EPA/440/6-85/001
ISSUE PAPERS IN SUPPORT OF
GROUND-WATER CLASSIFICATION GUIDELINES
Prepared For:
The Office of Ground-Water Protection
The Office of Policy Analysis
The Office of Solid Waste
October 1985
0 5 J987
ICF INCORPORATED International Square
1850 K Street, Northwest, Washington, D C. 20006
-------
ISSUE PAPERS IN SUPPORT OF
GROUND-WATER CLASSIFICATION GUIDELINES
Prepared For:
The Office of Ground-Water Protection
The Office of Policy Analysis
The Office of Solid Waste
October 1985
-------
CONTENTS
Executive Summary
Costs of Water Supply Systems
Class I: Empirical Evidence on the Implications of Various
Substantial Population Thresholds
Class I: Analysis of the Implications of the Current Economic
Test of Irreplaceability for Ground-Water Classification
Class III: Development of an Economic Test
Estimated Results of Ground-Water Classification at RCRA Facilities
and CERCLA Sites
Attachment: Data Sources for Analyzing Income and Substantial
Population Data
-------
EXECUTIVE SUMMARY
The papers compiled in this document identify major implications of
adopting alternative methods for classifying ground waters under the EPA
Ground-Water Protection Strategy. These papers provide estimates of the major
implications of including population and economic criteria for classifying
ground water as Class I or III. The remainder of this executive summary
presents the findings of each paper.
Scope of the Analysis
This analysis focuses on socio-economic criteria for classifying ground
water. The Agency has identified other criteria for Class I and Class III
designations.
Class I ground waters are resources of unusually high value. They are
highly vulnerable to contamination and are (1) irreplaceable sources of
drinking water and/or (2) ecologically vital. Ground water may be considered
"irreplaceable" if it serves a substantial population and if delivery of
comparable quality and quantity of water from alternative sources in the
region (i.e., within a reasonable pipeline distance) would be economically
infeasible or precluded by institutional constraints.
Likewise, a Class III ground water must be:
• contaminated by naturally occurring substances or
activities unrelated to a specific hazardous waste
disposal site and
• must not be treatable for drinking water purposes
using treatment methods reasonably available to public
water systems.
A ground water is also included in Class III if it is not a potential
source of drinking water due to salinity (i.e., greater than 10,000 mg/1 total
dissolved solids) or due to insufficient yield.
This analysis does not consider the effects of many of these criteria. In
addition, important assumptions concerning the number of contaminated ground
waters cannot be validated. The analysis focuses on the likely results of
using alternative economic feasibility and population criteria for
classification. The analysis does consider the sensitivity of the results to
the radial distance of the classification review area and to assumptions about
the extent of ground-water contamination. No attempt is made to estimate the
total number of Class I or Class III ground waters based on all the relevant
criteria.
-------
-2-
Costs of Water Supply Systems
The costs of replacing and/or treating water supplies have an important
influence on the implications of including economic criteria in the
classification of ground waters. This paper summarizes available cost data
and presents a method for estimating replacement costs for existing ground
waters used as drinking water supplies.
The costs of replacing existing ground-water supply systems were estimated
from detailed data on public water systems of various sizes throughout the
United States. The total average cost of supplying drinking water for
populations of various sizes was estimated, and costs were divided into the
following four major components: (1) acquisition; (2) treatment; (3)
distribution and transmission; and (4) support services. These costs were
used to develop three replacement scenarios: the extensive scope includes all
four components of costs; the moderate scope includes all but the costs of
building distribution and transmission systems (i.e., the pipes to people's
homes), which may already be part of the existing system; finally, the limited
scope includes only the cost of acquiring new sources of water.
The results of the analysis demonstrate the wide range of potential
replacement costs. Exhibit ES-1 shows that:
• replacement costs per household depend on both
the size of the population being served and
the extent of the measures necessary to deliver
replacement drinking water
• the scope of replacement more strongly affects the
costs than the system size.
These cost estimates are reasonable but subject to some uncertainty and
bias. The costs of replacing a water supply at a particular location will be
driven by numerous local factors -- such as the relative scarcity or abundance
of alternative water sources -- which cannot be fully reflected. It is also
likely that the existing supplies were developed first because they are the
low cost options'. Replacement of existing sources will require going to
second best sources, which will probably be more costly than the average of
the existing systems. The analysis attempts to compensate for these
limitations in the available data.
-------
-3-
EXHIBIT ES-1
ESTIMATED ANNUAL TOTAL COSTS PER HOUSEHOLD FOR
WATER SUPPLY REPLACEMENT
Replacement Scope
System Size
(Population served
by system)
750
2,000
Limited
$22.30
Moderate
$55.72
$21.76 | $52.25
I
7,000 | $14.20
12,500
$10.76
60,000 | $10.02
$35.49
Extensive
$111.44
$103.43
$70.98
$26.96 | $53.87
I
$24.96
i
$50.04
i
Empirical Evidence on the Implications of Various Substantial Population
Thresholds
EPA's Ground-Water Protection Strategy currently requires that in order
for ground water to be classified as Class I "irreplaceable," it must be used
by a "substantial population." The choice of a population threshold for
defining "substantial" significantly affects the results of the classification
scheme. As the threshold is increased, the number of ground waters likely to
qualify for Class I will decrease while a reduction in the threshold will lead
to more Class I determinations. This paper provides empirical data on the
implications of adopting alternative population thresholds.
Detailed data on the number of individuals served by public water supply
systems were summarized to identify the fraction of all the systems whose
entire populations served would be designated as substantial under thresholds
ranging from 1,000 to 50,000. Exhibit ES-2 summarizes these results and shows
that:
• 99.7 percent of all systems predominantly using ground
water serve fewer than 50,000 people and these systems
serve about 54 million people;
• The remaining 0.3 percent of the systems serve
approximately 30 percent of the individuals (about 23
million persons) served by public ground-water systems;
-------
-4-
A threshold of 2,500 would exclude ground waters from
Class I that are used by more than 90% of the systems
and about 20% of the users;
A threshold of 25,000 would exclude ground waters from
Class I that are used by more than 99% of the systems
and about 57% of the users; and
A threshold of 1,000 would exclude ground waters used
by 82% of the systems and 11% of the users.
EXHIBIT ES-2
IMPLICATIONS OF ALTERNATIVE
SUBSTANTIAL POPULATION THRESHOLDS
Percent of Percent of
Systems Persons
Excluded Excluded
POPULATION THRESHOLD from Class I a/ from Class I b/
> 1,000
> 2,500
> 5,000
> 10,000
> 25,000
> 50,000
81.6%
90.8
95.0
97.4
99.2
99.7
11.1%
20.3
29.6
40.2
57.1
69.8
a/ Based on 60,841 community public water systems
that use ground water.
b/ Based on 77.2 million persons served by these
systems.
Water systems may use more than one ground-water source. If a system's
users are "pro rated" among its ground-water sources, the number of potential
Class I ground waters may be lower than when the test is based on all system
users. For example, our data indicate that systems serving 5,000 persons use,
on average, 2.1 well groups. If each well group is considered a separate
ground-water source, then allocation of users among water sources means that,
on average, only 2,380 people (i.e., 5,000 users divided by 2.1 well groups)
would be considered for the test. Ground waters used by such systems,
therefore, are unlikely to be considered for Class I under either a 2,500
person or 5,000 person threshold.
At a 25,000 person substantial population threshold, considering the
number of users per ground-water source (instead of all system users)
-------
-5-
eliminates almost all public ground-water sources and therefore, the systems
using these sources, from Class I consideration. At a threshold of 2,500, the
effects are less pronounced. Ground waters serving an estimated 5% of the
systems, used by about 70% of the persons served by ground-water systems,
would satisfy the substantial population test, considering the number of
people served by separate ground-water sources.
Thus the 2,500 person criterion can be expected to allow between 5 and 10
percent of the ground-water sources serving public water systems as potential
Class I waters, encompassing between 70 and 80 percent of the users supplied
by such systems.
Data on the use of private water wells for drinking water were also
collected and analyzed. However, there appears to be no significant change in
the estimated percent of potential Class I waters when these users are
considered. Due to data limitations, the analysis must assume an even
distribution of private well users throughout each county where such users are
located. As a result of this assumption, the number of private well users in
a classification review area is estimated to be too small to significantly
affect the total number of people served by the ground water. However, within
an actual ground-water classification review area there may be high density
pockets of private well users that exceed the substantial population
threshold. Because the available data do not permit identification of such
situations, the results may underestimate the number of times that private
well use in a given area will be sufficient to consider the ground water for
Class I.
Analysis of the Implications of the Current Economic Test of Irreplaceability
for Ground-Water Classification
The proportion of ground waters eligible to be designated as Class I is
also driven by the determination of the waters as being "irreplaceable." The
definition of irreplaceable is currently: "replacement costs more than one
percent of the average annual household income of the affected population."
The choice of the percentage cutoff strongly influences the classification
results. This paper used available data on the cost of water replacement and
household income to identify the implications of alternative cutoffs.
A distribution of replacement costs was developed based on the results of
the Costs of Water Supply analysis (see above). Alternative scopes of
replacement were examined (see Exhibit ES-1). These costs were compared to
various percentages of a distribution of average household income, developed
for Zip Codes throughout the entire United States.
The results indicate that:
• The. number of potential Class I ground waters serving
public water systems is very sensitive to the economic
test cutoff chosen.
• Approximately 0.7 percent (143 ground waters) of all
ground waters used by public water supply systems that
-------
-6-
predominantly use ground water and serve at least 2,500
people are eligible to be classified as Class I at the
one percent economic test cutoff.
• A two percent economic test cutoff results in less
than 0.1 percent (22 ground waters) being eligible, and
a 0.5 percent cutoff makes about 7.6 percent (1,571
ground waters) eligible.
Because information is lacking regarding the necessary cost of replacement
for drinking water supplies around the country in the event current sources
were to become contaminated, estimates of the range of potential replacement
costs were used.
Development of an Economic Test for Class III Ground Water
Class III ground water is defined as ground water that is contaminated and
"cannot be cleaned up using methods reasonably employed in public water system
treatment." The current Agency guidance identifies water treatment processes
currently used by public water systems in specific EPA Regions. The guidance
also specifically identifies air stripping as a treatment process that should
be considered available on a nationwide basis. Within a given EPA Region, any
combination of used or available treatment processes may be considered in
evaluating the treatability of a contaminated ground water. This definition
looks primarily at technical feasibility and not economic feasibility of
ground-water cleanup. This paper identifies what would be required to develop
economic tests for evaluating whether treatment is economically feasible.
If an economic approach similar to the economic test discussed above for
Class I was implemented for Class III, the costs of treatment would be
compared to some measure of whether treatment was affordable. In general, the
cost components identified in the Costs of Water Supply Systems paper (see
above) are applicable, as adjusted to reflect the greater degree of treatment
needed to address contamination in Class III candidates. Developing an
economic test for Class III ground waters is difficult because, by definition,
Class III ground waters are not currently used. Consequently, there is no
clearly defined user population or market for which the water's value may be
estimated.
This paper develops several options that EPA could consider for defining
potential user populations for an economic test of ground-water treatability:
• Use of the concept of a substantial population
criterion for potential water users;
• Use of the existing population in the vicinity of the
ground water;
• Use of a proxy to represent potential water users; and
• Combinations of the above.
-------
-7-
Estimated Results of Ground-Water Classification at RCRA Facilities and
CERCLA Sites
The ultimate outcome of the Ground-Water Classification Strategy depends
on the interrelationships among the criteria for each ground-water class.
Once the rules for all three ground-water classes are defined, an integrated
analysis can determine how likely it is that ground waters will fall into each
class given certain assumptions and inputs. In order to comprehensively
assess alternative approaches to implementing the classification, a computer
model was used to estimate the results of classifying 4,330 RCRA hazardous
waste management facilities and 216 CERCLA NPL sites into ground-water
classes. The results of the model are sensitive to the assumptions and
parameters used, particularly assumptions about the extent of ground-water
contamination.
Using data from the 1980 Census and EPA databases, the model estimates:
(1) whether the ground water in the area surrounding the
facility is currently used for drinking purposes;
(2) if current users meet the substantial population
criterion and pass the replaceability test using
estimated replacement costs and household income;
(3) the appropriate treatment train and associated costs
depending on the type of contamination assumed to be
present if there are no current users; and
(4) treatability under various Class III tests.
For Class I, the analysis produced the following important findings using
available data:
• Assuming a review area of 2 miles radius1, about 70%
of the ground waters at RCRA or CERCLA sites, would be
considered currently used for drinking (and hence these
sites are Class I candidates). The remainder are Class
III candidate sites.
• At a 3 mile review area, about 83% of the ground
waters at RCRA or CERCLA sites would be considered
currently used for drinking purposes.
• Using a 2,500 person population threshold, a 1 percent
of.income economic test, and a 2 mile review area, .08%
of.the 4,330 RCRA sites fall into Class I. Of the 216
1 In the analysis, the location of a public water system within 2 miles
of a RCRA or CERCLA site is assumed to imply that the ground water is
currently used for drinking.
-------
CERCLA NPL sites, .55% would be Class I. These findings
are consistent with the results of the analysis of the
irreplaceability test which estimated the national
fraction of Class I ground waters among those currently
used as drinking water sources at less than 1 percent.
• Using a 2,500 person population threshold, a 1 percent
of income economic test, and a 3 mile review area, .1%
of the RCRA sites would be Class I. Of CERCLA sites,
.67% would be Class I.
• The number of Class I assignments is not extremely
sensitive to changes in the substantial population
threshold for values of the income burden factor above 1
percent.
• The number of Class I assignments is extremely
sensitive to the income burden factor for values less
than .2% (38% of the waters (RCRA and CERCLA alike) are
Class I at .1%).
For Class III the analysis compares estimated classification results for
several alternative tests including (1) the current guidelines reference
technology criterion, (2) a substitute economic feasibility criterion, and (3)
a combined available technology/economic feasibility test. For the latter two
tests, two different approaches were employed to characterize potential user
populations for ground water that is not currently used and is assumed to be
contaminated. Using available data and ground-water contamination scenarios
developed for the purpose of this analysis (based on case reports of
industrial, mining, and agricultural contamination), the Class III analysis
estimates that:
• Assuming that 50% of the ground waters that are not
currently used are contaminated, 6 percent of all of the
RCRA and 6 percent of all of the CERCLA facilities could
fall into Class III under the current guidelines
reference technology test.
• For both RCRA and CERCLA sites, between 3 and 8
percent could be Class III using an economic test, alone
or in combination with the reference technology test.
• For cost-related tests, the number of Class III
assignments at both RCRA and CERCLA sites is extremely
sensitive to the income burden factor for values between
.2% and .5%, and indifferent to the income burden factor
for-values above 1%. Below .1% of income, every not
currently used ground water considered to be
contaminated falls in Class III.
-------
-9-
• Adding an economic test to the reference technology
criterion increases the number of Class III waters by
30% percent at both RCRA and CERCLA sites.
• Populations in the vicinity of contaminated ground
water tend to be small. These populations cannot take
advantage of economies of scale in water acquisition,
treatment, and delivery costs. Thus, the use of a proxy
population -- instead of the existing population -- for
the Class III economic tests decreases the number of
Class III waters by 50 percent.
• Using a 5,000 substantial population threshold rather
than a value of 1,000 increases the number of Class III
assignments by 32% under the Economic Test, and by 14%
when an economic test is added to the reference
technology criterion.
The results of the analysis are subject to certain limitations. The Class
III results are driven by assumptions concerning the degree, type, and
prevalence of ground-water contamination. Also, the analysis for both Class I
and Class III must rely on zip code and county level data whereas
classification can be performed more accurately on a site-by-site basis.
Data Sources for Analyzing Income and Substantial Population Data
This attachment describes relevant information available from EPA
databases and the U.S. Bureau of the Census, including:
• locations of RCRA and CERCLA sites and public water
systems;
• use of private drinking water wells; and
• numbers of persons served by public water systems and
numbers of ground-water sources used by systems.
We used these data to analyze alternative substantial population and income
threshold definitions. The data are quite comprehensive, although certain
limitations have been noted.
-------
-------
COSTS OF WATER SUPPLY SYSTEMS
This paper presents and discusses cost data for water supply systems. The
data show the breakdown of costs among major costs components and the
variation in component costs and total costs among systems of various
characteristics.
The paper is divided into five sections. The first section describes the
data sources used for the analysis. The second section identifies the major
cost components of a water supply system and presents average cost structures
for small and large systems. The third section focuses on each cost component
in turn, describing the major determinants of cost and indicating the
variation in costs among systems. In the fourth section, the variation of
total costs among systems serving similar populations, and the variation of
average costs of systems serving populations of different sizes is discussed.
The fifth section investigates water replacement costs. An appendix discusses
potential sources and magnitude of cost mis-estimation.
1. Data Sources
The data presented in this memorandum are derived from studies conducted
by ACT Inc. for the Municipal and Environmental Research Laboratory of the EPA
and by Temple, Barker & Sloane Inc. for the Office of Drinking Water of the
EPA. The study by ACT Inc., conducted in 1977, is an analysis of the
economics of water delivery for 12 Class A utilities (revenues exceeding
$500,000 per year) and 23 small water utilities. The annual volume of
revenue-producing water supplied by the larger systems ranged from 2,300
million gallons to 63,700 million gallons. In 10 of the 12 cases, the volume
supplied exceeded 10,000 million gallons. The annual volume of
revenue-producing water supplied by the smaller systems generally ranged from
40 million gallons to 2,300 million gallons, however, one "small" system
supplied 5,600 million gallons to a small number of large users and municipal
utilities. The large systems were distributed throughout the U.S., while the
small systems were concentrated in five regions. The study by Temple, Barker
& Sloane Inc., conducted in the first half of 1982, surveyed financial and
operating characteristics of community water systems. Their random sample was
stratified on the basis of system size. A total of 1,056 water systems were
surveyed: 50 privately owned systems and 50 publicly owned systems in each of
12 size categories (except where there were less than 50 systems in the U.S.
in that size category) as well as 50 Native American systems.
2. Cost Components of a Water Supply System
Water supply system costs can be broken down into four major components:
(1) Acquisition;
(2) Treatment;
(3) Distribution and Transmission; and
(4) Support Services.
-------
-2-
24quisition costs are the costs of producing or acquiring water, and can be
thought of as the costs of getting the water to the treatment point. These
costs include the capital, operating, and maintenance costs of wells,
reservoirs, and aqueducts, and payments to suppliers for purchased water.
Treatment costs include the costs of treatment plants and equipment, and the
costs of chemicals that are added to the water. Distribution and Transmission
costs are the costs of pumping the water from the treatment points to the
service population, and the capital and maintenance costs of the piping
network. Support Services are the costs of administrative and customer
services that are not directly related to the physical process of delivering
water.
Exhibit 1 shows the average cost structure of the small and large systems
surveyed by ACT Inc. Costs are broken down into four major components, with
the exception of interest expenses which were not allocated to particular cost
components, and have been shown separately.
The data show that on average the largest single component of total cost
is for distribution and transmission. Treatment costs are on average the
smallest component of total costs. The cost structures of large and small
systems are quite similar, but the average unit costs are lower for the larger
systems than for the smaller systems.
3. Factors Determining the Major Components of the Costs of Water Supply
Systems
This section discusses the factors that determine the costs of
acquisition, treatment, distribution and transmission, and support services of
water supply systems and illustrates the variation of costs of each component
among water supply systems.
The data presented in this section were taken from the 1976 survey
conducted by ACT Inc. The service population for each of the systems was
estimated by dividing annual supply of revenue-producing water by 60,000
gallons, the average annual per capita consumption for utilities surveyed by
the American Water Works Association (AWWA) in 1981. The costs presented are
unit costs per million gallons of revenue-producing water, inflated to end of
1984 dollars using the Consumer Price Index (CPI). For each cost component,
the system unit costs (on the vertical axis) plotted against the estimated
service population (on a logarithmic scale on the horizontal axis) are
presented in a graphical exhibit.
Acquisition
Acquisition costs depend primarily on the characteristics of the water
sources for the system. At one extreme, these costs may be very low where
reliable natural sources are sufficient to meet the needs of the system. For
example, the Culpepper, Virginia, water utility acquired all of its water in
1976 from a free-flowing stream that fed directly into its treatment plant.
The flow of the stream was regulated by upstream reservoirs constructed to
control flooding using federal funds. Therefore the utility had no
acquisition costs. Other systems benefit from proximity to natural surface
-------
-3-
EXHIBIT 1
AVERAGE COST STRUCTURES FOR WATER SUPPLY SYSTEMS
Small Systems a/
Large Systems b/
Acquisition
Treatment
Distribution and
Transmission
Support Services
Interest Charges
Total
Cost per
Million
Gallons of
Water c/
$160
132
310
123
138
863
Percent
19
15
36
14
16
100
Cost per
Million
Gallons of
Water c/
$112
75
221
179
137
724
Percent
15
10
31
25
19
100
a/ Systems saving between 100 and 50,000 people.
b/ Systems serving over 100,000 people.
c/ Unit costs inflated to end of 1984 dollars using CPI,
SOURCE: ACT Inc..
-------
-4-
water bodies, or reliable stream or river sources. In these cases,
acquisition costs typically are limited to pumping water from the natural
source to the treatment plant, although payments for rights to use the water
may be required.
In general, acquisition costs may be incurred in the construction and
maintenance of dams and reservoirs to collect and store surface water, in the
development and operation of well fields to utilize ground-water sources, and
in pumping water from the source to treatment plants.
Exhibit 2 shows the unit acquisition costs of the systems surveyed by ACT
Inc. The highest unit costs among the 12 larger systems surveyed were
incurred by San Diego water utility. In 1976 this utility took about 90
percent of its water from the Colorado River and hence incurred substantial
costs in transporting the water via aqueduct to its storage reservoirs, and in
constructing and maintaining its storage reservoirs. The data indicate scale
economies in acquisition. Larger systems in general have lower unit
acquisition costs than the smaller systems. A few of the high acquisition
costs among small systems are biased because the acquisition costs are for
water purchased from nearby utilities; this purchased water has already been
treated and consequently these purchasers have no treatment costs.
Treatment
Treatment costs depend on the quality of the water source. The physical,
chemical, and bacteriological characteristics of surface water can vary on a
daily and seasonal basis, depending on rainfall, temperature, flow rate, and
the character of materials deposited over the run-off area or discharged
upstream. Daily monitoring and adjustment of the treatment process may
therefore be necessary. On the other hand, as for the New Haven Water Co.,
the run-off area for the source may be carefully controlled, water may undergo
natural filtration during run-off, and consequently require no treatment or
merely a small volume of chemicals added directly to the reservoir.
Similarly, ground water may require only limited treatment.
Exhibit 3 shows the unit treatment costs for the systems surveyed by ACT
Inc. Treatment costs typically range from 0 to 300 dollars per million
gallons, but in extreme cases may be as much as three times greater. The data
also indicate that there are scale economies in treatment.
Distribution and Transmission
Distribution and transmission costs depend primarily on the distance from
treatment points to the service population, the altitude of the population
served relative to the treatment points, and the density of the service
population. As the distance or height that water must be pumped to the
service population increases, the unit costs of distribution and transmission
increase. Similarly, increased dispersion of the service population raises
costs.
-------
-5-
EXHIBIT 2
CO
a. 2
O w
U. H-
CO
z >-
O CO
t >-
CO
D
Q.
0.
O CO
u. ui
O h-
CO
LLJ
53
D UJ
CO
O
O
00
o
O
(O
§
o
o
o
o
o
o
o
»
o
o
o
o
o
O *•»
UJ (D
> "5
K Q
_ u
ll
_i a
D 0»
a. o
o
o
o
o
o
o
o
o
o
CM
i!5
QC O W
a«2
?3o
§^?
o o u
u
c
O
<
-------
-6-
EXHIBIT 3
Q
UJ
H
O
111
_J
HI
CO
CO
QC LU
O •-
u. co
h-
z
UJ
>
CO
O
O
O
*
O
O
O
O
O
q
o
o
cr
UJ
CO
H
co
o
o
o
o
o
•I
o
o ~
UJ O
8 1
l
< b.
u eo
O 0»
o
o
o
o
— o
1
o
o
o>
1
0
o
00
1 1
o
o
(O
1 1 1
o
o
^>
1 1
o
o
ca
1 1
o
§Sj
o o
!-
O
<
o
u
3
O
CO
-------
-7-
Exhibit 4 presents the unit costs of distribution and transmission of the
systems surveyed by ACT Inc. The data clearly show scale economies in
distribution. It is likely that these are the result of the correlation
between service population density and service population size for these
systems. This component of cost is a substantial fraction of total costs for
water supply utilities and typically ranges between $200 and $500 per million
gallons supplied.
Support Services
The unit costs of support services are essentially independent of the
physical characteristics of the system. These costs are primarily
administrative and customer service costs and therefore depend primarily on
the characteristics of the service population and the level of service chosen
by the utility. The data in Exhibit 5 show that these costs are typically
about $200 per million gallons. The data point around $1,400 per million
gallons is anomalous because that system was built to serve a much larger
service population than suggested by the volume of water supplied at the time
of the survey.
4. The Level and Variation of Costs for Water Supply Systems Serving
Populations of Different Sizes
This section presents data on the total costs of water supply systems.
The data indicate the variation in costs among systems serving similar
populations, and the variation in average costs among systems serving
populations of different sizes.
Exhibit 6 shows the unit operating costs (including depreciation and
interest charges) for water supply systems in 12 size categories. Each point
on the graph represents the average unit operating cost (cost per million
gallons of revenue producing water) for a sample of water supply systems
serving the population shown on the horizontal axis of the graph. Unit costs
are shown on the vertical axis of the graph. These data are from a survey
conducted by Temple, Barker, and Sloane in 1982. Costs have been inflated to
end of 1984 dollars using the CPI. The square symbols indicate averages for
systems that predominantly use surface-water sources. The crosses indicate
averages for systems that use predominantly ground-water sources. Average
unit costs for ground-water systems in the two smallest size categories were
both greater than $3,000 and consequently have not been indicated on the
graph. Average unit costs for surface-water systems in the three smallest
size categories (represented by the three solid symbols in the lowest
left-hand corner of the graph) are averages for small samples (five
observations or less) and consequently are somewhat less reliable than other
data (ten or more observations). Similarly, the averages for ground-water
systems in the largest two size categories are averages of small samples.
Therefore the data focuses on systems serving between 2,000 and 300,000 (eight
size categories), which are the most reliable because they represent averages
of fairly large samples.
The data show that there are scale economies in system operation. Systems
serving populations of approximately 300,000 have average costs of about $600
-------
-8-
EXHIBIT 4
Z
o
CE^
§2
<§>
Z CO
2o
H- UJ
CD Q
DC UJ
CO <
CO^
°0
°£
I I
i i r
i i i
o
o
o
ft
o
o
o
o
o
o
f,
o
o
o
o
o
o ~
UJ O
_ o
O §
_J OB
D 01
Q. O
o
o
o
o
o
o
o
o
o
o
o
o
10
2 DC *
2PS
H < o
i
OC O "0
a. w <-
84
O O
o
c
o
<
-------
-9-
EXHIBIT 5
CC
£ co
CO UJ
UJ H
O CO
UJ
CO
Q.
D
CO
CC
o
Q.
Q- CC
D UJ
CO H
U- ^
O £
co Q
h- UJ
co i-
8§
H- UJ
o
o
IO
I I I I I I I I
o
o
o
T I I
o
o
q
o
o
o
o
o
o
•
o
o
o
o
o
o
o
o
o
h- o
o
o
in
o
3 <
tt •
o
O
o
c
CC
UJ
Q.
oo 5 r
o
O <
O O
UJ
O
CO
------- An error occurred while trying to OCR this image.
-------
-11-
pe.r million gallons, whereas systems serving populations between 2,000 and
20,000 have average costs in the range of $800-$1,400. Also, for systems
serving over 2,000, there appears to be little difference between the average
costs of systems that use predominantly ground water and systems that use
predominantly surface water. However, the data indicate that the average unit
costs of small ground-water systems are extremely high; much greater than the
average unit costs of the small surface-water systems surveyed.
For reference, based on annual usage of 163,800 gallons per household1
per annum, costs of $1,000 per million gallons imply annual costs of $163.80
per household served.
Because the data presented in Exhibit 6 are averages of samples of water
supply systems in a range of size categories, they do not show the amount of
variation in the costs of systems in each size category. Exhibit 7 presents
the data in tabular form showing the standard deviation of the costs in each
of the size categories. Inspection of the standard deviation of the costs
shows that, in general, there is a greater variability among the costs of the
smaller systems than among the costs of the larger systems both for
ground-water and surface-water systems.
