5 5
The results of the unit cost calculations were entered into
the treatment cost matrix of the PTm model. In that matrix,
“No Treatment” was also an allowable option, for those systems
not contaminated by VOCs or for those that the panel of experts
judged to be physically incapable of complying.
In addition to the cost and probability inputs just described,
PTm requires the user to specify detailed operating and financial
information on the water supply industry. These inputs, however,
do not change from run to run and are independent of the VOCs,
treatments, or costs being considered. The required data were
obtained from EPA’s Survey of 0oeratin and Financial Characteristics
of Community Water Systems (U.S. EPA, Office of Drinking Water,
1982).
To annualize the PTm outputs for purpo.ses of generating
regression equations, it was necessary to select an appropriate
interest rate. Two options existed. The first was the social.
interest rate, reflecting the opportunity cost of resources
assuming no risk for alternative investments. The second was
the market rate, reflecting the actual cost of capital to drinking
water systems, and therefore incorporating considerations of
risk and imperfect access to capital markets. The social rate
was• chosen for purposes of annualization, and the final annualized
costs of compliance were then adjusted appropriately to obtain
annualized market costs. The social interest rate was determined
from a one—year average rate for three—month treasury bills
(8.611 percent) adjusted for the same year’s inflation rate (11.2
percent) to yield a (rounded) value of 14.11 percent.. The market
rate of interest was obtained from current yields on municipal.
bonds and corporate bonds; adjusted for inflation, it was found
to be 7.5 percent for small systems (those serving fewer than
50,000 people) and 6.2 percent for large systems.
Also, for the annualization procedure, a useful lifetime
of 20 years was selected. This was believed to be appropriate
for the equipment used by drinking water systems.
5.2.3 Results
Exhibit 5—1 shows the national cost of meeting VOC maximum
contaminant levels using market—based interest rates. The estimates
V—6
-------
EXHIBIT 5-. 1
TOTAL ANNUAL MARKET COSTS OF VOC REMOVAL
(Figures in millions of 1983 dollars)
Small system interest rate (inflation free) = 7.5%
Large system interest rate (inflation free) = 6.2%
20 yrs.
Systes Size (Poo.): 25—500 501—3300 3301-50K 50K+ Total
10—5 Risk
Annualized Costs 3.2 5.5 10.3 10.9 299
Present Value 32.8 56.2 116.3 123.0 329.3
10—6 Risk
Annualized Costs 10.9 20.5 34.7 42.6 109.8
Present Value 111.3 209.5 391.4 481.2 1193.4
MCI = 0.5 ugh
Annualized Costs 18.0 35.0 57.4 64.1 174.5
Present Value 183.1 336.5 647.9 723.6 1911.1
MCI = 1 ugh
Annualized Costs 11.5 21.3 35.6 43.7 112.1
Present Value 116.8 217.6 402.1 493.2 1229.5
flCI5ughl
Annualized Costs 4.1 6.2 9.9 3.5 23.1
Present Value 42.0 63.1 111.3 39.6 256.0
MCI 10 ugh
Annualized Costs 2.6 3.3 5.5 0.7 12.6
Present Value 26.5 38.5 62.6 1.9 133.3
MCI = 20 ugh
Annualized Costs 1.7 2.6 3.4 0.4 8.1
Present Value 17.3 26.9 38.6 4.3 87.1
MCI = 25 ugh
Annualized Costs 1.5 2.3 2.9 0.3 7.
Present Value 15.2 23.3 33.0 3e5 75.1
MCI = 50 ugh
Annualized Costs 1.0 1.5 1.8 0.2 4.5
Present Value 10.0 15.1 20.5 2.0 47.6
MCI 100 ugh
Annualized Costs 0.7 0.9 1.1 0.1 2.8
Presenr Value 6.7 9.7 12.8 1.1 30.2
V—7
-------
are expressed in terms of annualized cost (i.e., the annual
payment that would cover operating costs and a 20 year mortgage
on the capital facilities at the stated interest rate) and present
value (i.e., capital cost plus the discounted stream of annual
0&M costs). The equivalent costs using a social interest rate
are presented in Exhibit 5—2. These costs (and those presented
in the remainder of this chapter) represent totals over all
of the regulated VOCs.
The number of systems in the nation requiring treatment
to attain a given MCL was computed by PTm for four specific
MCLs (0.5, 5, 20, and 50 ugh) and interpolated to other MCLs
using the regression technique described above. The results
are presented in Exhibit 5—3, rounded off to the nearest whole
system.
Exhibit 5 U shows the annual cost of removal (computed
using the market interest rates) per system requiring treatment.
Within each of the three smallest size categories, the average
cost per system is fairly stable over a].]. MCLs. This is because,
as the MCL decreases, the cost of treating any given system
increases, while new systems that are relatively inexpensive
to treat are “captured” by the regulation at a roughly compensating
rate. The largest size category is an exception, however.
Here, the addition of new systems to the population of those
that must treat does not compensate for the increased cost of
treating an individual system as the MCL decreases. Therefore,
‘the average per system cost rises with decreasing MCI..- i n this
size category.
Using statistics on water production per systenr and. per
household from the Survey of Ooerating and Financial Characteristics
of Community Water Systems (U.S. EPA, Office of Drinking Water’,
1982), the costs per system were transformed into annual costs
per 1000 gallons and per household. These are shown in Exhibits
5—5 and 5—6 respectively. The first of these exhibits shows
the annual treatment related increase in the cost of water production
to the utility, assuming that the average utility pays the market
interest rates used in the annualization computation. The’ second
exhibit gives the equivalent average increase in the yearly
water bill for homes within each size category, assuming that
all of the treatment related costs are passed on to the customer’
without any markup by the utility. The impacts on both the
utility and the consumer increase markedly with decreasing system
size.
The costs of an alternative regulatory strategy also were
analyzed. This alternative maintained the same MtLs as the
basic strategy, but allowed variances for all systems serving
fewer than 10,000 people. (Such a strategy is not permitted
under SDWA but was evaluated under Executive Order 12291.)
The estimated national costs for this strategy were approximately
60 percent of the costs for the basic strategy at MCLs of 0.5
and 1 ugh. At 5 ugh, the ratio dropped to 1414 percent. At
V—8
-------
g , • •i . —
TOTAL ANNUAL SOCIAL COSTS OF VOC REMOVAL
(Figures in millions of 1983 dollars)
interest rate (inflation free) 4, 14%
n 20 yrs.
Syste. Size Pop.): 25—500 501—3300 3301-50K 50K+ Total
10—5 Risk
Annualized Costs 2.6 4.9 9.3 10.1 27.0
Present Value 34.3 64,4 122.6 132.7 354.0
10-6 Risk
Annualized Costs 8.9 17.1 31.9 40.8 98.7
Present Value 116.8 224.0 419.1 535.5 1295.4
CL = 0.5 uq/l
Annualized Costs 14.7 28.3 53.0 61.5 157.4
Present Value 192.4 371.6 694.8 806.7 2065.5
CL 1 ugh
Annualized Costs 9.3 17.7 32.8 41.9 101.7
Present Value 122.6 232.6 430.4 549.8 1334.4
CL:5ug/1
Annualized Costs 3.3 5.6 8.9 3.3 21.2.
Present Value 43.8 73.9 117.3 43.0 278.0
CL 10 ugh
Annualized Costs 2.1 3.4 5.0 0.6 11.1
Present Value 27.7 44.1 65.9 8.4 146.1
MCI = 20 ugh
Annualized Costs 1.4 2.1 3.1 0.3 7.0
Present Value 18.3 28.0 40.6 4.5 91.4.
MCI = 25 ugh
Annualized Costs 1.2 1.8 2.7 0.3 6.0
Present Value 16.0 24.2 34.8 3.7 79,7
= 50 ugh
Annualized Costs 0.8 1.2 1.6 0.2 3.9
Present Value 10.6 15.5 21.6 2.0 49.6
MCI = 100 ugh
Annualized Costs 0.5 0.8 1.0 0.1 2.4
Present Value 7.1 9.9 13.4 1.1 31.1
v—9
-------
EXHIBIT 5—3
NUMBER OF SYSTEMS REQUIRED TO REMOVE VOCs
Systea Size (Pop.):
RISK or 25- 501— 3301— 50k+
MCI (ugh) 500 3300 50K Total
10—5 - 653 279 146 17 1095
10-6 2127 863 555 66 3610
.5 ugh 3483 1406 989 126 6004-
I ugh 2246 906 575 68 3794
S ugh 862 303 168 15 1347
10 ugh 506 182 100 7 795
20 ugh 332 119 61 4 516
25 ugh 290 104 52. 3 450
50 ugh 192 68 32 2 294
100 ugh 127 45 20 1 193 -
v—b
-------
EXHIBIT 5— 1
ANNUAL MARKET COSTS PER AFFECTED WATER SYSTEM
(Figures in thousands of 1983 dollars)
Sytte. Size (Pap):
RISK or 25- 501- 3301- 50K . Average
MCL (ugh) 500 00 50K
10-5 4.9 19.8 70.5 655.4 27.3
10-6 - 5.1 23.8 62.5 646.9 30.1
.5 ugh 5.2 24.9 58.1 501.8 29.1
1 ugh 5.1 23.6 62.0 645.8 29.5
5 uglt 4.8 20.4 58.7 239.9 17.6
10 ugh 5.1 0.7 55.6 93.0 - 15.8
20 ugh! 5.1 22.1 56.1 96.7 15 8
25 ugh! 5.1 22.0 56.2 98.0 15.6
50 ugh 5.1 21.6 56.5 105.7 15.2
100 ugh 5.2 21.0 56.6 108.5 14.7
v—u
-------
EXHIBIT 5—5
ANNUAL MARKET COSTS PER UNIT OF PRODUCTION
(Only water systems requiring VOC removal are included.)
(Figures are in $‘s/lOOO gal)
Syctea Size (Pop):
RISK or 25- 501— 3301— 50k+ veraqe
MCL (ugh) 500 00 50K
10—5 0.56 0.28 0.09 0.04 0.08
10—6 0.58 0.3 0.08 0.04 0.07
.5 ugh 0.59 0.35 0.07 0.03 0.06
I ugh 0.58 0.33 0.08 0.04 0.07
5 ugh 0.5’ 0.29 0.07 0.02 0.06
10 ugh 0.58 0.29 0.07 0.01 0.06
20 ugh 0.58 0.31 0.07 0.01 0.07
25 ugh 0.58 0. 1 0.07 0.01 0.07
50 ugh 0.58 0.30 0.07 0.01 0.08
100 ugh 0.59 0.29 0.07 0.01 0.08
1983 Ave. Cost -
of Production 2.84 1.83 1.23 0.89
V—i 2
-------
EXHIBIT 5—6
ANNUAL MARKET COST OF VOC REMOVAL PER HOUSEHOLD
(Only affected water systems are included.)
(Figures are in $‘s/household/yr.)
Syste. Size (Pop):
25-
501-
3301— SOK+
500
3300
50K
RISK or
CL tug/I)
10—5
Average
93.46
39.74
13.88
8.45 13.57
10-6
97.36
47.87
12.31
8.34 12.84
.5 ugh
97.79
50.00
11.43
6.54 11.11
1 ug/t
96.69
47.36
12.20
8.32 12.83
5 ugh
90.74
41.07
11.55
3.09 10.84
10 ugh
07.40
41.56
10.94
1.20 10.51
20 ugh
97.16
44.51
11.03
1.25 11.78
25 ugh
97.21
44.20
11.06
1.26 12.10
50 ugh
97.25
43.48
11.12
1.36 13.13
100 ug/L
97.75
42.19
11.14
1.40 14.11
v—i 3
-------
all higher MCLs, the ratio of alternate strategy costs to those
of the basic strategy was close to 30 percent.
Another alternative regulatory strategy is that of having
an additional TVOC MCL applicable to the multiple contaminant
case. Analysis of this concept showed that a nearly threefol
increase in total social c st would be produced at the 10
level while costs at the 10 level are unaffected because
is stringent enough to provide an equivalent level of VOC removal.
5.3 Uncertainties in Cost Assessment
There is a degree of uncertainty inherent in any estimate
of compliance costs. This uncertainty arises frám two general
sources.
o Only a small number of observations are available for
many of the factors that contribute to overall costs.
These factors include the following:
— the frequency and degree of contamination from any
given ‘JOC;
— the unit costs ofavailable treatment technologies;
— the likelihood that a system will select a particular
treatment technology;
and others; -
o Some of the factors that contribute to costs, such as
system growth rates, cannot be observed more than once.
The errors in the estimates of these input variables contribute
to the overall uncertainty in the computed total compliance
cost.
In order to estimate the magnitude of the uncertainty in
the final result, it is normally necessary to take account explicitly
of’ the errors in the independent variables used in the model.
.This has been done elsewhere for a hypothetical, but realistic,
application of PTm (see Uncertainty in the Re ulatorv Decision—makinf
Process , U.S. EPA, Office of Drinking Water, September 28, 198 4.
In the present case, however, the computation of total cost
is based on a regression equation that itself incorporates the
errors inherent in the PTm model results. Therefore, one can
compute the uncertainty In the predicted value of national cost
directly from the standard error of the prediction equation.
This is accomplished using the formula:
2 + lIT + (x - x)2, (xt - )2]
V—14
-------
where:
= standard error in the national cost
s standard error in the regression equation
X MCL for which a cost prediction is desired
x average value of all MCL.s
xt MCLs at which the observed MCLs were computed
T total number of observations.
This expression gives the following values of the standard errors
in the estimates of annualized social costs of compliance:
Total Annual Cost 4 Standard. Error* 4
MCL (u /l) ( millions. 1 8 1 ( millions .
0.5 157.14 14.7
1 101.7 13.14 .
5 21.2 12.1
10 11.1 11.8
20 7.0 12.2
25 6.0 10.2.
50 3.8 10.2
100 2.14 . 10.9
* From Exhibit 5.2
** The probability that the true value lies within
plus or minus one standard error of the estimate
is 68 percent. The probability that the true value
lies within plus or- minus two standard errors is
95 percent. These probabilities assume’ that the
error is normally distributed.
The standard errors shown are conservative, in that they assume
that all errors in the PTM model (on which the regressions are
based) are independent and uncorrelated. These conditions will
not hold if, for example, unit treatment costs are consistently
under estimated or over estimated, or if the occurrence data
are subject to bias. The figures do show, however, that the
relative magnitude of the uncertainty increases rapidly as MCI.
increases. This is presumably because these MCLs affect only
those systems that have high influent concentrations to start
with, arid the estimates of the numbers of such systems are subject
to large errors because they are rare.
v—i 5
-------
5.14 Cost to State Governments of Federal VOC Refulations
5.11.1 Puroose
The purpose of this analysis was to determine the start—up
and on—going costs to water supply departments in state governments
should EPA develop MCL.s for volatile organic compounds in drinking
water. Start—up costs are defined as those costs, including
labor and capital, incurred to initiate a state’ program. On—going
costs are those costs which states incur each year to implement
the program. Specific elements of both start—up and on—going
costs are discussed below. -
5.11.2 Data Sources
To determine the costs of an EPA MCL, informal telephone
interviews were conducted with representatives of state drinking
water programs in 12 states. Each respondent was asked to estimate:
o the costs of their state’s current program (if any)
for regulating VOCs in drinking water; and
o the likely additional costs their state would incur
should EPA institute MCLs to regulate- VOC.s in their
drinking water.
The twelve states were chosen to represent the’ range of -
severity of VOC contamination problems. In addition, the states
were selected to represent all regions of the country. The
states are listed below:
o California o New Jersey
o Florida 0 New York
o Maine • o Pennsylvania.
o Michigan o South Dakota
o Mississippi c Tennessee
o Missouri o Washington
5.14.3 Cost Categories
Respondents were asked to estimates the costs of the
current state program and the additional costs of an EPA program
in terms of labor (expressed in work years) and capital (expressed
in dollars). The cost categories are as follows:
Start—Un Costs
Review of Legislation or Guidelines
Preparation of Guidelines
Training of Staff
Conducting Baseline Surveys
Setting up Laboratory
Initiation of Technical Outreach
V.- 16
-------
Review of Monitoring Data
Review of Treatment Options
Review of Health Risks
Enforcement of State Actions
Ongoing Costs
Review of Utility Monitoring
Conducting of State Monitoring (if any)
Provision of Technical Assistance
Preparation of EPA Reports
Provision of Laboratory Analytical Services
Enforcement of State Actions
General Administration
State representatives also were asked to report the number of
water systems sampled for VOCs, the number of contaminated systems
discovered, and the schedule for testing of VOCs (if any).
5.Ll.L Findings -
Cost estimates for the 12 states are shown in Exhibits
5—7 through 5—9. Costs are presented for the current state
program, if one exists, and for additional activities if EFA
were to set MCLs for certain VOCs. Exhibit 5—T presents the
start—up and ongoing costs for each state. The costs are specified
in work years of. labor and .dollars of capital. The data show.
a range in current program and anticipated additional expenditures.
The wide variation in costs estimated by the states illustrates.
two phenomena: 1) the substantial unevenness in state programs;
and 2) the fact that states both measure and account for costs
differently. The cost data obtained from states were “best
estimates” from the person interviewed and do not reflect actual.
accountin.g data.
Total dollar figures for all 12 states are shown in Exhibit
5—8. A loaded labor rate of $1 O,OOO was used as the value of
an average work year.
Using these cost estimates, the state costs of additional
EPA regulations for the entire nation were calculated. Several
factors were analyzed to relate current and additional costs
to state characteristics. The factors considered were as follows:
o current level of state program (measured in dollars)
o state population
o state population density
o region of the country
o severity of VOC contamination problem
These factors in combination can adequately explain why
a particular state incurs its current costs or estimates a certain
additional cost. However, none of the factors was statistically
correlated with the costs of current or additional VOC programs.
V—i 7
-------
EXHIBIT 5—7
SUMMARY OF CURRENT AND ADDITIONAL COSTS
OF VOC REGULATION BY STATE GOVERNMENTS
NOTE: The data shown are based
‘reported show their best
incurred.
* Work years
** Work years per year
on informal telephone interviews. The
estimate for costs they may not record
STARTUP COSTS
ONCOINC COSTS
For
For
Additional
For
For
additional
Current
Current roram
EPA Re ulations
Capital
Labor
Capital
Labor**
Capital
Labor**
CapitaL
CaLifornia
18.75
0
4
0
15
0
L
0
Florida
.16
$ 3,000
0
0
.7
0
0
0
Maine
.75
$34,500
1.23
$134,500
1.26
0
.1.
0
Michigan
3.55
0
0
0
4
0.
0
0
Mississippi
.6
$215,000
2.25
0
.6
0
1
0
Missouri
1.1
0
4.8
$340,000
2 .5
0
5
0.’
New Jersey
6.5
$2 liii.
0
0
6.11
900,000
6
0
New York
25.92
$750,000
1.17
0
4.25
0
2.
0
Pennsylvania
2.1
0
0.05
0
1.5
0
0
0’
South Dakota
0
0
1.9
0
0
0
.25
0
Tennessee
.5
0
2.3
0
1.2
0
4.3
0
Washington
2.5
0
2
$500,000
1
0
1
0
costs state
and may not
officiaLs
yet have
v—i 8
-------
EXHIBIT 5—8
ESTIMATED COSTS FOR VOC REGULATION FOR SANPLE STATES’
For Current State Proçrams For Additional EPA Re u1at p
Start—Up Costs $5.5 million $1.8
Ongoing Costs $2.’ $0.8
EXHIBIT 5—9
ESTIMATED NATIONAL COSTS OF STATE GOVERNMENT
IMPLEMENTATION OF EPA REGULATIONS FOR VOCs
For Current State ?ro rar,1 For Addjtjpi,pi EPA Re uIatjo
Start—Up Costs $12.8 million
Ongoing Costs $ 5.6 $1.9:
*A loaded rate of $qO,000/wcrk year was used to determine labor costs.
V—I 9
-------
That is, it is not possible to predict additional costs due
to EPA regulations as a function of state characteristics since
there is no statistically significant relationship between additional
costs and state characteristics. Thus, it was not possible
to project costs to non—sample states on the basis of state
characteristics.
Instead, to estimate the costs of all 50 state governments
of additional VOC programs resulting from- EPA regulations, an
assumption was made that the states contacted represent a random
sample of states (weighted by population). The group of sample
states represents L 3 percent of the U.S. population, therefore
the costs of’ the sample states were assumed to represent -3
percent of the costs to all states for administering VOC programs.
Using this assumption, cost estimates were extrapolated to the
entire nation. Exhibit 5—9 shows that the predicted start—up
costs to the country for an additional EPA program would. be
approximately $14.1 million and on—going costs would be approximately
$1.9 million per year. The on—going costs are roughly e4uivalent
to one person year/state. States with a greater number of VOC
problems such as California, New Jersey, and Florida can be
expected to expend considerably more.
V—20
-------
6. SYSTEM LEVEL NET BENEFITS ANALYSIS
6.1 Introduction
Chapters L and 5 deal with total national benefits and
total national costs. This aggregate level of’ analysis serves
a fundamental and necessary purpose in evaluating national policy
alternatives for drinking water regulation. It is essential
to develop estimates of the total impact at a national level
and such estimates are required to comply with Executive Order
12291. However, these aggregate analyses are not, by themselves,
adequate to reveal all the subtleties of a proposed regulatory
strategy.
Some of those subtleties are quite important and are “subtle”
only in the sense that they are not readily apparent in an aggregate
level analysis. To evaluate these aspects of regulatory alterna-
tives, Executive Order 12291 also requires analysi of “net
benefits” which entails a more refined comparison of benefits
and costs. This chapter presents an analysis of the “net benefits”
of alternative regulations evaluated at the individual water
system level . It is emphasized that regulatory decisions under
SDWA cannot be based on benefit/cost analysis. This- analysis
is prepared therefore only to comply with Executive Order 12291.
Net benefits analysis may be viewed simplistically as a
mere reformulation-of the familiar concept of benefit/cost analysis
where net benefits are defined as the difference of benefits
less costs. Where in benefit/cost analysis it is desirable
to have a ratio exceeding 1.0, in net benefits analysis it is
desirable to have a difference that is positive. In both of
these formulations, the argument being made is no more complicated
than the common sense notion that in order for a regulatory
action to be worthwhile, the benefits should exceed the costs.
This is not an adequate distinction of the type of methodology
employed in this chapter, however. Net benefits analysis as
defined in the above paragraph could be carried out at the aggregate
level using the results of Chapters L! and 5. The distinction
being made here is that-of evaluating net benefits at the system
level .
The need for a system level analysis stems from the fact
that while the net benefits of a given regulatory action may
be positive in the aggregate, the aggregate figures may disguise
an unknown number of negatives that are “netted—out” and therefore
never seen in the aggregate results. The analysis of individual
systems of varying sizes can reveal such circumstances.
Sensitivity to system size is perhaps the most immediately
obvious, but - not the only distinguishing characteristic of the
analysis presented in this chapter. The other important feature
is that it is a “marginal” analysis of net benefits. The meaning
VI—’
-------
of this term is often obscure to non—economists because it is
often deeply embedded in the nature of the problem and therefore
difficult to define in abstract. For example, the need to pay
attention to the effect of system size is made explicit in terms
of: 1) the risk of setting an MCL that is too stringent; and,
2) the risk of setting an MCL that is not stringent enough.
These are the essential elements of a problem that lends itself
to “marginal” analysis —— a situation in which it is best to
evaluate a gradient of alternative actions searching for the
one that goes just far enough without going too far.
Specifically, “marginal” analysis in public policy problems
involves the study of society’s “willingness—to—pay” for social
benefits. If society behaved in perfect accord with the pre-
scriptions of economic theory, the upper limit of our will ingness—
to—pay would be defined by the equivalence of marginal social
benefit and marginal social cost. In terms of the above mentioned
gradient of regulatory action, this means that as long as a
dollar’s worth of expenditure brings more than a do].lir’s worth
of benefit, the buyer (the public, via EPA) should be willing
to continue making such bargains up to the break—even point.
Beyond such a point, the buyer should, of course, have rio further
interest in the transaction.
It is not easy to place a monetary value on benefits when
valuing human life. One approach to coping with this discomforting
situation is to cast the problem in the context of society’s
imolied wi1lin ness—to—oav . On the basis of past regulatory
actions, . it is possible to define the range of what society
has been willing to pay per death avoided in other similar instan-
ces. An inspection of EPA’s recent regulatory actions reveals
that values of life in the range of $300,000 to $7,000,000 have
been considered. The approach suggested by this finding is
simply to check regulatory alternatives to see if they fall
somewhere within the range of this imolied wi1lin ness—to—oav .
It is important to note, however, that this should be a “marginal”
comparison.
The importance of performing this comparison on a marginal
basis is made clear by consideration of the implications of
sole reliance on the results of aggregate analysis such as presented
in Chapters 4 and 5. The aggregate national totals presented
in the two chapters can be combined to produce estimates of
the “average” cost per case of cancer avoided. It may be misleading
to use these average cost figures to evaluate society’s willingness
to pay. The problem is that noted above;, the “aggregate average
cost” provides no guidance on whether the associated regulatory
action will err on the side of going too far or of not going
far enough.
It is easy to conceive of circumstances in which a regulatory
action may impose compliance requirements on a small segment
of the target group whose marginal cost of compliance happens
to be far in excess of the range society is normally willing
‘. .—2
-------
to pay. Because they are a minority, their costs are averaged
in with all other systems at the aggregate level and the regulation
appears to be within acceptable bounds. This is the situation
where marginal cost exceeds average cost. That fact, in itself,
is not bad, but average cost as an indicator gives no hint as
to how fast the marginal cost is continuing to rise with more
stringent increments of regulation or where the marginal benefit
stands with respect to society’s implied willingness—to—pay.
In other words, the average cost indicator does not provide
guidance on when to stop adding increments of stringency in
the regulatory action.
On the other hand, it is equally conceivable that the average
cost could conceal the fact that a regulatory action does not
go far enough. This is the circumstance where marginal cost
is less than average cost. This can easily come about in situations
where a large fixed capital investment is required for control
equipment capable of meeting a wide range of treatment standards.
In such cases, the average cost may be relatively stable over
a wide range of alte rnative treatment standards, indicating
no significant advantage in adding further increments of stringency
in the regulation. While the average cost is relatively stable,
however, the difference between marginal cost and marginal benefit
may offer attractive bargains over this same range. The danger,
therefore, is that of stopping short with the choice of regulatory
action because the average cost indicator provides no guidance
on how far to proceed. -
Ultimately, true willingness—to—pay can only be determined
at the level of individual water systems. It cannot be accurately
analyzed as a decision criterion on the basis of aggregate or
average cost data. The net benefits analysis presented in this
chapter provides the type of system level marginal indicator
that is needed to evaluate regulatory alternatives in terms
of willingness—to—pay. It makes possible an assessment of whether
a given regulatory action has gone too far or is not stringent
enough. -
6.2 Methodology
The total net benefits of alternative MCLs are evaluated
on a present value basis, computed via the following formula:
PVNB (PVF x [ (ARCxINFxEFCxPOPXV) — OMC]} — CAPC
Where:
PVNB the present value of total net benefits
PVF the present value factor: 1L.877, based on a 3 percent
discount rate anda 20 year period.
VI —3
-------
ARC the annual risk of cancer, using the same assumptions
as in Chapter LI.
INF the concentration of the VOC in the influent raw
water before treatment.
EFC the efficiency of removal of the VOC present, figured
as the percent removed.
POP the average population of a water system in a given
size category.
V = the value of life assumed; expressed in terms of
dollars per cancer case avoided.
OMC the annual operation and maintenance cost of the
treatment technology; based on data supporting the
analysis of Chapter 5. -
CAPC = the capital cost of the treatment technology; based
on data supporting the analysis of Chapter 5.
The treatment technology assumed for purposes of the cost
estimates used in this analysis was packed tower aeration.
It was chosen because it may be used for either partial or total
removal of VOCs. This permits evaluation of the widest possible
range of alternatives. Some other treatment options such as
granular activated carbon (GAC) are restricted to total removal.
Cost curves reflecting the effects of various levels of removal
were used as the input data based on the same analysis that
supports the aggregate cost estimates of Chapter 5.
The cost curves and the risk factors in the above equation
are the only variables that are held constant in the evaluation
of net benefits. All of the other variables are jointly varied
across their entire ranges to determine the effects on net benefits
produced by the different combinations of circumstances. The
dimensions of variation are outlined as follows: -
INF —— from high to low levels of influent concentration
EFC —— from high to low levels of removal, reflecting
more or less stringent MCLs
POP —— from large to small water system sizes -
V —— from high to low values per case of cancer
avoided
The framework for evaluating these dimensions of variation
is illustrated in Exhibit 6—1. The graph shows how PVNB changes
at MCLs of. varying stringency (EFC) given varying assumptions
about the value per cancer case avoided CV). Separate analyses
were prepared for each combination of high and low values of
VI—4
-------
EXHIBIT 6—1
PRESENT VALUE NET BENEFITS AS A FUNCTION
OF MCL AND THE ASSUMED VALUE PER CANCER CASE AVOIDED
Assumed Valu
I—. ‘i I ,! I—. I I I•—•
I ‘ I Per Cancer Ca
INF’LUENT 7 /I, PCJPUL4Th N CI35
4
3
11(1
— -
c i
E . -1
— .3
—4
—5
—4
2C’ 24
EffIu rit (.iT:i
VI—5
4
-------
the influent concentration (INF) and large and small water system—
sizes (POP).
The graph in Exhibit 6—1 illustrates the finding that there
is a maximum value of PVNB for any given value of V; this is
indicated by the fact that each of the PVNB curves has a maximum
point. The maximum point on each PVNB curve (e.g., point E)
represents the point at which the marginal addition to cost
for an additional increment of stringency is exactly equal to
the marginal addition to benefits. In other words, marjinal
net benefits are zero. When marginal net benefits are zero,
total net benefits are maximized. Thus, the maximum point defines
the optimal MCL for the value of V assumed and the other conditions
assumed for population and influent concentration.
It also is apparent from the graph that PVNB is not positive
for all values of V, not even at the maximum point. There is
a threshold level, V’, associated with the “break—ev n” point
—— the point where t ie maximum value of totat net benefits is
exactly equal to zero (point C in Exhibit 6—1). Levels of V
equal to or greater than V’ must be assumed in order to justify
regulatory action. This break—even point also defines the minimum
level of stringency required in an MCL in order to achieve positive
(or non—negative) total net benefits for the threshold. value
of V. Furthermore, all successive maximum points CD, , F,
etc.) are to the left of the break—even point, implying that
more stringent MCLs than that associated with the break—even
point are required to maximize totaL net benefits for values
of V above V’.
This family of PVNB curves can be collapsed into a more
compact form by sketching the single curve which connects the
maximum points of the individual curves. This produces a MAX
PVNB curve as shown in Exhibit 6—2. The point at which this
curve crosses from the negative to the positive zone is the
break—even point. It defines the minimum value of life required
to produce positive total net benefits (V’) and the associated
minimum level of stringency in MCL selection required to attain
maximum net benefits for any value of V above V’.