Exhibit 8 shows unit operating costs for individual water supply systems
surveyed by ACT Inc. Each point on this graph represents the unit operating
costs (costs per million gallons of revenue-producing water) of a particular
system. Costs have been inflated to end of 1984 dollars using the CPI. Unit
costs are shown on the vertical axis and the approximate population served by
the system is shown on the horizontal axis. (The horizontal axis figure was
calculated by dividing the annual volume of water delivered by the system by
60,000 gallons to give the approximate population served. Sixty thousand
gallons was the average annual water usage per person for the 1,397 utilities
surveyed by the AWWA in 1981.) The scale of the graph in Exhibit 6 is
identical to the scale of the graph in Exhibit 8.
The data in Exhibit 8 provide a very similar picture of the level and
variation of unit costs to the data in Exhibit 1. There is clear evidence of
scale economies, and the variation of unit costs among the small systems is
much greater than the variation of unit costs among the large systems.
Average unit costs are higher for the systems in Exhibit 1. This can perhaps
be explained by the difference between the time periods of the two samples and
the choice of using the CPI to inflate costs. The bundle of goods whose
prices are reflected in the CPI is likely to differ significantly from the
typical bundle of inputs used by a water supply system. The CPI has been used
in this exercise expediency. A more accurate picture of cost inflation for
water supply systems may be obtained by using a price index based on goods
that are more similar to the typical water supply input bundle.
1 Annual household usage was approximated by multiplying the average
annual consumption per person for the 1,397 utilities surveyed by the AWWA in
1981, 60,000 gallons, by the average number of people per household, 2.73,
from the 1985 Statistical Abstract.
-------
-12-
CO
CO
co
>CO
_l U
Q-N
Q- —
3) CO
CO
CO
a: =
LJ O
r-
(-
00
X
LJ
OLl-
O
CO
LJ CO
a.0
CO *J
S- CO O. C -4-> CO
Q- — J" >) > 4->
CTD W — CO W CUV3
C CO O (0 ON T3 O
— O O)— <- O
u) > ~ o -a -a
3— CUC4-1/) 0) S-JJ
•p woos- i->, re —
U)C CO— CO dlCU "DC
Eto < -0— -Q> C3
OC1-I » — C — Ei- CO
•"-> — O > — Q) O 33 •<->!>-
w E +j < e —-D z w co o
> O to
COI-D 3]
CM — 1
CO 1
00
CM lA VO
CM —
lA
VO VO r-
ON CM CM
lA
— J- —
lA — CM
VO
— .3- to
d- — cvj
VO
o r- o
lA CO LA
1^
CO CM VO
CM lA ro
f-
r^ vo vo
ro J- VO
ON
O 0 —
— lA h-
lA —
*
0 VO O
C\J lA 00
vo
•N
O CO ON
in vo —
m in
CO
O CM O
LA M J"
in ON
l~- —
c
i/> O
it- E
CD "D O V> 1) 4J
<- c a c 4J to
a. 3 o « — u)
O •>-> — d- X > •"
a*- w — 03 M cut/)
CO O 50 ON TD O
O Ol— <<- O
U) > "- O "D 'O
O— 1) C t— W CD S-iJ
•u aiooi- i->> re —
t/)C to-- CO CD!) "DC
Ere i -a— n> C3
Ci>C!- 03 — C — Ei- to
JJ.-QJ >— (J)O 33 *Jt._
W E *•> < e ^"D Z W CO O
> O to
tol-o s
o
a
c
re
o
CO
c«
t-
X
re
ffl
a
E
CD
LJ
O
cr
z>
O
CO
-------
CO
-13-
EXHIBIT 8
UJ
h-
co
>•
CO
Q.
a.
D
CO
DC
UJ
Q
HI
UJ
_l
LLJ
CO
u_
O
CO
h-
co
O
O
o
o
o
I I I I I I I I I
o
o
q
o
o
o
o
o
q
o"
o
o
o
o
o ~
UJ 0>
s g
£w
81
-J «D
I> O)
Q. O
2*
o
o
o
o
h- o
0
o
O
§
§
z
O
u.
cc o
o
c
O
0)
u
o
O
CO
-------
-14-
5. Use of Cost Data to Estimate Water Supply Replacement Costs
The data presented in the previous sections show total costs of water
supply systems and their breakdown among major cost components. This data may
be used to estimate typical costs of replacement of contaminated ground-water
supply sources under various scenarios. There are three basic possibilities:
(1) replacement of acquisition capacity only; (2) replacement or addition of
treatment capacity as well as acquisition capacity; and (3) replacement of
acquisition capacity, addition of treatment capacity, and addition of a
distribution and transmission system. The first of these cases might arise,
for example, when a utility replaces lost ground-water sources simply by
drilling new wells in the same aquifer outside the contaminated area. This
water is treated using the existing treatment plant and distributed through
the existing distribution network. The second scenario would arise if the new
water supply source differed in quality from the lost water source, and thus
required additional or different treatment processes. The third case could
arise when water from private wells serving individual households is replaced
by centrally provided water, necessitating the construction of a distribution
network for the service population.
Costs of water supply replacement depend on the scope of measures
necessary to provide an alternative potable supply. The cost component data
presented in this paper can be used to estimate typical costs of supply
replacement depending on the scope of measures necessary to provide the
alternative supply.
-------
APPENDIX
POTENTIAL SOURCES OF COST MIS-ESTIMATION
In the analyses, effort has been made to account for the full costs of
water supply and treatment facilities. However, there may be additional
expenses incurred in the construction of water supply systems that have been
overlooked. The following discussion will attempt to first identify and then
approximately quantify these additional cost sources in order to assess the
potential impact upon the aggregate cost estimates.
The first possible source of cost underestimation arises from the choice
of price inflator. The cost estimates used in the analysis were originally
compiled in 1981 dollars. In order to express the costs in real or current
dollars, the Consumer Price Index (CPI) was used to inflate the 1981 values.
Unfortunately, no price statistics are available that specifically deal with
the costs of water treatment plant construction and operation. Hence, the
choice of a price inflator is somewhat arbitrary. However, one may reasonably
question whether the CPI is an appropriate inflator for this application.
As a check against the CPI, the construction cost index (CCI) from the
Engineering News Record was also used. In a phone conversation with the
American Water Works Association Office in Denver, a member of the technical
services staff indicated that the CCI is appropriate for this application.
The comparison of the two indexes indicates that the choice of price
inflator makes little difference in the analysis. Using the CPI, an inflation
multiplier of 1.16 was derived, while using the CCI, this multiplication
factor would be set at 1.18. Therefore, if the construction cost index were
applied to this analysis, the cost estimates would be increased by less than 2
percent.
A second source of additional costs could arise from the legal and
administrative expenses associated with water acquisition, transport, and
treatment plant construction. In a 1979 study published by the Municipal
Environmental Research Lab of the U.S. Environmental Protection Agency2,
legal, fiscal, and administrative expenses were estimated as a function of the
total construction, engineering, and land costs for a water treatment
project. These estimates reveal that for a water treatment project with
construction costs of approximately $100,000, the legal and administrative
costs are expected to total approximately $4,000 or only 4 percent of the
capital expenses. For projects with construction costs in excess of
$1,000,000, the legal and administrative expenses decline to less than 2
percent of the capital expenditures. Thus, legal and administrative fees
should increase project expenses by only a small amount.
Charges for land acquisition and right-of-ways are another potential
contributor to project costs. The documentation accompanying the Temple,
Barker, and Sloane and ACT studies does not reveal whether these costs were
included in the reported costs. One could reasonably assume that expenditures
-------
-2-
for land would be included in the cost summaries; however, it is prudent to
examine land cost as a potential source of additional cost.
The EPA Municipal Research Lab estimates3 that a plant large enough to
supply water to approximately 30,000 people would require 2 acres of land.
Assuming the land is priced at $3,000/acre and the plant construction cost is
close to $3,000,000 (EPA estimate), the land acquisition would represent less
than 0.2 percent of the total cost.
Estimating right-of-way costs is a somewhat more difficult problem,
however, because no data is available concerning the price of easements for
water supply distribution systems. There is anecdotal information available
that discusses the costs of oil and gas pipeline easements. Attachment 1
lists the low, median, high, and estimated average costs of obtaining a
one-mile by 50 foot right-of-way. These costs are for easements containing
oil and gas pipelines ranging in size from 20 to greater than 42 inches. The
prices also include expected damages to the property value of the land during
construction of the pipeline.
To derive some measure of an average easement cost for water supply
systems, an average was taken of the low cost values (and median values when
the low values were not available for an area) for all the areas listed. The
low values were used because the piping required for a water distribution
network should on average be smaller than that used for oil and gas delivery.
Consequently, the land requirements should be lower for water supply pipes.
An average of $14,700 per mile of right-of-way was determined.
Values for the average length of distribution network .for different water
system sizes were taken from Temple, Barker, and Sloane (Attachment 2). These
distribution network distances exclude service connection piping.
The total cost for the right-of-ways for each system size was found by
multiplying the length of distribution network by the right-of-way cost.
These values were then multiplied by 0.5 to reflect cost reductions due to the
estimated reduction in land requirements needed for the piping of water
distribution systems. The total cost values are reported in Attachment 2.
These costs were then annualized by assuming 30 year depreciation and a 10
percent interest rate. The annual costs are reported in terms of millions of
gallons supplied.
Attachment 3 lists the annual right-of-way costs as a percentage of the
expected unit costs for each system size. This analysis estimates that for
small systems the impact of right-of-way costs could increase annual expenses
by approximately 20%. For larger systems, the impact declines to 10% of
annual costs.
2 Robert C. Gumerman, et al., Estimating Water Treatment Costs,
Municipal and Environmental Research Lab, U.S. EPA, 1979.
3 Ibid.
-------
-3-
These findings represent only very rough estimates of right-of-way
expenses. The true costs for water supply right-of-ways are unknown, and the
use of municipally owned land in some locations may reduce the expenses. In
addition, the amount of land required for the right-of-way may be less than
that used for this analysis for systems with small piping requirements.
Therefore, taking into account the inherent uncertainties in the analysis, an
increase of 15% over the estimated annual costs would appear to be a
reasonable upper bound estimate of the potential impact of expenditures for
distribution right-of-ways.
The above figure of 15% should be viewed only as an approximation of the
potential magnitude of right-of-way costs. Since the figure is primarily
illustrative, it will not be carried through the remainder of the analysis.
Another area of potential cost misestimation stems from the unknown impact
of federal grant support on the costs reported in the studies used for this
analysis. In the 1970s, federal grants and loans from the Department of
Agriculture, the Interior Department, and Housing and Urban Development aided
water treatment construction projects in many communities. In a 1977 study by
the University of Georgia Institute of Natural Resources*, it was reported
that for communities receiving Federal support, on average 36 percent of the
capital costs for water development were covered by Federal grants with
smaller communities receiving a larger proportion of support than larger
towns. One should also note, however, that according to an official of the
Farmers Home Administration in the Department of Agriculture, the level of
Federal funding available for water treatment projects decreased by over 2/3
between 1980 and 1985.
Although it is unknown how federal grants have been accounted for in the
water system cost estimates in the ACT and TBS studies, if the federal support
were added to the previously reported cost figures, the Georgia study suggests
that the total average annual cost for each municipality would still increase
by less than 5 percent. The Georgia study reports that the annual expenses
for operating and maintaining a water treatment facility are much greater than
the annualized costs of paying off the initial capital expenditures for
construction. (This conclusion was also borne out in the present analysis and
is discussed in greater detail in a following section.) Because the Federal
grants only fund construction expenses, the operating costs must be fully paid
by the local operators of each facility. As a result, if the construction
grant is depreciated over the life of the facility, the Federal dollars
represent only 4.2 percent of the annual operating budget of an average
Georgia water supply system according to the 1977 study. Thus, if the grants
were overlooked in the cost figures used for the preceding analysis, the
annual expenses for a typical system would increase only by approximately 4
percent.
* Ronald North, Financing and Cost Sharing of Municipal Water Supply
Systems, Institute of Natural Resources, University of Georgia, 1977.
-------
-4-
The final area of concern deals with inherent cost differences between
existing and replacement facilities. We would expect that existing systems
represent least cost solutions to water supply needs and that, on average,
replacement systems would have higher costs than existing systems.
Unfortunately, this factor cannot be addressed with existing data.
Furthermore, new systems may use different techniques and technologies than
existing systems. The effect of such technical changes on water supply system
costs is also unknown. Therefore, an arbitrary figure of 15 percent was
chosen as the factor to account for the difference in total costs between
existing and replacement systems.
In order to determine the total impact of the potential cost
underestimates, one must first note that many of the cost factors that were
discussed would primarily affect capital costs. Only the choice of price
inflator index and the inherent differences between existing and replacement
systems discussed here also have an effect upon operation, maintenance, and
other expenses. Therefore, the impact of the possible misestimates must be
evaluated in light of the level of capital expenses relative to other expenses.
As previously mentioned, operation and maintenance account for a majority
of the expenses for a water supply system. Temple, Barker, and Sloan report
that, for the water systems they surveyed serving more than 2,500 people,
between 67 percent and 78 percent of the average annual expenses for each
facility were taken up by operation and maintenance costs. Overall,
depreciated capital expenses never accounted for more than an average of 18
percent of the annual expenditures. Furthermore, ACT reports that, over a
ten-year period from the mid-1960s, the facilities they studied experienced a
faster increase in operation and maintenance (O&M) expenses than in capital
costs. Annual O&M expenditures rose 161 percent while capital costs increased
only 117 percent. Thus, annual payments on capital expenses appears to be a
relatively small and declining portion of annual costs.
Legal and administrative costs together with land costs appear unlikely to
exceed 5 percent of total capital costs. As annualized capital expenses are
on average no more than 18 percent of total annual expenses, these costs on an
annualized basis are unlikely to exceed 1 percent (5 percent multiplied by 18
percent) of annual expenses. The Georgia study indicates that on an
annualized basis Federal grants and loans amount to about 4 percent of total
annual costs. These factors, therefore, combine to about 5 percent of annual
costs. Furthermore, the misestimation of inflation is unlikely to result in
more than a few percent of costs. Rights-of-way costs appear to be possibly
more significant, based on estimates for oil and gas pipelines. However,
direct information on the level of these costs has not been analyzed.
Potentially most significant are inherent cost differences between existing
and replacement systems.
The approximate magnitude of various components of the potential cost
mis-estimation are summarized in Exhibit A-l. Overall, we have inflated
existing system costs by 20 percent to estimate replacement system costs to
account for all of these factors in the subsequent analysis.
-------
-5-
EXHIBIT A-l
APPROXIMATE MAGNITUDE OF SOURCES
POTENTIAL COST MIS-ESTIMATION
Source of Cost Mis-Estimation
Legal and administrative costs
Land costs
Rights of way
Federal loans and grants
Inherent differences between
existing and replacement systems
Total allowance for cost mis-
estimation in subsequent analysis
Potential Overall Impact as a
Percentage of Total Annual Costs
~ 1 percent
~ 15 percent
~ 4 percent
Unknown
20 percent
-------
ATTACHMENT 1
RIGHT OF WAY COSTS PER MILE
FOR OIL AND GAS PIPELINES
Location Area
NORTHEAST
New York
New Hampshire
New Jersey
Connecticut
Massachusetts
Pennsylvania
Penn/NJ Combo
NY/Penn Combo
Maine/NH/Mass/RI Combo
Low $
2,900
32,000
44,000
48,000
3,900
1,150
27,800
20,800
Medium $
11,300
50,000
43,000
21,000
70,000
High $
85,000
32,000
216,000
100,000
96,250
49,000
39,000
24,000
Estimated
Average $
22,000
85,000
41,250
20,750
Comments
2 (80,000)
costs were
w/o damage
SOUTHEAST
Delaware
Florida
Missouri
Virginia
Tennessee
Kentucky
Maryland
West Virginia
WV/Ohio
Tenn/Vir
SOUTH CENTRAL
Louisiana
Oklahoma
Texas
Kansas
Arkansas
Okl/Texas Combo
Okl/Kansas Combo
Tenn/Ark Combo
Ark/Louisiana Combo
Louisiana/Miss Combo
2
8
2
3
8
8
3
16
14
2
,900
,000
970
,500
,300
,800
,000
,000
,000
,950
,600
8
5
55
6
10
1
40
14
14
15
18
15
18
16
38
600
,200
,100
200
,500
,500
,700
,200
,000
,400
,400
,200
,400
,200
,000
,000
,500
8
9
6
5
12
52
17
67
38
17
2
,300
,400
,400
,500
,000
,000
,900
,500
,500
,900
,800
8
5
6
39
13
26
18
15
,200
,100
,400
,000
,250
,500
,000
,200
10" pipe
4" pipe
1/2 mile
line
-------
ATTACHMENT 1 (continued)
Location Area
NORTH CENTRAL
Minnesota
Ohio
Wisconsin
Michigan
Iowa
Iowa/Minn Combo
Mich/Wise Combo
Illinois/Iowa Combo
Indiana/Ohio Combo
Illinois/Indiana Combo
Ohio/Penn Combo
WEST
Colorado
Wyoming
N. Dakota
New Mexico
Montana
Arizona
Utah
Wyoming/Utah Combo
Colorado/Kansas Combo
Col/Utah/New Mexico Combo
Low $ Medium $ High $
Estimated
Average $
Comments
4,800
700
6,400
6,500
13,500
12,200
7,200
9,000
6,700
21,500
32,000
27,000
41,500
32,200
29,000
12,500
6,400
17,500
17,800
563,000
7,200
9,000
37,000
2,600
1,100
6,400
7,750
7,000
11,300
5,000
4,500
13,200
3,200
12,200
10,500
22,000 9,250
21,400 8,000
7,300
10,500
1. Figures obtained from:
Oil & Gas Journal Nov. 26, 1984
Oil & Gas Journal Nov. 28, 1983
Oil & Gas Journal Nov. 22, 1982
Oil & Gas Journal Nov. 23, 1981
-------
ATTACHMENT 2
CHARACTERISTICS OF AVERAGE DISTRIBUTION NETWORKS
AND RIGHT-OF-WAY COSTS
System Size
(Avg. Population
Served)
63
300
750
2,150
6,650
17,500
37,500
62,500
87,500
300,000
750,000
1,000,000
Avg. Miles of1
Distribution
2.9
9.8
18.0
39.3
87.0
146.8
280.1
379.8
420.8
883.8
1,835.3
2,283.9
Total $ Cost for2
Right-of-Way
21,315
72,030
132,300
288,855
614,450
1,078,980
2,058,735
2,791,530
3,092,880
6,495,930
13,489,455
16,786,665
Annualized3
Right-of-Way
$Cost/Million Gallons
480
694
452
278
200
112
92
74
58
60
32
10
1 Source:
2 Source:
Temple, Barker, and Sloane.
Oil and Gas Journal Nov. 26, 1984
Oil and Gas Journal Nov. 28, 1983
Oil and Gas Journal Nov. 22, 1982
Oil and Gas Journal Nov. 23, 1981,
and ICF analysis.
3 Source: Temple, Barker, and Sloane, and ICF analysis.
-------
ATTACHMENT 3
RIGHT-OF-WAY COSTS AS PERCENT OF UNIT COSTS
PER MILLION GALLONS SUPPLIED
Annualized Right-of-Way
System Size $ Cost as Percent of
(Avg. Population Served) Unit $ Cost Per 10s Gallons
63 6.4
300 20.8
750 26.8
2,150 18.4
6,650 21.4
17,500 15.4
37,500 12.2
62,500 11.6
87,500 9.0
300,000 10.0
750,000 6.2
1,000,000 1.2
Source: Temple Barker, and Sloane, and ICF analysis.
-------
CLASS I: EMPIRICAL EVIDENCE ON THE IMPLICATIONS
OF VARIOUS SUBSTANTIAL POPULATION THRESHOLDS
The U.S. Environmental Protection Agency's Ground-Water Protection
Strategy defines three classes of ground water. Class I ground waters are
defined as those that are highly vulnerable to contamination because of
hydrogeological characteristics and that are either irreplaceable for a
substantial population or ecologically vital. This paper focuses on the
implications of different thresholds for the substantial population test.
The Strategy does not currently define a size cut-off for the substantial
population criterion. The threshold ultimately chosen for this test has the
potential to significantly affect the results of the classification scheme
because as the threshold is increased, the number of ground waters likely to
qualify for Class I will decrease while a reduction in the threshold will lead
to more Class I determinations. The purpose of this paper is to analyze the
issues associated with the choice of a particular substantial population
threshold and to present available empirical evidence about the results of
different thresholds. Section 1 first presents an overview of the policy
issues associated with the choice of a substantial population threshold.
Section 2 describes an analysis of alternative population thresholds based on
public water systems while Section 3 presents an analysis based on both public
water systems and private wells. Section 4 concludes with a summary of the
analysis.
1. ISSUES ASSOCIATED WITH THE SUBSTANTIAL POPULATION TEST
In order to implement the Ground-Water Protection Strategy, it may be
useful for the draft guidance to define precisely what is meant by a
substantial population. Without such guidance, it may prove difficult to
achieve a uniform and consistent implementation of the Strategy. In general,
the issues connected with the substantial population test can be analyzed by
considering the two extremes. This section thus describe the effects of both,
a very low and a very high threshold. The issue of precisely which water
users -- public, private, or both -- ought to be included in the test is
examined next. This section concludes with an overview of the need for the
analysis that is presented in Sections 2 and 3.
Low Substantial Population Threshold
EPA may choose to implement a relatively low population cut-off for the
substantial population test. The result of doing so is likely to be a
substantial number of Class I designations. As noted in the data analysis in
Section 2, the number of public water systems falling into Class II appears to
grow disproportionately as the system size threshold is reduced. Thus, a
small reduction in the population threshold can produce a substantial increase
in the number of systems above the population cut-off and thus eligible for
Class I.
EPA may further endeavor to avoid a low population threshold in order to
preserve the notion that Class I ground waters are somehow special and thus
-------
-2-
deserving of extraordinary protection. As the population threshold is
lowered, more ground waters would be classified as Class I and the argument
for special protection is somewhat weakened. Simultaneously, the number of
Class II ground waters would decrease. EPA may consequently create a higher
level of environmental protection than is justified by the circumstances of a
particular case (i.e., the policy may "over-protect" certain ground waters).
High Substantial Population Test
In addition to the disadvantages of an especially low population
threshold, there are two key drawbacks to a particularly large population
cut-off. First, as the threshold is increased, the number of ground waters
eligible for Class I will decline. If the threshold is set "too" high, the
number of Class I ground waters may be insufficient to ensure the level of
environmental protection intended by the Ground-Water Protection Strategy.
Second, many water systems draw water from more than one ground-water source.
If a system's users are "pro-rated" among its sources, the number of users of
any one source who are evaluated for the substantial population test may be
significantly smaller than the whole user population. (For example, in a
system with 100 users and 2 well groups, the 50 users per well group would be
used for the test.) If this method is adopted for the test, a particularly
high population cut-off may lead to few, if any, Class I determinations.
Public System Users and Private Well Users
There has been some discussion of drawing a distinction in the substantial
population test between the number of persons using a public water system and
the number using individual private wells. The rationale for doing so appears
to reflect the fact that the cost of providing water to households not
currently connected to a distribution system may prove larger than the costs
of securing a new source of water for households already hooked up to a
central system. For the following reasons, it may be the case that EPA ought
to set the same population threshold for both types of water users:
• The cost elements of alternative water systems are
already incorporated in the irreplaceability test.
• It may prove difficult to justify differential
treatment of similar individuals (i.e., residential
water users).
Furthermore, in many instances, private well users and public water sources
are in close proximity and draw from the same ground water. When evaluating a
particular ground water, it seems reasonable to consider all users regardless
of whether the well is owned publicly or privately.
Need for Analysis
In short, there are disadvantages to setting the substantial population
threshold at either a particularly low or especially high level. EPA must
therefore select some moderate level for the test that minimizes the
-------
-3-
disadvantages of the two extremes. Consequently, five alternative substantial
population thresholds are analyzed in this paper to determine the effects of
each. Due to data limitations, the primary focus of the analysis is on public
water systems; however, a somewhat limited analysis based on public and
private ground-water users is also presented.
2. CHARACTERISTICS OF WATER SUPPLY SYSTEMS
For the purposes of this analysis, two types of data have been obtained.
The first, drawn from the Federal Reporting Data System (FRDS), describes the
characteristics of public water systems. Using FRDS data, it is possible to
analyze various issues associated with the substantial population test as it
relates to public systems. FRDS does not, however, contain data on private,
individual water wells. Information on such wells was obtained from a second
source: the 1980 U.S. Census. Census data was combined with FRDS data to
produce estimates of combined public/private water usage in selected areas.
The following sections describe in more detail how the data were analyzed and
provides a brief characterization of water supply systems. The remainder of
Section 2 investigates public water supply systems while Section 3 summarizes
the results of our combined analysis.
2.1 Characteristics of Public Water Systems
There are a number of ways to characterize public water systems. To begin
with, we have divided the set of public water systems predominantly using
ground water into twelve groups based on the number of persons served by
each. For each size group, three key characteristics have been determined:
• The number of systems for which ground water is the
principle source,
• The number of well groups used by these systems, and
• The number of persons served by these systems.
Each of these characteristics will be used to assess the results of
alternative substantial population thresholds. First, however, a brief
overview of the water supply industry in terms of these characteristics is
presented.
Number of Systems: As indicated by Exhibit 1, there are over 47 thousand
public water systems in the United States for which the principle source of
water is ground water. The vast majority of these systems are quite small.
In fact, more than 33,000 public water systems (about 71 percent) serve less
than 500 people while only 45 systems serve over 100,000 users.
-------
-4-
EXHIBIT 1
ESTIMATED NUMBER OF GROUND-WATER AREAS
USED BY PUBLIC WATER SYSTEMS
Number
of People Served
by the System
25-100
101-500
500-1,000
1,001-2,500
2,501-3,300
3,301-5,000
5,001-10,000
10,001-25,000 .
25,001-50,000
50,001-75,000
75,001-100,000
over 100,000
TOTAL
Number
of Systems1
Using Groundwater
18,155
15,782
4,855
4,389
891
1,103
1,143
819
273
67
15
45
47,537
Average
Number of Well2
Groups/System
1.0"
1.0*
1.3
2.2
2.2
2.1
2.1
2.1
2.6
7.4
5.5
27.8
Number
of3
Well Groups
18,155
15,782
6,312
9,656
1,960
2,316
2,400
1,720
710
496
83
1,251
60,841
1 From Federal Reporting Data System.
2 From Temple, Barker, and Sloane.
3 Calculated by multiplying the number of systems using ground water by
the average number of well groups for each system size category.
4 Systems with a single well appear to have been reported as having no
well groups. Thus, the average number of well groups per system may be biased
downward. Indeed, for the two smallest system sizes, the reported average
number of well groups per system is less than 1.0. For all calculations in
this paper, we have assumed that the average number of well groups for systems
of this size is equal to 1.0.
-------
-5-
Number of Well Groups: The number of well groups is important because it
can be used as a proxy for a "ground water." The rationale for doing so is
that well groups are likely to be distinct from one another and will be
located in different aquifers or in separate parts of the same aquifer. As
such, each well group constitutes a distinct water source and potentially
could have a unique ground-water classification assigned to it. As shown in
Exhibit 1, systems serving between 500 and 1,000 people use, on average, 1.3
ground-water sources while systems serving between 50,000 and 75,000 persons
use an average of 7.4 well groups. As the rightmost column of Exhibit 1
indicates, most of the 60,841 well groups used by public water systems are
used by small systems.
Population Served: Despite the fact that small systems dominate the
industry in terms of the number of systems and the number of well groups, such
systems actually serve only a small fraction of the people receiving their
water from public water systems using ground water. Because small systems
each serve so few people, the total number of persons supplied by these
systems is quite low when compared to the number of people receiving water
from big systems. As indicated in Exhibit 2, the 45 largest public water
systems using ground water serve about 18 million people while the more than
18,000 smallest systems which each serve less than 100 customers provide water
to barely more than 1 million persons.
Also shown in Exhibit 2 is the average number of persons served by each
well group in systems of different sizes. Because it was computed by dividing
the average number of persons served by the average number of well groups,
variations among systems in the number of persons per well group may be not be
evident; we believe, however, that the numbers presented are a reasonable
approximation of the true value. The number of persons per well group may
prove particularly important if the draft guidance requires that the system
user population be prorated among its water sources. For example, consider a
system that serves 7,500 people using five well groups. In applying the
substantial population test to a ground water that includes only one of the
five well groups, it must be decided whether to use the entire population of
7,500 or to prorate the users among the well groups and estimate the
population at 1,500 (i.e., 7,500 users divided among 5 well groups). As
Exhibit 2 shows, some of the dissimilarity between large and small systems is
eliminated when using the persons per well group measure.