One might be inclined to ask how there can be a minimum
level of stringency required to justify regulatory action.
Would not some measurable benefit result from any level of action?
The answer is yes, but below this minimum level of stringency,
the net benefits will be either negative or; if positive, less
than the maximum attainable for the given value of V.
Plotted above the MAX PVN curve in Exhibit 6—2 are the
marginal and ave rage cost curves associated with it.. The relation-
ship between these two curves is familiar in light of the earlier
discussion of marginal and average cost. The diagram confirms
that the average cost curve is fairly flat over most of the
relevant policy range (i.e., the range of MCLs being considered).
Thus one cannot be certain on the basis of average cost alone
whether a given MCL is above or below the break—even point.
VI-6
-------
EXHIBIT 6—2
MAXIMUM PRESENT VALUE NET BENEFITS AS A FUNCTION
OF MCL AND RELATION TO MARGINAL AND AVERAGE COST
verage Cos
Lrginai Co
AX PVNB
TRI C H LO R 0 ETHYL EN E
(1
C
‘—I
10
7
4
—1
3
I
10 14
u nt (& ••i)
1 22 26 .30
VI—7
-------
There is no direct theoretical link between the average
cost curve and the MAX PVNB curve. A somewhat indirect link
is provided by the intersection of the marginal and average
cost curves. Principles of differential calculus mandate that
the marginal curve will always intersect the average curve at
the minimum point of the average curve. In this particular
problem, the location of this intersection will always be somewhere
to the left of the break—even point on the MAX PVNB curve.
The intuitive reason for this is embedded in the definition
of the break—even point; it is the point at which enough benefit
is being realized to offset the initial fixed cost of the treatment
technology. It stands to reason that the average cost will
steadily decrease as one adds successive increments of benefit
against the same fixed cost. Minimum average cost is not required
to achieve the break—even point, but it is not surprising to
find it nearby.
This might suggest that analysis of regulatory options
using average cost per case estimates developed in Chapters
U and 5 may lead to choices having positive net benefits when
a “1east...cost per ...unit ...of .benefit” or “minimum average cost”
criterion is applied. If, however, the average cost. curve is
very flat (or if ’ there are large uncertainties that affect the
perceived shape of the curve) in the vicinity of’ the minimum
point, the potential for error may be great and options having
negative net benefits or excessive, costs at the :margin could
be selected by this procedure. It is further important to note
that the average cost per case implied by the analyses of Chapters
U and 5 is an aggrega€e of the average costs- of all system size
categories whereas the average cost curve shown in Exhibit 6—2
is based on the system level average cost. While the minimum
average cost may be easy to spot on a diagram at the level of
an individual water system, there is no straightforward method
of defining a point of minimum average cpst in the aggregate..
By contrast, the marginal cost curve in Exhibit 6—2 is
continuously upward sloping. Because the points on the MAX
PVNB curve represent the points at which marginal cost and marginal
benefit are equal, the marginal cost curve also may be regarded
as a marginal benefit curve. So defined, it may be directly
interpreted in terms of the willingness—to—pay decision criterion
mentioned earlier. The marginal cost/marginal benefit curve
and the MAX PVNB curve may be used together to define the solution
space; the boundaries within which all MCL choices are associated
with a maximum level of’ positive net benefits for values of’
V above V’ but below the upper limit of society’s implied willing-
ness—to—pay.
The minimum acceptable MCL or break—even level of stringency
is defined by the point at which the MAX PVNB curve crosses
into the positive zone. Extending a vertical line up from this
point to intersect the marginal cost/marginal benefit curve
defines the value per cancer case avoided (V’) which must be
VI—8
-------
assumed in order to make positive net benefits achievable.
In some cases, this value may be outside the affordable range,
thus dictating no viable regulatory action.
The maximum level of stringency is defined by extending
a horizontal line from the $7,000,000 mark on the vertical axis
across to Its intersection with the marginal cost/marginal benefit
curve. The MCL corresponding to this point (directly below
it) represents the maximum level of stringency that is within
the affordable range.
Summarizing, the analysis developed in this chapter may
be used to answer three essential questions for any given set
of population and influent conditions:
o What is the minimum value per case of cancer avoided
(V’) that must be assumed in order to obtain positive
net benefits? And, is this within the affordable
range based on society’s implied willingness tO—PaY?
o Assuming V’ Is within the affordable range, what is
the minimum level of stringency required for an MCL
in order to maximize net benefits?
o What is the most stringent MCL that is within the
affordable range based on society’s implied wilLingness.
to pay?
6.3 Results and Discussion
The analysis described above was performed for a sampling
of the VOC contaminants being considered for regulation to provide
a check on the validity of conclusions inferred from the results
•of the aggregate analyses .of Chapters 4 and 5. The following
cases were studied across twelve population size categories:
o TrichloroethYlefle (TCE)
— influent 10 ugh
— influent 75 ugh
o Benzene
—. influent 10 ugh
— influent 75 ugh
o Carbon Tetrachioride
— influent 10 ugh
— influent 75 ugh
o Multiple Contaminants —— “Average” Mixture
— inf].uent 10 ugh
— influent 75 ugh
o Multiple Contaminants —— “Weak” Mixture
— influent 10 ugh
— influent 75 ugh
vI—9
-------
Exhibit 6—3 presents the results of these analyses, providing
answers to the three questions listed above. The table presented
in this exhibit requires some explanation prior to interpretation.
First, results are presented in terms of risk levels achieved
instead of the equivalent MCLs concentrations. This lends more
meaning to the interpretation by facilitating equal comparisons
between contaminants and by stating the result in terms of what
it is that society is willing—to—pay for; the attainment of
a lower leve of cancer risk. Risk levels are presented in
term3 of 1O and 10 individual lifetime cancer risks. (It
is noted that under SDWA, MCLs cannot be set on the basis of
a risk—based approach. This analysis merely evaluates MCLs
in terms of risk out of analytical convenience and to comply
with Executive Order 12291.)
Two risk levels are presented for each case studied. The
first represents the risk associated with the least, stringent
MCL that will produoe positive net, benefits for the break—ev n
value of V. Sometimes this is 1O and other times it is 1O .
The second risk level listed for each case represents the risk
associated with the most stringent MCL that is within the affordable
range. In many cgses the break—even MCL and the most stringent
MCI.. are both 10 . An MCL more stringent than 1O would be
outside the affordable range in most instances.
The adjacent column of the- table presents-a summary. of. .
the results on what society would have to be willing- to pay
in order to achieve the associated risk levels. These values
are presented as a range which represents the variation across
different water system sizes. In general, the low end. of the
range is representative of the marginal cost per case avoided
to the largest systems and the high end of the range is represen-
tative of the marginal cost per case avoided to smaller systems.
Incorporation of the very smallest size categories would
require a much broader range in the cost column of the table.
For this reason they are not included in the ranges given.
Instead, the footnotes to the table indicate places where the
cost ranges applicable to the large and mid—size systems are
insufficient for the smallest size categories.
Interpretation of these results must be prefaced with a
brief classification scheme; all VOCs were not evaluated but
those that were are intended to represent the others, as follows:
trichioroethylene —— relatively low risk group
benzene —— moderate risk group
carbon tetrachioride —— relatively high risk group
multiple contaminants —— a special case
This classification is purely relative and reflects broadly
the variation in risk indicated in Exhibit —3.
VI—lo
-------
EXHIBIT 6—3
SUMMARY OF RESULTS OF
PRESENT VALUE NET BENEFITS ANALYSIS
E4 fluent Value Per Case 0+
Risk Levels Cancer Avoided CV)
Contaminant Associated That Must Be
& Influent With Break—Even Assumed To Obtain
Condition MCL And With Most Break—Even MCLa
Stringent MCL And
A44ordab le Most Stringent MCLa
C$ Millions)
Trichioroethylene
10 ugh Break—Even MCLs 10 —6 30 — 40 *
Most Stringent MCL: 10 —ó 30 — 40 0
73 ugh Break—Even MCL: 10”—S 4 — 7 ••
Most Stringent MCL: l0 —6 — 25 cc.
Ben:ene
to ugh Breal.—Even MCLi 1O ’—b 10 — 30 Cc.-
Most Stringent MCL: L0’—ó 10 — 30 cc.
75 ug/l Break—Even PICL: 10- ”-! 4 — a cc
Most Stringent MCL: 10 ’—6 14- — 18 * d-
Carbon Tetrachioride
10 ugh Break—Even MCL: 10’—ó 4 — B cc
Most Stringent MCL: 10 ó 4 — B co
75 ugh Breab—Even MCL: 10’—5 2 — 5 c c
Most Stringent MCL: 10’—6 5 — 9 cc
Multiple Case -
“Average” Mt;ture
10 ugh Break—Even MCL: t0 —ó I — 3 c.*.-
Most Stringent MCL: l0”—6 — 3 coil.
75 ughl Break—Even MCL: l0’—6 1 — 3 ccc ’ s -
Meet Stringent MCL: 10—ó 1. —
Multiple Case
“Weak” Mj turo
10 ugh Break—Even MCL: 10—6 10 — 15 .‘
Most Stringent MCL: 10’—ó 10 — IS .
75 ug/l Break—Even MCL: 10’—5 4 — 7 cc
Most Stringent MCL: 10’—ó 10 — 14 ...
* The four smallest system si:e classes would require values of V
even further outside the af+ordable range to achieve this MCL.
.0 The two smallest system si:e classes would require values o4 V
outside the af4crdable range to achieve this MCL.
ccc The two smallest system size classes would require values o4 V
e’eri Fi.zrther outside the affordable range to achieve this MCL.
**- d- The smallest syet m size category w u1d require values 04 V
in the neighborhood of Z7 to sit) million to achieve this MCL.
VI—”
-------
The most obvious conclusion to be drawn from the data in
Exhibit 6—3 is the fact that it does not appear to be possible
to produce positive net benefits in the two smallest of the
twelve system size categories for any of the cases studied.
These categories consist of systems having 25—100 people and
101—500 people. The populations at risk are simply not large
enough to produce a level of benefits equal to the cost of control
measures.
The results for carbon tetrachioride, representing the
relatively high cancer risk category of VOCs, indicate that
a 1O risk level can be achieved for both the high and low
influent concentratic n ranges at costs per case ranging from
$L _9 million. A 10 risk level could also be achieved with
positive net benefits for the high influent concentration case,
but 10 is the minimum level required to produce positive net
benefits in the low influerit concentration case.
The results for trichioroethylene (TCE) and benzene, represent-
ing low and medium cancer risk categories of VOCs, indicate
that positive net benefits are available only when influent
concentrati ns are high. With influent concentrations of 75
ugh, a 1O level of removal is within the “affordable” range,
requiring a cost per case of $14—7 million to; TCE and $1_8 million
for benzene. However, removal to the 10 level is not within
the affordable range, costing $15—25 rnillion per case• for TCE
and $14—i8 million per case for benzene.
For the more important case of low influent concentrations
positive net benefits cannot be attained for TCE and benzene
except at the 10 level of removal. The cost per case would
be beyond the affordable range, however, costing $30—140 million
for TCE and $1O—30 million for benzene. This finding is significant
because most occurrence of these VOCs is characterized by low.
influent concentrations. The implication is that it is not
worth it to regulate them. This conclusion will be reversed,
however, upon consideration of the multiple contaminant case, -
discussed below.
The multiple contaminant case is evaluated in- terms of
two different mixtures. One reflects the total risk in terms
of a weighted tTaveragett of all possible constituents including
two very high risk VOCs — — vinyl chloride and 1,1—dichioroethylene
—— which occur only in the presence of other VOCs. The other
“weak” mixture represents a mix of TCE, tetrachioroethylene,
and 1,1,1— trichioroethane which are very low risk, but very
ubiquitous VOCs.
Results for the “as erage” mixture indicate that there is
no question but that a 1O level of removal is in the affordab1e!v
range, at both high and low influent levels, costing only $t—3
million per cancer case avoided, This result, however, is really
nothing but a reflection of the fact that there are high cancer
risk VOCs assumed to be present in this mixture. When the difference
VI—12
-------
in risk between the high and low risk chemicals is in terms
of one or two orders of magnitude, as is the case here, the
presence of only a small amount of a high risk chemical is sufficient
to drive the analysis to the same conclusion as when the high
risk chemicals are considered individually. In other words,
there is a negligible dilution effect.
Results for the “weak” mixture are similar to the results
obtain%d for TCE and benzene except that the cost ranges for
the 10 level of treatment are somewhat closer to the affordable
range. The cost per case for high influent concentrations is
estimated to be $10—l’ million while the cost for low ipfluent
concentrations is estimated to be $1O—15 million. A 10’ level
of treatment produces positive net benefits in the high influent
concentration case at a cost per case of $lt_7 million. In the
low influent concentration range, positive net benefits cannot
be achieved except at the 1O level.
A z discussed in Chapters L and 5, the option of establishing
an MCL for Total VOC concentration (TVOC) has been considered
as a means of dealing with the multiple contaminant problem.
The above analysis suggests that the choice of an optimal level
for such an MCL would be complicated by uncertainty over which
contaminants re present. To err on the side of safety and
specify a 1O requirement would produce negative net benefits
in cases where high risk VOCs are not found.
It appears that the approach çf setting individual MCLs
for each of the contaminants is more efficient for several reasons.
The comparative volatility of the different VOCs favors an approach
in which the low and moderate risk VOCs are removed to some
degree to assure removal of t ,e high risk VOCs as well. For
example, removal of TCE to a 1O level b y packed tower aeration
also will remove vinyl chloride to a 1O level during the same
treatment process.
This feature of the treatment technologies takes on special
significance in consideration of some evidence which suggests
that certain VOCs, including the very potent vinyl chloride,
may be found primarily in the multiple contaminant case because
they may be produced as biological degradation products of the
other VOCs. If this is indeed the case, then the above results
•for low and moderate risk VOCs would be incorrect. A corrected
analysis would have to impart a portion of the risk posed by
compounds such as vinyl chloride to the parent compounds. A
result more like that obtained for the “average” multiple contaminant
case would then be expected.
VI— 13
-------
7; REGULATORY FLEXIBILITY ANALYSIS AND PAPERWORK ANALYSIS
7.1 Regulatory Flexibility Analysis
The Regulatory Flexibility Act (RFA) was enacted on September
19, 1980 and requires all executive agencies to explicitly consider
small entities in their regulatory design and implementation
process. The purpose of RFA is to encourage regulatory agencies
to try and minimize the disproportionate burden that falls on
small entities. The three specific objectives of the RFA are
listed below:
1. To increase agencies’ awareness of their regul atory
impact on small entities; I
2. To compel agencies to explicitly analyze, explain,
and publish regulatory impacts on small entities; and
3. To encourage agencies to provide regulatory relief
to small entities while still accomplishing their statutory
mandates.
These objectives are accomplished through th& requirements of
regulatory flexibility analyses for all existing and proposed
regulations. If a regulation does not have a “significant”
impact on a substantial number 0-f small entities,.then the r egulatory
flexibility analysis will consist of a certification to that
effect.
Prior to conducting a regulatory flexibility analysis,
a regulatory agency such as EPA must define a small entity. (ft.
should be noted that, under the Safe Drinking Water Act, EPA ’s
Office of Drinking Water employs a different definition of small
water systems from that used here. The analyses presented in
this section are prepared only for compliance with the RFA.)
The RFA defines small entities as including small businesses,
organizations, and governments (PL 96_35L, Section 601(6)].