2.2 Results: Implications of Different Thresholds
Based on the data described above, it is possible to estimate the results
of using different substantial population thresholds. The analysis omits
consideration of private well users and is based on only those persons served
by public water systems. This omission is attributable to limitations
inherent in the available data. To a limited extent, the analysis in Section
3 overcomes this shortcoming by combining data from two sources. In general,
however, it is unlikely that the omission of private well users from the
analysis will affect our results significantly because only a small fraction
of ground-water users obtain their water from private wells rather than a
public system.
-------
-6-
EXHIBIT 2
SERVICE POPULATIONS FOR PUBLIC WATER
SYSTEMS OF DIFFERENT SIZES
Number
of People Served
by the System
Average Number1
Of Persons Served
By Each System
Average Number2
Of Persons Served
By Each Well Group
Total Number3
Served
(in Millions)
25-100
101-500
500-1,
1,
2,
3,
5,
10,
25,
50,
75,
001-2,
501-
3,
301-5,
001-
10
001-25
001-
001-
001-
over
TOTAL
50
75
000
500
300
000
,000
,000
,000
,000
100,000
100,000
1
2
4
7
15
35
61
86
398
55
249
743
,638
,901
,121
,219
,878
,917
,313
,326
,570
55
249
572
745
1,319
1,962
3,438
7,561
13,814
8,286
15,696
14,337
1
3
3
7
2
4
8
13
9
4
1
17
77
.002
.927
.608
.162
.585
.545
.252
.004
.805
.108
.295
.936
.229
1 Calculated by dividing the total population served for systems of each
size by the number of systems of each size. Data from FRDS.
2 Calculated by dividing the average number of persons served by each
system size by the average number of well groups for each system size. Well
group data from Temple, Barker, and Sloane.
3 From FRDS.
-------
-7-
As noted, a key issue affecting the outcome of the substantial population
test is whether all system users are considered part of the population or
whether only some prorated fraction is used in measuring the population.
Accordingly, these two options have been treated as separate scenarios and
results are presented for each scenario.
Scenario 1 - All System Users Considered: The results of this scenario
are presented in Exhibit 3 and are perhaps best understood by working through
an example. If a population threshold of 2,500 is selected and all system
users considered for purposes of the test, then the ground waters used by
roughly 91 percent of the public water systems could not be classified as
Class I. (This result occurs simply because 91 percent of the systems supply
fewer than 2,500 people). Similarly, because these systems use about 82
percent of all well groups, only the remaining 18 percent of the well groups
exist in areas potentially classifiable as Class I. Finally, because of the
fact that small systems serve relatively few persons, a substantial population
threshold of 2,500 users would only exclude 20 percent of the population which
is supplied by public water systems using ground water.
The results of other substantial population thresholds can similarly be
estimated from Exhibit 3. In general, the key finding of the analysis is that
while even a vary low threshold (for example, 1,000 persons) is likely to
exclude from consideration for Class I more than two-thirds of both water
systems and well groups, the total fraction of people whose water would be
excluded from Class I is much smaller.
Scenario 2 - System Users Prorated Among Well Groups: If the substantial
population test is not based on the total number of people served by a system
but is instead based on some prorated fraction of system users that are
allocated to each well group, the results of the analysis may differ
significantly. As the data presented in Exhibit 2 illustrate, while large
systems tend to serve significantly more people from each well group than do
small systems, the difference is not proportional to size. Consequently, the
substantial population test, when applied to only some fraction of a system's
users, will exclude more areas from Class I than when applied to all users of
a public water system.
The results of the analysis for Scenario 2 are presented in Exhibit 4.
The results indicate that even for a threshold of 2,500 persons, water sources
used by about 95 percent of the smallest systems would be excluded from
consideration as Class I. This occurs because small systems, by far the vast
majority, rarely supply more than 2,500 persons from each well group. As
noted before, small systems as a group do not supply water to a relatively
large number of people. Consequently, despite the fact that a 2,500 person
per well group test excludes 95 percent of the systems, fewer than 30 percent
of the persons using ground water would have their water source excluded from
consideration for Class I. Furthermore, under Scenario 2, both of the two
-------
-8-
EXHIBIT 3
IMPLICATIONS OF DIFFERENT SUBSTANTIAL POPULATION THRESHOLDS
BASED ON NUMBER OF PEOPLE SERVED BY EACH PWS
Population1
Threshold
1,000
2,500
5,000
10,000
25,000
50,000
Percent of2
Systems Excluded
81.6
90.8
95.0
97.4
99.2
99.7
Percent of3
Persons Excluded
11.1
20.3
29.6
40.2
57.1
69.8
Percent of*
Well Groups Excluded
66.2
82.0
89.1
93.0
95.8
97.0
1 Based on the service population for each public water system as
reported by FRDS.
2 Based on a total of 47,537 public water systems for which ground water
is the principle source as reported by FRDS.
3 Based on a total of 77.229 million persons supplied by public water
systems for which ground water is the principle source as reported by FRDS.
* Based on a total of 60,841 well groups used to supply public water
systems for which ground water is the principle source. The total was
estimated by ICF based on data from FRDS and Temple, Barker, and Sloane.
-------
-9-
EXHIBIT 4
IMPLICATIONS OF DIFFERENT SUBSTANTIAL POPULATION THRESHOLDS
BASED ON NUMBER OF PEOPLE SERVED BY EACH WELL GROUP1
Population2 Percent of3 Percent of" Percent of5
Threshold Systems Excluded Persons Excluded Well Groups Excluded
1,000 90.8 20.3 82.0
2,500 95.0 29.6 89.1
5,000 97.4 40.2 93.0
10,000 99.3 62.4 96.6
25,000 100.0 100.0 100.0
50,000 100.0 100.0 100.0
1 The number of persons per well group was computed for each system size
by dividing the average number of users by the average number of well groups.
As noted in the text, this approach may be somewhat imprecise. The above data
must therefore be interpreted carefully. In any given size category, there
are likely to be some systems serving more than the average number of users
from each well group and some that serve less. In particular, for those
thresholds that appear to exclude 100 percent of the systems, it is likely
that some systems serve more persons from each well group than the threshold
value and consequently that the ground waters used could be classified as
Class I. Given the available data, it is not possible to refine the analysis.
2 Based on the service population for each public water system as
reported by FRDS.
3 Based on a total of 47,537 public water systems for which ground water
is the principle source as reported by FRDS.
* Based on a total of 77.229 million persons supplied by public water
systems for which ground water is the principle source as reported by FRDS.
5 Based on a total of 60,841 well groups used to supply public water
systems for which ground water is the principle source. The total was
estimated by ICF based on data from FRDS and Temple, Barker, and Sloane.
-------
-10-
largest thresholds considered (i.e., 25,000 and 50,000) result in an almost
complete exclusion all ground waters from Class I.1
3. Combined Public and Private Water Use
The foregoing analysis does not include private individual ground water
wells. There is only limited evidence on the use of such wells but the 1980
Census does contain some information. Selected respondents to the Census were
asked whether they received their water from "individually dug or drilled
wells." Unfortunately, responses to the question are only available
aggregated at the zip code level. It is still possible, however, to draw some
tentative conclusions. This section briefly summarizes how results of the
Census were combined with FRDS data to address the substantial population
issue. The section concludes with an analysis of the implications of the data
analysis.
In order to combine public water system data from FRDS with Census data on
private well users, it was necessary to analyze the nation on a zip code by
zip code basis. From the Census data, we obtained the number of private well
users in each zip. Next, we determined how many public water system users
received water from a source in each zip code.2 We then added the number of
private well users to the service population(s) of the public water system(s)
located in each zip code. The results of the analysis are presented in
Exhibit 5.
There are more than 35,000 zip codes in the United States; of these, about
2,200 contain no ground-water users. The remaining 32,837 zip codes contain
anywhere from a few users to more than 100,000. Unfortunately, zip codes are
not an especially meaningful measure of ground-water areas. For example, a
zip code in a rural area may cover a very large area and include more than one
ground-water source. Similarly, an urban zip code may cover a very small area
in which little or no ground water is available. Zip codes do, however, allow
1 The results in this section must be interpreted carefully. The number
of systems excluded from Class I is based on the number of ground waters that
appear to serve less people than the threshold value. The number of persons
per ground water was calculated based on both the average number of persons
and the average number of well groups for each system size. Because of the
averaging process, the results may therefore be somewhat imprecise. For
example, even though a threshold of 50,000 appears to exclude virtually all
systems, there may in fact be some systems which serve more than 50,000
persons from a single well group. Given the available data, these
imprecisions are unavoidable.
2FRDS contains zip code information only for the main treatment and
distribution plant for each system and does not have zip code information for
water sources. Based on a suggestion from an official of the Office of
Drinking Water, we made the assumption that the ground-water sources used by a
public system are usually in close proximity to the main treatment and
distribution plant operated by the system. Given this assumption, we also
assumed that public wells are located in the same zip code as the plant.
-------
-11-
EXHIBIT 5
PUBLIC AND PRIVATE GROUND-WATER USE
IN ZIP CODE AREAS1
Frequency Cumulative
(# of zip codes) Percent Percent
Population Category
(both public and
private water users)
0 2 2198
1-100 5057 15.4 15.4
101-500 8897 27.1 42.5
501-1000 5062 15.4 57.9
1001-2500 5902 18.0 75.9
2501-3300 1466 4.4 80.3
3301-5000 1620 4.9 85.3
5001-10000 1920 5.8 91.1
10001-50000 2221 6.8 97.9
50001-75000 261 0.8 98.7
75001-100000 122 0.4 99.0
100001+ 309 0.9 100.0
1 Based on the number of private well users as reported in the 1980
Census and the number of public system users as reported by FRDS.
Approximately 10,000 public water systems in the FRDS data base did not
contain zip code information and have been eliminated from the analysis.
2 Zip codes that did not contain any public water system users were not
included in the percentage calculations. The percentages reflect only the
proportion of potentially classifiable subpopulations in each population
category.
Source: ICF analysis of information obtained from the Federal Reporting Data
System and the U.S. Bureau of the Census.
-------
-12-
us to make tentative estimates of combined public/private water ground water
use.
Generally speaking, our analysis confirms the results presented in Section
2. As indicated by Exhibit 5, if the substantial population threshold is set
at 2,500 (and each zip code area is analogous to the area that would be
considered for the purposes of the classification), then roughly 76 percent of
the zip codes would be excluded from classification as Class I. If the
threshold is increased to 5,000 persons, 85 percent of the zip codes would be
excluded. In short, our analysis suggests that substantial areas of the
country (though not necessarily a substantial portion of the population) may
be excluded from Class I with the use of even a relatively low threshold.
4. SUMMARY
There are a number of ways of analyzing the implications of various
substantial population thresholds. This paper has used three methods:
• Assume the substantial population test is applied to
the entire user population of public water systems and
determine the number of systems, the number of persons,
and the number of well groups excluded from Class I by
various population cutoffs;
• Assume the substantial population test is applied to
the user population for each well group of public water
systems and determine the number of systems, the number
of persons, and the number of well groups excluded from
Class I by various population cutoffs; and
• Analyze both public water system users and private
well users in each zip code.
A summary of these results is presented in Exhibit 6. In general,
relatively low thresholds of, say, 1,000 or 2,500 people, have the potential
to exclude significant areas of the country from designation as Class I. For
example, a threshold of 2,500 is likely to exclude 91 to 95 percent of the
public water systems (depending on whether all system users or only well group
users are evaluated) and 76 percent of the zip codes. A threshold of 2,500
also would exclude 82 to 89 percent of the well groups from Class I.
Increasing the threshold to a high level such as 25,000 eliminates almost all
systems from Class I. If all system users are considered for the test,
however, a few systems which collectively serve more than 42 percent of
persons using ground-water would remain eligible for Class I under a 25,000
user threshold.
Another important finding is that the fraction of persons affected by
these thresholds may be significantly lower than the fraction of systems or
well groups affected. For example, the 5,000 person threshold is likely to
exclude ground waters used by only 30 to 40 percent of all ground water users
from consideration for Class I despite the fact that between 95 and 97 percent
-------
-13-
>co
I- CO
< UJ
cc
UJ "
CO O 3)
— a.
X WO
x zo-
LJ O
O
O CO
c— in
0 — 3)
ja
"O 3 CD
co a- -P
U) CO
CO °O >
CO CD ™
C !-
4J._ 0-
U] J3
CO E "D
HOC
O ro
Q
3
o
o o
3 CO
0-2
C S-
O CO
0.
•o
0) in
3)
in E
0) CO
h- -P
00
^
CO
in
t_
CO
00
3>
E
CO
4J
U)
CO
o
•—
—
J2
3
Q-
—
-^
<
C
O
T3
CO
to
CO
CD
4_)
U)
CO
1—
—
d~
o in 13 o)
CO CO W
•PT3 T3 CO
C O 3 —
0)O — O
o o
i- ax E
0) — UJ O
a. M t-
d-
d- in
o o.'o en
3 CD in
•P O TD CO
C S. 3-
0>O — O
o o
<-- X E
CO — UJ O
Q. CD 1-
2 d-
C|-
O "D in
in a) in
•P CT3 CO
C O 3 —
CD W — O
O S- 0
!- 0) X E
0) a. uj o
a. i-
<+.
c^
O T3 in
in to u>
•P ET3 CO
c CO 3 —
CD -P— O
0 W 0
1- XX E
CO CO UJ O
a. s-
d-
Q- ul
O O.-D Ifl
3 CD in
•P OT3 CO
C t. 3 —
CDO — O
0 0
! X E
CO — UJ O
Q. CU i-
2 d-
d-
o -a in
us a> in
J-> CT3 CO
C O 3 —
0) in — o
O 1- O
i- CO X E
CO cu UJ O
a. t_
d-
o -o in
in CD in
P £-a co
c u 3 —
CO 4->— O
o m o
<- >>X E
<0 CO UJ O
Q. i-
Q
_J
O
_1_
CO
UJ
C£
X
H
O
H
_l
Q.
0
Q.
fe^
Os
1
r-
m
^5
o
*
CSJ
00
^5
ro
O
CNJ
65
00
.
0
fe^
csj
SO
VO
6^
-_
*
__
—
fe^
VO
.
_
co
0
o
0
OS
.
in
h-
—
.
ON
CO
vO
.
Os
CM
O
in
OS
o
•
CSJ
CO
ro
.
0
OJ
CO
0
Os
O
o
in
CSj"
ro
•
in
00
O
.
ro
C7v
CM
*
O
^f
^
•
r—
Os
—
•
Os
CO
vO
•
ON
CSJ
0
in
ON
0
o
o
in
—
—
ON
vO
•
vO
ON
j3"
•
CM
vO
CO
•
ON
ON
O
•
ro
ON
CSJ
O
Jt
.
f*_
OS
O
O
0
o
1
1
o
.
o
0
0
.
o
o
—
o
o
o
—
CO
in
OS
_
•
r-
in
CNJ
Os
ON
O
o
o
in
CSJ
Os
•
p^
Os
O
.
0
o
o
•
0
o
—
o
t
o
o
—
0
•
j —
Os
CO
Os
so
r-
.
Os
ON
o
o
0
o"
in
co
C
CO
CO
p
CO
•a
c
o
4-J
CO
C
CO
a
x
tu
JZ
X
CO
a:
•o
c
CO
X
0>
p
a:
CD
CO
-------
-14-
of the systems would be excluded. In short, while a low threshold may
eliminate a substantial fraction of the nation's ground waters from inclusion
in Class I, the proportion of persons affected is likely to be much less.
-------
CLASS I: ANALYSIS OF THE IMPLICATIONS OF THE CURRENT ECONOMIC
TEST OF IRREPLACEABILITY FOR GROUND-WATER CLASSIFICATION
This paper examines the implications of the economic test of
irreplaceability for ground-water classification. The EPA intends to use the
test, for those ground-waters highly vulnerable to contamination, to determine
which ground-water sources are designated as Class I or irreplaceable water
supplies. The EPA's draft guidance for Ground-Water Classification defines a
ground-water to be irreplaceable if reliable delivery of water of a comparable
quality and quantity from an alternative source in the region would be
economically infeasible or institutionally precluded. An alternative source
of water is economically infeasible if the annual cost of a typical user would
exceed one percent of the average household income in the area. In addition,
ecologically vital ground waters may also be designated as Class I; such
ground waters are not analyzed here.
The economic burden test designates a ground-water source as Class I when
it is an irreplaceable source of drinking water for a substantial population.
Current EPA draft guidelines define a substantial population to be 2,500
people, and define a ground water to be irreplaceable when incremental annual
replacement costs would exceed one percent of the household income of the user
population. In the analysis, the fraction of ground waters that are currently
used as drinking water sources in public water systems likely to be classified
as Class I was estimated by comparing a frequency distribution of expected
water supply replacement costs with a distribution of household incomes in
order to estimate how often replacement costs exceed a given proportion of
household incomes. The frequency distribution of replacement costs was
created by considering various replacement scenarios based on the size of the
system to be replaced and the scope of replacement measures necessary.
The paper consists of five sections. Section one describes the data
sources used in the analysis. The methodology for estimating the proportion
of ground-water sources that are deemed irreplaceable under the economic test
is described in the second section. The third section of the memo presents
the results of the base case analysis, and includes an estimate of the number
of Class I irreplaceable ground waters, based on the results of the economic
test. In the fourth section, the sensitivity of the results to changes in
certain parameters used in the estimation procedure is discussed. Section
five presents conclusions based on these results.
1. DATA SOURCES
The data presented in this paper are derived from studies conducted by ACT
Incorporated for the Municipal and Environmental Research Laboratory of the
EPA, and by Temple, Barker, and Sloane, Inc. for the Office of Drinking Water
of the EPA. Information was also obtained from the Federal Reporting Data
System (FRDS) database, and from the U.S. Bureau of the Census.
The ACT Inc. and Temple, Barker, and Sloane, Inc. studies evaluate the
economics of water delivery and the financial and operating costs of community
water systems, respectively, for water supply systems of various sizes. The
-------
-2-
ACT Inc. study, conducted in 1977, analyzes a sample of 12 large water
utilities (revenue greater than $500,000 per year), and 23 small water
utilities, and was used in this study to obtain a breakdown of costs for major
water supply system components. The Temple, Barker, and Sloane, Inc. analysis
of 1982 is based on a survey of 1,056 water systems. These 1,056 systems
included 50 publicly-owned and 50 privately-owned water utilities selected at
random from each of 12 size categories (except where there were less than 50
systems in the U.S. in that size category) as well as 50 Native American
systems. Data from this survey was used in this analysis to estimate average
unit costs and variance of unit costs for water supply systems of various
sizes.
The FRDS database provided data on the number of ground waters serving
water supply systems in each size category. Finally, data from the 1980
Census provided information on the distribution of average household incomes
throughout the counties and zip codes of the U.S.
2. ANALYTICAL METHODOLOGY
This section describes the methodology used to estimate the fraction of
Class I irreplaceable ground waters. The section begins by describing the
components of water supply system costs. The three factors affecting
ground-water irreplaceablility under the economic test are then outlined. The
next three subsections describe these factors in detail. The final subsection
explains how these factors are combined to evaluate irreplaceability.
Cost Components of a Water Supply System
Water supply system costs can be broken down into four major components:
(1) Acquisition;
(2) Treatment;
(3) Distribution and Transmission; and
(4) Support Services.
Acquisition costs are the costs of producing or acquiring water, and can be
thought of as the costs of getting the water to the treatment point. These
costs include the capital, operating, and maintenance costs of wells,
reservoirs and aqueducts, and payments to suppliers for purchased water.
Treatment costs include the costs of treatment plants and equipment, and the
costs of chemicals that are added to the water. Distribution and Transmission
costs are the costs of pumping the water from the treatment points to the
service population, and the capital and maintenance costs of the piping
network. Support Services are the costs of administrative and customer
services that are not directly related to the physical process of delivering
water.
Exhibit 1 shows the average cost structure of the small and large systems
surveyed by ACT Inc. Costs are separated into four major components, with the
exception of interest expenses which were not allocated to particular cost
components, and have been shown separately.
-------
-3-
EXHIBIT 1
AVERAGE COST STRUCTURES FOR WATER SUPPLY SYSTEMS
Small Systems a/
Large Systems b/
Acquisition
Treatment
Distribution and
Transmission
Support Services
Interest Charges
Total
Cost per
Million
Gallons of
Water c/
$160
132
310
123
138
863
Percent
19
15
36
14
16
100
Cost per
Million
Gallons of
Water c/
$112
75
221
179
137
724
Percent
15
10
31
25
19
100
a/ Serving between 300 and 50,000 people.
b/ Serving over 100,000 people.
c/ Unit costs inflated to end of 1984 dollars
SOURCE: ACT Inc.
-------
-4-
The data show that on average the largest single component of total cost
is for distribution and transmission. Treatment costs are on average the
smallest component of total costs. The cost structures of large and small
systems are quite similar, but the average unit costs are lower for the larger
systems than for the smaller systems.
Factors Affecting Ground-water Irreplaceability Under the Economic Test
Three factors must be considered in evaluating the irreplaceability of a
ground-water supply on the basis of replacement costs under the economic test:
(1) The scope (or extent) of replacement measures needed if
the water supply is lost, that is, the system
components that must be replaced or constructed to
re-supply the affected community;
(2) The scale of replacement, that is, the number of people
affected by the loss of the water supply or capacity
loss; and
(3) The amount of money available for replacement costs in
the affected community.
These factors are considered below.
The Scope of Replacement Measures
Three basic possibilities arise when a ground-water source is contaminated
and requires replacement: the first possibility is that only the acquisition
component of the system would require replacement; the second is that the
acquisition components would require replacement and new treatment capacity
would need to be added; the third is that, in addition to acquisition and
treatment components, a transmission and distribution network would need to be
constructed. We refer to these situations as limited, moderate, and extensive
system replacements, respectively.
Acquisition costs only would be incurred when existing treatment and
distribution capacity could be used with the replacement source. Source
replacement may include such measures as locating and drilling a new well
field in an area of uncontaminated ground water, or switching from a
ground-water source to a surface-water source.
Both acquisition and treatment costs may be incurred when a difference in
water quality between old and new water sources require that additional
treatment processes be added in order to meet water quality standards. For
example, the ground-water supply for a community may require no treatment
other than chlorination; however, switching to a nearby surface water supply
may require the addition of unit processes such as coagulation, flocculation,
sedimentation, and filtration to the existing treatment plant.
Distribution and transmission costs may be incurred in situations where
the installation of a new distribution system is necessary in order to
-------
-5-
re-supply an affected community with drinking water. Such extensive
replacement would generally be required in situations where a ground-water
source supplying a number of private wells is contaminated, thus requiring the
installation of a centrally located water supply system. This situation is
particularly applicable to rural settings.
The estimation of the cost implications for these various alternatives was
based on the typical percentage breakdown of total costs among the major
system components from the data reported in the ACT Inc. study (presented in
Exhibit 1). For this analysis, 20 percent, 50 percent, and 100 percent of
total system costs were taken to be representative of the replacement costs as
a proportion of total system costs under the limited, moderate, and extensive
system replacements, respectively. The data from the ACT Inc. study indicate
that the breakdown of costs between small and large water supply systems is
similar, and at the level of resolution of this analysis any differences in
cost breakdowns between the two are not likely to be significant.
The Scale of Replacement Measures
In evaluating the irreplaceability of a ground-water supply on the basis
of the economic test, it is important to consider the size of the system
affected by the loss of that ground water (i.e., the scale of replacement),
because different system sizes will lead to different unit replacement costs.
To quantify this effect, we used data from the 1982 Temple, Barker, and
Sloane, Inc. study to estimate unit costs of replacement for a range of water
system sizes (measured by size of population served).
The data presented in Exhibit 2 show the unit operating costs (including
depreciation and interest charges) for water supply systems in 12 size
categories. Average unit costs for surface-water systems in the three
smallest size categories are averages for small samples (five observations or
less) and consequently are somewhat less reliable than other data (ten or more
observations). Similarly, the averages for ground-water systems in the
largest two size categories are averages of small samples.
The data show that there are scale economies in system operation. Systems
serving populations of approximately 300,000 have average costs of about $600
per million gallons, whereas systems serving populations between 2,000 and
20,000 have average costs in the range of $800-$1,400. Also, for systems
serving over 5,000 people, there appears to be little difference between the
average costs of systems that use predominantly ground water and systems that
use predominantly surface water. However, the data indicate that the average
unit costs of small ground-water systems are extremely high; much greater than
the average unit costs of the small surface-water systems surveyed.
The data presented in Exhibit 2 also show the amount of variation in the
costs of systems in each size category. Inspection of the standard deviations
of the costs shows that, in general, there is a greater variability among the
costs of the smaller systems than among the costs of the larger systems both
for ground-water and surface-water systems.
-------
-6-
CO
LJ
h-
CO
CO
>co
_IUJ
Q-N
CL —
13 tO
CO
CO
££ O
UJ O
I --
< CC.
2<
O
CO
LJ CO
-JZ
Q.O
•o
CD
0)
CO
c
o
+J
to
3
a.
o
o.
CO
X
O
Q.
a
05
N
E
01
I/I
>,
CO
x
LJ tO <
O O-
o
CO O.
I—
CO O
OZ
CJ —
I- CC
— LJ
' zco
LJ
O
LU
O
o
ON — vo
0 J
0
0
O
o o oo o
OX. — CM
O -3-
in
r-
o
o
O in O\ O
O ro C\J
0 VO
ro
0
o
m — o a-
O CM CM
r- vo
CO
o
o
IA 00 C\J VO
— CO CM
c\j r~
vo
0
O
m — r^ CM
VO CM .3-
r- r~
ro
0
o
iTi VO CM O
O C\J ro
r- oo
o oo in —
lA [^ CM VO
VO 0
VO —
0 0 O 0
m o — vo
CM' — "
VO IA (M
O O LH
in co
r~-
fO ro C7x
or- -a-
o r-
ro
ro CM CTs
-3- —
ro —
vo
C
l/l O
i a) t- E —
a) o CD -u
1- CO O- C W CO
O- Ci_ o f — W
s- *J — J- >> > w
CTI! to — OO W 0)U)
C CO O CO OX T3 O
— O O)— <*- O
l/J > — O "D "D
3— 0!C 01004- i->> CO —
«C CO— CO CUOJTJC
SCO ! -D— ^2> C3
CDC'- CU — C — El- CO
jj.-jj > ._ 0 o 33 .pc,.
U5 E J-1 < E ^T3 Z W CO O
> O TO
wl-02
CM — 1
00 1
CO
CM lT\ VO
CM —
in
vo vo f-
OX CM CM
in
— cr —
in — CM
VO
— st ro
J- — CM
VO
o r- o
in ro in
r-
CO CM VO
CM in ro
r~
r— vo vo
ro J- VO
OX
o o —
— m r—
m —
^
O vo O
CM m co
vo
O 00 ON
m vo —
ro m
ro
0 CM O
m ro ^
m ox
r- —
c
w O
I (- E
CO T3 CU t/5 CU -P
s- c ac *J co
a- 3 o v> — 05
0 -U — J- >, > JJ
pi- co — co « a>w
CO O CO OX T3 O
O O>— *- O
uj > -~ o -a T:
Z>— CUCd-U) 0) S--U
4J OlOOS- !->> CO —
i/)C co— eo cum -oc
Eco ' 13— J2> C3
0) C s- CU— C— E t- CO
•U — CU > — CUO 33 -PC*.
w E -P < E —-a z « co o
> 0 CO
crt-a2
o
c
0)
c
CO
o
•a
c
CO
CO
CD
a
e
U
O
a:
3
o
CO
-------
-7-
The cost data presented in Exhibit 2 are for existing systems. In order
to estimate the full costs of replacement systems these cost estimates were
inflated to allow for factors that could lead to underestimation of
replacement costs using this data. These factors include cost components that
may not have been included in the data for existing systems such as legal fees
and costs of rights of way, federal grants and loans that reduced costs faced
by the systems surveyed from the full costs of system, and inherent
differences between the costs of existing and replacement systems. In
summary, these factors would in general to lead to an underestimation of
capital costs. However, on an annualized basis capital expenses are a small
fraction of total annual expenses --on average less than 18 percent for the
systems surveyed by Temple, Barker and Sloane Inc. Therefore underestimation
of capital expenses by even a large fraction will lead to a small
underestimate of total annual costs. Potentially most significant are
inherent differences between the costs of existing and replacement systems.
Because existing systems will, in general, represent a least cost solution to
water supply needs, replacement systems will tend to be more expensive than
existing systems.
These factors are discussed in more detail in the appendix of the paper
entitled "Costs of Water Supply Systems." For this analysis, these factors
have been crudely taken into account by adjusting costs of existing systems
upwards by 20 percent to estimate costs of replacement systems.
Available Funds
The remaining factor to be considered in evaluating the irreplaceability
of a ground-water source using the economic test is the income available for
replacement costs in an affected community. This factor may be considered a
combination of two sub-factors: (1) the maximum percentage of household
income available for water supply costs; and (2) the mean household income
level of the community. Maximum percentage of household income available for
replacement costs for this analysis was set at one percent, the level
currently used in the draft classification guidance, and was converted to a
unit cost (i.e., percent of household income on a per million gallon basis) to
allow direct comparison with unit costs of water replacement for various
system sizes.
One percent of household income must be considered with regard to actual
mean household income in the affected community in order to determine the
amount of money available for water supply replacements. This factor was
taken into account in this analysis by using 1980 Census data to compute mean
household incomes for five economic groups representing the range of incomes
in the U.S. shown in Exhibit 3. The exhibit shows the distribution of average
household incomes by county and by zip code. The distribution of average
incomes by zip code is more dispersed than the the distributions of average
incomes by county because the average size of zip codes is smaller. Thus, by
using zip code rather than county level data, extreme values for mean
household income levels are not lost by averaging. For this analysis, the
distribution of income by zip codes was used to reflect the possibility that
potentially affected populations may be much smaller than average county
population. The income levels were inflated from 1979 dollars to 1984 dollars
-------
-8-
EXHIBIT 3
DISTRIBUTION OF AVERAGE HOUSEHOLD
INCOMES BY ZIP CODE AND BY COUNTY
Mid Point Income Percentage Groups with
Inflated to Respective Mean Household Income
Income Level 1984 Dollars Zip Codes Counties
0-5,000 3,420 .6% 0%
5-10,000 10,250 4.0% .1%
10-15,000 17,090 28.4% 27.9%
15-20,000 23,930 40.9% 56.1%
> 20,000 30,760 26.1% 15.8%
Source: ICF analysis of file STF 3B, 1980 U.S. Census.
-------
-9-
for the analysis. These categories were then used to determine the funds
available for water supply replacement in communities at different economic
levels.
The percentage of household income threshold for irreplaceability can be
translated into a unit replacement cost threshold (in dollars per million
gallons) for each income level, as follows. The annual household water
consumption was assumed to be 60,000 gallons, based on information from the
USGS 1983 National Water Summary. This implies that the household income
threshold per gallon is 1 percent of income divided by 60,000 gallons. This
figure multiplied by one million gives the threshold income per million
gallons. For example, if household income is $10,000, then $100 (1 percent of
$10,000) is the threshold cost for 60,000 gallons. This implies a threshold
of .167 cents per gallon ($100 divided by 60,000), or $1,650 per million
gallons. Exhibit 4 shows the unit cost thresholds for five income level
ranges.
Evaluation of Ground-Water Irreplaceability Using Alternative
Replacement Scenarios
In order to evaluate the irreplaceability of a ground-water supply for a
specific set of circumstances, the three economic factors outlined in the
previous section were used to construct a series of replacement scenarios.
These scenarios consisted of three components: (1) the scale of the affected
system; (2) the scope, or extent, of water supply replacement; and (3) the
average household income. The irreplaceability of a ground water supply for a
scenario was evaluated by comparing cost distributions for the various levels
of replacement for a given population size with a unit cost threshold (or
cut-off) for a particular household income. This procedure is summarized in
Exhibit 5, and may be broken down into three steps:
(1) Estimation of cost distributions for the three scopes
of replacement for each of the system sizes;
(2) Determination of a unit cost threshold for the
household income level considered; and
(3) Determination of the proportion of cases for which
replacement costs exceed the allowable percentage of
household income by comparing the unit cost threshold
with the cost distribution for the scenario.
Cost distributions for the three scopes of replacement (limited, moderate,
and extensive) were estimated using means and standard deviations of unit
costs for systems of various sizes from the Temple, Barker, and Sloane, Inc.
study. (Exhibit 2) It was assumed that the costs in each case were normally
distributed about the estimated mean with a standard deviation as indicated.
This assumption facilitates combining the available data to construct cost
distributions.
Unit cost thresholds for the different levels of average household income
were determined according to the method described in the previous section of
-------
-10-
EXHIBIT 4
UNIT REPLACEMENT COST THRESHOLD
FOR 1 PERCENT OF INCOME
Percentage of Zip Codes
with Corresponding Mean
Income Level Threshold I/ Household Income Level
$ 3,420 $ 570 .6%
$10,250 $1,708 4.0%
$17,090 $2,848 28.4%
$23,930 $3,988 40.9%
$30,760 $5,127 26.1%
I./ Cost per million gallons assuming annual household
consumption of 60,000.
-------
-11-
EXHIBIT 5
Distribution of unit replacement costs
in a particular replacement scenario
Percentage
of Cases
proportion of cases
that replacement costs
exceed given percentage
of Household Income
mean
unit cost
threshold
standard
deviation
mean and standard deviation determined by:
• size of replacement (affected population)
• scope of replacement measures
unit cost threshold (per million gallons)
= 1% x average household income x one million gallons
consumption per household
-------
-12-
the paper. An example calculation for this parameter appears in the footnote
to Exhibit 5. As illustrated in Exhibit 5, the unit cost threshold was then
compared to the unit cost distribution for the particular scenario, and the
area under the distribution curve to the right of the threshold was the
estimated percentage of cases for that scenario for which water supply
replacement costs would exceed available replacement funds.
In order to apply this methodology to the total "universe" of ground-water
sources in the U.S. three factors must be considered. They are: (1) the
relative frequency of each of the three scopes of replacement; (2) the
percentages of potentially affected population groups falling into each of the
five income categories; and (3) the percentage of ground waters supplying
systems in the each of the system size categories.
The general absence of data on the frequency of each of the three
replacement levels required that we make assumptions about these percentages.
The assumptions made are presented in Exhibit 6. As shown in the exhibit, we
assumed that extensive replacement measures are more likely to be needed when
small systems are affected than when large systems are affected.
Data from the 1980 Census was used to obtain a distribution of the average
household income of the potentially affected populations, and has been
summarized previously in Exhibit 4.
Percentages of the total universe of ground waters supplying systems of
different sizes were estimated using data from the FRDS database and the
Temple, Barker, and Sloane survey. The FRDS database gives the number of
public water systems of each size that use predominantly ground water; these
are assumed to be the potentially affected systems. The Temple, Barker, and
Sloane survey gives the average number of ground waters supplying systems in
each of the size categories. (For this analysis, it has been assumed that a
"group of wells' referred to in the Temple, Barker, and Sloane survey
corresponds to a ground-water source.) These numbers have been combined to
estimate the number of ground waters supplying systems in each of the system
size categories. This data is shown in Exhibit 7. Only potentially affected
populations greater than the minimum substantial population (taken to be 2,500
people on the basis of current draft classification guidance for the
Ground-Water Protection Strategy) were considered in the "base case"
analysis. Therefore only systems serving populations of more than 2,000 were
considered in the "base case" analysis. (2,000 is the system size closest to
2,500 among the representative system sizes considered.) Thus only the
estimated 20,592 ground-water sources supplying larger systems out of the
estimated 55,632 ground-water sources supplying public water systems are
considered in the analysis. The estimated fraction of Class I ground waters
is a fraction of the 20,592 ground waters supplying such public water systems.
These three factors were multiplied to obtain weighting factors for the
various scenarios. The weights were applied to the proportion of Class I
ground waters for each of the scenarios, and the results were added to
estimate the total number of Class I ground-water sources in the U.S.
-------
-13-
EXHIBIT 6
ASSUMED PROBABILITIES OF ALTERNATIVE
SCOPES OF REPLACEMENT
System Size Probability of Scope of Replacement
(Approximate Populations Served) Limited Moderate Extensive
300 .25 .25 .50
750 .25 .25 .50
2,000 .33 .33 .33
12,500 .33 .33 .33
60,000 .50 .50
300,000 .50 .50
1,000,000 .50 .50
-------
-14-
EXHIBIT 7
ESTIMATED NUMBER OF GROUND WATERS SUPPLYING
PUBLIC WATER SYSTEMS IN ALTERNATIVE SIZE CATEGORIES
System Size
(Approximate Populations Served)
300
750
2,000
12,500
60,000
300,000
1,000,000
Total
Total Supplying Systems Serving
More Than 2,000 People
Number of
Ground Waters
33,937
6,312
11,616
7,146
496
733
601
60,841
20,592
Percentage of
Ground Waters
Serving Systems
Supplying 2,000
People or More
56.4%
34. 7%
2.4%
3.6%
2.9%
100%
a/ 2,000 is the system size closest to the substantial population cut-off
of 2,500 people among the representative system sizes considered.
Source: FRDS, and Temple, Barker & Sloane Inc.
-------
'15-
Although several assumptions are inherent to this methodology, the
approach allows the impact of these assumptions to be readily evaluated with
sensitivity analyses. The results of the "base case" analysis (i.e., the
analysis in which the previously outlined values and assumptions are used) is
described in the following section. Sensitivity analyses for various critical
parameters are outlined in Section four of this paper.
3. RESULTS OF THE BASE CASE ANALYSIS
The analysis was implemented on a microcompter using LOTUS 1-2-3
spreadsheet software. The results of our analysis for the proportion of Class
I ground-water sources serving public water systems are found in Exhibit 8,
which shows a matrix of probabilities corresponding to all possible
combinations of components of the economic test: scope of replacement, scale
of water system affected, and income-determined unit cost threshold. There
are seven different representative water system scales, ranging from systems
serving about 300,people to those designed for over 1,000,000 people. Within
each scale, there are three divisions for scope of replacement ranging from
limited to extensive. Each scope and scale of replacement has an associated
unit cost whose distribution is compared to the distribution of incomes which
are shown at the top of Exhibit 8. If the cost of water system replacement
exceeds an area's income-determined unit cost threshold, the area's ground
water is categorized as Class I.
The probabilities for each category of water source being categorized as
Class I may be easily understood by comparing the distribution of the unit
cost of replacement with the unit cost threshold. For example, an extensive
replacement for a water system serving approximately 2,000.people is estimated
to cost on average $1,730 per million gallons. A community with an average
income of $10,250 has a unit cost threshold of $1,708 per million gallons.
Thus, on average, the unit replacement cost of such a system would be greater
than the unit cost threshold and therefore such a water source would be be
classified as Class I.
Illustration of Probability Computations
To illustrate the computational methodology used to derive the results in
Exhibit 8, it is helpful to explain the discrete steps taken to derive one of
the probabilities. Let us take the probability for a extensive scope of
replacement for a system serving approximately 2,000 people with an average
income of $10,250. This water source has 55 percent probability of being
classified as Class I. This high probability is plausible in light of the
fact that the average cost for this scope and scale of replacement is $1,730,
whereas the unit cost threshold for the area's income group is $1,708.
The average unit cost for replacement is derived from Temple, Barker, and
Sloane estimates and is represented by the following formula:
Weighted Average of Proportion of System
Total System Costs Per X Necessary for Scope of
Million Gallons of Replacement
-------
-16-
CO
CO
o
CO
Q
LJ
C£
COr-
LJCO
a LJ
CO
so
CO O
> z
CO O
o
tr LJ
UJ
I— LJ
oo
— a.
CO co
- 00
X Z —
X — en
LJ ><
ex z
LJ LJ
CO O
CO
CO
LJ LJ
0>
cc —
3) (-
0<
CO Z
cc
CC LJ
LJh-
o —
z
o
oz
o
ce
o
a.
O
CC
O-
o
-J II
o or-
ZLJNOCM
u s: r —
ff\ f—\
Vi t«J ^ •>
3> O O LA
O Z ro
I —
O
-I II
o ooo
X LJ ro 00
ui s: osos
CO O "i •>
3 O ro ro
O Z CM
X —
a
i M
O Oeo
XLJ00.3-
LJ S COOO
C/5 O «i «i
3) O C-CM
o z —
Q
J II
— J II
O Oeo
X LJ lAO
LJ S CM r-
3> O O —
O Z —
x —
o
—1 II
O OO
X LJ CM r^
LJ Z.3-LA
C/J O *
n o ro
OZ
x —
II
i-a
C0_l
OO
ox
CO
1— LJ
— CC
zx
p
.
—
•— £0
.a co
CO CO
o o
£_
CL.
X
.— _
— .
— (0
.a en
CO CO
00
a.
4J
•—
(_
a.
CC
z
OO
CC —
< K-
o ^
z —
1— LJ
CO O
CD
LJ O
O 1— LJ
< 0
CC r— «3l
LJ CO _l
>O o-
< O LJ
CC
U. 1 CO
O LJ LJ
O 1- CC LJ
LJ< Z3> CL.
0 1 LJ CO O
O c^ S ^ O
O LJ LJ CO
CO CC E
LJ —
r- CD TO
CO -P 0)
> CO >
CO E s-
— CO
u. X co
o o
c_ .
LJ Q.C
— i a Q
« 0
O - - A
CO
fe^&^fe^
OOCM
OOOs
00-
fe^&^^^
OO —
OO J"
OOrO
&?&§&?
O ro ro
O — O\
OOlA
S^S?S5
O CO 00
O OOOv
1— O\
fe^fe^fe^
— OS O
OSOSO
IAOS O
OsOsO
—
Osovo
— oo\
— ro lA
CO lA ON
r- j- oo
r^ ON co
n •*
— ro
CD
CD >
0) CO W
P <- C
— CO CO
E-O P
— 0 X
_l Z LJ
O
O
ro
&^fe^fe^
OOO
000
ooo
fe^fe^^?
ooo
ooo
ooo
S565S5
ooo
ooo
ooo
SS65S5
O O 00
ooo
O Ovo
OS
&^^^fe"5
ooo
ooo
ooo
OO
O -31 ON
ro 3- OO
ro CM J"
ro OSOO
^
—
CO
CO >
•a -P —
CO CO U)
4-> !- C
— CD CD
E T3 -P
— OX
_l S LJ
o
r-
fe^&^fe^
ooo
ooo
ooo
&^fe^&5
ooo
ooo
ooo
S3&3&3
ooo
ooo
ooo
&S&S&S
OOCM
ooo
OOlA
LA
fe^^^fe^
OVOO
ooo
OOSO
ONO
—
lAlAO
ro CO f—
—
NO -3° O
J C^« CO
ro 00 P-
•,
—
CD
CD >
•a -P —
CO CO «
P i- C
— CD CD
— o x
— 1 S LJ
0
O
0
«
CM
S^&^&^
000
ooo
ooo
fe^&^&^
ooo
ooo
ooo
&3S3&3
ooo
ooo
ooo
S3&3&3
ooo
ooo
ooo
&^&s&^
ooo
ooo
ooo
0
—
NO NO CM
— ro
o
ooino
— J" OS
CD
CD >
•a P —
CD CO «
P S_ C
— CO 1)
— 0 X
_l S LJ
0
o
lA
.,
CM
&S6-5&^
ooo
ooo
ooo
&?&^&^
ooo
ooo
ooo
S3 63 63
ooo
ooo
ooo
&56S&S
ooo
ooo
ooo
fe^&^£^
000
ooo
ooo
0
—
lA CNJ ro
— CM
r-NO J-
NO — ro
— J-00
to
CO >
•a P —
CD CO U)
•P 4- C
.- CD CD
E -a P
— ox
-ISLJ
o
o
o
o
NO
fe^&^fe^
ooo
ooo
ooo
&^&5&?
ooo
ooo
ooo
&?S5S5
ooo
ooo
ooo
S5&565
ooo
000
ooo
&s
ro
fe^fe^ro
O O CM
ooo
ooo
-3" O OS
— —
•^ — ~ •
J" NO CNJ
— ro r-
CO
CO >
•a P —
CO CO CO
P 1- C
— CD CO
E-o -P
— 0 X
_i s: LJ
o
o
o
..
o
o
ro
^^fe^fe-S
ooo
000
ooo
^Sfe^fe^
ooo
ooo
ooo
fe5&?fe?
ooo
ooo
ooo
&9S5S^
ooo
ooo
ooo
fe^&^fe^
000
ooo
ooo
J-OOS
— —
— CM ro
OlAO
— CM LfS
CO
0) >
•o P —
CD CO CO
*J i- C
— CO CD
E-O P
— 0 X
_l £ LJ
o
o
o
o"
o
o
•l
—
a:
LJ
>.
o
NO
O
CO
CO CO
•a
u co
— P
— CO
.a c
3 01
a—
m
en CD
— c
a o
a—
~z P
CO CO
CO 'Z
*- a.
CD o
p a
CO
T3 CO
c —
3 P
O C
i- CO
O1P
CO
O 3
CO
C
O Ol
— c
p —
O 1-
a co
o i/>
a. co —
E
— 0) 10
CO P CO
P CO CO
o >,—
I- co O
O
p
O
CO
•o
CO
CO !-
CO O
.c c
Ol O
CO P
> CO
CO-
CO CO
CO
CO f_
CO CD
<--a
3 C
Ol 3
CO
CO-
CO p
CO C
.C 0)
r- P
O
\ a.
col
CO
>,
CO
c
CO
o
CO
o
C-
o
CO
-------
-17-
For a system serving a population of approximately 2,000, the average water
system cost is $1,730. In this case of extensive scope replacement, the total
per million gallon water system cost is also $1,730 (i.e., because the whole
system would be replaced).
The unit cost threshold per gallon is:
(Percent of Income Threshold X Average Household Income)
Gallons of Water Used in Household
Under the base case assumptions, the unit cost threshold per million gallons
for a water system with the above mentioned economic characteristics is:
((1% X $10,250)/60,000) X 1,000,000 = $1,708
The average cost to replace this scope and scale of water system has an
assumed normal distribution with a mean of $1,730 and a standard deviation of
170. Exhibit 9 shows this distribution and the relationship of the unit
income threshold ($1,708) to the average unit replacement costs ($1,730). In
this case, the unit cost threshold is .13 standard deviations to the left of
average system unit costs. The number of standard deviations the unit income
threshold is from the mean, the "z-statistic," is used to find the probability
that the threshold is greater than the average per unit system replacement
costs. The z-statistic tells us what portion of the normal curve distribution
for system replacement costs, which has a total area equal to one, is to the
right of the unit income threshold. This area, shown by the shaded region of
Exhibit 9, is equal to 0.55, meaning that there is a 55 percent chance that
this water may be categorized as Class I Irreplaceable.
Estimated Proportion of Public System Water Sources Classified as
Irreplaceable
The 55 percent probability that the example water source is Class I may
also be thought of as the proportion of all water sources for this scope,
scale and income group in the universe of water sources that will be
classified as irreplaceable. So, if there are 100 water sources with these
economic characteristics in the U.S., 55 of them will be of the Class I type.
While the proportion of Class I ground waters with the characteristics
described in the foregoing example is large, Exhibit 8 shows that, for most
combinations of characteristics (scope, scale, income group), the proportion
of Class I ground waters is low. This implies that the overall proportion of
Class I ground-water sources is low. To find the proportion of all water
sources in the universe of water sources serving public water systems that
will be categorized as Class I, it is necessary to weight the proportion of
Class I water sources for each scope, scale and income group by the proportion
of that group within the universe of water sources. The result is a weighted
average of probabilities of Class I water sources and an estimate of the
overall proportion of Class I ground waters.
This concept may be explained by continuing with our example of extensive
scope water system replacement for a system serving approximately 2,000 people
with a mean household income of $10,250. Under the "base case" assumptions,
-------
-18-
EXHIBIT 9
ILLUSTRATIVE DERIVATION OF PROBABILITY OF
DESIGNATION AS CLASS I FOR A PARTICULAR
REPLACEMENT COST AND INCOME SCENARIO
55 Percent Of
Area Under Curve
Unit Cost
Threshold = 1,708
Mean = 1,730
Replacement Cost
Standard Deviation = 170
-------
-19-
each of the probabilities of extensive scope of replacement is taken to be 33
percent. It has been estimated that 56.4 percent of ground waters supply
systems that serve populations of approximately 2,000 people. It has been
estimated from Census data that approximately 4 percent of potentially
affected populations have mean household incomes of $10,250. Each of these
percentages (33 percent, 56.4 percent and 4 percent) are multiplied together
to find the weight (0.0075) for ground waters in this category.
To find the total proportion of Class I ground waters, the proportion of
Class I ground waters in each category is multiplied by the weight of the
respective category. These weighted probabilities may then be added to yield
the final result for the proportion of ground-water sources which could be
classified as Class I Irreplaceable: 0.7 percent.
This means that, under the base case assumptions and given the proposed
components for ground water classification, 0.7 percent of ground-water
sources supplying systems that serve at least 2,500 people will be classified
as Class I. If there are about 21,000 such ground-water sources in the U.S.,
about 150 of them will be deemed of the Class I type.
In the next section, we discuss the results of analyses in which we change
our assumptions with regard to the key parameters.
4. SENSITIVITY ANALYSIS
Using the spreadsheet program, it is possible to examine the effect of
changes in critical parameters on the number of Class I water sources. For
example, if per household water usage is taken to be larger than 60,000
gallons, or if further investigation shows that water system costs are greater
relative to income, then the effects of these changes on the analysis can be
easily determined. The impact of changing requirements for ability-to-pay
criteria on the number of Class I water sources can also be assessed. The
sensitivity of the results have been examined in light of changes in three
different variables: per household water usage, percent of income threshold,
and water replacement costs.
Per Household Water Usage Sensitivity
For the base case analysis, the annual per household water usage has been
taken to be 60,000 gallons (which equals 164 gallons per day). If household
usage is greater, but the income available to pay for water supply remains
unchanged, then the per gallon ability to pay is reduced and so the unit cost
threshold for classification as irreplaceable declines. For example, if the
annual household usage is 100,000 gallons, then households with average
incomes of $10,250 can pay only $1,025 per million gallons if 1 percent of
income is available for water supply, rather than the $1,708 available when
usage is only 60,000 gallons.
Increasing the annual per household usage estimate reduces unit cost
threshholds for irreplaceability and therefore increases the number of ground
waters classified as irreplaceable. To quantify the effect of changes in
estimated usage on the number of ground waters classified as irreplaceable,
-------
-20-
the analysis was repeated using alternative estimates of 40,000 gallons,
80,000 gallons, 100,000 gallons, and 120,000 gallons.
The results show that the percentage of ground waters classified as
irreplaceable is reduced to 0.2 percent when usage is estimated as 40,000
gallons, but increased to 1.1 percent, 4.1 percent, and 7.6 percent for usage
of 80,000 gallons, 100,000 gallons, and 120,000 gallons respectively.
Income Requirement Sensitivity
Changing the limit for the percentage of household income that a community
would have to pay for a replacement source also has a large impact on the
proportion of Class I water sources. If the ability to pay from household
income is reduced from 1% to 0.5%, the proportions of water sources classified
as irreplaceable increases to 7.6%. Increasing the threshold to 2% would
result in 0.1 percent of all water sources being considered as Class I
Irreplaceable.
Cost Sensitivity
Cost is an important variable to consider in the sensitivity analysis
because of the uncertainties inherent in inferring replacement costs from the
costs of existing systems. Therefore, the analysis was repeated with all
costs scaled up and down by a range of factors to assess the sensitivity of
the results to changes in estimated replacement costs. The analysis shows
that assuming that replacement costs are 50 percent greater increases the
estimated proportion of Class I ground waters to 2.0 percent. With costs 100
percent higher, the proportion of Class I ground waters is.5.1 percent.
Further cost scaling sensitivities are presented in Exhibit 10.
Substantial Population Cut-Off Sensitivity
The final sensitivity examined is the effect of changes in the substantial
population cut-off. Reducing the substantial population cut-off will increase
the number of ground waters eligible for designation as Class I under the
economic test. Conversely, increasing the substantial population cut-off will
reduce the number of eligible ground waters.
For this sensitivity analysis, to simulate a reduction in the substantial
population cut-off, ground waters supplying systems in the size class serving
about 750 people were included in the analysis. Inclusion of these systems
adds an estimated 6,312 to the 20,592 eligible ground waters, totalling 26,904
ground waters. The analysis shows than an estimated 1.09 percent of these
will be designated Class I under the test, i.e. approximately 290 ground
waters. This is equivalent to about .5 percent of all 60,841 ground waters.
To simulate an increase in the substantial population cut-off, ground
waters supplying systems in the size class serving about 2,000 people were
excluded from the analysis, reducing the number of eligible ground waters by
11,616 from 20,592 to 8,976. Of these, .15 percent are estimated to be
designated as Class I, i.e. approximately 14 ground waters. This is
equivalent to less than .02 percent of the total of 60,841 ground waters.
-------
-21-
EXHIBIT 10
SENSITIVITY TO KEY PARAMETERS OF THE ECONOMIC TEST
OF THE PROPORTION OF GROUND WATERS SUPPLYING
PUBLIC WATER SYSTEMS DESIGNATED AS CLASS I
Sensitivity Run
Base Case
Alternative Assumptions about
Household Water Usage
(gallons/year)
40,000
60,000 (base case)
80,000
100,000
120,000
Alternative Percentage of
Household Income Used in the
Economic Test
.25 percent
.50 percent
.75 percent
1.00 percent (base case)
1.25 percent
1.50 percent
2.00 percent
Alternative Estimates of
Replacement Costs
(percent of cost estimated
in base case)
100 percent (base case)
150 percent
200 percent
300 percent
Proportion of
Ground Waters Supplying
Public Water Systems
Serving Populations
Over 2,000
Designated as Class I
.7 percent
.2 percent
0.7 percent
1.1 percent
4.1 percent
7.6 percent
29.6 percent
7.6 percent
1.1 percent
0.7 percent
0.3 percent
0.2 percent
0.1 percent
0.7 percent
2.0 percent
5.1 percent
19.0 percent
Number of Class
Ground Waters
143
47
143
235
845
1,571
6,088
1,571
235
143
59
47
22
143
409
1,053
3,918
-------
-22-
EXHIBIT 10 (continued)
SENSITIVITY TO KEY PARAMETERS OF THE ECONOMIC TEST
OF THE PROPORTION OF GROUND WATERS SUPPLYING
PUBLIC WATER SYSTEMS DESIGNATED AS CLASS I
Proportion of
Ground Waters Supplying
Public Water Systems
Serving Populations
Over 2,000 Number of Class I
Sensitivity Run Designated as Class I Ground Waters
Substantial Population
Exclude ground waters .2 percent a/ 14
supplying systems serving
approx. 2,000 people
Include ground waters 1.1 percent b/ 290
supplying systems serving
approx. 750 people
a/ A proportion of ground waters supplying systems serving approximately
12,500 people or more (8,976 ground waters).
b/ A proportion of ground waters supplying systems serving approximately
750 people or more (26,904 ground waters).
-------
-23-
5. CONCLUSIONS
Based on the assumptions used in this analysis, the proportion of all
ground-water sources serving public water systems with more than 2,000 users
which are likely to be classified as Class I Irreplaceable is 0.7 percent or
about 150 ground waters. A key reason for this result is the
income-determined unit cost threshold as defined in the proposed
classification guidance. One percent of household income is a fairly large
sum relative to water system costs for all but the poorest communities.
Varying the income requirement changes the proportion of Class I waters
significantly. Exhibit 10 summarizes the sensitivity analyses included in
this paper.
-------
ATTACHMENT
SENSITIVITY ANALYSIS
-------
S^- e» i»-
en — i c-4
I
_
_J „ O. g ^
S S - " S
f es ^ uj
IM il
s -s
: =
O j-i
U Ul
s. a
in (— o
" ~- S gj
se 2 " s
o N- uj
S il
5 _ I !
§ " *- i i'
is "1
O "O —
b
8 §8
8 §S:
gg
gg
0 O 0
§» ^ w
OB 00 OD
000 K> I
s ss s
i O r-. f- o
l
888
8 8 g
—
•g ° s-
•s s
— o o o
w. »«*£*••« M £4 •»! »* £* r* »* £* y»* r+ ** **
S--S gig SiS 818 g8g Si8
1 *•* •*• Ef *•*
-mSZ^Soo ooo ooo o g o Sgs
•Mu)'«.caoOo o o o o o o o ° o <» «* o
? s
*i
i""
£«-
l
*-«»•* 1-4 »* C* )*f 1** »-4 >»« M »•*
oo ^- w .«• .o -a .0 CM £2 o
og 2 —'2 S^M e^»*g
fi
g
gg
o< »-I>*K >««w»rf K*»f*-4 >^t!*^ >*Si*a
c: <3go o o o ogo ooo ^g^
ooo
o o o
0 0 0
^ o o
o
| S
£• § -
1-stJ
§<=•«» O O o ^^^ f^lOro
O ^ •» O O OD ™ OD ~O *g -O
S 8
O O <±> ^-Od OOo
88 ggg
88 g 88
o o o
o o» c>
g S §
-••5.111 §11
•~ » "S s
.888
§§ 88
00 00
8 § 8
8 g 8
ggg
0 0 0
0 «=» 0
g
888
OD IT) H"* ITi CO CO O
-O •*• •*" O>
r- ••• » *• ^ O- ^
o ooo ooo
O OB O
O OO QD
»^ >-» »-•»-• i^« ?** *~t +* t~t
o^o oc> o
O^- OOCM oO«, O
oo goS ooo o
r^ »-* »-«
•»**••=>
(^ CM O
-
88 S
g 8 8
n-> r^ O
en °° o
In « o
0 <=» 0
g— e ^o^ o9o o^oi^ <=>o <=> «=» c> o «> o oo*a <=.<=><=>
— - • — . _ — _, o o O1^^ ^>—•—• «^«-O <^^O
i
^— a,
I 5-
^SS SSS SSK KSS SSg
Oo <»O<> ai CD CD r^f^r^. r-^»O
ogo .» <=> o ^j -o ,0 ^ — ~
oo^ — —._ .0^0 o
0
o S S —
oo o o
S 00 S
0 <=> 0
o o
» - r
S S g
O 090 o-OO
o o o o o o- o
g "=88 ^SS « ^
g
sl
8 •=• " 8
8 8 R
o ® o
•=><=•»
"S » S Ha.