Small businesses are defined as any business which is independently
owned and operated and not dominant in its field (15 USC, Section
632). Small organizations are defined as any non—profit enterprise
which is independently owned and operated and is not dominant
in its field. Finally, small government entities are: defined
as those city, county, town, township, village, school district,
or special district governments serving a population of less
than 50,000 persons [ Regulatory Flexibility Act, PL 96—35U
Sections 601( 4) and 601(5)].
Community water systems can be divided into three ownership
categories for purpose of RFA analysis: 1) publicly owned,
2) investor owned, and 3) ancillary systems. Publicly owned
systems are those owned by governmental entities; investor owned
systems are privately owned; and ancillary systems are those
small systems that are ancillary to other enterprises such as
vu—i
-------
mobile home parks or hospitals. According to EPA’s 1980 Survey
of Operating and Financial Characteristics of Community Water
Systems, there are 26,1424 publicly owned community water systems.
Of this total, 98 percent serve fewer than 50,000 persons (see
Exhibit 7—1).
Investor owned water systems are firms primarily engaged
in production and distribution of water to consumers (SIC 149141).
These companies are considered to be small businesses if their
annual receipts are less than $3.5 million (Federal Register,
Vol. 149, No. 28, p. 5035). Applying the Consumer Price Index
for water and sewage maintenance for the period February 1980
to February 1981; to this figure, the upper limit for a small
water utility would be $2.14 million in 1980 dollars. EPA’s
Survey of Operating and Financial Characteristics of Community
Water Systems indicates that a population of 50,000 persons
is roughly the cut—off for systems generating revenues of $2.14
million. Revenues for investor owned water systems serving
25,000—50,000 persona averaged $1.97 million in 1980. F or investor
owned systems serving 50,000—75,000 persons, revenues in 1980
averaged about $3.16 million.
There is some question ai to whether investor owned water
utilities serving fewer than 50,000 persons qualify as smalL
businesses. Many of these utilities are not individually owned,
but are owned and controlled by large holding companies such
as American Water Works S.e.rvice. Co.and Gener .al. Mater Works..
In addition, every investor owned utility operates in a franchised
area and thus oonstitutes a natural monopoly. This raises the
question of whether domination in a limited geographic area
is the same as dominance in a field of enterprise. The Small.
Business Administration considers dominance to mean on a national.
basis; therefore, no individual water utility can be dominant
in the marketplace.
All ancillary community water systems serve fewer than
500 persons according to EPA’s 1980 survey. These could be
considered small entities; however, the main activity of the
enterprise may be sufficiently large to disqualify the organization
as a small entity. It is not possible to determine how many
of these systems constitute small entities because of the lack
of data.
7.1.1 PurDose of Regulation
Under the Safe Drinking Water Act, EPA is authorized to
set maximum contaminant levels (MCLs) for those contaminants
in drinking water which may have any adverse effect on the health
of persons. The purpose of regulations for VOCs is to limit.
human exposure to these chemicals via drinking water (i.e.,
both from inhalation and ingestion) and thereby reduce the health
risk posed by this class of chemicals. Regulations to control
VOCs in drinking water are likely to affect a number of small
water systems. It is estimated that one or more VOCs will be
VII—2
-------
EXHIBIT 7—1
NUMBER OF COMMUNITY WATER SYSTEMS BY POPULATION SERVED
Pooulation Served
3301
2c —cOO 5O1— OO O.0OO co,ooo+ TOTAL
Publicly Owned 8,932 11,51414 5,1455 1493 26,14214
Privately Owned 12,591 2,239 802 108 15,7140
Ancillary 16,g07 a ______ 0 16, OT
Total 38,1430 13,783 6,257 601 59,071
Source: EPA, Survey of Operating and Financial Characteristicsof Community
Water Svstem , 1982.
VII—3
-------
detected in over 6700 systems serving fewer than 50,000 persons
(See Exhibits 2—6 and 2—7).
7.1.2 Number of Systems Affected
Exhibit 7—2 shows the number of water systems that would
be affected if EPA sets MCLs at eight different levels. The
exhibit also shows a distribution by size category.
EPA guidelines on compliance with the Regulatory Flexibility
Act (April 12, 1983) indicate that, in general, a substantial
number of small entities is more than 20 percent of the total.
Exhibit 7—2 shows that a total of 600L systems serving 50,000
or fewer persons would be affected by a 5.0 ugh MCL. The total
population of systems in this size range is approximately 59,000.
Therefore, the maximum proportion of systems likely to be affected
is about 10 percent.
Even this number is overstated since the 600L number includes
those systems in which there is multiple occurrence of VOCs;
thus some double counting probably has occurred. A less stringent
MCL would affect even a smaller proportion of total systems.
Therefore, by the 20 percent rule, VOC regulations would- not
affect “a substantial number” of small water utilities.
7.1.3 Economic Im acts of VOC Regulations on Small Water
Svst ems
Exhibit 5 —Li shows annual cost of VOC removal ‘on a per system
basis (costs are for treatment only; see also section 7.2. 4
regarding monitoring costs). These data indicate that at an
MCL of 5.0 ug/l, very small systems (serving 25—500 people)
would incur an annual cost of about $ 48C0 per year. For those
systems serving 501—3300 persons, annual costs would amount
to about $20,’400 annually. Systems serving 3301—50,000 persons
would incur average annual costs of abou,t $58,700. Annual costs
at l-ess stringent MCLs would be about the same or marginally
less.
Exhibit 5—5 shows that for an MCL of 5.0 ugh, average
water production costs would increase by $0.5 4 per 1000 gallons
for affected systems in the 25—500 population served range.
For those systems in the range of 501—3300 population served,
average production costs would increase by about $0.29/l000
gallons. Systems serving 3301—50,000 persons would experience
average production costs of’ about $0.07/1000 gallons. These
costs would represent increases in production costs of’ about
nineteen percent in systems serving 25 to 500 persons, about
sixteen percent in systems serving 501—3300 persons, and six
percent in systems serving 3301—50,000 persons.
VII—4
-------
EXHIBIT 7—2
SYSTEMS REQUIRED TO TREAT AS A FUNCTION OF MC ! .
System Size Cateiorv (PoDulatthn Served )
3301—
2c—cu O cp1 — oo cO. 000 O.OOO+ TOTAL
MCL (ugh)
0.5 3L 83 1U06 989 126 6O0
1 22U6 906 575 68 37914
5 862 303 168 15 13147
10 506 182 100 7
20 332. 119 61 14 5.16
25 290 1O 4 52 3 1450
50 192 58 32 2 2914
100 127 145 20 1 193
VII—5
-------
7.2 Psoerwork Analysis
7.2.1 Paoerwork Reduction Act
Among the purposes of’ the Paperwork Reduction Act (Public
Law 96—511; 9k STAT 2812) are as follows:
o minimization of the federal paperwork burden for individuals,
small businesses, state and local governments, and other
persons; and
o minimization of’ the costs to the federal government
of collecting, maintaining, using, and disseminating
information.
Water utilities and state water supply agencies wi].l be required
to maintain records on monitoring of VOCs and report results
to the EPA; this is likely to be the largest component of paperwork
associated with establishment of’ federal VOC regu].atibns. The
Paperwork Reduction Act is intended to minimize the burden imposed
on utilities and states as they strive to protect the public
health by implementing the provisions of the SDWA.
7.2.2 Reoufrements of the Paoerwork Reduction Act -
EPA is required to submit to the Office of Management and
Budget COMB) proposed information collection_r.eque.s.ts.. ...EPA.
also must submit a copy of proposed rules containing an information
collection requirement. These proposed rules mu t be submitted
no later than publication of a notice of proposed rulemaking
in the Federal Register. When a final rule is published in
the Federal Register, EPA must explain how any information collection
requirements have been designed to be responsive to public comments.
0MB determines the necessity, practicality, and utility of the
information being requested, and if approval of the request
is made, 0MB will issue a control order.
Under the Safe Drinking Water Act, EPA is authorized to
regulate contaminants in drinking water to protect the public
health. VOCs are known to constitute a health risk. To determine
whether a specific water system exceeds an MCL for VOCs, or
to determine whether VOCs are present in drinking water- supplies,
EPA must require water systems to collect and analyze samples
•and report results to the relevant primacy agent (i.e., either
EPA or the states). In the case of VOCs, EPA, the states, water
utilities, and the public would use monitoring information to
determine whether VOCs are present. More importantly, this
monitoring data would allow appropriate action plans and removal
decisions to be made by affected utilities.
7.2.3 Number of Systems Affected
Exhibit 7—3 shows the number of water systems having one
or more VOCs present. There are about k070 very small systems
VII—6
-------
EXHIBIT 7—3
WATER SYSTEMS WITH VOLATILE ORGANIC COMPOUNDS
System Size (Pooulation Served )
3301—
2 — OO SO1— 0O co.ooo co 1 ooo+
Number of Systems 4067 1574 11Z$14 181
with VOCs’
Percentage of Systems 10.5 11.3 18.0 30.8
with VOCs
Total Systems in U.S.” 38,736 13,985 6,352. 58T
* One or more VOCs with at least one having a concentration greater than
or equal to 0.5 micrograms per liter. In the case of vinyl chloride
the minimum concentration is 1.0 microgram per liter, and in the case
of 1,1_dichloroethYlefle the minimun concentration is 0.2 micrograms
per liter.
** totals are different than those shown in Exhibit 7—1; the difference
is due to fluctuating inventory numbers.
VII—7
-------
having VOC occurrence, amounting to 10.5 percent of all systems
in this size category. The number of systems with VOC occurrence
decreases with system size, but the proportion increases due
to the smaller number of systems in larger size categories.
All systems probably would be required to monitor at least
once to determine whether VOCs are present. Those systems con-
taminated with VOCs, or which are thought to be vulnerable to
VOC contamination, would in all likelihood be required to monitor
on a regular basis.
7.2.4 Resoondent Burden (Monitoring Reauirements )
Respondent burden will largely be a function of monitoring
and reporting requirements. Compliance monitoring requirements
will be proposed by EPA for the purpose of determining if public
water systems are distributing drinking water that meets the
MCL. As a class of chemicals, VOCs are included in the second:
tier of the three ttered approach presented in the Phase II
ANPRM published on October 5, 1983 (48 FR 45502). The tiers
are as follows:
Tier I those which occur with sufficient frequency and which
are of sufficient concern to warrant national regulatiorr
(MCLs) and consistent monitoring and reporting.
Tier II those which are of sufficient concern to warrant’ nation I
regulation’ (MCLs) but which occur at limited frequency,
justifying flexible national minimUm- monitoring require-
ments to be applied y state authorities.
Tier III those which would not warrant development of a regulation
but for which non—regulatory health guidance could.
be provided to States or water systems.
EPA considered the following factors in. the development.
of VOC compliance monitoring alternatives for community water
systems: -
o differences between ground and surface water systems;
o the collection of samples which are representative of
consumer exposure;
o the economic burden associated with the sampling and
analytical costs; and
o the limited occurrence of VOCs and the need for states
to take an active role in requiring increased monitoring
over the federal minimums.
EPA has determined that the sampling and analytical costs are
reasonable and that there are sufficient analytical laboratories
capable of handling sample analyses provided the initial monitoring
VII -8
-------
requirements are phased—in over a period of several years.
Surface and ground waters have been considered separately because:
(1) the sources and mechanisms of contamination for these systems
are different, (2) the quality of ground waters tends to change
more slowly with time than does the quality of surface waters,
and (3) ground water contamination is usually a localized problem
confined to one or several wells within•a system.
For groundwater systems, sampling will be performed. at.
entry points to the distribution system since VOC contamination
of the water reaching the consumer is not expected to increase
within the distribution system. However, source monitoring
results may be used to decrease the number of samples taken
at entry points to the distribution system or to reduce the
frequency of monitoring for the determinatiOn of compliance
with the MCL. Determination of a reduction in the number of
samples or frequency of monitoring will be allowed at the option
of the primacy agency.
Cnmpliance Monitoring Requirements
The fundamental questions considered by EPA in developing.
proposed compliance monitoring requirements are as follows:
o How can monitoring regulations be developed to provide
states with an active, role such that resources are effic-
iently utilized?
o What minimum requirements should be set?
o What distinctions should be made between ground and
surface water systems?
— What locations for sampling?
— Number of samples per system?
— One—time monitoring or monitoring over a period of
time? Should minimum repeat frequencies be established?
What frequency and upon what basis?
— How much time should be allowed for public water systems
to complete the monitoring requirements?
— What is the cost of monitoring per system?
o What sampling requirements should be set?
o What follow—up actions may be needed to assist the public
water systems and the states when positives are reported?
— Follow—up confirmation sampling?
— Health and treatment advisories?
o What reporting and public notice requirements should
be set?
VII—9
-------
EPA believes that all Community Water Systems (CWS) should
conduct at least one initial round of monitoring to determine
the extent of contamination of water supplies and to provide
maximum consumer protection. EPA also believes that there should
be minimum requirements for repeat sampling since the vulnerability
of a system to VOC contamination may change with changing land
and water use and waste disposal practices. The repeat sampling
requirements should reflect the potential for contamination
of the system (i.e., the most vulnerable systems should monitor
the most frequently). The states should recertify the vulnerabilitY
status of each system on an annual basis. Systems should notify
the state whenever a significant change takes place that could
affect the vulnerability of the system (e.g., change in water
source, new VOC—based industry nearby, or a positive VOC analysis).
Several approaches to monitoring requirements have been
considered by EPA. Three specific options are outlined below.
In each option, requirements are displayed for (1) an initial
round of monitoring, and (2) repeat monitoring. rn addition,
different requirements are set within each option for ground
water systems (about 145,000) and surface water systems (about
15,000). The primary differences between the options relate
to the extent of specific sampling requirements and the provision
of state discretion. In each option, monitoring for vinyl chloride
would not be required for all systems. Ground water systems
would be required to analyze for vinyl chloride only when other
chlorinated 2—carbon VOCs -(TCE, PCEI, 1.,2 DCA or i.,1 1 .1—TC’A) had
been detected and no requirements would be set for vinyl chloride
monitoring in surface water systems. This is because EPA has
concluded that the most likely explanation for vinyl chloride
detection in ground waters is from j.. situ transformation.
In each option, monitoring requirements are proposed to
- be phased in depending upon the size of the systems. Phasing
in requirements over four or five years allows public water
systems and states sufficient time to efficiently allocate the
necessary resources to conduct the monitoring. Systems that
are most vulnerable to VOC contamination should sample first;
while EPA studies have not shown a clear relationship between
potential sources of contamination and actual VOC contamination
that could be used to pinpoint specific systems that would be
vulnerable to VOC contamination, the Ground Water Supply Survey
(GWSS) found the best correlation was between the size of systems
and VOC contamination. Therefore, monitoring requirements are
proposed to be phased in by system size with the largest systems
sampling first.
EPA is proposing that Option 2 be selected as the minimum
federally enforceable monitoring requirement. Option 2 provides
for reasonable minimum federaL. requirements while also providing
for state discretion in their application. While the requirements
are phased in by size of system, states will be encouraged to
sample vulnerable systems as early as possible.
vu—b
-------
Option 1 . This option would require all sy5tems to monitor
at least once over a four year period. The federally mandated
monitoring requirements would be relatively stringent under
this option. The monitoring requirements would be phased—in
based on the size of the population served by the CWS, as follows:
System size C moleted by
(population served)
>10,000 End of 1 year
3300—10,000 End of 2 years
<3300 End of 1 years
o Ground water systems would be required to sample quarterly
for one year at each entry point to the distribution
system. Confirmation of positive samples also would
be required.
o Surface water systems would be required to sample quarterly
for one year. The minimum number of samples would be
one sample per source in the distribution system. -
All systems (i.e-., ground and surface) would sample quarterly
for- one year, and would be required to resample any positLves.