.2 a* £ 2 oi
u. o . §
u.5^ Sr,
3: £ S: ct' "3 *
m a. a!
— o ut
g
s
O i/l O
I
^
"5
I
o
§•
fe.
01
j=:
M
Of
.. §•
.
f -
•n c *-
V 'B «•
o j= a
r;i
S »
=• §
-
sis«
01 OJ —
« =
>- 01 O
Wl ^ A.
§.«
£)
r*
•3
"^ o 2
<=• S LJ
Ul — •«
III
I U> S
i t 8 S
O 3 - —
i^ cj r« t-J
in *
-------
w
§
H
i-l
«J
Cn
o
o
o
w
D
0)
-P
td
O
0)
in
s
(fl
3
C
s
S J c,
s £ a
s s -
ca o w
kJ t/>
I*. CO
S — u
S „ S K S £
S rw o S .«
£ S£ 0 " 1 r--
S§" = S
»—
i * i « 35
I S »— U
'M i!
" ~ g
gsr
5_ls
tj tn
ll
g s §
§ § g
- 2-" ~ jj ~
S £ S R3R S S
S 5 S 555 p; ^
§ :
o o o
o m
k. Oi
a. A
0 "5
•§ I-
«
S«j: v ^o-o^i CNI c»* ,
— w
-, « 1/1 o O c-4
_. — «» S O O
s
s
g § g
•S ° &G
"o "3 »
s -
I
S § S
:S g §
•o o
g
ss
8g
1 O O •* O
r s
So
O
g f 3
000
l§
I g
o <=»
g g
gg
S :
S
II
g g
gg
g g
*
r- ^ o in "G
G> ^ C> o "=>
o o *=» o
o o o o
•5" I -
~ "!• tn
•S o g.^
•g ^"
g
S :
i II
OOO
O O .£,
020
o o o
g O g
o' o o
S ~
S g g g
s s g s g
: g
•g ?
—
t §
<=• o
go o
O o
O « O
i 111
O 000
o »• o
5 £
£ •=
g
s
' . S _' 2 _
i £ _j £ Tm "S
i tn O- a. ^ > o
= 11 := i i =
^11
f 5 2
* a
m
__
2 t g
.2 .3 ~-
i
5 - g.
I
u*
t
I
|
i
o.
^
« » s
Sis
•s s "
s — t*
» 3
III
"oi tS c w
O. X •• OJ
" p. •"..'"
•i o
- 5s-
s
°
- >
11
X .£
•- S o -
^ •. s
- -s. .
in a .
At •* (A
3 •=> —
f-t £ Ol
S 2 I
^ o S
? Q
Qi h.
o ^= -
Z S 2
u Oi O*
« £ »
Qt *~ a,
£ -S fe
•
2 - g
tj => O
<
« s s
b * Q.
------- An error occurred while trying to OCR this image.
------- An error occurred while trying to OCR this image.
------- An error occurred while trying to OCR this image.
------- An error occurred while trying to OCR this image.
------- An error occurred while trying to OCR this image.
-------
" £ S
w -S
g
g " Z S o £
Z UJ — t_' Z -
uj £ PO en rs
C/1 O C-^ »— UJ
i
en
^
UJ U~J '
11 £
-
£
I
u> tn
UJ O
y"S >— I
" O i
1 G LJ ;
—
IS
i
-
•g .S1" Z!
^ I SI.
S g
gg
•-*»•*•-»
Kt JO tO
S g
gg
M o
g:
§:
S —
" 5. i-)
€ s •»
o o o
o o o .
g
§
g - •= £ g
= S 2 £ fe g
* n» -* cr- ~ -
C _ O. — LJ O
.O O* CJ •• Qi *B
o — —• at u.
g ggg g:
O OOo
000 O <=* G 000
^" -o -a -o rvtr-io
OD ^ "^ ^ —« ~" O
- ^-
e^ s:
m o
£ I1: IKE
- _ - £ g o S
o o o o o G o
§0,0 o o o o
S; o » G o o
«* O OOo O
1 as
^§--s~fe:
^ Ij: i.51 -s
- -
O O o
0<=>0
CJ O* O
o o o
O O O
0-00
1,-S
o in
g I g
g
§
§-£> IT)
"™—
o -o K,
O <=» GO
o OO-O
§ -
oOo oor-4 ^o-^, ooo
o o o ooin ioc>«, ooc>
§Oo ^O-^ O-C-c- ^O-^>
^O oOf— «>^O -C>^O
O O C, C.
o o ^
- fe
.C —'
» •-»
s^
-- « - ^ ^ * ^
O3 CO CD WDIOtn CD OP o,
«»t*'^' -o-o-o «-^TO
r^»^r*- *r^-«- oo
OOC- OOO OOO
g g S
«><=»<=>
SSI
g
C1 ^ O O —« O- '—• o. o O c,
So> «>O«-j o<=>o o OOo
g -
g I s §
^* >^ »-»
1? ?
O O C'
£•£>!>
ooo — — —
o o o o o c"
o *-- c-
ooo
h'jOO O~Oo OOo C' —' -C^ O^"C>
OOO ''••So- r— Oo ''J<=. O — c.
hoC»o —* ^ ^ r--O>^> CP~™ O oCDo
CDOO — ^ o- eoOo oino o'^-o.
^^g
o •=- o
o o o
Ck.LL.
s °
u_ es . o
° K 3 5 is
« " S: S:3
" -
• fc
u • a
w •*- u*
Mis
J! 5=1.
« Ip
in ^, Q- rvi
* i»
(U ,-. Ol 01
I •: =
S 6 S "
•ti *a
• ea 51 *
"e "** — u
r sL st i
§ 111-
o o< ^ o
e Ei^
I I
u e
: I-.
= Cr
fc « ~
-
5 =
— 01
tsT^
*o ••••»•
— s « =
""»- IJT -S £
i - * o r
o- — ui —' •»
S £ i
^- S «- S
•5 S
19 0.
" f.
tu ^-oo
t — = °
* S £
o
f— LJ rw (-) ~*
------- An error occurred while trying to OCR this image.
-------
> « •** w? ari ***
i f* & S ,0
i ifi: cT tJ in r-T
O f> *— m
5 i i
i
£
5£
_
—i ii <=> en
S w
SS
S 5
£ " "i g i
=c LU o- S :
i
> O O '
I S o o o o
i§ Hi
«> S o
S 5
f U
r-z
£
.•? .!
w ^
*.«*-•«
•S ° g.
"o 'in
' O 55 f- S
1 O »-. r*- o
O K> •"» O
o o <=> o
g
I g g
g g g
g
5 -a .- 5 o S, SgS
"o^SjJIS^gog § « § « * £. * ° ^ 5. °! ^ * "^ a. ^ ^ *
• c « "5. .^ i_iooo o^o ooe> ooo o "• ~ ~" — ~ ~ — ^'
g
8
5 fe
»~i t*c **« **»-*+* M i-* »*e ysi r»r »j »* »-« *•* +* *•* *J +++*
o «• o o S o ^ H ^" -*» •* -o CM r^ o m DO o in u^
O O O O^*O m °° OD ^ W ^- « —" o -«O ~O C- Kl *•*>
o O o O C> O .i, —' — r-> '"•' r-o o- *»- O *•> f** o &-O-
o o o e> o c> -a "O -o r- **• r-» w •«- o t^.'^-o u->m
o «•
g g
1 C» O «• C
O 000
o o e; o S? e, c> •=> o o <=> o o o o o o ooo o ^* o- o ^" "S" -^ "S* o o o o e>
•4.-*wiai^-k_G.oc' c»2^ ^o^ o £ o e> '=> c> ^Oo ooc>
°'»i'l'«lC «• o «. o o o c-c>^ o^o ci^c. ooo
• C^M^^iLJO^O OO<^ oOo C>^O O*Oo «O •—* -O O^O
° 3"£ * '
^ = s s ^
^ o
fs
*"•
fi ro c> «n ""*
I irt m to **"> ro o^c> Ci^o o<^o
>
ci O
Cv O
o o o o
o ooo
C- O O »-r y^i-rfi^ »^*^^r j-J >-"
r^sj^r--* «=.oo tnu->o. ^>^-
-o^-o ooo —- ™ c- c>o>
>- C *-**-* >-r »^M)_r »-/»-«
*- o— r-"P-o o>«^o o""1-
— — -O f-J ^> O^'O OOB^. C'^'f-'J -_._
— ^* tf rj oo —. <^ o
i-f^^i^m o-o-ci oo-o o^o c>^in c>o-
raOo>" ino-cJ ou^o o*ro oOr^ oo
^> — <_j r^o-e- ff-o too o
o tn — — —
k. at IP
a. ^ *•
OO— C>Ov ^O^ OO'C'
— -—•—- — "•- - C» «• C •=- «• O
o o c-- o o o
^ C- ^ C. O C"
o «• o o <=> o
5 k- c * t- c • w e
M(UOJ ^ of
_|"O*j _j"O*- _i "O —i
O»T O >• o x
8: £ =
=• 11 CX
tn ~ o
S. a
o. V o °
5" e c
£ ~ S S
£ H 1: I.
r iig
^ a*
s sa
s «a
s I" 1
at tn
O C/l i- ^
*- £
C£l CT ^ -D
= fc- **
O > O rB
§• . §•
^ U1 CL
W> w **
a* ^T —
— ro >
> S a>
^ in
C in
* L. &
* o o.
o S »
— » 2
5 t §
™ «
•§ = s -5
» ="
i •"
C IT
g s
o tn —
srs
Q_ — a.
ui „ ft* m fc-
^ w oi O*
J= ,_• Ul C *
S S. a, C a,
m ui if 01 k-
------- An error occurred while trying to OCR this image.
-------
b !E *>*
c/i o t~> *~
1* I
^ — ^
K m 3 a
9* ^ a v>
"
s gs
Sr o>"
- . l-sl
I !
i H
ill
i •»• "? ^ •
! " « e i
g "5
'M 50
M M<
o o ;
•
• e ^ **i •• LJ
-a &" z a ^
|SSjr
oa &
oaoa
CBOBI *•»***•••» r* r*
11
S « S
»* MMM M ••* M M »* M MI
.a - ""•
S—• e O ^ <> «s .
•O -• 4 O ^ ** '
•SS S =£ stS8 S:
3:
S
§
ss
>-^M ^•>^^f*» 9-»>
<«%e~O **r~t^ ^v
SS
1
• S
II
O
lit ls§ 1s
S 2 ? § a o So
°
IS SS
i it
ssi
S
0
sss
SS S
55g
^ O ^
£ s s
•8 S
in iii iii isi H
5,11
i i;
^« M M M tj >* **£** *^ 51 *"* *^ S "^
288 °S8 «S8 ="8 ^^S
*"
S »i —
= t £ = t £ =
-1 1 -s -• -5 " -1
•iUJ KU4
IS ^il ^12
K ie fc.- S
s 5 s- a *
w> — y *
3 i
5 -
s I i
•if ja M>
£ "at
I S5*
fe. i 2; a
-5 -r:s
§ -5 k S.
— 3 *•" in
I »1s
ss
^ n " •* ** *• e
SI - 5 s- * j5
?S I 2S fcjl
It." 3-5 -.3
^ f 11 j * I *• i
•8 o § J
£ 13*
. .S a .-
5 it.
£ £-5
a * i-
s w - c
u m rft £
5 =12
I ii?
^ M ^ Ol
.» s «. s
Jl!
Iri
= •
|"i. fl fc?
1=I_ *^ *5S
,?i= i a^i
------- An error occurred while trying to OCR this image.
------- An error occurred while trying to OCR this image.
-------
CLASS III: DEVELOPMENT OF
AN ECONOMIC TEST
1. INTRODUCTION
In this analysis, attention is focused on an economic test for the
determination of Class III ground waters. Initially EPA defined Class III
ground waters as those with total dissolved solids (TDS) concentrations of
greater than 10,000 mg/1, or waters that were contaminated with industrial or
agricultural waste and could not be adequately treated using processes
commonly employed in municipal water treatment.
However, initial analyses revealed that the processes and process
configurations employed for municipal water treatment vary widely among
different geographic regions in the U.S. Thus, what may be considered an
unreasonable practice in one location may be perfectly feasible in another
location. For example, water treatment plants in areas of abundant, high
quality water sources may include only conventional unit processes such as
coagulation, flocculation, clarification, sand filtration, and disinfection.
However, in regions where water sources are generally not of high quality,
more sophisticated and/or expensive technologies such as reverse osmosis,
granular-activated carbon (GAG) filtration, and air stripping may be widely
employed.
Also, the type and level of ground water contamination may vary widely,
from waters containing only elevated levels of nitrate to waters containing
high levels of a wide variety of organic and inorganic contaminants. Ground
waters of the latter type may be treated only by relatively complex
arrangements of unit processes, or treatment "trains". For example, a ground
water contaminated by a leaking industrial landfill may contain volatile
organic compounds, heavy metals, and PCBs, and may be treated adquately only
by a treatment "train" comprised of air stripping followed by GAC filtration
followed by precipitation/sedimentation/filtration.
Therefore, a decision was made to interpret the "adequately treated by
processes commonly employed in municipal water treatment" criterion of the
Class III definition as allowing consideration of whether the ground water in
question could be adequately treated by any treatment "train" of unit
processes available either on a regional or national scale. Costs for water
treated by these sophisticated "trains" could also be evaluated, and compared
to some local, regional, or national cost benchmark, or economic cut-off
point, to decide whether the ground water should be classified as Class II or
Class III.
This paper outlines an approach for developing an economic test for the
determination of Class III ground waters. The situations being considered in
this test are somewhat more complicated than those considered in the Class I
economic test that is currently being evaluated, in that there are more
site-specific parameters to consider. They include:
-------
-2-
• type and level of contamination varies widely among
contaminated aquifers, making the selection of treatment
alternatives difficult;
• the waters are not currently being used as beneficial
water sources, thus making it difficult to identify a
potential user population;
• hydrogeologic parameters, most particularly well
yield, may vary widely among different aquifers, thus
affecting costs; and
• the difficulty in identifying a potential user
population makes such factors as transmission costs and
water supply scale economies difficult to incorporate.
Also, it is necessary to establish what a reasonable cost benchmark, or
economic cutoff point, for comparison to the cost of treated water from the
aquifer in question would be. Thus, there are a number of options to be
considered in any Class III economic test, and any approach adopted must be
flexible enough to incorporate these various options.
The following sections of this paper identify several population and
costing options that need to be considered in a Class III ground-water
economic test, and present an approach for developing such a test. In Section
2.0, the problem of deciding upon a potential user population for a
contaminated ground water is addressed, and several options with regard to
selecting a population are discussed. Section 3.0 considers the implications
of including various costing options in the economic test, and in particular
focuses on variables such as transmission costs, the costs of well field
development, and the costs associated with various treatment alternatives.
Section 4.0 outlines the approach developed, based on the various population
and costing options considered, for determining Class III ground waters using
an economic test. Ground-water treatment cost scenarios are developed in
Section 4.1, by combining different population and costing options. In
Section 4.2, the development of a cost benchmark for comparison to the cost of
treated ground water from a particular aquifer is discussed, and two
alternative methods for computing such a benchmark are presented. Section 4.3
presents a hypothetical example classification, based on the approach outlined
in the previous sections of the analysis.
2. POPULATION ISSUES IN USING A CLASS III ECONOMIC TEST
By definition, a Class III ground water is contaminated and is not
currently used for drinking purposes. The treatability component of the test,
however, implies use by some hypothetical population. Three population
factors are critical to the results of the economic burden test:
(1) the size of the potential user population, because of
economies of scale and cost-spreading;
-------
-3-
(2) the distance to the site of the potential user
population, because water transmission costs, including
rights of way costs, vary with distance; and
(3) the average household income, which would be used to
determine the cost benchmark.
Two conceptual issues need to be addressed in considering the use of an
economic burden test for Class III ground waters:
(1) Does the concept of substantial population have any
relevance?
(2) How should a potential user population be identified?
It may be helpful to illustrate potential Class III situations. Of key
importance is the requirement that Class III waters not be currently used. As
defined by the current draft guidance, this means that not a single well
supplying drinking water is found in the Classification Review Area. Thus,
any people living in the Classification Review Area are supplied through water
systems with sources outside of the area. This situation would arise because
outside sources of water are or were cheaper, because of institutional
constraints, technical constraints, yield limitations, etc. Alternatively,
there may be no residences in the Classification Review Area itself;
populations may be located in varying proximities to the ground water, outside
the boundaries of the Review Area. This complicates the application of a
burden criterion in the Class III context.
For simplicity, consider a typology of nine (9) situations representing
combinations of populations P within the Classification Review Area (CRA), and
populations PI outside of this area but within a distance that is not uncommon
for water transport for populations of that size. Either P or PI can (1)
equal zero, (2) be less than the substantial population threshold N, or (3) be
equal to or greater than N. These situations are shown in Exhibit 1. The
likelihood of these situations is affected, to some degree, by the definition
of N, given a relatively fixed definition of the CRA, in terms of square miles.
It is possible to make some basic observations about the situations listed
in Exhibit 1:
• Situation 1 represents a western scenario (e.g.,
isolated mining area in the western U.S. or Alaska);
• Situations 2 and 3 represent a ground-water in a
non-residential area (e.g., industrial park) near a
population center of some size;
• Situations 4 and 7 are relatively unlikely; the larger
the value of N, the less likely is situation 7 (e.g.,
for N=25,000, the population within a 12.5 square mile
CRA is at least 2,000 persons per square mile);
-------
-4-
EXHIBIT 1
ILLUSTRATIVE TYPOLOGY OF EXISTING POPULATIONS
IN PROXIMITY TO CLASS III GROUND WATERS
Situation
1
2
3
4
5
6
7
8
9
Population
Within CRA
0
0
0
N
>N
>N
Population
Outside CRA
0
N
0
N
0
N
-------
-5-
• Situations 5 and 6 represent likely scenarios in
populated areas; note that adding the populations within
and outside the CRA in situation 5 could produce a total
greater than N; and
• Situations 8 and 9 are more likely than 7, and
represent relatively densely settled areas, depending on
the value given N.
Further data analysis could quantify these observations somewhat. For
policy purposes, it would seem consistent with the Class I approach, to
consider only "substantial populations" in a Class III economic burden test.
In essence, the Agency would consider imposing higher cost requirements (Class
II standards) in light of the greater potential benefit, as measured by the
number of people potentially involved. On the other hand, one can argue that
if a population less than N can shoulder the burden of water treatment, then
the Agency should not consider a Class II designation. A "pure" ability to
pay approach, regardless of the number of people involved, could appear to
favor wealthy pockets of residents (in resort or high rent districts).
Potential User Populations
Consider two types of potential user populations: existing and proxy. An
existing population is a potential user of water if it is located within a
"reasonable distance" from the site. Consistent with the Class I ground water
guidelines, "reasonable distance" may be defined as not exceeding the uncommon
pipeline distance in the EPA Region for that population category. The size,
distance to the site, and average household income of an existing population
can be known fairly unequivocally. A proxy potential user is an imaginary
population whose size and distance satisfy the reasonable distance criterion.
The size can be assumed equal to the average or median size of populations
served by ground water in the region. The distance to the site may then be
taken as the uncommon pipeline distance- in the EPA Region for that population
category; this is a conservative assumption since most existing users would be
located closer to the ground water. Finally, the average household income in
the Region may be used in an economic burden test for such a population. In
either case (existing or proxy), the size of the potential user population
should be constrained to a maximum value given by the maximum sustainable
yield of the ground-water aquifer. For example, if the maximum sustainable
yield of the contaminated ground-water aquifer is one million gallons per day,
the size of the potential user population should be assumed to be no greater
than 16,667 persons, assuming an individual consumption of 60 gallons of water
per day.
A given ground-water site may be characterized by one of six population
scenarios. These scenarios will be used to assess the implications of
alternative configurations of the economic test. These scenarios, shown in
Exhibit 2, are defined by the answers (yes or no) to the following three
questions:
(1) Is there a substantial population living in the vicnity
of the site?
-------
-6-
Exhibit 2
POPULATIONS SCENARIOS FOR
THE ECONOMIC TEST
Potential
User Population
Exists ?
Existing
Population
Passes Test?
Proxy
Population
Passes Test?
Proxy
Population
Passes Test?
Proxy
Population
Passes Test?
SCENARIOS
Option 1
Option 2
Option 3
Option 4
II
II
II
II
II
II
III
II
III
III
II
II
III
III
III
III
III
II
II
II
HI
III
III
III
-------
-7-
(2) If there is, would the site pass the economic test
using this existing population as a potential user of
treated water?
(3) Would the site pass the test using a proxy potential
user population? (Note that a proxy population is
always available, by definition)
Scenarios 1 to 4 are for sites where there exists a potential user
population. Conversely, no potential user population can be identified in
sites falling under Scenarios 5 or 6. Scenario 1 is for a site that passes
the economic test using the existing population as well as the proxy
population. Conversely, Scenario 4 is for a site that fails the test using
the existing population, and fails it also using the proxy population.
Scenario 2 corresponds to a ground-water site which fails the economic test
using the existing population while it passes the test using the proxy
population while Scenario 3 is simply the reverse (i,e., the site passes the
existing population test and fails the proxy population test). Finally, a
site that has no nearby substantial population and which passes (respectively,
fails) the test using the proxy population is a Scenario 5 (respectively, 6)
site.
The outcome of the Class III versus Class II decision process using an
economic burden test will depend on the type of potential user population
considered. There are four options for deciding which populations to consider:
(1) existing only
(2) proxy when no existing
(3) proxy only
(4) both existing and proxy.
1. Option 1: Existing Only
The "existing only" option is one in which only existing, co-located
potential user populations may be considered in the economic treatability
test. The advantage of this option is that it considers real situations.
Thus when no potential user population exists in the vicinity of the site, the
ground water is automatically classified as Class III, regardless of its level
of contamination; absent a potential user population, the ground water is not
a potential source of drinking water (of course, we assume in this memo that
the ground water satisfies the other Class III criteria, i.e. it has few
beneficial uses and is not connected to Class I or Class II ground water in a
way that would allow contaminants to migrate to these waters).
The "existing only" option implies that only present situations are
accounted for, regardless of future developments. In particular, it doesn't
allow for the possibility that a substantial population settlement may develop
in the future in the vicinity of the site, thereby reducing the chances that
any such demographic evolution will ever take place. This is consistent with
the philosophy of the strategy to acknowledge the lead role of the States and
localities in land use planning. This option may also be justified on the
grounds that demographic expansion has reached a steady state which precludes
-------
-8-
the possibility of development of new population settlements in the USA.
Option 2 assumes the opposite point of view.
Under this option, a site will be classified as a Class II ground water if
it falls under Scenarios 1 or 2; that is, if it is located "near" a
substantial population and if it passes the economic test using that
population as a potential user of treated water. Any other site will be
classified as Class III. Whether a proxy population would pass the test or
not has no bearing whatsoever on the outcome of the classification process.
2. Option 2: Proxy When No Existing
In the "proxy when no existing" option, a proxy population is considered
when there is no existing potential user population. Unlike Option 1, this
option assumes that some population may eventually settle in the area where no
substantial population currently lives. However, it doesn't abstract away
from the demographic environment by using real populations when they exist.
Because there is always a potential user population, this option will always
test the "economic" treatability of the ground water. In a sense, it gives
more weight in the classification process to the ground water itself.
Consider, for example, a contaminated ground water for which no potential
user population exists (Scenarios 5 and 6). In an economic test using
Option 1, this ground water would automatically classify as Class III,
regardless of how contaminated it is or how difficult it is to exploit. Under
Option 2, however, the economic treatability of the ground water can be tested
against a surrogate population, despite the absence of substantial populations
that could use the water after treatment. The ground water is in effect given
another chance to classify as Class II. If indeed, the use of a proxy
population still results in a Class III decision, it would mean that the
characteristics of the ground water are such that it is economically
infeasible to develop the resource for a hypothetical future population. If,
on the other hand, a proxy population passes the test, the ground water will
classify as Class II; the treatability characteristics of the ground water
have prevailed over the absence of potential user populations.
3. Option 3: Proxy Only
In the "proxy only" option, only proxy populations are considered for the
economic test. Unlike Options 1 and 2, this option completely abstracts away
from the immediate demographic environment by using surrogate populations
consistently. Because surrogate populations are specified using regional
averages for size, distance, and household income, the classification of a
ground water is made on the basis of the ground-water characteristics as
tested against a "typical" user population in the EPA region. Whether the
existing population (if there is one) is typical or not makes no difference in
the classification outcome. Thus, this option takes the logic of Option 2 one
step further by putting all the weight on the ground water itself and
completely ignoring the actual distribution of populations around the site.
The "proxy only" option shows a striking parallel with the reference
technology criterion where a treatment train may be applied if it uses
-------
-9-
technologies available in the EPA Region. Whether the technologies are
available elsewhere in the nation makes no difference in the outcome of the
classification using the reference technology criterion. This approach limits
the scope from the nation to the EPA region. Symmetrically, the emphasis
under Option 3 is shifted from the immediate vicinity of the site to the EPA
region.
4. Option 4: Both Existing and Proxy
Under Option 4, both existing and proxy populations may be considered in
the economic test. In other words, a ground water would classify as Class II
if it could pass the economic test using either population criterion. This
option attempts to pull together the advantages of the previous ones while
eliminating their drawbacks. Indeed it considers existing populations (like
Option 1) while still allowing for future growth (like Option 2). Also it
consistently tests the ground water using a typical regional population (like
Option 3).
There are two scenarios for which Option 4 would result in a Class III
determination, as opposed to 3 scenarios for Options 2 and 3, and 4 under
Option 1. Adopting Option 4 would hence be consistent with the intent of
having fewer Class III ground waters. The scenarios that fail under Option 4
represent those instances in which a proxy population would fail the test and
either a substantial population does not exist or it does but fails the
economic test. They can be thought of as worst cases since they classify for
Class III under all four options.
Choice Among the Options
The choice of one of the four options depends on the relative frequencies
of the six scenarios, which in turn depend on the probability of a positive
answer to each of the three questions posed to construct the scenario tree of
Exhibit 1. In the absence of any information on these probabilities, one can
imagine various situations and assess the relative merits of the 4 options for
each. Assume, for example, that Scenario 3 has a very low probability of
occurrence (e.g. lower than a tenth of the second lowest occurrence
probability). Then, Options 3 and 4 are so similar that the use of Option 3
might be preferred to Option 4 due to its simplicity (the proxy population is
specified once and for all by EPA Region). Assuming another instance where
all sites fall under Scenarios 1, 4, or 6, the four options would yield
identical classes. In this case too, it would make sense to adopt Option 3
because of its simplicity. The data analysis performed for RCRA and CERCLA
sites sheds some light on the relative frequencies of the six scenarios. (See
attached issue paper.)
3. COSTING OPTIONS
The costs of providing drinking water consist of four major components:
(1) Acquisition;
(2) Treatment;
-------
-io-
Delivery; and
(4) Support Services and Interest Charges.
In deciding which cost items to include in the economic test, we should
keep in mind the purpose of the test which is to identify those situations
where providing drinking water from the contaminated source to some population
will be too costly to envisage. The relative cost is defined in reference to
a benchmark which incorporates the same cost items initially used to compute
the "cost of treated water". For example, if we ignore water distribution in
our costing mechanism, we should also disregard it in constructing the
benchmark.
Because the aquifer is contaminated, the cost of treating water to a
drinking level is likely to be substantially higher than that currently
observed in the region, or even nationwide. The example developed in Section
4.3 of this paper shows a treatment cost of $4.21 per thousand gallons to be
compared to the average treatment cost for small systems of $.132 per thousand
gallons, as reported in the issue paper entitled "Cost of Water Supply
Systems". We could therefore expect to find unit treatment costs that are up
to 1.5 orders of magnitude higher than the average unit treatment costs.
If treatment costs are expected to vary widely, they should be included in
the economic test in order to reflect site specific conditions. Conversely,
if the regional distribution of costs for a given item is such that the
average cost and higher cost are very close (say, their ratio is less than 2),
then this item need not be used in the economic test. We can justify the
decision by saying that the cost contributions of this item on both sides of
the test inequality are so close to being equal (in comparison to treatment)
that the inclusion of this item is not likely to affect the test results.
3.1 Acquisition Costs
Acquisition costs are the costs of producing and acquiring water; they
include the cost of land, rights of way, and well field development costs.
The latter may vary widely depending on the hydrogeologic conditions, in
particular the depth to the acquifer and the nature of the geologic formations
overlying them. We therefore recommend including them in the economic test.
3.2 Treatment Costs
Treatment costs include the costs of treatment plants and equipment, and
the costs of the chemicals that are added to the water. For a water of given
quality, the unit costs of treatment depend on the rate and/or quantity of
water treated and the treatment technologies used.
Due to scale economies, it costs less per gallon to treat a million
gallons of water per day than to treat a hundred thousand gallons per day,
other things being equal (i.e., applying the same treatment process to water
of the same quality). The capacity of the treatment plant is determined by
the size of the potential user population.
-------
-11-
Th e other factor that determines treatment costs is the treatment process
used. Depending on the level of contamination of the aquifer, a single
treatment technology may not be sufficient to upgrade the water to drinking
water standards. In this case, a combination of treatment technologies is
required. EPA may decide that a treatment train is applicable if it is:
(1) already in use in the EPA region;
(2) already in use in the nation; or
(3) a feasible combination of applicable treatment
technologies that results in the lowest cost.
Most water treatment technologies are used by public water systems
individually in all EPA regions, while combinations of those technologies are
less commonly used, if ever. For example, while ion exchange, carbon
adsorption, and aeration are all used individually to treat water in EPA
Regions II to VIII, a treatment process combining these three technologies is
used only in EPA Region II. For this reason, using only the currently
existing combinations would be very restrictive. For an economic burden test,
we recommend using a feasible combination of available treatment technologies
that would result in the lowest cost.
3.3 Delivery Costs
Delivery costs include transmission and distribution costs. Transmission
costs are the costs of pumping water from the treatment plant to the main
distribution network, and apply when the service population is not currently
served by a centralized water supply system. We expect transmission costs
within an EPA region to be distributed widely enough to warrant inclusion of
transmission costs in the economic test, for populations that are currently
obtaining drinking water from private wells. These costs should not be
included in the economic test for populations currently being supplied by
centralized water systems.
Distribution costs include the capital, and operations and maintenance
costs of the piping network (transmission pipes excluded), and should also be
included in the economic test for populations currently served by private
wells.
3.4 Support Services and Interest Charges
Support services are primarily administrative and customer service costs
and therefore depend primarily on the characteristics of the service
population and the level of service chosen by the utility. Interest charges
include interest on outstanding debt and depend on how the utility finances
its investments. The unit costs of both support services and interest charges
are essentially independent of the characteristics of the site and may thus be
ignored outright in the economic test (see the issue paper entitled "Cost of
Water Supply Systems").
-------
-12-
3.5 Summary of Costing Options
Treatment should always be considered in an economic test while the cost
of support services and interest charges would always be ignored. Depending
on whether acquisition or delivery costs are included in the economic test,
there are four options for estimating the costs of providing treated water.
Option 1 is when only treatment costs are considered. Option 2 uses both
acquisition and treatment costs while Option 3 uses delivery costs in addition
to treatment costs. Option 4 includes acquisition, treatment, and
distribution costs in the economic test. In each option, the benchmark is
adjusted accordingly as shown in Section 4.2 of this paper.
4. DEVELOPMENT OF ECONOMIC TEST FOR CLASS III GROUND WATER
This section reviews the various options for conducting an economic burden
test. It then defines two different benchmarks and shows how to construct
them under various test options. Finally, it presents a hypothetical example
of an economic burden test.
4.1 Test Options
The economic test must be based on decisions about what kind of
populations to look at and which cost items to consider. Section 2 of this
paper identified 4 populations options: existing, proxy if not existing,
proxy, both existing and proxy. It also described four options for costing
the treatability of the ground water. By combining population options with
the costing options, there are 16 options for conducting the economic test.
One such option is, for example, to use existing potential user populations
and consider only treatment costs.
4.2 Comparison of Treatment Cost Scenarios to Cost Benchmarks
Development of Cost Benchmarks
As outlined in the preceding section of this memo, a number of
ground-water treatment cost scenarios may be developed, each involving a
specific aquifer, a population that will potentially be served by that
aquifer, and treatment technologies that may be used to restore the quality of
the ground water in question. In order to determine which of these ground
waters should be designated as Class III, it is necessary to compare the cost
involved in using this aquifer as a water supply to a cost benchmark. Ground
waters for which clean-up costs exceed this benchmark would not be considered
feasible sources of drinking water. The cost benchmark would thus provide an
economic cut-off for the determination of Class III ground waters.
In setting a cost benchmark, it is important that the methodology used in
computing the benchmark be consistent with the ground-water treatment cost
scenario developed, otherwise a comparison between the two is not valid. For
example, if the costs of well development are not considered in a particular
scenario, then they should not be considered in the cost benchmark either.
Thus, the cost benchmark vs. treatment cost scenario may be considered as an
-------
-13-
inequality (or inequation), and in order to obtain a valid comparison all
parameters considered must appear on both the right and left sides of the
inequality.
A cost benchmark for water treatment could be defined by one of two
methods. One method would involve a comparison of treatment costs to the
average water cost for the region, and would require that an incremental
increase to the average cost be set as the economic cut-off. For example, the
average cost of drinking water in a region may be $863 per million gallons.
Based on this average cost, a cost benchmark could be set by estimating that
incremental increases of more than 20 percent are not reasonable. Thus,
contaminated ground-waters that required treatment resulting in a total cost
of more than $1,036 ($863 -I- $173) per million gallons would be considered to
be beyond the cost benchmark and would be designated as Class III.
A second method of setting a cost benchmark would be to use household
income in the affected area as a measure of the community's ability to pay for
additional water supply costs. This method would then involve estimating a
percentage of household income that would be reasonably spent for water
supply, and using this value to obtain a unit cost threshold, or economic
cut-off, for the affected community. This method is directly analogous to the
unit cost threshold method currently under consideration for the determination
of Class I ground waters.
In the following sections each of these two methods is discussed in
greater detail. In particular, we focus on key features and the advantages
and disadvantages of each method for setting a cost benchmark, and discuss
what data would be required for each type of analysis. Available data are
also presented and analyzed.
Setting a Cost Benchmark Using Average Regional Water Costs
A cost benchmark, or economic cut-off, for Class III ground waters may be
set on the basis of average water costs for the area of interest. Given an
average water cost per million gallons within the area of interest, one can
estimate an incremental increase for water costs beyond which treatment is no
longer economically feasible. This value can then be used to determine
whether a ground water should be designated as Class III.
Information on the average cost of water at state, regional, or national
levels is necessary to carry out this approach, and may be obtained from the
1981 AWWA Water Utility Operating Data National Survey.1 Current
information on water supply costs for systems of various sizes is also
available from two sources: the 1982 Temple, Barker, and Sloane, Inc., study
1 AWWA, 1981 Water Utility Operating Data, American Water Works
Association, Denver, Colorado, 1981.
-------
-14-
for the EPA Office of Drinking Water,2 and the ACT, Inc. study for the EPA
Municipal and Environmental Research Laboratory.3 Using these data, one
could make an informed decision, based on economic considerations and policy
guidelines, as to what incremental increase would be set as the cost
benchmark, or unit cost threshold.
It is important to note that the ability to account for the various
options included in a particular scenario is inherent in this methodology for
computing a cost benchmark. In particular, this methodology allows for: (1)
the adjustment of the cost benchmark for systems for which different supply
options are included; and (2) the adjustment of the cost benchmark for
different-sized water supply systems, thus accounting for scale economies in
water supply.
For example, information contained in the ACT study may be used to obtain
the percentage of total water supply costs accounted for by treatment in a
small or large water supply system. This information may then be combined
with regional water cost data from the AWWA survey to obtain an average cost
of water treatment only (rather than water supply) for that particular region,
and the cost benchmark may then be set accordingly. The same approach may be
applied to situations in which transmission costs or transmission and well
development costs are also to be considered.
This type of benchmark does not account for differences in economic levels
between communities, but rather assumes that the average regional water cost
is a satisfactory indicator of the ability of a community to pay for water
supply. Although this approach has not been tested, one would intuitively
assume that any classification system using such a cost benchmark would be
particularly sensitive to the "reasonable" incremental cost increase decided
upon, and therefore a standardized method for obtaining this value would be
desirable. One way to obtain a value for this increase would be to construct
a distribution for water supply costs for the region. Assuming that costs
were normally distributed about the mean, 95 percent of water supply costs
would fall within two standard deviations of the mean. A cut-off value could
then be chosen from this distribution; for example, the 95th percentile (two
standard deviations above the mean) could be selected as the maximum
reasonable cost for water supply in the state or region of consideration. In
setting this cost threshold, specific information on situations where
contaminated ground water is now being treated and used as drinking water
would be especially useful.
Another factor to consider in this approach is that the dependence of the
methodology on average water costs per region means that the reasonable
2 Temple Barker and Sloane, Inc. Survey of Financial and Operating
Characteristics of Community Water Systems. Prepared for the EPA Office of
Drinking Water, 1982.
3 ACT, Inc., Economics of Water Delivery for Small and Large Systems.
Prepared for the EPA Municipal and Environmental Research Laboratory, 1977.
-------
-15-
incremental cost increase, if set at constant value, will result in different
amounts of available funds for different regions. For example, average water
costs may be high in one region, where a treatment process such as
desalination is used, while costs in another region, where high quality water
sources are abundant, may be significantly less. Thus an incremental increase
of 20 percent will provide more money for contaminated ground-water treatment
in the first region than the second region. As a result, depending on the
regions they are located in, two ground waters of identical chemical
composition may be classified differently using this system.
Setting a Cost Benchmark Using Population Income
A cost benchmark may also be established based on the average household
income of population, which may be used as a measure of the community's
ability to pay for the increased costs of using a contaminated water source
for drinking water supply. Thus, for a particular household income level, the
determination of whether a ground water is Class II ,or Class III could be made
by comparing the cost of treatment to a pre-established percentage of
household income. If the cost of treatment is higher than the designated
fraction of household income, the ground water in question would be designated
as Class III.
As mentioned previously, this method for computing a cost benchmark is
directly analogous to the unit cost threshold method currently under
consideration for the determination of Class I ground waters. However, in
this case, it is the cost incurred in supplying a community with treated water
from a contaminated source, rather than the cost of obtaining a replacement
water supply, that is compared to the cost benchmark.
Information on average household incomes in the U.S. at the national,
state, county and zip code levels may be obtained from the Federal Census
Bureau (1985 Statistical Abstract). In using this information to determine a
cost benchmark, four factors must be considered. They are:
• Whether an existing or proxy population is being used;
• The percent of household income that should be
considered to be available for water supply costs;
• Whether the mean or median household income level
should be used; and
• The population size, or area of consideration, that
should be used in the analysis.
All of these factors will have a significant effect on the outcome of any
classification analysis, and thus must be considered carefully. The Class I
economic test currently being considered is particularly sensitive to the
percent of household income considered to be available for water supply;
therefore, one would anticipate that the Class III test outlined in this paper
would also be affected significantly by the choice of this parameter, since
-------
-16-
the tests are somewhat similar. Preliminary EPA policy guidelines have
established a figure of 1 percent of household income for this parameter.
In most cases, it would seem that the mean value for household income in a
population would be used to compute the benchmark, although in some instances
the median would perhaps be a more useful indicator of economic conditions in
a particular community, particularly if a proxy population is used. In
implementing a methodology such as the one outlined here, it would be very
important to test both the mean and median values and to determine the
relative merits of using one in favor of the other.
This approach does account for economic conditions in the selected
populations, but does not take into consideration the average costs for water
in the region in which the community is located. Thus, although the
community's ability to pay is accounted for, there is no way in using this
method to determine whether the costs allowed for using a contaminated source
for drinking water are reasonable compared to what is typical for that area.
It should also be noted that this methodology for establishing a cost
benchmark, like the average regional water costs methodology, can be adjusted
to maintain consistency with treatment cost scenarios being considered, by
assuming that the percentage of household income available represents the
total amount available for water supply. Thus, if only treatment costs are
considered, and treatment represents 15 percent of total water cost, then the
treatment cost for the contaminated source should be compared to 15 percent of
the available percentage of household income to establish whether the cost
threshold has been exceeded.
4.3 An Example Classification - Comparison of Costs for Treated Water
from a Hypothetical Treatment "Train" to Average Water Cost and
Percentage of Household Income Benchmarks
In this section of the analysis, an example classification is presented.
This example compares the cost of treated water for a hypothetical treatment
"train" (or reference technology) to the cost benchmarks developed in the
previous section of the paper. By presenting this example, the interaction
between the different components of the economic test may be emphasized, and
the various options outlined in Sections 2 and 3 may be considered.
In the example, a ground water considered to be typical of that which may
be found in the area of CERCLA sites or poorly operated RCRA facilities is
considered. The ground water contains a non-volatile, non-biodegradable
organic compound, a set of volatile organics, heavy metals, and elevated levels
of suspended and total dissolved solids. In order to use this contaminated
ground water as a drinking water source, it is necessary to employ a
relatively sophisticated treatment "train", consisting of (1) air stripping;
(2) precipitation using lime addition and clarification; (3) rapid sand
filtration; and (4) reverse osmosis. The costs for treated water utilizing
this "prototype" ground water as a source are summarized in Exhibit 3.
In determining whether this "prototype" ground water should be classified
as Class II or Class III, the population and costing options outlined in the
-------
-17-
EXHIBIT 3
COSTS PER GALLON OF WATER TREATMENT FOR A
PROTOTYPE CLASS III GROUNDWATER
(Costs in million dollars)
Total Annualized Total
Process Stage Capital Capital O&M
Pumping to plant N/A N/A N/A
Storage N/A N/A N/A
Air stripping 1.5 0.2 0.45
Precipitation 1.3 0.1 0.33
Sand filtration 1.5 0.2 0.25
Reverse osmosis 1.2 0.1 0.70
"Totals" 5.5 0.6 1.73
Sources: Costs from U.S. Environmental Protection Agency, Treatability
Manual, Volume III, Technologies for Control/Removal of Pollutants,
EPA 600/2-82-001C, September 1981.
Assumptions: Plant designed to treat 1.5 million gallons per day (i.e., to
serve approximately 10,000 people).
Capital costs annualized using the capital recovery factor
approach with a 10 percent discount rate.
Untreated groundwater contains:
a non-volatile, non-bio-degradable organic;
a set of volatile organics;
dissolved salts, primary chlorides; and
suspended solids and metals, a few dissolved solids.
-------
-18-
previous sections of the paper must be considered. For the purposes of this
example, the scale economies in water supply will be taken into account by
identifying the potential user population as a small population, and thus the
comparison of treated water costs to average water costs will be for small
systems.
The average water cost benchmark has in this case been developed by
constructing a cost distribution for small water systems on a nationwide
basis. This was accomplished by aggregating the cost distributions from four
states considered to be representative of the four major geographic provinces
of the U.S. These were: New York (Northeast); Texas (South); Illinois (North-
Central); California (West). A cut-off value of 1 million gallons per day (1
MGD) total water produced was used to distinguish between small and large
systems.
Distributions for mean income levels on a nationwide basis were obtained
from 1980 census data at both the county and zip code level. Mean household
income groups considered were: under $5,000; $5,000 to $10,000; $10,000 to
$15,000; $15,000 to $20,000; $20,000 to $25,000; $25,000 to $30,000; and over
$30,000. The percentage of household income available for water supply was
set at 1 percent, based on preliminary EPA guidelines, and adjusted to the
unit comparison volume of 1,000 gallons of treated water using an average
annual consumption of 60,000 gallons per household (based on information from
the U.S. Geological Survey 1983 National Water Summary").
The results of the comparison of these cost benchmarks to the cost of
treated water from the previously outlined reference technology are presented
in Exhibit 3. As illustrated, the cost for treated water -from the reference
technology, at $4.21 per 1,000 gallons, is considerably higher than the
average national water cost for small systems, and in fact does not even fall
within the 98% limits of the distribution for these costs. In order for this
ground water to be classified as Class II using the average national water
cost benchmark, the allowable incremental increase would need to be set at
approximately 300 percent of the mean national cost of $1.42 per 1,000
gallons. Also of note is the fact that the treatment cost estimate of $4.21
per 1,000 gallons does not include the costs of acquisition and delivery.
This is despite the fact that actual water charges used to establish the
benchmark reflect these costs. If the benchmark or costing method is
consequently adjusted, it becomes even less likely that the ground-water would
be classified as Class II. Thus, using this classification system,
essentially all "prototype" contaminated groundwaters would be classified as
Class III.
If the county or zip code mean income distribution (adjusted to 1 percent
of household income and per 1,000 gallons) is used, the results are
significantly different. The percentages of cases for which the percentage of
household income available for water supply is greater than the cost for
u USGS, 1983 National Water Summary - Hydrogeologic Events and Issues,
U.S. Geological Survey Water Supply Paper 2250, 1983.
-------
-19-
treated water are 9.1 percent and 2.9 percent, for.zip code and county mean
income distributions, respectively. These regions are illustrated by the
shaded area on the graph in Exhibit 4. Thus using the percentage of household
income benchmark, 91-97 percent of the "prototype" contaminated ground waters
would be classified as Class III, depending upon which distribution (county or
zip code) was chosen.
This example is meant to be purely illustrative, and any conclusions drawn
from it should be considered tentative. However, the implications in choosing
one type of cost benchmark as opposed to another are clearly indicated, as is
the fact that the costs of using a contaminated source for drinking water
supply will probably greatly exceed average water costs except in cases where
(1) the nature of the contamination is such that it is easily and cheaply
treatable, e.g., requires only the addition of a single unit process such as
air stripping; and/or (2) high quality drinking water sources in the area are
rare and thus costs for treated water in the area are relatively high. The
percentage of household income benchmark avoids these problems but may not be
realistic in that 1 percent of the mean national household income appears to
,.be a much larger amount than the average U.S. family is now paying for water,
based on a comparison of the distributions in Exhibit 4.
-------
-20-
Exhibit 4
COMPARISON OF HYPOTHETICAL REFERENCE
TECHNOLOGY COST TO WATER COST
AND HOUSEHOLD INCOME BENCHMARKS-
AN ILLUSTRATIVE TEST FOR CLASS III GROUNDWATER
60 T
Percentage so
of Cases
40 ••
30--
20- •
10
0
Avg. National Water
Cost Distribution
County Mean
Income Distribution
Zip Code Mean
Income Distribution
5/1000
2.
Gallons)
50
vyy
rss».~
5.
Cost of Treated Water Utilizing
Selected Reference Technology3
Water Costs or Percentage of
Household Income Available for
Water Costs 1>2
1 Avg. National Water Cost Distribution from AWWA
1981 National Survey. Zip Code & County Income
Distributions from 1980 Census Data.
2 National Water Costs in this Example are for Small
( <1 MGD) Systems.
3 Reference Technology (Treatment "Train") is Described
in Exhibit 3. Cost Per 1000 Gallons is $4.21.
-------
ESTIMATED RESULTS OF GROUND-WATER
CLASSIFICATION AT RCRA FACILITIES AND CERCLA SITES
This paper presents the results of an analysis of alternative substantial
population and economic tests for ground-water classification. The paper
summarizes the key concepts used in the analysis, the findings, and important
assumptions.
Available data from the Census Bureau and EPA sources (including the
Office of Drinking Water's Federal Reporting Data System) permit some
quantitative analysis of alternative definitions of substantial population,
irreplaceability, and treatability. Previous papers described empirical
evidence and simulated implications relating to alternative socio-economic
criteria for Class I ground waters. However, as described in the previous
issue paper "Class III: Development of an Economic Test," a more complex
methodology is required to evaluate alternative Class III tests.
Specifically, the Agency wanted to analyze the implications of alternative
Class I and Class III tests at RCRA facilities and CERCLA sites. The Class
III tests to be evaluated include:
(1) The current reference technology approach;
(2) An economic burden test; and
(3) A combination of the reference technology and economic
burden tests.
In addition to testing alternative economic burden levels (e.g., 1%, 1/2%
of household income), the Agency also wanted to compare two options identified
as ways to define potential users of Class III ground water. These two
options are: (1) existing populations within the vicinity of the ground
water, and (2) proxy hypothetical populations.1
A computer program was developed to simulate the classification of RCRA
facilities and CERCLA NPL sites under alternative Class I and Class III
tests. We assumed that the same substantial population threshold and
household income cut-off are used for both Class I and Class III. All other
things being equal, use of a relatively higher substantial population
threshold will produce fewer Class I designations and more Class III waters.
However, use of a higher income cutoff will produce both fewer Class I and
fewer Class III ground waters. Changes in both parameters, therefore, are
more easily evaluated with a computer model as described here. Using this
computer program, the Agency can compare estimates of the distribution of
ground-water classes at RCRA and CERCLA sites under the different policy
options.
1 These options are fully described in the attached issue paper, "Class
III: Development of an Economic Test."
-------
-2-
Although this analysis focuses on socio-economic criteria for classifying
ground water, the Agency has identified other criteria for Class I and Class
III designations. For example, a ground water that is ecologically vital and
highly vulnerable to contamination may be Class I. Similarly, in addition to
being highly vulnerable, a Class I ground water may be irreplaceable because
it serves a substantial population and, within a reasonable pipeline distance,
there is no alternative of comparable quantity and quality that is
economically feasible and available (i.e., not institutionally precluded).
Likewise, a Class III ground water must be contaminated by naturally occurring
substances or activities unrelated to a specific hazardous waste disposal
site, in addition to being not treatable for drinking water purposes using
treatment methods reasonably available to public water systems. Thus, because
of these additional criteria, the numbers developed here are not comprehensive
estimates of the total number of Class I and Class III ground waters.
This paper is divided into three sections. The first section defines key
concepts used in evaluating Classes I and III for this analysis. The second
section presents the results of the simulation model and analyzes the
implications for ground-water classification. The third section concludes
with an overview of the analytic methodology.
1. OVERVIEW OF KEY CONCEPTS
This section provides a brief overview of the key concepts involved in the
analysis.
Proximity Factor. The proximity factor represents the area of review
around a facility where the existence of a public water system is assumed to
imply that the ground water is currently used for drinking purposes.
Empirical observations suggest that in 95 percent of the cases, leachate
contaminants are not found beyond a 2 mile distance from the point of
release. For this reason, this analysis uses a 2 mile proximity factor in the
baseline.
Class I: Irreplaceable for Substantial Population. This test is used to
determine which ground waters fall into Class I. In order for a ground water
to be designated as Class I, two criteria must be met. First, the particular
ground water must be the source of drinking water for at least as many users
as required by the substantial population test. Second, the cost per
household of an alternative water source must exceed the fraction of income
that has been set for the economic test.
Class III: Reference Technology Test. The current draft guidance uses a
"reference technology" approach to determine whether ground water can be
cleaned up using "treatment methods reasonably available to public water
systems." Under the current reference technology approach, a ground water
would not be Class III if any one or more treatment methods considered to be
available in the EPA Region was capable of treating water to comply with
relevant federal standards and guidelines for drinking water. The result of
applying this test will differ somewhat by EPA Region because not all
technologies are considered available in every Region. Given the set of
available technologies, their removal efficiencies, and federal drinking water
-------
-3-
standards and guidelines, it would be theoretically possible to infer types of
contamination (i.e., pollutants and concentrations) which would qualify as
Class III in each Region. For the purposes of this analysis, we have done so
only for six contamination scenarios.
Class III: Economic Test - Existing Population. Under this test, one of
two criteria must be met for a ground water to be classified as Class III.
The first criterion is that the potential user population (i.e., the nearby
residents) that could make use of the ground water must be smaller than the
substantial population cut-off. Otherwise (i.e., if there is a substantial
potential user population), the cost of treating the water to meet drinking
standards and supplying it to the population must exceed some fraction of the
potential users' household income. (For the purposes of this analysis, we
assumed that the substantial population and household income thresholds would
be identical for both Class I and III.) Of note is that the required
treatment train need not be considered available in the same Region where the
ground water is located; rather, it is sufficient if the treatment train is
considered available in any Region of the country.
Class III: Economic Test - Proxy Population. This test is quite similar
to the one just described. The difference is that instead of analyzing the
existing population (i.e., nearby residents), a proxy population is used. The
size and income of a proxy population are important inputs to the analysis can
be defined in a number of different ways. To distinguish this test from the
previous test, a regional proxy is used: here, the proxy population is
assumed to have the mean size of a population served by a public water system
(PWS) in the EPA Region, and the average household income in the same Region.
Further, the proxy population is assumed to be substantial and therefore the
substantial population criterion is not tested. Thus, the only criterion for
Class III is whether the contaminated ground water can be treated for less
than some fraction of household income. If not, the ground water is
designated as Class III.
Class III: Combined Reference Technology and Economic Test2 -
Existing Population. For this test, the reference technology and economic
tests are combined. If the potential user population exceeds the substantial
population threshold, the required technology is available in the Region, and
the costs of applying it to the contaminated water do not exceed the specified
fraction of the area's household income, then the ground water is considered
Class II; otherwise, it will be classified as Class III.
Class III: Combined Reference Technology and Economic Test - Proxy
Population. This test is very similar to the combined test for an existing
population. The only difference is that the regional population size and
2 The Combined Reference Technology and Economic Test (existing or proxy
population) is not, strictly speaking, a combination of the Reference
Technology Test and Economic Burden Test, because it applies the economic test
to the least-cost regionally available technology, not the least-cost
treatment train available anywhere in the nation.
-------
-4-
income are substituted for the characteristics of the residents in the local
area. Further, it does not test the substantial population criterion.
2. SIMULATION RESULTS
Section 2.1 presents the results of the baseline simulation. The
implications of changing the basic socio-economic parameters of the
classification scheme (i.e., the substantial population threshold and income
burden factor) as well as certain environmental parameters (i.e., the
proximity factor and the probability of ground-water contamination) are also
analyzed.
2.1 Baseline Case
The baseline is characterized by:
• a 2,500 person substantial population threshold,
• a 1 percent income burden factor,
• a 2 mile proximity factor, and
• a 50% probability of ground-water contamination.3
2.1.1 RCRA Facilities. There are 4,330 facilities whose geographic
location was available in the database (latitude, longitude, and zip code
area), For 3,031 of these facilities, ground water in the area of review is
considered to be currently used for drinking purposes. In other words, 70% of
the RCRA facilities are Class I candidates while the remainder (i.e., 30%) are
Class III candidates. Out of the 4,330 RCRA facilities tested, approximately
6% (i.e., 239 sites) fall into Class III under the Reference Technology Test,
5% (or 196 sites) under the Economic Test for the existing population (3% or
108 sites, for the proxy population), and 7% (or 315 sites) under the Combined
Test for the existing population (6% or 239 sites for the proxy population).
Only 4 sites (.08%) are assigned to Class I in the baseline. These results
are presented in Exhibit 1A.
Of note is that the facilities that fail the Reference Technology Test
(and thus become Class III) are the same as those that fail the Combined Test
for the proxy population. This result suggests that in the baseline, any
facility that passes the Reference Technology Test also passes the Economic
Test for the proxy population using the least-cost treatment train available
in the EPA Region. In other words, whenever the ground water can be treated
using treatment techniques currently available in the EPA Region, then the
proxy population can always pay for treatment using the least costly among
these techniques. This is not true for the existing population, which may
3 The basel-ine assumes that half the ground waters at RCRA facilities
and CERCLA sites that are not currently used as sources of drinking water are
actually contaminated. This is approximately 15% of the total ground waters
included in the analysis. These ground waters are assumed to be contaminated
according to one of the six contamination scenarios defined in Section 3.3,
with equal probabilities.