Costs are based on an assumption that the rate of positive samples
will be. 20 percent.
Repeat monitoring would be based on prior monitoring results
and the vulnerability of the system to VOC contamination. The
repeat monitoring frequency would be as follows: -
Status* - Frequency
VOCs not detected and not Repeat in 5 years**
vulnerable
VOCs not detected and vulnerable Repeat in 3 years*C
VOCs detected Monthly
•States would annually recertify the vulnerability status of
systems.
**Surface water systems sample during four consecutive quarters.
The estimated costs of this option at $150 per sample are
as follows:
Initial round $25 million/year (average) for 14 years
Repeat monitoring $63.7 million/.year
VII—1]
-------
Optinn 2 . The federally mandated monitoring requirements
would be less stringent under this option than in Option 1.
All systems would monitor at least once over a four- year period
but fewer samples would be required than in Option 1. Implementation
of the monitoring program would be the same as in Option 1,
phased in based on the size of the population served by the CWS:
System size ComDleted by
(population served)
>10,000 End of 1 year
3300—10,000 End of 2 years
<3300 End of 1 years
o Ground water systems’ would be required to sample at
each entry point to the distribution system. The minimum
number of samples for ground water systems wouLd. be-
one sample per entry point to the distribution system,
per quarter for one year. If no VOCs are detected in
the initial sample and the system is not considered
vulnerable to contamination, states would have discretion
to reduce the sampling requirements to that- initial.
sample.
o Surface water systems would sample at points representative
of each source. The minimum number of samples. would
be one samp.le. .per.. source, per. quarte-r” for one- year.
States would have discretion on requiring conf irmati’on-” - - - ‘
samples for positive results.
All systems would be required to conduct repeat monitoring
except for surface water systems that were not vulnerable and
did not detect any VOCs in the first round of sampling. The
frequency of such monitoring would be based on prior monitoring,
results and the vulnerability of the system to VOC contamination.
The monitoring frequency would be as follows:
Status* Ground Water Surface Water** -
VOCs not detected Repeat in 5 years State discretion
and not vulnerable
VOCs not detected Repeat in 3 years Repeat in 3 years**
and vulnerable -
VOCs detected Quarterly Quarterly
‘States would annually recertify the vulnerability status of
systems. -
‘Surface water systems sample during four consecutive quarters.
Vu—i 2
-------
The estimated costs of this option at $150 per sample are
as follows:
Initial round $9.3 million/year (average) over 14 years
Repeat monitoring $17.14 million/year
Option . More state discretion is provided under this
option than the previous options. Al]. ground water systems
would monitor at least once over a five year period. Monitoring
of surface water systems would be at state discretion based
upon vulnerability. The monitoring program would be phased—in
based on the size of the population served by the CWS as described
in the previous options except that systems serving less than
3300 people would have five years from the date of promulgation
to complete the initial monitoring, as follows:
Size of System Com lete by
(population served)
>io,ooo End of 1 year
3300—10000 End of 2 years
<3300 End of 5 years
Ground water systems would be required to sample. in the
distribution system at points representative of each well at
least once during the initial monitoring period; if VOCs were
detected, - three add-i.tional quarterly samples wQuld be required.-
States would have discretion on requiring confirmation samples..
Specific requirements for surface water systems would be up
to state discretion based upon a vulnerability assessment.
Repeat compliance monitoring requirements would only be for
those systems that detected VOCs in the initial monitoring round.
States would have discretion in the frequency of monitoring
for those systems where VOCs were not found. 1 The monitoring
frequency would be as follows:
Status Ground Water Surface Water
VOCs not detected State discretion State discretion
and not vu3.nerable
VOCs not detected State discretion State discretion
and vulnerable
VOCs detected Annually State discretion
VOCs > MCL Quarterly State discretion
‘States would annually recertify the vulnerability status of
systems. -
VII—13
-------
The estimated costs of this option at $150 per sample are
as follows:
Initial round $3.8 million/year (average) 5 years
Repeat monitoring $2.9 million/year
EPA is proposing Option 2. This option provides minimum
monitoring requirements for all systems. For those systems
using both ground and surface waters, the monitoring requirements
would apply individually to each type of source. The monitoring
frequency includes sampling for four consecutive quarters during
the monitoring period for surface water systems since variability
of surface waters is expected to be influenced more by seasonal
and weather conditions. Ground water systems would be required
to collect four quarterly samples unless the first sample did
not detect VOCs and the system was not considered vulnerable;
in these cases states could waive the additional three samples.
States are expected to have a major role in implementation of
these monitoring requirements; assessments of vulnerability,
extent of contamination, and individual system factors will
determine the amount of monitoring properly conducted at each-
system.
Monitoring for Unregulated Contaminants
Because similar analytical procedures for the nine V0C s
can also measure numerous other VOCs at relatively small. ad.dLtionaL.
cost, monitoring regulations will be proposed for other V0Cs .
Monitoring for most pesticides and other SOCs is more costly
and additional time is needed to develop analytical methods.
and baseline data (i.e., which pesticides should be monitored
and in what locations) such that directed monitoring requirements
can be developed (i.e., systems vulnerable to contamination
would be required to monitor). The National Pesticides Survey,
currently being conducted by the Agency, will provide much of
thispreliminiarydata , Threeoptionswereconsidered forunregulated
•VOCs, similar to the VOC compliance monitoring requirements;
the options presented range from an extensive federally mandated
specific monitoring program to a monitoring program whose specifics
(e.g., repeat monitoring frequencies) would be largely determined
by the states. . The middle option will be proposed by EPA; a
monitoring program providing reasonable minimum federal requirements
with provisions for state discretion in their application.
Insofar as possible, the monitori.ng requirements for unregulated
VOCs will be similar to those proposed for compliance monitoring
under the NPDWR so that systems will be allowed to use the same
samples for analysis of both the unregulated VOCs and the nine
VOCs for which MCLs are proposed. In addition, provisions for
“grandfathering” previous data of acceptable quality are included.
The three options for minimum federally mandated monitoring
requirements outlined below generally correspond to the three
options described earlier for compliance monitoring for the
nine VOCs; Option 2 will be proposed by EPA.
VII .-14
-------
Option 1 . This option proposes relatively stringent monitoring
requirements and includes minimum repeat monitoring for all
systems. The monitoring program will be phased—in over a four
year period based on the size of the population served by the
CWS in a similar manner as described under Option 1 of the proposed
compliance monitoring requirements. Ground water systems would
be required to sample once at the well head rather than in the
distribution system. Surface water systems would be required
to sample quarterly for one year in the distribution system
at points representative of each source. All systems would
be required to resample positive samples. All systems would
be required to repeat monitoring every 10 years.
ODtion 2 . This option is the same as Option 1 above except
that it provides for state discretion on resampling positive
results and on repeat monitoring requirements. Exhibit 7—
presents the proposed requirements.
EXHIBIT 7_Z
PROPOSED MONITORING FOR UNREGULATED VOCs
Initial Monitoring
o All systems monitor once over four years
o Requirements are by system size:
Size of System Comolete by End of
>10,000 1 year
3300—10,000 2 years
(3300 1 years
o Ground Water Systems: One sample per well at the well head
o Surface Water Systems: Quarterly samples per each source’
for one year at points in distribution
system representative of each source
Reneat Mon1torin
o State discretion for repeat sampling; dependent upon vulnerability
and results of first round of monitoring.
Option . Under this option all systems would monitor
once over a five year period. Monitoring would be phased—in
by the size of population served by the CWS as described under
Option 3 of the compliance monitoring requirements. Ground water
VII—15
-------
systems would be required to sample only 25 percent of their
wells and the sampling would be done at the well head. The
State would have discretion on whether to require confirmation
samples. Also, States would have discretion on whether to require
surface water systems to monitor based upon a vulnerability
assessment. There is no repeat monitoring requirement under
this option.
EPA considered two approaches in the selection. of unregulated
VOCs to be included in a monitoring regulation. The first and.
most comprehensive approach is to include approximately 60 VOCs
that are detected using purge and trap gas chromatography tech-
niques. The second approach considered is to include only those
VOCs which may be of concern because of their possible occurrence
in drinking water supplies and potential adverse health effects.
EPA believes that the monitoring efforts should be limited to
the chemicals that have been detected or are likely to occur
in drinking water and that may pose an adverse health risk.
EPA is proposing that the 50 VOCs listed in Exhibit 7—5 beconsidered
for a monitoring regulation as part of this proposal. The compounds
have been selected based on present regulatory interest and
available occurrence information. The compounds listed include:
o four trihalomethanes in the November 29, 19T9 Federal
Resister ;
o additiona.l. .VOCs. be.in.g co.nsidered for later phases of
the .Revised Regulations; and
o VOCs not included above but detected in the Ground Water
Supply Survey and various Federal and State surveys;
o VOCs that have potenti.al for occurrence in drinking
water; VOCs detected in waste waters, surface or ground
waters or have widespread dispersive use pattern5 and
high production.
The VOCs in Exhibit 7—5 can be measured in a single analysis
by GC/MS or by four separate analyses using GC. Estimated costs
are $150 to $200 for the GC/MS and about $100 per GC analysis.
The four analyses include:
A Purgeable halogenated hydrocarbons
B Purgeable aromatics
C Highly volatile substances
D Low sensitivity (or low limits of detection required)
Analysis C of the highly volatile substances can be incorporated
Into the procedures for the purgeable hydrocarbons through minor
adaptations (e.g., c ian e the trapping device). Analysis D
(EDB and DBCP) is estimated to cost an additional $50.00 per
sample. Monitoring for EDB and DBCP will only be required for
systems considered to be vulnerable to EDE or DBCP contamination.
VII— 16
-------
EXHIBIT 7—5
UNREGULATED VOCs CONSIDERED FOR MONITORING
A Chloroform
A Bromodichioromethane
A Chlorodibromomethane
A Bromoform
A trans—i ,2—Dichloroethylene
A Chlorobenzene
A m—Dichlorobenzene
A Dichioromethane
A cis—1 ,2—Dichloroethylene
A o—Dichlorobenzene
A 1 ,2,L _Trichlorobenzene
C F].uorotrichloromethane
C Dichiorodifluoromethane
A Dibromomethane
D 1,2—Dibromoethane (EDB)
D i ,2 —Dibromo —3 —ch].oroproPafle (DBCP)
B To].uene
B p—Xylene
B o—Xylene
B m—Xylene
A 1,1 —Dichloroethane
A 1 ,2 —Dichloropropane
A 1,1 ,2,2 —Tetrach].oroethane
B Ethylbenzene
A 1 ,3 —dichloroproparie
B Styrene
A Bromobenzene
NOTE:
A Chioromethane
A Bromomethane
A Bromochioromethane
A 1 ,2,3—Trichloropropane
B 1 ,2,3 —Trichlorobenzene
B n—Propylbenzene
A 1 , 1 , 1 ,2 —Tetrachloroethane
A Chioroethane
A 1,1 ,2—Trich].oroethane
A Pentachioroethane
A bis—2 —Chloroisopropy]. ether
A sec—Dichioropropane
B 1 ,2,A4_Trimethylbenzene
B n—Butylbenzene
B Napthalene
B hexachiorobutadiene
B o—Chlorotoluene
B p —Chlorotoluene
B 1,3,5 —Trimethylbenzene
B p —Cymene
A 1 ,1 —Dichloropropane
B iso —Propylbenzene
B tert—Butylbenzene
B sec—Butylbenzene
A — Can be analyzed using the Purgeable Halogenated Hydrocarbon
Method
B — Can be analyzed using the Purgeable Aromatic Hydrocarbon Method
C — Highly volatile substances
D — Low sensitivities (requires a low limit of detection)
VII—17
-------
Analyses are to be conducted in certified laboratories
if such data are to be used for compliance with MCLs for the
nine VOCs in this proposal. Because the monitoring for unregulated
contaminants will be required before full certification programs
can be implemented, interim certification will be provided to
those laboratories presently certified for trihalomethane analyses
and, those that analyze performance evaluation samples for additional
VOCs within acceptable limits.
Estimated costs for the three options are shown in Exhibit 7—6.
EXHIBIT 7-6
COSTS FOR MONITORING UNREGUL&TED VOC3
Option 1 ODtion 2 Option
Initial Round $2.7 $2.3 $0.5
million million million
Repeat Monitoring $2.7 0 0
million
VII—18
-------
8. SUMMARY OF COSTS AND BENEFITS
8.1 Introdu ticn
This chapter summarizes the results of the cost and benefit
analyses (Chapters L and 5) performed at the aggregate or national
level and of the net benefits analysis (Chapter 6) performed
at the level of individual water systems of varying sizes.
Together, these analyses provide a complete evaluation of the
regulatory alternatives.
8.2 Probable Actions. Health Benefits and Costs
As discussed in Chapter 3, alternative EPA actions fall
into three categories: 1) no action (other than existing health
advisories); 2) monitoring and reporting requirements (in addition
to health advisories); and, 3) MCLs, RMCLs, and monitoring and
reporting requirements. MCLs are the most probable action based
on the results of benefit and cost analyses.
The aggregate results indicate that such regulatory action
is warranted due to the nature and extent of VOC occurrence
and the carcinogenic potential of most of the chemicals under
study. At present, the baseline level of exposure to VOCs in
drinking water is estimated to be responsible for approximately
50 cases of cancer per year in the United States (see page IV—2).
The extent to which the health effects of VOC exposure
can be reduced by regulatory action will vary depending upon
the levels chosen for the MCLs. There are, of course, trade—offs
between the level of stringency required in such regulation
and the costs incurred. Exhibits 8—1 and 8—2 present summaries
of the aggregate cost and benefit estimates in a manner which
allows a comparison of the total national cost on an annual.
basis versus the number of cases of cancer avoided.
The diagrams presented in Exhibits 8—1 and 8—2 are designed
to take explicit account of the uncertainties involved in making
such estimates at a national level. The sources and magnitudes
of the uncertainties in both the benefit and the cost analyses
were analyzed in detail in Chapters L and 5. In these Exhibits,
the ratio of the standard error in the cost estimate to that
of the benefit estimate for each MCL was entered into a table
of circular error probabilities to obtain the radii of the circles
that include respectively, 68 percent and 95 percent of the
volume under the joint probability surface. Thus, the “true”
values for each MCL t s cost and benefit fall inside the circles
with either 68 percent or 95 percent probability.
The diagrams in Exhibits 8—1 and 8—2 display two clusters
of circles: one relatively high cost cluster representing MCLs
of 0.5 and 1.0 ugh and another relatively low cost cluster
representing MCLs of 5.0, 10.0, and 20.0 ugh. The obvious
VIII— ’
-------
EXHIBIT 8—i
68 CONFIDENCE PLOT OF CANCER CASES AVOIDED
AS A FUNCTION OF ANNUALIZED COST
(True values are expected to
with a probability
fall within the circlea
of 68 percent.)
4)
4
I-I
IS)
I—4
U
II
4)
U
C
4)
U
.1
4)
C
C
4
Annualized Cost of Compliance (a4llions ot 1983 do11a a)
100
80
60
40
20
20 40 60 8Q 100 20 140 160 180
200
(MCLs shown iq ugh)
-------
EXHIBIT 8—2
95 CONFIDENCE PLOT OF CANCER CASES AVOIDED
AS A FUNCTION OF ANNUALIZED COST
(True values are expected to fall within the circles
with a probability of 95 percent.)
40 60 80 100 120 140 160
Annualized Cost of Compliance (millions of 1983 dollars)
(IICLs sho in ugh)
100
I-I
•0
, 80
V
.4
0
4
60
U
Ii
a ’
(1
C
o 40
‘a
C
4
20
4- ÷
5
‘ 10
20
20
180
200
-------
implication of this diagram is that the high cost cluster implies
costs that are four to five times as high as the low cost cluster
whereas the difference in benefits is only on the order of a
factor of two. On the basis of aggregate costs and benefits,
the lower cost cluster appears to be a better bargain.
Results of aggregate analysis must be supplemented with
analysis of costs and benefits at the level of individual water
systems. The net benefits analysis in Chapter 6 performed such
analysis. The system level analysis showed that in terms of
the marginal cost per cancer case avoided, MCLs in the vicinity
of 1.0 and 0.5 ugh are very expensive for VOCs that pose low
to moderate cancer risks. Net benefits of such levels of regulation
would be positive only for the relatively high cancer risk VOCs
such as vinyl chloride, carbon tetrachioride, and 1,1—dichioro—
ethylene.
The system level analysis also showed that MCLs in the
vicinity of 5.0, 10.0, and 20.0 ug/l do not achieve positive
net benefits for the VOCs posing only low and moderate degrees
of cancer risk. Control of these “weaker” carcinogens to these
levels does not produce benefit estimates high enough to equal
the high initial fixed cost of treatment equipment.
However, nearly 50 percent of all VOC occurrence is multiple
occurrence where several VOCs are present together. Vinyl chloride
arid 1,1—dichioroethylene occur with particular frequency. in
such mixtures making the average risk posed by these mixtures
relatively high. There is suspicion that vinyl chloride may
exist in some of these circumstances as a by—product of the
biological breakdown of the other VOCs. The presence of higher
risk mixtures may greatly increase the net benefits of controlling
all VOCs.
Exhibit 8—3 presents other summary statistics characterizing
the impacts and benefits of VOC control.
VIII—4
-------
EXHIBIT 8—3
SUMMARY OF IMPACTS OF THE REGULATORY OPTIONS
Regulatory Options
I u /l 5 ugl]. 10 ugL] .
Systems Impacted 3,800 1,300 800
National Cost of’ Control
Total ($ millions) $1,300 $ 280 $ 150
Annual ($ millions) 100 21 11
National Cost of Monitoring
Compliance ($ millions) $ 1L
Unregulated Contaminants 3
Annual Cost per Family ($/year)
System Size (people served):
Very Small (25—500) $ 96 $ 91 $ 90
Small (501—3,3001 147 L 1 L2
Medium (3,301—50,000) 12 12 - 11
Large (over 50,000) 8 3 1
Typical Rate Increases ($/1000 gal)
Very Small (25—500) $ 0.58 $ 0.5 4 $ 0.58
Small (501—3,300) 0.33 0.29 0.29
Medium (3,301—50,000) 0.07 0.07 0.07
Large (over 50,000) 0.0k 0.02 0.01
Annual- Cancer Cases Avoided
Total 42 32 31
Attributable to Vinyl Chloride 37 29 27
Average Cost/Case Avoided
(mill ions)
Very Small (25—500) $ 10 $ 5 $
Small (501—3,300) 7 3 2
Medium (3,301—50,000) 2 0.6 0. 4
Large (over 50,000) 2 0.2 0.0k
VIII—5
-------
APPENDIX A -— HEALTH EFFECTS OF INDIVIDUAL CHEMICALS
BENZENE
Benzene is a chemical that effects multisystems, but the
hematopoietic and immune systems appear to be most sensitive.
Maltoni has published several papers that demonstrate that benzene
causes leukemia as well as hard tissue cancers in animals.
Leukemia has been associated with exposure to benzene in humans.
A suggested Adjusted ADI was calculated based on data from
a subchronic gavage study in rats in which leucopenia was observed
at specific dose levels. A value of 0.025 mg/i was calculated
using a no—observed—adverse—effect level of 1 mg/kg and an uncer-
tainty factor of 1,000, since a study with less than lifetime
exposure was used.
The EPA’s Carcinogen Assessment Group (CAG) calculated
projected excess cancer estimates with the linearized non.threshold
multistage model and data from an epidemiologic study of workers
exposed to benzene vapors on their jobs. An increased risk
of one excess cancer per 1,000,000 people corresponds to lifetime
exposure to a benzene level of 0.67 ugh in drinking water.
Benzene is placed in category 1 by the IARC and category
A by the EPA, because the strength of evidence for its carcinogeni—
city is sufficient in humans with supportive evidence in animals.
CARBON TETRACHLORIDE
The main toxic effects of carbon tetrachloride in humans
occur in the liver, kidney, and lung. Toxic effects from carbon
tetrachioride exposure in animals include kidney and lung damage
and fatty infiltration and necrosis in the.liver. Carbon tetra—
chloride has been shown to be carcinogenic in rats, mice and
hamsters.
A suggested AdjUsted ADI was calculated based on data from
a 12—week gavage study in rats in which liver toxicity was evident
at specific dose levels. A value of 0.025 mg/i was calculated
using a no—observed—adverse—effect level of 1 mg/kg and an uncer-
tainty factor of 1,000, since a study with less than lifetime
exposure was used.
The EPA’s Office of Health and Environmental Assessment
calculated projected excess cancer estimates with the linearized
non—threshold multistage model and the geometric mean of four
cancer studies in animals (Nd, 1976 — mice, Nd, 1976 — rats,
Edwards et al., 19L 2 — mice, Della Porta et al., 1961—hamsters).
A—i
-------
An increased risk of one excess cancer per 1,000,000 people
corresponds to lifetime exposure to a carbon tetrachioride level
of 0.27 ugh in drinking water.
Carbon tetrachioride has been classified as a 2B and B2
carcinogen by the IARC and EPA, respectively, based on conclusions
of sufficient evidence in animals and inadequate evidence in
humans.
p — DI C H LO ROB EN Z E Nt
In animals, p_dichlorObeflZefle has induced liver and kidney
damage, porphyria, pulmonary edema and congestion, and. splenic
weight changes. In humans, exposure to dichlorober3ZefleS has
been reported to result in anorexia, nausea, yellow atrophy
of the liver, and blood dyscrasias.
A suggested Adjusted ADI was calculated based on toxicity
data from a subchro iC gavage study in rats. A value of 3.75
mg/i was. calculated using a no_observed_adverse—effect level
of 150 mg/kg and an uncertainty factor of 1,000, since a study
with less than lifetime exposure was used.
Because there is no evidence for carcinogenic effects,
a strength of evidence judgement and a risk assessment are not
applicable for p_dichlOrObenZefle.
1 2 _ DICHLOROETHANE
NoncarcinOgefliC effects observed in animals and humans
include liver and kidney damage, central nervous system depression,
gastrointestinal distress, adrenal and lung effects, and circulatory
dist rbanCe5. 1,2_dichloroethafle has been shown to significantly
increase tumor incidences at several sites in both rats and
mice when administered by gavage, but not following inhalation
exposure.
A series of inhalation studies where several animal species
were exposed to 1,2_dichiorOethane were selected for calculation
of an Adjusted ADI. Several toxic effects were observed in
these studies. A value of 0.26 mg/i was calculated u . sing a
no_observed_adVerSe effeCt level of 100 ppm (405 mg/mfl and
an uncertainty factor of 1,000, since a study with less than
lifetime exposure was used.
The CAG calculated projected excess cancer estimates with
the linearized non—threshold multistage model and NCI carcinogeniCitY
bioassay data. An increased risk of one excess cancer per 1,000,000
people corresponds to lifetime exposure to a 1,2_dichioroethafle
level of 0.5 ugh in drinking water.
1,2_DichiorOethafle has been categorized as a group 2B and
Group B2 carcinogen by the IARC and EPA, respectively, based
on conclusions of sufficient evidence in animals and inadequate
evidence in humans.
A—2
-------
1 . 1 —DICHLOROETHYLENE (VINYLIDENE CHLORIDE )
Toxic effects of 1,1—dichioroethylene in animals include
liver and kidney damage, central nervous system depression,
and sensitization of the heart. Although 1,1—dichioroethylene
has been reported to be a renal carcinogen in one mouse study
and positive in mouse skin as an initiator with phorbol esters
as the promoter, most of the carcinogenicity studies have failed
to demonstrate significant carcinogenic activity for this agent.
A suggested Adjusted ADI was calculated based on data from
a chronic drinking water study in rats in which toxic liver
effects were found with specific dose levels. A value of 0.35
mg/i was calculated using a no_observed—adverse—effect level
of 10 mg/kg and an uncertainty factor of 100.
The CAG calculated projected excess cancer estimates with
the linearized non—threshold multistage model and renal adenocar—
cinoma data in a carcinogenicity study with mice exposed to
1,1 —dichloroethylefle by inhalation. An increased risk of exposure
to a i,1—dichloroethylene level of 0.24 ugh in drinking water.
However, the EPA’s Science Advisory Board has questioned the
validity of the study on which this risk assessment is based.
1,1—Dichloroethylefle is placed in category 3 by the IARC
and category C by the EPA, because the strength of evidence
for its carcinogenicity ..islimited in animals and inadequate
in humans.
TE TR AC HL OR ETHYL NE
Principal noncarcinogenic effects of tetrachioroethylerte
in humans and animals include liver and kidney damage, congestion
and edema in lungs, hyperemia of the kidney and lungs, and centra]
nervous system depression. TetrachloroethYlene is positive
as a liver carcinogen given to mice by gavage.
The Adjusted ADI has been revised as a result of public
comments. In lieu of what is in the June 12, 198L , Federal Register
proposal (0.085 mg/i), the Adjusted ADI is presently 0.68 mg 1
based on a no_observed adverSe—effeCt level of 70 ppm (475 mg/mfl
in a subchronic inhalation toxicity study in rats and an uncertainty
factor of 1,000, since a study with less than lifetime exposure
is used.
The CAG calculated projected excess cancer estimates with
the linearized non—threshold multistage model and liver carcinogefli—
city data for mice given tetrachloroethYlene by gavage in a
NCI bioassay. An increased risk of one excess per 1,000,000
people corresponds to lifetip e exposure to a tetrachloroethYlefle
level of 1 ugh in drinking water.
Tetrachloroethylefle has been classified as a Group 3 and
a Group C carcinogen by the IARC and EPA, respectively, based
A—3
-------
on conclusions of limited evidence in animals and inadequate
evidence in humans.
1 .1 . i—TRICHLOROETHANE (METHYL CHLOROFORM )
The principal toxic effects of 1,1,1—trichloroethane at
high doses in animals and humans are depression of the central
nervous system, increase in liver weight and cardiovascular
changes.
Liver toxicity is the most sensitive end—point with respect
to adverse health effects. An inhalation study (McNutt et al.,
1975) which examines exposure of mice to 1,1,1—trichioroethane
is used to calculate a suggested Adjusted ADI of 1.0 mg/i.
This study demonstrates changes in the livers of mice at dose
levels.
Two animal carcinogenic bioassays by the National Toxicological
Program have been conducted in rats and mice (1977; 1983).
In the earlier bioassay, 3 percent of the animals had survived
to the end of the experiment (chronic murine pneumonia, etc.).
Because of this, it was concluded that carcinogenicity could
not be determined from this study. A repeat carcinogenesis
bioassay of 1,1,1—trichloroetharie was conducted in which doses
of 3,000 or 1,500 mg/kg were administered by gavage to both
sexes of mice, and rats were given doses of 750 or 375 mg/kg
in corn oil. In the preliminary report of this study, 1,1,1—tr i—
chioroethane was carcinogenic in both male and female mice showing
an increased incidence of’ hepatocellular carcinomas but not
in rats; however, these intitial results have been questioned.
IARC has not classified i,1,1—trichloroethane for carcinogenic
potential and EPA cancer guidelines has classified 1,1,1—trichioro—
ethane Group D (inadequate date to classify).
TRICHLOROETHYLENE
Trichioroethylene has been shown to exhibit noncarcinogenic
effects in humans as well as in dogs, rabbits, guinea pigs,
rats and mice. Major effects include liver and kidney damage,
central nervous system effects and myocardial contractility. Tn—
chioroethylene was reported as carcinogenic in mice.
A suggested Adjusted ADI was calculated based on data from
a 1L _week inhalation study in rats in which elevated liver weights
were observed with each exposure level. A value of 0.257 mg/i
was calculated using a lowest observed—adver5e—effect level
of 55 ppm (300 mg/rn 3 ) and an uncertainty factor of 1,000, since
a no_observed adver5e—effeCt level was not identified and a
study with less than lifetime exposure was used.
The CAG calculated projected excess cancer estimates with
the linearized non—threshold multistage model and liver carcinogeni—
city data for mice given tnichioroethylerle by gavage in a NCI
A—4
-------
bioassay. An increased risk of one excess cancer per 1,000,000
people corresponds to lifetime exposure to a trichioroethylene
level of 1.8 ugh in drinking water.
Trichioroethylene has been classified as a Group 3 carcinogen
by the IARC based on conclusions of limited evidence in animals
and inadequate evidence in humans. On the- basis of evidence
obtained after the IARC review, the EPA has raised the strength
of evidence in animals to sufficient which consequently places
1 trichloroethylene in the B2 category in the EPA classification
scheme.
VINYL CHLORIDE
Vinyl chloride is toxic in lungs, kidneys and liver in
animals. Exposure to vinyl chloride has been reported to induce
acroosteolysis, pulmonary insufficiency, and disturbances in
the cardiovascular, gastrointestinal, and central nervous systems
in humans, liver angiosarcomas and tumors in the brain, lung
and hematopoietic systems have been associated with vinyl chloride
exposure in humans. Vinyl chloride is carcinogenic in rats,
mice and hamsters, with major tumor types including liver angio—
sarcomas, hepatocellular carcinomas, brain tumors, and lung
tumors.
The Adjusted ADI has been revised in light of a new carcino—
genicity study on vinyl chloride in rats. In lieu of what is.
in the June 12, i98 4, Federal Register proposal- (-0.06- mg/i)-,
the Adjusted ADI is currently O.OL 6 mg based on a no—observed—
adverse—effect level of 0.13 mg/kg in a carcinogenicitY study
in which liver toxicity was found at specific dose levels in
rats given vinyl chloride monomer in the diet and an uncertainty
factor of 100.-
The CAG calculated projected excess cancer estimates with
the linearized non—threshold multistage model and carcinogenicitY
data for rats given vinyl chloride monomer in the diet. An
increased risk of one excess cancer per 1,000,000 people corresponds
to lifetime exposure to a vinyl chloride level of 0.015 ugh
in drinking water. The CAG is reviewing this risk assessment
in view of a new carcinogenicity study with vinyl chloride in
rats.
A—5
-------
EXHIBIT B—i
?IOIMILZTT IT5T 0? 93114 1120 10) W230 92113 1J U 1UT 00 2T)ATI0) IWCTI20 U41 1S450120 301 £ CIVIl
(Cr.i10 ru.)
00IIMUMIVII T7i O00l1VTt20I. 1.1 .1.TLZ .OI0tTII411. ?tl1 1á001111tLl . 015203 11TI MIIX .
I- I • D&o31.r.oelp1. • I • 1—O& 0011 S. M)Oflt . • IVPICS& ISILZ&P%Z OC5IUL
1 S 1400 3315u40
e..MnU Isø)
( pb)
I
054
P.0104
T r
U ,...d
Mr
I V.?
rsiL
I
40
lIelg
I 1 ..
I M11
l0s1 4— souls
10046 sgu.