-------
-5-
EXHIBIT 1A
GROUND-WATER CLASSIFICATION
AT RCRA FACILITIES, a/
BASELINE CASE b/
Class I c/d/ Class II Class III c/
e/ (//) 1%) e/ 0/1 (%) e/
Reference Technology Test 4 .08 4,087 94 239 6 f/
Economic Test, Existing Population 4 .08 4,130 95 196 5
Economic Test, Proxy Population 4 .08 4,218 97 108 3 g/
Combined Test, Existing Population 4 .08 4,011 93 315 7
Combined Test, Proxy Population 4 .08 4,063 94 239 6 f/
a/ Results based on 4,330 RCRA facilities.
b/ Substantial Population threshold = 2,500
Income Burden Factor = 1%
Probability of Ground-Water Contamination = 50%
Proximity Factor = 2 miles.
cj 70% of the ground waters within 2 miles of RCRA facilities are
currently used for drinking purposes by public water systems, while 30% of the
ground waters are not currently used for drinking purposes by public water
systems.
d/ Class I assignments are determined by the "irreplaceability to a
substantial population" criterion, regardless of the type of test used for
Class III assignment.
e/ The percentages for Classes I, II, and III may not sum to 100 due to
rounding.
f/ Any facility that passes the Reference Technology Test also passes the
Economic Test for proxy populations using the least-cost treatment in the EPA
Region.
g/ 2.5 percent (rounded up to 3) of the RCRA facilities are Class III
candidate sites and have ground-water contamination assumed to be not
treatable using water treatment techniques available nationwide. The proxy
population can afford to pay for treated water in all other contamination
scenarios. -
-------
-6-
not satisfy the substantial population criterion or for which the cost of
providing treated water sometimes exceeds its ability to pay. Indeed, in the
Combined Test for existing populations, 315 ground waters are assigned to
Class III, 76 more than in the Reference Technology Test (i.e., 32% more).
Note also that 2.5% of the ground waters (rounded up to 3% in Exhibit 1A)
are assigned to Class III in the economic test for proxy population. This
percentage corresponds to those Class III candidate ground waters which are
assumed to be so contaminated that they are not treatable even using water
treatment techniques available nationwide (i.e., of the 30% of the ground
waters not currently used for drinking purposes, 50% are simulated to have
ground-water contamination and, of the latter, one sixth are assumed to be not
technically treatable; that is, .025 = .3 x .5 x 1/6). This result suggests
that the proxy population can always assume the financial burden of cleaning
up the ground water for drinking purposes provided the ground water can
actually be treated using techniques currently available nationwide. As
before, this result does not hold for the existing population because the
latter may not be a substantial population or it may not have the financial
means to pay for treated water. Consequently, using the Economic Test for the
existing population leads to 5 percent of the sites (i.e., 196 sites) being
designated as Class III while use of the proxy population results in only 3
percent of the sites being classified as Class III (i.e., 108 sites). Hence,
approximately 2 percent of the ground waters at RCRA facilities (or 88 sites),
while treatable using nationally available techniques, are such that either
the existing potential user population is not substantial or it cannot afford
to pay for treated water.
2.1.2 CERCLA Sites. There are 216 CERCLA sites whose geographic
location was available in the database (latitude, longitude, and zip code
area). For 147 of these sites, the ground water in the area of review is
estimated to be currently used for drinking by public water systems. In other
words, 68% of the CERCLA sites are Class I candidates. Out of the 216 CERCLA
sites tested, 14 sites fall into Class III under the Reference Technology
Test, 12 under the Economic Test for the existing population (7 for the proxy
population), and 18 Class III sites under the Combined Test for the existing
population (14 for the proxy population). Only 1 site is assigned to Class I,
that is .55% of the total. These results are presented in Exhibit IB.
With respect to Class III, the results for CERCLA sites, in percentage
terms, are virtually identical to the results for RCRA facilities. The other
observations noted in Section 2.1.1 above apply equally to CERCLA sites.
The CERCLA Class I sites represent a proportion of the CERCLA Class I
candidate sites equal to .81%.u It is interesting to compare this ratio to
its equivalent for RCRA facilities where .11% of Class I candidate facilities
" While 68% of the total CERCLA sites are Class I candidates, only .55%
of the total CERCLA sites classify for Class I; that is a percentage of the
Class I candidate sites equal to .S5/.68 = .81%.
-------
-7-
EXHI-BIT IB
GROUND-WATER CLASSIFICATION
AT CERCLA SITES, a/
BASELINE CASE b/
Class I c/d/ Class II Class III c/
1£1 (%1 (#) (%) e/ (#) (%) e/
Reference Technology Test 1 .55 201 93 14 6 f/
Economic Test, Existing Population 1 .55 203 94 12 6
Economic Test, Proxy Population 1 .55 209 96 6 3 g/
Combined Test, Existing Population 1 .55 197 91 18 8
Combined Test, Proxy Population 1 .55 201 93 14 6 f/
a/ Results based on 216 CERCLA sites.
b/ Substantial Population threshold = 2,500
Income Burden Factor = 1%
Probability of Ground-Water Contamination = 50%
Proximity Factor = 2 miles.
c/ 68% of the ground waters within 2 miles of CERCLA sites are currently
used for drinking purposes by public water systems, while 32% of the ground
waters are not currently used for drinking purposes by public water systems.
d/ Class I assignments are determined by the "irreplaceability to a
substantial population" criterion, regardless of the type of test used for
Class III assignment.
e/ The sum of percentages for Classes I, II, and III may not sum to 100
due to rounding.
f/ Any site that passes the Reference Technology Test also passes the
Economic Test for proxy populations using the least-cost treatment in the EPA
Region.
£/ 2.7 percent (rounded up to 3) of the CERCLA sites are Class III
candidate sites and have ground-water contamination assumed to be not
treatable using water treatment techniques available nationwide. The proxy
population can afford to pay for treated water in all other contamination
scenarios.
-------
are finally assigned to Class I in the baseline.5 This discrepancy between
the two ratios may have two origins which are related to the size and average
household income of the populations currently using the ground water in the
area of review.
• On average, a ground water at a Class I candidate RCRA
facility and one at a Class I candidate CERCLA site have
slightly different chances of being currently used by a
substantial population.6 The size of the user
population hence does not entirely explain why there are
more Class I CERCLA sites relative to the number of
Class I candidate sites than there are Class I RCRA
facilities relative to the Class I candidate facilities.
• The remainder of the explanation is to be found in the
average household income of the current user
population. The data show that a user population around
a CERCLA site will fail the economic test more
frequently than a user population around a RCRA
facility. This finding indicates that CERCLA sites are
likely to be located in sparsely-populated low income
areas of the country, although some of these areas might
still be inhabited by a substantial population (e.g.,
2,500 persons).
2.2 Effects of a Change in the Substantial Population Threshold
If the substantial population threshold is set at a higher level, fewer
populations will be considered substantial. As a result, there will be fewer
Class I assignments and more Class III (for existing population tests). If
the ground water is irreplaceable, but not currently used by a substantial
population, then it cannot be Class I. For Class III, if the "existing"
potential user population is not considered substantial, then the ground water
cannot be considered for Class II and must be Class III instead. In sum, when
the substantial population threshold is increased, the number of Class I
ground waters goes down while the number of Class III ground waters increases
(for existing population tests). The effect on Class II assignments depends
on the relative number of transfers between Classes I and II and between
Classes II and III. These relationships are confirmed by the results shown in
Exhibits 2A and 2B.
5 While 70% of the total RCRA sites are Class I candidates, only .08%
of the total RC-RA sites classify for Class I; that is a percentage of the
Class I candidate sites equal to .08/.70 = .11%.
6 The data show that at 81% of Class I candidate RCRA facilities and 77%
of Class I candidate CERCLA sites the ground water in the area of review is
currently used by a substantial population (at least 2,500 persons).
-------
-9-
EXHIBIT 2A
GROUND-WATER CLASSIFICATION
AT RCRA FACILITIES, a/
EFFECT OF THE SUBSTANTIAL POPULATION THRESHOLD b/
Class I c/ Class II Class III c/
(HI !%1 d/ (#) (%) d/ (#) (%) d/
Substantial population = 1,000
Reference Technology Test 4 g/ . 10 4,087 94 239 6 e/
Economic Test, Existing Population 4 .10 4,156 96 170 4
Economic Test, Proxy Population 4 .10 4,218 97 108 3 f/
Combined Test, Existing Population 4 .10 4,030 93 296 7
Combined Test, Proxy Population 4 .10 4,087 94 239 6 f/
Substantial Population = 2,500
Reference Technology Test 4 g_/.08 4,087 94 239 6 e/
Economic Test, Existing Population 4 .08 4,130 95 196 5
Economic Test, Proxy Population 4 .08 4,218 97 108 3 f/
Combined Test, Existing Population 4 .08 4,011 93 315 7
Combined Test, Proxy Population 4 .08 4,087 94 239 6 f/
Substantial Population = 5,000
Reference Technology Test 3 .07 4,088 94 239 6 e/
Economic Test, Existing Population 3 .07 4,102 95 225 5
Economic Test, Proxy Population 3 .07 4,219 97 108 3 f/
Combined Test, Existing Population 3 .07 3,991 92 336 8
Combined Test, Proxy Population 3 .07 4,088 94 239 6 f/
a/ Results based on 4,330 RCRA facilities.
b/ Income Burden Factor = 1%
Probability of Ground-Water Contamination = 50%
Proximity Factor = 2 miles.
c/ 70% of the ground waters within 2 miles of RCRA facilities are
currently used for drinking purposes by public water systems, while 30% of the
ground waters are not currently used for drinking purposes by public water
systems.
d/ The percentages for Classes I, II, and III may not sum to 100 due to
rounding.
e/ The Reference Technology Test does not use the substantial population
criterion.
f/ The proxy population is not subject to the substantial population
criterion.
g/ The same absolute numbers of Class I assignments correspond to
different percentages due to rounding.
-------
-10-
EXHIBIT 2B
GROUND-WATER CLASSIFICATION
AT CERCLA SITES, a/
EFFECT OF THE SUBSTANTIAL POPULATION THRESHOLD b/
Class I
£/
d/
Class II Class III c/
(#) (%) d/ (#) (%) d/
Substantial population = 1,000
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
Substantial Population = 2,500
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
Substantial Population = 5,000
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
1 S/.59
1 .59
1 .59
1 .59
1 .59
.55
.55
.55
.55
.55
1 S/.46
1 .46
1 .46
1 .46
1 .46
201
204
209
198
201
201
203
209
197
201
201
201
209
196
201
93
94
96
91
93
93
94
96
91
93
93
93
97
91
93
14
11
6
17
14
14
12
6
18
14
14
14
6
19
14
6 e/
5
3 f /
8
6 f/
6 e/
6
3 f/
8
6 f/
6 e/
6
3 f/
9
6 f/
a/ Results based on 216 CERCLA sites.
b/ Income Burden Factor = 1%
Probability of Ground-Water Contamination = 50%
Proximity Factor = 2 miles.
c/ 68% of the ground waters within 2 miles of CERCLA sites are currently
used for drinking purposes by public water systems, while 32% of the ground
waters are not currently used for drinking purposes by public water systems.
d/ The percentages for Classes I, II, and III may not sum to 100 due to
rounding.
e/ The Reference Technology Test does not use the substantial population
criterion. :
f/ The proxy population is not subject to the substantial population
criterion.
g/ The same absolute numbers of Class I assignments correspond to
different percentages due to rounding.
-------
-11-
2.2.1 RCRA Facilities. When the substantial population threshold is
increased from 1,000 to 5,000 persons, the number of Class I assignments goes
down from four sites to three (Exhibit 2A). At the same time, the number of
Class III ground waters goes up from 170 sites to 225 sites (i.e., a 32%
increase) for the Economic Test (for existing populations) and from 296 to 336
(14% increase) for the Combined Test. The effect of a change in the
substantial population threshold is less acute when a Combined Test is used
because of the screening effect of the Reference Technology Test.
2.2.2 CERCLA Sites. As shown on Exhibit 2B, the results for CERCLA
sites are comparable to the results for RCRA facilities. Note that while the
percentage of Class I ground waters declines, the absolute number is always
shown as being equal to one site. This happens because of rounding.
2.3 Effects of a Change in the Probability of Ground-Water Contamination
In this analysis ground-water contamination is relevant only for Class III
candidate sites; that is, when the ground water is not currently used for
drinking purposes by public water systems. The probability of ground-water
contamination thus affects the ultimate classification of Class III candidate
sites but does not change Class I assignments.
Exhibits 3A and 3B present the outcome of the classification analysis for
three different values of the probability of ground-water contamination, while
the other parameters (substantial population threshold, income burden factor,
and proximity factor) are set at their baseline value.
2.3.1 RCRA Facilities. Exhibit 3A confirms that very few facilities (4)
will fall into Class I independently of the extent of ground-water
contamination. The results of the baseline case are also confirmed on the
Class III side. Whether the probability of ground-water contamination is
equal to .5, .25, or .1, the two following propositions hold:
1) The Reference Technology Test and the Combined Test for
proxy population yield the same classification outcome,
and
2) The Economic Test for proxy population assigns to Class
III only those sites where the ground water is
contaminated at such a level that it is not treatable
even using water treatment techniques available in the
nation.
It is also noteworthy that the ratio of Class III assignments relative to
the number of ground waters that are considered contaminated does not depend
on the probability of ground-water contamination. Consider the Reference
Technology Test- for example. For a 50% chance of ground-water contamination,
239 facilities fall into Class III, out of 650 Class III candidate sites where
-------
-12-
EXHIBIT 3A
GROUND-WATER CLASSIFICATION
AT RCRA FACILITIES, a/
EFFECT OF THE PROBABILITY OF CONTAMINATION b/
Probability of Contamination = 50%
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
Probability of Contamination = 25%
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
Probability of Contamination = 10%
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
4
4
4
4
4
.08
.08
.08
.08
.08
4,087
4,130
4,218
4,011
4,087
94
95
97
93
94
239
196
108
315
239
6
5
3
7
6
Class
(#)
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
I
(%)
.08
.08
.08
.08
.08
.08
.08
.08
.08
.08
.08
.08
.08
.08
.08
c/ Class II
(%) d/
Class III c/
d/
4
4
4
4
4
.08
.08
.08
.08
.08
4,193
4,228
4,273
4,158
4,193
97
98
99
96
97
133
98
53
168
133
3
2
1
4
3
4
4
4
4
4
.08
.08
.08
.08
.08
4,277
4,287
4,306
4,259
4,277
99
99
99
98
99
49
39
20
67
49
1 e
1
.46
2
1 e
/ JL \
id
239
196
108
315
239
133
98
53
168
133
49
39
20
67
49
<
6 e,
5
3
7
6 e,
3 e,
2
1
4
3 e,
1 e,
1
.46
2
1 e
a/ Results based on 4,330 RCRA facilities.
b/ Substantial Population = 2,500
Income Burden Factor = 1%.
Proximity Factor = 2 miles.
c/ 70% of the ground waters within 2 miles of RCRA facilities are
currently used for drinking purposes by public water systems, or 30% of the
ground waters are not currently used for drinking purposes by public water
systems .
d/ The percentages for Classes I, II, and III may not sum to 100 due to
rounding.
e/ The Reference Technology Test and the Combined Test for proxy
populations always yield the same classification outcomes.
-------
-13-
EXHIBIT 3B
GROUND-WATER CLASSIFICATION
AT CERCLA SITES, a/
EFFECT OF THE PROBABILITY OF CONTAMINATION b/
Class I c/ Class II Class III c/
I%I d/ (#) (%) d/ (//) (%) d/
Probability of Contamination = 50%
Reference Technology Test 1 .55 201 93 14 6 e/
Economic Test, Existing Population 1 .55 203 94 12 6
Economic Test, Proxy Population 1 .55 209 96 63
Combined Test, Existing Population 1 .55 197 91 18 8
Combined Test, Proxy Population 1 .55 201 93 14 6 e/
Probability of Contamination = 25%
Reference Technology Test 1 .55 207 96 8 4 e/
Economic Test, Existing Population 1 .55 208 97 63
Economic Test, Proxy Population 1 .55 211 98 42
Combined Test, Existing Population 1 .55 205 95 10 5
Combined Test, Proxy Population 1 .55 207 96 8 4 e/
Probability of Contamination = 10%
Reference Technology Test 1 .55 213 98 2 1 e/
Economic Test, Existing Population 1 .55 213 99 21
Economic Test, Proxy Population 1 .55 214 99 1 0.25
Combined Test, Existing Population 1 .55 212 98 32
Combined Test, Proxy Population 1 .55 213 98 2 1 e/
a/ Results based on 216 CERCLA sites.
b/ Substantial Population = 2,500
Income Burden Factor = 1%.
Proximity Factor = 2 miles.
c/ 68% of the ground waters within 2 miles of CERCLA sites are currently
used for drinking purposes by public water systems, or 32% of the ground
waters are not currently used for drinking purposes by public water systems.
d/ Class I assignments are determined by the "irreplaceability to a
substantial population" criterion, regardless of the type of test used for
Class III assignment.
e/ The Reference Technology Test and the Combined Test for proxy
populations always yield the same classification outcomes.
-------
-14-
the ground water is assumed contaminated.8 The ratio of Class III ground
waters to contaminated, not currently used ground waters is then 239/650 (37%)
for a .50 probability of ground-water contamination. Similarly, this ratio is
equal to 133/325 (41%) for a .25 probability of ground-water contamination and
49/130 (38%) for a .10 probability of contamination. These results suggest
that, given the values of the three other parameters, a contaminated, not
currently used ground water within 2 miles of a RCRA facility has
approximately a 40% chance of classifying as Class III under the Reference
Technology Test. Following the same reasoning, a contaminated ground water
that is not currently used has a 30% chance of falling into Class III under
the Economic Test for existing populations (16% under the Economic Test for
proxy population), and a 50% chance of classifying as Class III under the
Combined Test for existing populations (40% for proxy population).
2.3.2 CERCLA Sites. The same results generally apply to the
classification of CERCLA sites, as shown in Exhibit 3B. Tests involving
existing populations -- instead of proxy populations -- produce more Class III
assignments because these potential water users may fail the substantial
population criterion or the economic burden test.
2.4 Effects of a Change in the Income Burden Factor
If the income burden factor is set at a higher percentage level, more
money is assumed to be available for either replacement or treatment. As a
result, more Class I candidate sites and more Class III candidate sites will
end up in Class II (for cost-related tests). In sum, when the income burden
factor goes up, there are more Class II assignments and less Class I and III.
2.4.1 RCRA Facilities. The data in Exhibit 4A show that for all
cost-related tests, the number of Class I and III ground waters increases when
the income burden factor is decreased. For a 1/10 percent economic burden
factor, 38 percent of the ground waters (i.e., 1,650 sites) under RCRA sites
are assigned to Class I while at the other economic burden levels (1 percent
and 1/2 percent), very few sites fall into Class I (4 and 7 sites,
respectively).
The percentage of sites that fall into Class I should be compared to the
proportion of sites that are Class I candidates, that is those sites where the
ground water is currently used for drinking purposes by public water systems.
According to this analysis, 70% of the RCRA facilities fall into this category
(i.e., 3,031 sites out of a total of 4,330 sites are Class I candidates).
While only 4 of these sites are assigned to Class I for a 1 percent of
household income burden factor, 1,650 fall into Class I for a 1/10 percent
factor (i.e., approximately half the Class I candidate sites). Exhibit 5A
'displays the percentage of Class I assignments as a function of the income
burden factor. It shows that the number of sites falling into Class I is
8 30% of the ground waters (1,299 facilities) are not currently used for
drinking purposes by public water systems. Of these 1,299 Class III candidate
facilities, 50% are assumed contaminated, which corresponds to 650 sites.
-------
-15-
EXHIBIT 4A
GROUND-WATER CLASSIFICATION
AT RCRA FACILITIES, a/
EFFECT OF THE INCOME BURDEN FACTOR b/
Class I c/
Income Burden Factor = 1%
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
Income Burden Factor = .5%
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
Income Burden Factor = . 1%
Reference Technology Test
Economic Test, Existing Population
Economic Test, Proxy Population
Combined Test, Existing Population
Combined Test, Proxy Population
Class II Class III c/
(//) (%) d/ (#) (%) d/
4
4
4
4
4
.08
.08
.08
.08
.08
4,087
4,130
4,218
4,011
4,087
94
95
97
93
94
7
7
7
7
7
.15
.15
.15
.15
.15
4,084
4,076
4,215
3,949
4,035
94
94
97
91
93
1,650
1,650
1,650
1,650
1,650
38 f/
38
38
38
38
2,441
2,030
2,030
2,030
2,030
57
47
47
47
47
239
196
108
315
239
239
650
650
650
650
6
5
3
7
6
239 6
247 6
108 3
374 9
288 7
6
15 e/
15 e/
15 e/
15 e/
a/ Results based on 4,330 RCRA facilities.
b/ Substantial Population Threshold = 2,500
Probability of Ground-Water Contamination = 50%
Proximity Factor = 2 miles.
c/ 70% of the ground waters within 2 miles of RCRA facilities are
currently used for drinking purposes by public water systems, or 30% of the
ground waters are not currently used for drinking purposes by public water
systems .
d/ The percentages for Classes I, II, and III may not sum to 100 due to
rounding.
e/ Contaminated ground water that is not currently used for drinking is
not economically treatable.
f/ Approximately half of the Class I candidates fall into Class I.
-------
-16-
EXHIBIT 4B
GROUND-WATER CLASSIFICATION
AT CERCLA SITES, a/
EFFECT OF THE INCOME BURDEN FACTOR b/
Class I c/ Class II Class III c/
(%) d/ (#) (%) d/ (#) (%) d/
Income Burden Factor = 1%
Reference Technology Test 1 .55 201 93 14 6
Economic Test, Existing Population 1 .55 203 94 12 6
Economic Test, Proxy Population 1 .55 209 96 63
Combined Test, Existing Population 1 .55 197 91 18 8
Combined Test, Proxy Population 1 .55 201 93 13 6
Income Burden Factor = .5%
Reference Technology Test 1 .55 201 93 14 6
Economic Test, Existing Population 1 .55 200 92 15 7
Economic Test, Proxy Population 1 .55 209 96 63
Combined Test, Existing Population 1 .55 194 90 21 10
Combined Test, Proxy Population 1 .55 200 93 15 7
Income Burden Factor = . 1%
Reference Technology Test 82 38 f/ 120 56 14 6
Economic Test, Existing Population 82 38 99 46 35 16 e/
Economic Test, Proxy Population 82 38 99 46 35 16 e/
Combined Test, Existing Population 82 38 99 46 35 16 e/
Combined Test, Proxy Population 82 38 99 46 35 16 e/
a/ Results based on 216 CERCLA sites.
b/ Substantial Population Threshold = 2,500
Probability of Ground-Water Contamination = 50%
Proximity Factor = 2 miles.
c/ 68% of the ground waters within 2 miles of CERCLA sites are currently
used for drinking purposes by public water systems, or 32% of the ground
waters are not currently used for drinking purposes by public water systems.
d/ The percentages for Classes I, II, and III may not sum to 100 due to
rounding.
e/ Contaminated ground water that is not currently used for drinking is
not economically treatable.
f/ Approximately half of the Class I candidates fall into Class I.
-------
-17-
o
o
CM'
o
00
2
UJ
Q
D _
CQ ift
UJ ii
SB*
8 Is
2 cj ^
i*l
°2l
(A
JS
U
o
— o
Z
UJ
a
oe
3
03
UJ
Z
O
u
Z
u.
o
.0
• fM
£
u
OC
UJ
a.
til H (J
2 2 > tt
U < t -
S 2
S S 3
O D <
D5^g
C/D O
^" CC
« ^
c/>
c^
-J
U
1 1
3 0
1
O
1
O
1
o
1
o
c
o
o
M
M
*2
O H
H 35
z -
UJ
U tf)
5 «5
UJ <
*•£
-------
-18-
extremely sensitive to the income burden factor for values less than .2
percent. Exhibit 5A further suggests that at 13% of the RCRA facilities, the
ground water is currently used by a population whose size does not exceed the
substantial population threshold (2,500 persons). Thus, while 70% of the
sites 'are Class I candidates, only 57% of RCRA facilities are assigned to
Class I for a 0% income burden factor.3
Similarly, Exhibit 6A presents, for each of the five tests, the percentage
of Class III assignments as a function of the income burden factor. Exhibit
6A shows that each of the four cost-related tests results in a 15 percent
Class III assignment for values of the income burden factor below .10%. This
percentage-corresponds exactly to those ground waters that are not currently
used for drinking by public water systems and are assumed contaminated.10
This result means that for such a low income burden factor (^ .10%), all
contaminated ground waters are not economically treatable, irrespectively of
whether an "existing" or "proxy" population is assumed to incur these costs,
or whether the least-cost treatment train available in the EPA Region or in
the nation is applied.
The number of Class III assignments varies less with the income burden
factor than does the number of Class I sites (the curves in Exhibit 6A are
less steep than in Exhibit 5A). Furthermore, ranges of values of the income
burden factor where this variation is observed do not coincide: when the
income burden factor is lowered below .2 percent, the number of Class I
assignments increases abruptly while the number of Class III waters is at a
ceiling value. On the other hand, both Class I and Class III assignments are
unchanged when the income burden factor rises above 1 percent. The difference
is that while the number of Class I sites ultimately drops to 0, the number of
Class III assignments levels out at a finite value (the number of Class III
waters cannot be lower than the number of ground waters which are not
currently used and which are too contaminated to be treatable).
2.4.2 CERCLA Sites. The same types of results are observed for CERCLA
sites as for RCRA facilities. In particular, the ranges of values of the
income burden factor where the Class I or III assignments change drastically
coincide with those for RCRA facilities. See Exhibits 5B and 6B.
For about 16% of all CERCLA sites (compared to 13% for RCRA facilities)
the ground water'in the review area is currently used by a population whose
size does not exceed the substantial population threshold (2,500
9 The hypothetical 0% income burden factor simulates a fictional
scenario where Jiot a single user population can afford to pay for replacement,
regardless of its size. Any ground water that is currently used by a
substantial population then becomes a Class I ground water.
10 30% of the sites are Class III candidate sites; of these, 50% are
assumed contaminated.
-------
-19-
Ifl
V
+*
1/5
2
U
Q
QC
3 _
CQ in
U
§ o 2
^ o o
u § c
5 « %
MS
+ r t
(0
U
CQ
-O UJ
2 o-
x s,
uu
CQ
2
U
S
2
O
c/5
CD
0. o (j
s^s
125
2
LU
a
cc
D
CO
UJ
2
O
U
2
u.
O
H
2
UJ
U
K
UJ
a.
o
&
D-
CD
<
J
U
1
1 I 1
o o o
*O: ift rt
U.
0
H
2
UJ
O
DC
U
0.
1 1
0 0
M N
C/3
UJ
H
00
V)
uo
<
u
o
0
1 1 d
0 0
1-H
------- An error occurred while trying to OCR this image.
-------
-21-
z
UJ
Q
CC
D
CQ
UJ
S
O
U
z
u*
O
H
Z
Su
^ K
+J flj
— ***
.•£ ^
2 >,
UJ ffl
C/3
H
Z
UJ
£
Z
O
en
U
1/5
o 2
o O
35
Z^
o*~
« < en
H H UJ
< Z H
O en
u
u
D
a, - ^
Ps2
o u
£*
O (j
AL
ANT
TY
(
H J
0
v u
CC UJ
1
1
k
1
1
J
-«»^1
» y
r
I
, 1.
! 1
00 ^0
e
o
•**
(0
g
Q.
0
CL
>i
X
h«
CL
v.
U
i
o
c
o
u
UJ
1
1
1
1
J*
/^
\
*t
c
0
" c
(0 O
a «
«? 3
&• a
?^°
•••
** ^>
(0 X
— o
X u
UJ Q.
^ \
T5 -0
V &
C C
_O ft
•* J0
E E
0 0
U U
/
1 /
1 /
/
/
/
^' //
/»'<•*
gfe^— -
^^^
1 1
M O C
i
i
i
i
!
i
1
•
i
/i
;
•
9
i
1
//
//
.*•'
m " ^^^
\
o
' *
»
»
\
.'
\
o
1
!
s
/
s
\
rf
1
1
1
1
1
1
1
1
1
!
1
i
1
M C
O
.— O
CM'
o
_ 00
1^
0
— *o
^H
o
_ ^
^^
o
— 04
i^
O
— o
>^
o
— 00
o
o
— v£
6
o
— rt
O
O
- 04
O
o
6
>
Z
UJ
G
CC
D
CQ
UJ
o
U
Z
u.
O
H
UJ
O
CC
UJ
a.
UJ
H
z s
UJ -
W W3
Pi «5
U
O
-------
-22-
persons).11 Put differently, when a ground water at a CERCLA site is
currently used for drinking, there is a 77°m chance that it is used by a
substantial population.
On the Class III side, at sixteen percent of all CERCLA sites the ground
water in the area of review is not used for drinking and is considered
contaminated.12 (See Exhibit 6B, income burden factor = 0%). For low
values of the income burden factor (< .10%), these ground waters fall into
Class III under all cost-related tests.
2.5 Effect of a Change in the Proximity Factor
A key classification criterion is whether the ground water in the area of
review is currently used for drinking purposes, in which case the site is a
Class I candidate. Otherwise, the site is a Class III candidate. If a larger
area of review is considered, there will be more Class I candidate sites (and,
conversely, less Class III candidate sites); there will subsequently be more
Class I sites and less Class III sites in the final classification. For RCRA
facilities and CERCLA sites alike, the percentage of Class I candidate for a 3
miles proximity factor is approximately 84% of the total, as compared to 70%
for 2 miles. As a result, the percentage of Class I waters using the 3 mile
distance would be expected to increase by 7/77 = 970, while the percentage of
Class III waters would go down by 14/30 = 50%.