103as I
12000110
V.o U 0.3— ) 0.3
(23-SOOp o ) 3
20
30
.54
0
0
0
.03
0
0
0
0
0
0
.03
0
0
0
0
0
0
0
0
0
0
0
0
0
.03
0
I
I
0
S
0
0
3.0
1.0
1.0
2 0 04
3
30
30
.3)
.53
0
0
.10
.03
0
0
5
0
0
0.
. 45
0
0__
0
0
0
0
0
0
0
0
I
0
.03
I
S
.30
JO
S
S
0
0
1.0
1.0
30—30 04
3
35
30
.jS
.33
.3)
0
.40
.10
.43
0
0
0
S
0
0
S
.41
0
0
0
S
0
0
0
S
0
0
S
S
0
.0 5
. 03
. 03
0
.30
.30
.30
0
0
0
0
1.0
I 30. 04
I 3
20
I 30
.5)
43’
.5)
.33
.10
.10
.03
.01
S
0
0
0
0
0
.01
.01
0
0
0
0
0
5
0
0__
0
0
0
0
.43
.43
.0)
.03
.30
.30
40
3O
0
0
0
0
I 9.3—) 04
I(S01—3 )00p.,) 3
30
.10
0
0
.40
0
.0
0
0
0
.03
0
0
0
0
0
p
0
0
p
0
0
•Q3
0.
.10
0
0
p
3.0
3.0-
30 —
20 04
3
20
0
.33
.10
0
0
.40
.40
0
0
0
0
0
0
—T
.03
0
0
p
0
0
0
o
0
0
0
O
0
0
0
.03
.03
S
0
.40
.40
0
1.0
0
0
1.0
30
20-30 04
3
30
30
0
.10
.10
.10
0
0
.40
.40
0
0
0
0
0
0
0
0
.03
.03
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.0 )
.0)
.03
0
0
.6)
.40
.40
0
1.0 I
0
0
0 I
1.0
30. 0.3
3
30
30
.10 .40
•10 .40
.10 .40
.10 1 .48
0
0
0
0
0
. 43
.03
.01
0
0
0
0
0
S
I I
0 1
0
0
5
0___
.03
.0)
.0)
.01
.13
.40
.40
.10
0 I
0
0.
0
I d 1l. 0.5—) 04
(3301—10.000 S
pop) 20
30
.10 .20
0 0
0 0
0 0
.10
0 0 ,
0 0
0 ( 0
0 .10 1 0
0 .30. 0
0 0 I 0
0 1 0 0
5-
S
0
.30
- 5
0
0
0
1 .50
1.0
3.0
3.20 04
I S
I 30
30
.10
.10
0
0
.40
.20
0
0
0 0
0 .10
0 0
0 0
0 1 I
0 I .10
0 .13
0 .10
0
0
0
0
0
0
0-
0 -
.30
JO
.13
.10
0
0
.10
.50
20.30 0.3
1
30
— 30
.10
.10
.10
0
.40
.33
.33
0
0 0
0 .0)
0 I .03
0 0
0
0
0
0
0
5
0
.10
0
0
0
Q (
.10 .40
5 .30
0 .30
0 I .10
0
0
0
.50
0.5 T
3
20
30
.10 .40
.10 I .33
.10 I .33
.10 L -“
0 j 0
0 I .03
0 I .03
0 L 0
. 0 0
0 0
0 I 0
0 L 0
0 .10 1 .40
0 I 0 I Jo
0 I 0 1 .30
- 0 L 0 L • ‘
0
0
0
0
Ia n. 0.3— I - 04
(10.001 — S
100.000 p ) 30
30
3—10 0.3
I
30
50
•
.10
0
0
L 0
T .10
I .10
0
0
.20
0
0
0
.33
.20
I
0
.03
0
0
0
.03
.03
0
0
.03
0
0
0
— 0
.45
5
0
0
0
0
0 -
0
0
0
0
.30
.20
0
0
.10
.30
33
33
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
30
0
0
0
.40
.30
.05
0
I
.30
1.0
1.0
0
0
.60
.63
0
0
.30
0
0 I
0
0
20—30 0.3
3
20
.ao
•10
•1O
.40
.30
.20
0
.43
.03
S
.0)
.43
0
0
I
.10
.30
0
S
0
0
0
.40
.40
.30
sO
0.3
3
20
30
0
.10
.10
.10
.30
0
.40
.30
.20
.20
0
0
I .41
I .0)
j .01
0
0
.0)
.43
.0)
0
0
0
0
0
.35
0
I .30
I .30
.20
0
6.......’ ..
0
0
0
0
0
I 0
0
L 0_
.05
.40
.40
.30
.30
V. ,y Iai s 0.3.) 0.3
(100,000-poS) 3
20
50
.10
0
0
0
.20
0
0
0
0
0
0
0
.10
S
0
0
0
0
0
0
1 .60
.60
0
0
0
0
0
0
0
0
S
0
0
0
I
0
0
.40
1.0
1.0 I
0
0
.10
.30
0
0
0
.30
0
0
0
J 0
3 -20 0.3
5
20
30
.10
.10
0
0
.3)
.20
0
0
.03
. 43
0
0
.03
0
0
0
0
0
0
.30
.40
.30
.30
S
0
0
0
0
0
I S
0
.20
.20
0
0
0.3
3
20
30
0
0
0
0
•
.45
.30
0
0 -
.01
.03
0
0
0
.03
0_
0
0
0
0
.30
.30
.44
.30
0
0
0
0
0
0
0
0
.20
.20
.20
0
30. 0.3
3
20
30
0
0
0
0
. 50
.43
.30
.30
0
.05
.03
.01
0
0
.05
.03
0
0
0
0
.20
.30
.64
.40
0
0
0
0
0
0
0
0
.20
.20
.20
I .20
Note: the probabilities have been revised so that a system always selects
“do nothing” if the MCL is not exceeded
B—i
-------
EXHIBIT 3—2
10014311.85 ITlibs OP 01vta flU 303 Vh S CZYIN 3011.0103 0103 T1A53OU SCLIC?13C 03 UA ‘ t n pot £ C l i i i .
(Ce...d 10S)
tn?30 110111i V1V01. 01100100’ • I .3-o1 030TS n, *ml IYPIC*1 imue c&st ,vt 3ipi . ot u
sy.i lii. 3033...*
-14ratLus (p0)
(ppb)
fv ,r,ri.ii pj. -i 0.3
tIC
.40
P
T I
.03
00Its
Mo
?r.y
20.c4
.0)
P40
M I
IIsU l os
ss—at PrssU—
0
l.a3—
luutt
£11.,—
at&t
•btM.i I
3 -S00pip) 3
30
0
0
0
0
0
0
0
0
S
S
0
0
9
0
.03
0
0
.33
S
0
3.0
30
5-30 04
S
30
30
0
.3 )
.S3
0
0
0
.30
.03
S
0
0
0
0
0
0
0
.03
0
0
0
S
0
0
0
0
S
0
0
0
0
0
0
0
.0 )
. 13
0
0
0
.30
.30
S
3.0
3.0
0
0
1.0
20-50 0 3
3
30
.33
.3)
.33
.30
.10
.0)
0
0
0
0
0
.03
0
0
S
0
0
S
0
0
0
0
.1U
.0 )
.03
S
.30
.30
1.0
0
0
I 50
I 0.3
3
30
0
.53
.33
.33
0
.10
.03
0
•
0
0
0
0
0
.03
0
0
0
0
0
p
0
0
0
0
0
,
. 0 )
Jo
0
,
.30
0
1.0
0
50
3 .. LI 0.3—3 0.3
(303—3200 pm,) 3
20
50
.33
0
0
0
0
.03 0
.30 0
0 I 0
0 I 0
- 0 j 0
.03
.03
0
0
0
0
0
0
0
0
p
0
0
0
0
p
0
0
0
.03
.os
0.
0
0
.30 0
.30 0
. o f p
e
0 3.0
0
3-20 04
3
30
50
0
0
0
0
.33 -i 0 0
.30 I 0 .03
S 0 0
0 0 0
0
0
0
0
0
0
0
0
0
0
0
0
. 0)
.03
0
0
.40
.00
0
0
0
0
1.0 I
1.0
0
0
0
1.0-
20-50 0 3
3
20
50
0
0
0
0
.30
.30
.30
0
0 0
0 .05
0 .03
0 0
0
0
0
0
0
S
0
0
0
0
0
0
.03
. 03
. .03
0
.4 )
.40
.40
0
50 . 04
3
20
50
0.3—3 3.5
0
0
0
0
0
.30
40
.30
. 50
. 3 0
0
0
0
0
0
0
.03
.03
.03
.30
0
0
0
0
0
0
0
0
0
0
0
0
F
.03
.03
. 03
.03
.4)
.40
.40
,40
0
0
0
0
(3301—10.000 3 I 0
p ’p) 20 I 0
50 0
0
0
0
0
0
0
0
0
0
0 •
0
0
.10
.10’
0
0
0
0
0
0 .30
0 I S
0 S
0
0
.40
1.0
1.0
6
0
.70 I
.10
0
0
0
3— 20 0 3
3
20
50
0
0
0
0
. 30
. 50
0
0
0
0
0
0
0
.10
0
0
0
0
0
0
0
.10
.33
0
0 0
0 0
0
.30
.30
.33
20-50 0.3
S
30
0
0
0
.30
. 43
.45
— T
0
0
0
.0)
.0)
0
0
0
0 1
0 0 7
0 10
0 I 0
0
.40
0
0
.10
.40
.30
.30
50
30. 0.3
- 3
20
I
0.3— 3 0 .3
(10.000 — 5
100.000 p.p.) 20
30
3-20 0.3
3
30
50
0
0
0
0 I
•
0
0
0
0
0
0
0
0
0
.30
. 41
.43
.50
S
0
0
.43
.30
0
0
0
0
0
0
•
.03
0
0
0
.03
.03
0
0
0 j 0
i— — 1 0
.03 I 0 I
.03 I 0 I
•°‘ L 0 j
.05 0
0 0
0 0
0 0
0 0
.03 0
0 0
0 0
.10
0
0
0
0
.30
.20
0
0
.30
.50
.3)
.35
0
0
0
0
0
0
0
0
0
d
0
0
0
0 .10 L
To .‘0
0 .30 I
0 .30 I
0 .50
0 I 40
0 0
0 0
0 0
p .40
S .30
S .0 1
0 0
•0
0
0
0
0
I
.io
3.0 i
1.0
0
0.
.40
30—50 0. 3
3
30
50
0
0
0
0
.40 0 0
.40 I .05 I .03
JO .051 .03
0 0 0
0
0
0
0
0
.10
.30
.3)
0
I
0
0
0
I
5
0
.40
.40
.30
.01
0
0
.0
.40
30 , 0.3
5
30
50
0
0
0
0
.40
.40
.30
.30
0
.03
.01
.01
0
. 0 3
.03
.05
0
0
0
0
0
.10
.30
.30
0
0
I 0
0
0
0
0
0
.40
.40
JO
.30
0
0
0
0
Y.r7 L qs 0.3—5 0.3
I (100.001. psi) 5
30
50
0
0
0
0
.50
S
S
0
0
0
5
0
.40
0
0
0
0
0
0
0
.40
.40
0
0
0 0
I 0 I 0
0 5
0 0
0
0
0
0
p
.
1.0
1.0
3 .-go 0.5
3
20
50
0
0
0
0
.45
. 30
0
0
.05
.03
0
0
0
. 03
0
0
0
0
0
0
.20
.40
.10
.50
0
0
0
0
0
0
5
0
.10 T
.20
5
0
a
0
.30
20-30 0.3
3
20
I 50
I 30 03
3
0
0
0
0
0
0
.30
.43
.30
0
.30
.4)
0
.05
.03
0
0
.03
0
0
.03
0
0
0
0
0
0
0
0
0
.30
JO
.40
.30
.30
.30
0
0
0
0
0
0
0
0
0
0
0
0
1
•30 0 —
.30 0
.20 I 0
0 j_.so
.20 7 0
.20
30
30
I
0
.30
.50
. 0 ) .03
.01 J .03
I
0
.40
.40
0 0
0 _0
.2*
.20
0
0
0
• i .i ,tsyI ctI.r44. , 030 3. L.cI.d. 040 ptv s4IU1L.. .8..23 .31.. i.d is p .ck.d i . rau...
Note: the probabilities have been revised so that a system al ys selects
“do nothing’ 3.f the MCL is not exceeded.
B—2
-------
EXHIBIT 3—3
PWIMSLITT ftfl CIYtJ $113 £30 V$ c1v1j UI1UJ$ T 00II TIATI00 IILNISIC U10 30*345103 500 £ 05V03
(Cl iiid rss)
00TMU0*3T$s PAaA01 &030
0001U&0
$ .I.s Has ZaS1..U
ussIM130 (ppl)
( b)
vary L1 04-3 04
C2 5-300p ) 3
N
030
— IT
0
S
?sck.d
T. r
.03
0
0
Us
T
S
0
?10
M,s14
.03
0
0
t H
P10 k si
0
0 0
0 0
‘‘
?5555sU.i
I M111
•.I1 sL—I 551115
lash.. I
.0 * . 1 1
0 0
S S
I
ULs
o
2.0
2.0
5.0
o
0
2.0
5.0
d
0
0
5.0 I
0
0
0 I
0 I
0
0
S
50
— 3-l I 0.3
1
N
50
0
45
.60
0
0
0
.40
.03
5
0
0
0
S
5
0
0
0
5
S
0
0
0
0
5
0
0
0
S
0
0
0
0
S
0
0
0
f
. 03
5
S
5
.30
.30
o
0
2 0 - 3 0 , 0.3
5
20
10
43
.33
.60
0
.10
.10
.31
0
— T•
0
0
0
o
0
0
0
o
0
5
S
o
0
0
S
o
0
S
0
.o
.0 5
.03
0
.
.30
.30
5
30. 0.3
5
N
50
.33
.33
.40
.30
.40
.10
.05
.05
0
0
5
0
0
0
0
0
0
0
0
0
0
0
S
0
S
5
0
. 03
.05
.05
.01
.30
.30
.30
.30
0.3—5 0.3
(30L-3300 p) 1
N
50
3-20 0.3
.43
0
0
0
.13
.40
0
. S
0
.40
0
0
0
0
T
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
. 0 .3
5
5
0
.03
5
0
0
.40
0
1.0
*.s
5.0
0
3
20
50
.10
0
0
.40
0
0
0
0
0
.03
0
0
0
S
0
0
0
0
0
5
0
.03
0
0
.45.
5
0
0
4.0
5.0
2 0-30 0.3
5
0
50
.10
.43
.10
0
.30
.40
.40
0
0
0
0
0
0
0
. 03
0
0
0
0
0
0 0
0 0
0 IS
0 0
.03
.03.
.03.
0
.43
.30
. 40
0
0
0’
0
5.0
30. 0.3
5
20
30
—T5
.13
.40
.10
.30
.40
.30
.40
0
0
5
0
0
0
.0 5
.05
0
0
0
0
0
S
0
0
0
0
0
0
.03
.03
. 53
.01
.43
.10
.30
.40
0
0
0
0
I4 , ’ 0.3—S Oi S
(3301-10.000 1
p.p.) 20
30
.13
0
0
0
.23
0
0
. 0
0
0
0
0
0
0
0
0
0 .10
0 .10
0 0
0 1 0
0
0
0
0
0
0
0
0
.30
0
0
0
0
.10
1.0
1.0
3— 40 0.3
S
20
30
—. 40-30 04
.10
.10
0
0
.10
.40
.30
S
1 0
T .40
0
0
0
0
0
0
.10
0
0
0
0 f 0
0 I .50
0 .13
0 .50
0 0
0
0
0
0
0
0
0
0
0
T4—
.30
.30
.13
.50
.40
0
0
.70
.10
0
0
0
.30
0
0
0 I
0
5
20
30
.10 I .33
.10 $ .33
0 L 0
0
0
0
.03
.03
a
0
0
0
0
0
.10
0
0
0
0
0
0
.30
.30
.50
50. 0 4
5
20
50
.10 ’T .40 0
.10 I .35 0
.10 .33 I 0
.50 .35 0
0 0
.03 0
.03 I 0
.05 L 0
0 0 .40
0 0 0
0 I 0 I S I
0 0 0
.30
.30
.30
.30
sr I. 0. 3 -3 0.3
(10.001 — 5
00.000 p ) 20
.15
0
0
.2)
0
0
I I
0
0
U
5
0
0
0
0
.30
. 20
0
0
0
0
0
S
0
.30
S
0
0
,ao
5.0
50 0
3—20 0.3 .41
S I .10
20 0
— 20 0
20-50 04 .25
5 .10
20 .10
50 0
0
as
.20
S
0
0
F
.03
0
0
0
0
.05
0
0
0
0
0
0
0
0
— T5•
.30
.33
.11
0
0
0
0
0
0 1
0 1
S I
S
0 j.0
.èo 0
.30 I 0
.05 .30
.41
.33
.40
.30
0
0
0
.03
0
0
0
.03
0
0
0
0
0
0
•10
Jo
.31
0
S
0
0
0
S
S
0
.30
.34
.30
.05
0
0
0
-
30. 0.3
3
20
50
.21
.10
.10
.50
.33
.40
.20
.20
0
0
.01
.01
0
0
.03
.05
0
S
0
0
0
.50
.30
.30
0
0
0
0
0
0
S
0
.40
.30
.30
30
0
0
0
0
VsryIs . 04—3 0.3
C100.000 .p. ,) 1
20
50
.13
0
0
0
.25
0
0
0
0
0
0
0
•
0
0
0
0
0
0
0
0
•io
.40
0
0
o
0
0
0
•
0.
0
S
&
S
0
0
o
.40
5.0
1.0
3-40 04
3
N
10
.43
.10
0
0
.33
.20
0
0
0
.03
0
0
0
. 03
0
0
0
0
0
0
.30
.40
.50
.50
0
0
S
0
S
0
S
0
.20
40
S
0
0
0
.10
so
2 0 -30 04
S
.10
.10
.40
.33
0
.03
0
0
0
0
.20
40
0
0
0
0
.30
.20
0
0
20 0 .30
10 0 I 0
04 .10 .40
.10 .35
3 0 p 0 .30
50 0 .30
.03
0
T
.05
.01
.05
.03
0
0
0
.01
.01
0
0
0
0
0
0
.40
.3O
.30
.30
.10
.40
0
0
0
0
0
0
5
0
i
5
5
.20
0
. 20
.20
40
.20
0
. 20
0
0
0
0
Note: the proba i1ities have been revised so that a system always selects
“do nothiria” if the ZICL is not exceeded.
B-3
-------
EXHIBIT B—4
PtosAIU.11T 1T2T5 1 C? CIVIl 2120 A V210 CIVIl 1lltbU? 002C20! 15T202 IWC?NC 00 U.2 1 *2111102 201 5 01102
(1. .tu. ru.)
00 112*11111 1 1 Z EZVTt11IE. I • 1 • 3-I12 10E11IA*. TCT2 0ETVTU*. CM Tt1I 101 , 1 •
I • 1—0I AS0LNYUI11. U*V . PAa —O1 Ib1&11120I, *1VTL02I 0002301, 2110 1IHCAI IUI2Ut* X00lUJ
t&6200 I tl1
17S1 L i la 1s11..U — 1
e%1rsS1 (ppb) I
( oØ)
111
P,sk
T r
0111w
U,
?Ve
I.lI.s
WeL l
fluid 1s ne
NC Pap 11 PMlastlie
I
1s. 11
a.Lgeg—
lilius
ufU
111q
V.r u..L2 0.5—3 0.3
(25-ScOpe ,) 5
a
30
.60
0
S
0
.03
S
0
0
0
0
0
0
.05
0
0
0
S
0
0
S
0
0
0
0
0
0
.03
0
0
0
.23 I
0
S
0
o
(.0
1.0
(.0
I 3 .10 0 . 3
I 3
10
so
.35
.33
0
0
.10
.03
0
0
0
0
S
0
0
.05
0
0
S
5
0
I
0
S
0
0
.1
.10
5
0
.01
.01
S
0
. 20
.20
0
0
0
0
1.0
l.a
10—30 0.3
3
a
30
0
5
•
0
- 0
0
o
0
0
0
5
0
0
0
0
0
0
S
•
0
0
S
S
0
1.0
1.0
*.s
0
0
0
0
5
0
0
o
0
0
0
s
J.0
10. 0.3
3
a
so
0
0
o
o
0
0
o
0
0 0
0 0
o o
0 1 0
0
0
o
0
0
S
o
0
(.0
1.0
&.o
1.0
0
0
o
0
0
0
o
0
o
5
s
o
04— 5 0.3
C301—3200p.p) 5
10
30
.03
0
0
0
.30
0
0
0
0 1 0
0 I 5
0 I 0
0 0
0 0
S 0
0 0
0 0
.30
.03
0
0
.0)
0 .
0
0
.30
0
0
0
0
.15
1.0
1.0
3-20 0.3
3
20
so
.03
S
0
0
.30
.10
0
0
0
.10
0
0
0
0
0
o
0 1 0
5 0
0 0
0
. 1 0
.6)
S
0
. 03
.03
0
0
.30 0
.25 I 0
0 I 1.0
0 [ l.a
2 0— 50 0.3 0
3 0
10 I 0
I 50 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
S
0
0
1.0
1.0
1.0
0
0
0
0
0
0
0
0.
0
To
0
a
1.0
I 30 . 0.3
I S
20
30
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(.0
1.0
(.0
1.0
0
0
0
0
0
0
0.
0
0 I
0
0
0
0.3— 5 0.3
(3301—50 ,000 3
pop) 20
30
.03
0
0
0
.03 .10
0 0
0 0
0 I 0
0
0
0
0
.03 0
0 I I
0 0
0 0
. 29
.131
0
0
.03.
I 3
0
0
.30..
0
0
1. 0
0
i .65
1.0
l.a
0
0
1.0
(.0
0 I
0
0
l.a
3-20 0.2
S
20
30
.03
0
0
0
.03
0
0
0
T
.10
0
0
0
0
0
0
0
.10
0
0
0
I
0
0
. 1 0
.10
0
0
0
0
0
0
0
0
0
0
30-30 0.3
3
10
30
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
I
0
0
5
S
0
1.0
1.0
1.0
0
0
0
0
0
0
0
0
0
20. 0.3
I 0
20 I 0
L 0
0
0
0
0
0
0
0
L 0
0
0
0
0
0
0
0
0
0
S
0
0
1.0
1.0
1.0
1.0
0 0
I I 0
5 I 0
0 L 0
0
0
0
0
1 .ot u 0. 3 -5 0.3
( (0.00$ — 3
100.000 pe,l 20
50
.03
0
0
0
.13
0
0
0
.10
0
0
0
0
0
0
0
.13
.10
0
0
0
0
0
0•
.33
.20
0
0
0
0
I 0
[ 0
0
0
0
0
S
.70
1.0
(.0
5-30 0.3
3
20
30
0
.02
0
0
.10
.02
0
0
0
.10
.03
0
0
0
0
0
0
.10
.03
0
0
0
0
0
.10
.10
.10
0
1 0
I 0
0
0
0
0-
0
0
0
0
.20
1.0
I 3 0 - 50 0.3
I 5
20
50
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
0
0
1.0
1.0
1.0
.30
0
0
5
0
0
0
0
0
0
0
0
.30
10. 0.3
5
I 20
I 50
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
0
1.0
1.0.
(.0
1.0
0
0
0
0
0
0
0
0
0
0
0
0
Very I ar o 0.3—5 0.3
I (100.000. po ,) 3
20
10
.03
0
0
0
.13
0
0
0
.10
0
I
0
0
0
0
0
.13
.10
0
0
0
S
0
0
.33
.20
0
0
0
0
0
0
0
0
0
0
0
.10
(.0
1.0
0
I 0
I .10
[ .60
T 0
Is
(0
.301
0
0
0
0
3-20 0.3
3
20
50
0
.02
0
0
.10
.02
0
0
0
.10
.05
.03
0
0
0
0
0
.10
.0 5
.03
0
5
0
0
. 10
.10
.10
.10
0
0
0
0
0
0
0
0
20-50 0.3
5
I 20
30
0
0
I
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
S
0
0
0
S
0
(.0
(.0
1 .0
.30
0
0
0
0
0
0
0
0
0 4
3
20
30
S
0
0
0
1 0
I 0
0
0
0
I
0
0
j 5
0
0
0
0
0
0
0
0
0
I 0
0
1.0
1.0
(.0
(.0
0
0
5
0
0
0
0
0
Note: the probabilities 1 ave been revised so that a system always selects
“do nothing” if the MCL is not exceeded.
B—4
-------
EXHIBIT B—5
NOIMLUTI C l STile. 01 CIlia 1333 A V I I CITe. UIPUJ&I? 00 e.TIA7I 1e &c ue qt3 VW III A CITe.
(SsrI.c. Isiftil
haiL 00t0110C.. I.2—0lO0.0200200a1 . & 1TPICS& 100)? CAlL ‘SäL?ZPIi sm
$ it Ui•
L L1a ,sai
C..s iait
(pps)
1
(p0) I
I
C I I
P.sk
5e
20I1
Ml
Sr.,
M1111
PLC
u
11.1 0 I
sie (
-
‘I’sSisl$
i t
kI s1.I sill,. I
1 1U1 siuas
PLaI 1.
1 v.r .13
(2 3-S O Opsp)
0.3- )
0.3
3
30
10
.10
0
5
0
.03
I
S
0
T
S
S
0
— T
S
S
0
T
S
5
5
*
0
I
0
0
I
0
S
.0)
5
S
0
.33
S
0
Q
0
1.0
1.0
1.0
1 -20
0.1
.33
.10
0
0
5
0
.30
.01
• F
p
3
30
.3)
I
.03
S
0
S
.0)
I
5
S
S
0
.10
I
.03
S
.30
0
0
3.0
I
30
0
0
0
0
5
0
0
0
0
3.0
3 0-30
9.3
0
0
0
0
0
0
3.0
0
0
0
S
0
S
I
S
5
5
5.0
I
0
0
30
10
0
0
I
•
S
0
0
0
0
0
0
0
1 . 0
0
S
0
5
0
0
3.0
I
0 .3
5
0
0
0
0
0
1.0
0
0
0
S
30
50
0
0
0
5
5
0
5
0
0
5
5
0
0
5
0
5
0
0
3.0
1.0
5.0
S
S
0
5 IS
0 to
0 10
1 2L
0.3-3
0 4
0
.33
0
0
0
0
.30
.03
40
To
C 101 .5300 o)
S
5
5
0
5
5
5
.0 5
I
S
.13
30
5
S
0
S
S
0
0
5 -
S
3.0
10
0
0
0
0
0
0
0
0
0
3.0
1—20
0.3
0
.3)
0
0
0
0
.30
.4)
.30
0
I
3
0
.15
.10
I
S
0
. 03
.0 5
•10-
0
30
0
5
S
0
0
0
S
I
0
3.0
30
0
0
0
0
0
0
0
0
0
3.0
0
0.3
0
0
0
0
0 -
0
1.0
0
0
0
1
0
0
0
0
5
0
1.0
S
S
S
30
5
0
5
0
0
0
3.0
5
5
0
3 0.
50
0 4
0
0
0
0
0
0
0
0
0
0
0
0
0
3.0
0
0
0
0
t.0
0
1
0
0
0
0
0
0
1.0
5
0
0
20
0
0
0
0
0
0
5.0
0
0
0
50
0
0
0
0
0
0
3.0
0
0
0
d&i
0.3— )
0.3
0
.10
.1 0
0 -
.03
0
.30
.4)
.30
0
(3303.30.0 0 0
psi)
5
20
0
0
0
0
0 I 0 I 0
0 I S I 5
0
0
.1)
0
0 I S
0 I S
. 13
3.0
50
0
0
0
0 J
0
0
0
0
0
L l . a
3 —20
0. )
0
.30
0
0
0
0
.10
0
0
T 0
1
0
0
.10
0 I 40
0
.10
0
I
S
20
0
0
S
0 I 0
0
0
0 I 0
3.0
10
0
0
0
0
0
0
0
0 j
0
3.0
20-30
04
0
0
T.....
0
0
0
1.0
0 7
9
0
S
30
I
I
5
5
0 5
5 I 0
S
S
0
0
1.0
3.0
S
0 I
I- 0
5
0
0
i
10
0
0
0 j
0
0
0
0
0
0
3.0
30
04
0
0
0 1
0
0
0
3.0
0
0
0
3
5
‘
I I 0
0
5
1.0
0
5
0
.
30
0
0
0 0
0
0
1.0
S
0 10
rss
(10.001 —
I 300.000 pip)
I
5.3— )
50
0.3
S
20
10
0
0
0
5
0
.30
5
5
0
0
.10
0’
S
0
0
0
0
0
0
0
.13
.10
S
0
0
0
0
0
0
3.0
.3)
.20
5
0
S
0
0
0
o
0 0
0 0
5
0 I 3.0
0 i.e
3-20
0.)
0
.30
0
0
0
0
JO
5
0
0
3
I
.10
.10
5
.10
0
. 70
5
5
0
30
5
0
.01
S
.03
0
.35
0
5
. 50
10
0
0
0
0_
0
0
0
0
0
1.0
30
0.)
0
0
0
0
0
5
1.0
0 7
0 jo
I
0
0
5
0
0
0
3.0
5
0
0
20
5
0
0
5
0
0
3.0
I
0
0
30
0
0
S
0
0
0
.30
S
0
.30
30.
0.3
0
0
0
5
0
0
3.0
S
0
0
1
S
5
0
0
0
0
1.0
0
0
0
30
10
0
5
0
0
I
0
0
0
0
0
0
1.0
3.0
0
0_
0
0
0
0
lary r.
(1X.X0 .psp)
0.3-)
04
5
30
0
I
S
.40
S
5
.1O
0
S
0
0
0
.13
.10
S
0
0
0
.3)
.30
0
o
0
S
a
0
5
a
.70
5.0
30
0
0
0
0
0
0
0
0
0
1.0
I
I
I
3-20
0.3
1
20
30
S
S
0
0
• 30
.30
0
0
0
.10
.0)
.01
0
0
0
0
0
.30
.0)
.OS_
0
0
0
0
.30
.70
.10
. 30
5
0
5
0
0
0
5
0
0
0
.10
.50
3 0
0.3
0
0
0
0
0
0
3.0
0
0
0
5
0
0
0
5
S
0
1.0
5
0
S
.
20
0
0
0
0
0
0
3.0
5
0
0
30_
0
0
0
0
0
0
.20
0
0
.30
30.
04
0
0
0
0
0
0
2.0
0
0
0
I
S
S
0
S
S
0
3.0
5
0
0
30
5
0
0
0
0
0
3.0
0
0
0
30
0
0
0
0
0
0
3.0
0
0
30
• ., vispI sAlarll ., C II Is s.LsiiM. CLI pr sS11I&.. sL 3 W 41siu.4 is psü..i 1r sirsilsi.
Note: the probabilities have been revised so that a systen always selecr.z
“do nothing” if the MCL is not exceeled.
B—5
------- |