2.5.1 RCRA Facilities. Exhibit 7A shows the number of RCRA Class I and
Class III candidate sites for different values of the proximity factor ranging
from 1 to 6 miles. The analysis assumes that a change of one minute in
latitude or longitude is equal to a one mile change. Although 43% of the RCRA
facilities have a public water system (PWS) within a one mile distance, as
many as 95% of the sites have at least one PWS within a 6 mile distance.
2.5.2 CERCLA Sites. The results are generally similar for CERCLA sites
except in very close proximity to the sites (Exhibit 7B). There is less often
a PWS within one mile of a CERCLA site (32%) than within one mile of a RCRA
facility (43%).
3. GROUNDWATER CLASSIFICATION SIMULATION MODEL
The results presented in this paper are based on a computer model which
combines stochastic and deterministic methods to estimate relative frequencies
of Class I, II, and III ground-water assignments. The universe of sites which
is evaluated includes all active RCRA treatment, storage, and disposal
facilities (data are from the Hazardous Waste Data Management System (HWDMS)),
and all CERCLA facilities on the National Priority List (data are from Mitre's
11 68% of the CERCLA sites are Class I candidates while only 52% of the
total sites end up in Class I for a 0% income burden factor (see Exhibit 5B).
12 32% of CERCLA ground waters are not currently used, of these, 50% are
contaminated.
-------
-23-
EXHIBIT 7A
EFFECT OF THE PROXIMITY FACTOR
ON THE NUMBER OF CLASS I AND CLASS III CANDIDATES,
RCRA FACILITIES a/
Class I Candidates Class III Candidates
111 (#) (%
Proximity Factor (miles) b/
1 1,862 43 2,468 57
2 3,074 70 1,256 30
3 3,637 84 693 16
4 3,940 91 390 9
5 4,070 94 260 6
6 4,113 95 217 5
a/ Results based on 4,330 RCRA facilities.
b/ A change of one minute in latitude or longitude is assumed to be equal
to a one mile distance.
-------
-24-
EXHIBIT 7B
EFFECT OF THE PROXIMITY FACTOR
ON THE NUMBER OF CLASS I AND CLASS III CANDIDATES,
CERCLA SITES a/
Class I Candidates Class III Candidates
(#)
Proximity Factor (miles) b/
1 69 32 147 68
2 147 68 69 32
3 179 83 37 17
4 192 89 24 11
5 201 93 15 7
6 205 95 11 5
a/ Results based on 216 CERCLA sites.
b/ A change of one minute in latitude or longitude is assumed to be equal
to a one mile distance.
-------
-25-
Hazard Ranking System (HRS) database) whose location is specified in the data'
base. The results of the model consist of estimates of the proportions of
these facilities falling into Class I, Class II, and Class III. The data
inputs are supplied to the model by the user and can be easily changed for
further policy analyses.
The model analyzes one facility at a time and, based on a series of tests,
assigns it to one of the three ground-water classes. Section 3.1 outlines the
sequence of tests used to classify each site. Section 3.2 describes in detail
how the program determines the outcome of these tests. Section 3.3 concludes
with an overview of the contamination scenarios that were used in evaluating
Class III.
3.1 Outline of the Program Logic
Because the model analyzes each facility based on several criteria, its
operation can best be described by a series of questions that are asked about
each facility. For each facility, the program inquires about the following:
1) Is the ground water in the area of review currently used
for drinking purposes?
If it is (i.e., it is potentially Class I)
Yl) What is the size and average household income
of the user population?
Y2) Is the user population a substantial population?
Y3) If so, can the user population pay for
replacement?
If it is not (i.e., it is potentially Class III)
Nl) Is the ground water assumed to be contaminated
and, if so, which contamination scenario is
applied?
N2) Can the water be treated using treatment
techniques currently available in the EPA
Region?
N3) What is the size and average household income
of the "existing" potential user population
(or, depending on the policy being analyzed,
: the proxy potential user population)?
N4) Is the "existing" potential user population a
substantial population?
-------
-26-
N5) Can the existing (or, alternatively, proxy)
population pay for treatment using the
least-cost treatment train available in the
nation?
N6) Can the existing (or, alternatively, proxy)
population pay for treatment using the least-
cost treatment train available in the EPA
Region?
The flow chart sketched in Exhibit 8 presents an overview of the model's
operations and indicates the interrelations among the model's analytic
components.
3.2 Detailed Description of the Tests
This section describes in some detail how the program determines the
answers to the questions raised in the previous section.
3.2.1 User Population Test
For each RCRA/CERCLA facility, the presence or absence of a user
population within the area of review is determined (Question 1). If a user
population is not present, the "Class III routine" is called. This routine
executes the Reference Technology Test, the Economic Test for an existing and
proxy population, and the Combined Economic Reference Technology Test for an
existing and proxy population, and assigns the facility to either Class II or
III for each test. If a user population is present, the "Class I routine" is
called. This routine tests for a substantial population and, if a substantial
population exists, for the replaceability of the ground water. The "Class I
tests" in this routine result in either a Class I or Class II assignment. The
next two sections describe the Class I and Class III routines in more detail.
3.2.2 Class I Routine
After the existence of a user population is determined, the total size and
average household income of the user population is estimated (Question Yl).
The number of public water system users is estimated from available data. The
relevant private well user population is estimated using a two step
procedure. First, the number of housing units per county using private ground
water sources is converted to the number of persons using such sources
(assuming an average of 2.75 persons per housing unit). Second, a percentage
of these private ground-water users, where the percentage equals the review
area divided by the county area, is assumed to be equal to the number of
private well users in the review area. The total Class I "existing"
population equals the sum of the private well user population plus the public
water system user population. The average household income is estimated based
on the mean household income in the zip code area.
Substantial Population Test. The substantial population test compares
the estimated user population against the substantial population threshold
(Question Y2) defined in the model input. If the population exceeds the
-------
-27-
Exhibit 8A
GROUND-WATER CLASSIFICATION FLOW CHART
Is
roun
Water
Currently
Used For
Drinkin
Call
Class III"
Routine
Call
Class I '
Routine
PWS
Population
Is the
Ground
Water
Contam-
inated?
Private Well
Population
Existing
Population
Existing
Population
Substantia
Water
Supply
Irreplaceable
Yes
See Exhibit 8B
Class II
-------
-28-
Exhibit 8B
'Class III" Routine
asses
Reference
Technology
Test
]
r
Reference
Technology
Test
\
Economic
Burden
Test
\
f
Economic
Reference
Test
Yes
Class II
Yes
Class III
asses
Reference
Technology
Test
Potential
User
Population
Exists
See Exhibit 8C
otentia
User
Population
Exists
Class III
No
See Exhibit 8C
-------
-29-
Exhibit 8C
"Class III" Economic Burden Test
Yes
Ground- water
Class under:
Existing Population Test II
Proxy Population Test II
Potential
User Population
Exists ?
Existing
Population
Passes Test?
Proxy
Population
Passes Test?
Proxy
Population
Passes Test?
Proxy
Population
Passes Test?
II
III
III
II
III
III
III
II
III
III
-------
-30-
threshold, the replaceability of the ground water is tested. If the user
population is less than the substantial population cutoff, the site is
assigned to Class II.
Replaceability Test. If a substantial population exists, the
ground-water replaceability routine determines whether or not the user
population can pay for replacement, A major input to the routine is the cost
of water, which varies according to the user population size and extent of
replacement. The model uses national average costs of delivered water, for
four population categories: less than 1,000 persons, between 1,000 and 3,300,
between 3,300 and 10,000, and more than 10,000. These costs are assumed to be
comprised of three components: acquisition costs (20%); treatment costs
(30%); and distribution and transmission costs (50%). One hundred percent
(all three components) of the cost of water is used in determining private
well water replacement; fifty percent (acquisition and treatment) of the cost
is used in estimating replacement costs for public water systems. These data
are described in more detail in the attached issue paper, "Costs of Water
Supply Systems."
The ability to pay for treated water by households (Question Y3) is
represented by multiplying the average household income by a household income
burden factor. (In the baseline, this factor is 1%.) If the water
replacement costs are less than or equal to the available financial resources,
the water is considered replaceable, and the site is classified as Class II.
Otherwise, the water is determined to be irreplaceable, and the site becomes a
Class I site.
3.2.3 Class III Routine
The Class III routine tests each facility that was determined to have no
current user population. The model probabilistically assigns a ground water
contamination scenario (Question Nl). These scenarios describe the condition
of the ground water surrounding the facility. In the bas,eline, 50 percent of
the facilities are assumed to have no ground water contamination. The
remaining 50 percent are modelled as if ground-water contamination is
present. For these contaminated ground waters, the model selects one of six
equally probable contamination scenarios. These scenarios are described in
detail in Section 3.3.
The model next characterizes the "existing" and "proxy" potential user
populations (Question N3). The "existing" potential public water system (PWS)
user population is based on the total number of public water system users
served by systems located within 5 miles of the ground-water being
classified. The "existing" private well user population is computed in the
same way as the Class I private well user population (i.e., by computing the
fraction of the county area represented by a single area of review and
multiplying ths.fraction by the total number of private well users in the
county). The average household income in the zip code area is assumed to
represent that of the "existing" population.
The "proxy" population test ignores private well users and estimates the
size of the proxy population as the mean of the populations served by public
-------
-31-
water systems in the EPA Region. Furthermore, the "proxy" test uses the
average household income in the EPA Region for determining affordability.
The program then executes the five following tests:
1) Reference Technology Test
2) Economic Test, with:
existing population
proxy population
3) Reference Technology and Economic Test, with:
existing population
proxy population
Reference Technology Test. The results of the reference technology test
for each of the six contamination scenarios is input to the program. If a
treatment train suitable for the particular contaminant mix is available in
the EPA Region, the facility is assigned to Class II; otherwise, the facility
is considered Class III (Question N2).
Economic Burden Test. Under the economic burden test, the program first
tests whether the existing population passes the substantial population
criterion (Question N4). If the existing population is not considered
substantial, the site automatically becomes a Class III. The substantial
population criterion is not tested against the proxy population (i.e., all
proxy populations are assumed to be substantial).
The program then tests whether the existing population (or, alternatively,
proxy) can pay for treatment using the least-cost treatment train available in
the nation and capable of treating the water to drinking water standards
(Question N5). The costs of treatment, which depend on the ground-water
contamination scenario and the size of the service population, were developed
using engineering cost techniques (see Section 3.3). In addition to the costs
of treatment, the model includes acquisition costs (for the total population)
and delivery costs (for private well populations). Acquisition costs are
estimated as 20% of the national average water cost, while delivery costs
represent 50% of the national average cost. A population is assumed to be
able to pay for treatment (i.e., the water is considered treatable and thus
Class II) if the cost of providing treated water to the average household does
not exceed a specified percentage of average household income.
Reference Technology and Economic Test. Under this option, a site is
assigned to Class II if and only if it is a Class II site under the Reference
Technology Test-and the Economic Test. Of note is that the Economic Test
alone considers all treatment trains available throughout the country while
the Combined Test uses the least-cost treatment train available in the EPA
Region.
-------
-32-
3.3 Contamination Scenarios
This section describes the six contamination scenarios that are used by
the model to assess the costs of treating ground water.
These scenarios have been developed with reference to documented instances
of ground-water contamination. Each scenario is presented with a brief
description of the potential source type, the results of the Reference
Technology Test, and the "treatment trains" which should be used in the Class
III tests. The allowable concentration of each contaminant under the Safe
Drinking Water Act (SDWA) is also presented. This section concludes with a
discussion of analytic limitations.
Contamination Scenario 1
This scenario is based upon reported contaminants in an area located below
heavy industrial use, including a variety of manufacturing facilities and an
industrial waste landfill. The levels of contamination were developed without
reference to known contaminant concentrations in the ground water.
Contaminant Level SDWA Level
Constituents (mg/1) (mg/1)
Trichloroethylene 5 0.075
Tetrachloroethylene 25 0.02
1,1,1-Trichloroethane 50 1.07
Carbon Tetrachloride 300 0.02
Benzene 300 0.07
Toluene 50 0.34
Chloroform 250 0.1
Cadmium 25 0.01
Under the Reference Technology Test, this scenario would be Class III in
Regions 1, 2, 5, 7, 9 and Class II in Regions 3, 4, 6, 8, 10.
Contamination Scenario 2
This scenario represents a condition of mining or refining waste disposal
which has caused contaminant release to the ground water. This type of
contamination is likely to occur in any area which currently or in the past
has had mining, smelting, or refining operations. The types of metals present
is a function of the type of ores used and of the final product generated.
Contamination levels will fluctuate widely.
Contaminant Level SDWA Level
= Constituents (mg/1) (mg/1)
Arsenic 5.0 0.05
Selenium 5.0 0.01
Under the Reference Technology Test, this scenario would be Class III in
Regions 1, 2, 5, 7, 9 and Class II in Regions 3, 4, 6, 8, 10.
-------
-33-
Contamination Scenario 3
This scenario represents a Superfund site which includes a now-closed
hazardous waste landfill. Although organic contamination was reported by the
specific reference upon which this scenario is based, the identity of the
organic contaminant was not reported; therefore this contamination scenario
addresses only inorganic contaminants.
Contaminant Level SDWA Level
Constituents (mg/1) (mg/1)
Nitrate 500 10.0
Fluoride 3 1.4
Selenium 1 0.01
Barium 1 1.0
Cadmium 10 0.01
Chromium 10 0.05
Under the Reference Technology Test, this scenario would be Class III in
Region 1, and Class II in Regions 2 through 10.
Contamination Scenario 4
This scenario represents a rural or suburban situation in which
multiple-source nitrate contamination has occurred (e.g., from septic tanks
and fertilizers) as well as some agricultural contamination of the ground
waters.
Contaminant Level SDWA Level
Constituents (mg/1) (mg/1)
Endrin 0.02 0.0002
Lindane 0.10 0.004
2,4-D 0.20 0.1
2,4,5-TP 0.20 0.01
Trihalomethanes 1.0 0.1
Nitrate 200 10.0
Under the Reference Technology Test, this scenario would be Class II in all
regions.
Contamination Scenario 5
This scenario was generated as a composite of the many types of volatile
organic contamination reported, with little or no indication of the true
source of contamination. All values given are somewhat greater than the
average or median values reported. This scenario was developed to represent
the common contamination levels currently found.
-------
-34-
Contaminant Level SDWA Level
Constituents (mg/1) (mg/1)
Trichloroethylene 1.0 0.075
Toluene 35.0 0.34
Tetrachloroethylene 1.5 0.02
1,1,1-Trichloroethane 5.5 1.07
Benzene 1.0 0.07
Carbon Tetrachloride 1.0 0.02
Under the Reference Technology Test, this scenario would be Class II in all
Regions.
Contamination Scenario 6
This scenario, which is an entirely fabricated situation, is intended to
represent the type of conditions in which it would not be technically feasible
to clean up the ground water. This particular situation could occur at a
pesticide manufacturing facility or a pesticide landfill.
Contaminant Level SDWA Level
Constituents (mg/1) (mg/1)
i
Fluoride 150 0.4
2,4-D 50 0.1
Dichloromethane 250 0.15
Toluene 1,500 0.34
Trichloroethylene 250 0.075
Chlordane 2.5 0.0075
Toxaphene 5.0 0.005
Endrin 2.5 0.0002
Under the Reference Technology Test, this scenario would be Class III in all
regions.
Limitations
Each scenario is representative of a limited set of conditions, and the
total set of six scenarios is representative of only a very small percentage
of the large number of types of ground-water quality which exist throughout
the country. Some of the limitations associated with analysis of this limited
set of scenarios are listed below:
(1) Scenarios representing naturally occurring low-quality ground
water are not included. Such situations, which may include
saline ground waters, naturally high levels of metals, and high
levels of naturally occurring organic and inorganic compounds,
could potentially be a substantial percentage of the ground
waters which are eventually determined to be Class III.
-------
-35-
(2) The scenarios developed are not likely to occur at equal
frequencies; however because the development of a statistically
valid distribution of levels of contamination was not the focus
of this task, the analysis assumes equal probabilities of
occurrence.
(4) Except for Scenario #1, all contamination scenarios are based on
documented cases which report a contaminant plume. Under Class
III, an entire ground water must be contaminated; otherwise the
ground water is by definition Class II. Thus, the scenarios
extrapolate from plumes to entire ground waters.
(5) Except for Scenario #6, all scenarios are based on contamination
events which were detected through the use of the ground water
as a drinking water source. By definition, Class III ground
waters are not currently in use. Thus, these contamination
scenarios may not be fully representative of the ground waters
adjacent to RCRA facilities and CERCLA sites.
Treatment Costs
This section presents the results of an analysis of the least-cost set of
treatment technologies which will adequately treat each of the six
contaminated ground waters described in previous section. Exhibit 9
summarizes the costs of serving 500, 2,500, 5,000, and 25,000 persons. These
costs include the costs associated with each treatment process, with raw water
(influent) pumping and storage, with finished water pumping, storage, and with
administrative and laboratory requirements. Distribution and acquisition
costs are not included in these costs, nor is a profit factor (in other words,
these are the base engineering costs). Most cost values were estimated using
the cost curves available in EPA's Handbook of Water Treatment Costs,
developed by Gulp, Wesner, Gulp. Costs were adjusted to 1985 levels, and were
annualized. Exhibit 10 describes the specific set of treatment processes used
in the least cost treatment trains for each scenario.
-------
-36-
Ld
>
LU
O
<
to
T3
o
s.
0)
U)
3
o
X
t_
0>
ft
U)
o
CJ
—
ro
3
C
C
"*
0 0
CO O
as
00
o in
o •
m —
CM
»~
aa
0 O
0)2
a
0
O (O
o •
oo
in
o
_ >— .
an
00
CO E
a
in
o —
o •
CM
— a
ao
OS
Q)
aro
o
0 •
oo
c
0
Dl
cc
—
4-1
U) U)
(0 0
(01-
o
c
10 O
c —
— t-
£ ro
to c
*J OJ
C 0
O C/5
O
O
[^
•
CO
\Q
o
CO
.
0
in
CM
o
in
•
CM
CM
ON
i
\
01
m
LJ
—
OON
c —
* .
\COCM
"Ol^O ro
—
OO
-3-VO
. .
\cor~
•aim .a-
— CM
o —
CO .3-
. .
\ O ro
*o|in so
CM ro
OO
in —
• .
— CM —
•OlCM ro
ON —
Z
X
— "
> X
•v %
- = >
01
1-
^
CD
LJ
O
m
.
CM
CM
O
ON
.
—
in
vy
o
—
•
.3-
O
ro
m
CM
CM
i
CO
LJ
00
oo
lf\ ON
.
\ C\J if\
•DIOJ CO
00
. .
x^ — —
"Dim J"
{ft —
OO
. .
-S^^f \Q
"D|N- CO
^^
oo
ro O
. .
\ ro -3"
•OlCM 'O
CM J
X
— "
> X
- ,>
r-
a:
v^
m
LJ
0
m
.3-
ro
t/>
0
ro
—
CO
O
ro
*
VO
CO
ro
^
CO
LJ
ro
O O
mcM
\ J 00
•OlroCM
—
OO
ro r^-
. *
^— r-
•OlCOro
CM
OO
. •
^~roCM
•OICM in
— ro
OO
ro in
. •
\*o o
•oloom
ro —
V> ^
^
X
\
>
-
-
-X
H
K
CD
LJ
O
—
f-
J-
O
•
f-
ON
o
00
.
•3-
O
ro
.
.3-
ON
J
i
CD
LJ
.3-
O OO
— oo oo
• . .
r-oo
-3-mrn
o oo
•3" ^ -3"
. • .
r~- ro m
ONOO
— —
o oo
CO CO CO
...
J1 -3" ^
^ NO vO
«>
0 — 0
ro ON m
• . •
J" ON r*~
ON ON ro
j in in
X
—
>x
= ->
1-
cc
CO
Ld
O
ro
.
.3-
O
^
^O
00
o
ON
.
0
ro
O
^
•
— .
—
m
^
to
LJ
in
o
ro
.
.3-
j-
O
J-
.
~&
00
o
ON
.
0
ro
o
f^.
*
— -
—
in
i
i—
cc
CO
LJ
\
•ol
"^^
•ol
-s^
•ol
v^
•ol
^
CO
Ld
VO
\
•ol
"s^
•ol
\
"01
vv
•ol
i
1—
a:
CO
Ld
-------
-37-
o
O t/5
(-O
ec
co <
zo
•o ce. i-
0 i—to
3
C l->
Z h-
4J LJ —
c s: _i
o
u
— u or
o:
ON KBS
LJ
I- <
— CO <
CD 02
— O
X I O
X I-Z
UJ CO —
o:
•
•o
—
o
X:
0
U)
3
0
X:
L.
0
a
V)
c
0
«
1.
0
a
m
r~
rJ
c
o
•o
0
in
CCS
o
•o
—
O
£
0
W
3
O
JZ
1_
0
a
4->
V)
o
o
col
1.
CD
x:
4-1
0
£
>
li-
CS
tn
U)
0
T3
i.
CO
en
o
i_
0
—
X!
CO
•—
CO
>
CO
c
CO
!_
4-1
4-1
C
0
E
4-1
CO
0
1_
4J
4J
W
O
O
1
4-1
in
CO
0
—
0
XT
4J
C
o
•o
0
tn
CO
X3
0
1.
CO
4J
U>
0
4-1
C
0
T3
^
3
XI •
C
o o
._ ._
E 01
O 0
C i-
O
O 0
0 £
4-1
0
x: c
>
3 —
4->
U5 C
4-1 0
V> <_
O 1-
O 3
U
X3|V)
•••
4J
—
in
.—
4-1
CO
x:
4-1
C
•—
CO
t_
4-1
4-1
C
0
E
4-1
CO
0
l_
4J
4-1
in
o
O
4J
in
CO
0
—
0
x:
4-1
c
0
•D
0
V)
CO
-Q
0
(.
CO
4J
cn
0
4-1
>>
Ol
0
o
c
jC
o
0
4-1
0
0
c
0
I_
0
It-
CD
t_
^
C
0
T3
3
XI
0
•—
E
0
C •
O C
a o
o —
Ol
0 0
x: s-
4-1
0
ex:
— 4-1
•o c
0 —
cn
3-D
0
(/>
•W 3
V)
O >>
0-
4J
\ C
0|0
1_
&_
3
O
>1
4-1
C
0
u
!_
3
0
U)
—
in
C
._
CO
t_
4J
4-1
C
0
E
4-1
CO
0
£_
4-1
0
^
4-1
O
0
d.
d-
0
0
x:
4-1
Q-
o
0
c
0
c
0
in
3
CO
o
0
xs
c
0
cn
0
L.
in
x:
4-1
C
0
X!
CO
4J
CO
0
4-J
4J
O
c
tn
•- >
O
1.
CO
c
0
o
in
•" c
c o
CO —
c 01
— CO
E <-
CO
4J 0
ex:
O 4-1
u
c
\ —
•01
•a
0
tn
3
-------
Class III Test
Scenario 1
• Economic Test
• Combined Test
Scenario 2
• Economic Test
• Combined Test
Scenario 3
• Economic Test
• Combined Test
Scenario 4
• Economic Test
• Combined Test
-38-
EXHIBIT 10
DERIVATION OF LEAST COST TREATMENT TRAINS
Least-Cost Treatment Train a/
Scenario 5
Both Tests
Scenario 6
1, 2, 3, 6, 7, 8
Regions VI, VIII, X -- 1, 2, 3, 4, 6, 7
Regions III, IV -- 1, 2, 3, 6, 7, 8
7, 8
Regions III, IV -- 7, 8
Regions VI, VIII, X -- 4, 7
3, 7
Regions II, III, IV, V, VI, VII, VIII, X -- 3, 7
Regions IX -- 2, 3, 4, 5, 6
2, 5, 7, 8
Regions III, IV -- 2, 5, 6, 7, 8
Regions I, II, V, VII, IX -- 2, 3, 5
Regions VI, VIII, X -- 500 people -- 2, 4, 7
-- 2,500; 5,000; or 25,000
people -- 2, 3, 5
500; 2,500; or 5,000 people -- 1, 2, 5
25,000 people -- 1, 2, 6
No treatment trains available.
*#******:fr***************#**y?****^^
a/ Treatment Process Components:
1. Air Stripping/Aeration
2. Carbon Adsorption
3. Chemical Precipitation (includes clarification)
4. Desalination
5. Flotation
6. Granular Media Filtration
7. Ion Exchange
8. Ozonation
-------
DATA SOURCES FOR ANALYZING
GROUND-WATER CLASSIFICATION ISSUES
This paper describes the principal data sources relevant to analysis of
EPA's National Ground-Water Protection Strategy. Most of the information we
obtained relates to the income and substantial population criteria, though
some of the issue papers included in this package made use of other data
sources which are also briefly reviewed here. Part One of this paper
describes data from the 1980 Census of persons in the United States, the types
of information it offers, as well as the results of an overlay of income and
population data. Part Two provides a description of data available from the
Federal Reporting Data System (FRDS). Part Three concludes with a brief
overview of other data sources that have been used in one or more of the issue
papers that have been prepared.
PART I: CENSUS DATA
The principal data base that can be used in an analysis of "substantial
population" and income thresholds as criteria for defining Class I ground
waters is the 1980 Census of All Persons in the United States, specifically
Summary Tape File 3B (STF 3B). Types of information available on STF 3B that
are relevant to the classification of ground waters are the number of
households in a given area, the distribution of household income (which allows
for computation of median income), mean income, and the distribution of
households using various types of water sources (which allows for computation
of the percentage of households using individual ground-water wells). Thus,
it is possible to derive from the census data base the following types of
information: the distribution of mean and median incomes at the county, SMSA
(Standard Metropolitan Statistical Area), and zip code levels; the
distribution of income thresholds at each geographical level of aggregation;
the number and percentages of households using private wells; the number of
counties, SMSA, and zip code areas that use groundwater in a low, moderate, or
high percentage range (0-30%, 30-70%, and 70%-100%, respectively). Finally,
given these types of data it is possible to make the correlation between
household income and ground-water use. This is achieved by dividing the
different county, SMSA, or zip code areas into income groups and determining
the percentage of groundwater use in each group.
Because data on STF 3B is organized on the county, SMSA, and zip code
levels it allows for considerable flexibility of analysis. For example, while
the average household income at the county level may not trigger the income
threshold (which determines the ability of a community to pay for replacement
water sources), at the zip code level, where average income may be
considerably lower and the costs of replacing the water supply may exceed the
income threshold, the ground-water source serving the area may be described as
an "irreplaceable", or Class I, ground-water source.
The remainder of this section will focus on census data for population,
mean income, and ground-water use at the county and zip code levels
respectively.
-------
-2-
1. POPULATION
As there are many more zip code areas than counties in the U.S. (35,035
zip codes as compared with 3,137 counties), it follows that population levels
for zip code areas are considerably smaller than counties. Exhibit 1 shows
that 63.8 percent of all zip code areas have household populations of 1,000 or
less with 85.5 percent having household populations of 5,000 or less, while
only 35.1 percent of all counties have household populations of 5,000 or
less. At the same time, less than .2 percent of all zip code areas have
populations of 30,000 or more whereas 8.6 percent of all counties have
household populations over 55 thousand. These data becomes important when
population and income are figured together to estimate an area's "ability to
pay" for the costs of a replacement water supply.
2. INCOME
As indicated in the introduction of this paper, considerable differences
in mean income level can occur between county and zip code level. Exhibit 2
shows that while there are no counties with a mean household income of zero to
$5,000 in the U.S., there are 204 zip code areas that fall within that range,
representing .59 percent of all zip code areas. Further, while 0.13 percent
of all counties have mean household incomes between $5,000 and $10,000, 4.0
percent of all zip code areas fall within that range. At the other extreme,
although only 0.61 percent of all counties have mean household incomes over
$30,000 annually, 3.8 percent of all zip codes have incomes of $30,000 or
greater. Thus, while the average mean household income for all zip code areas
remains about the same for as all counties, a larger number of zip code areas
can be considered "poor" or "wealthy" relative to counties. This "flattening
out effect" or level of aggregation, as depicted in Exhibit 3, is critical to
determining the data used for the "ability to pay" exercise and therefore
could greatly influence the number of ground waters estimated to be in the
Class I category.
3. GROUND WATER USE
The percentage of housing units in a zip code or county using private
wells (i.e., dug or drilled wells) can be grouped into three categories within
a high (70 to 100 percent), moderate (30-70 percent) and low (0-30 percent)
range.
At the county level, the level of aggregation is such that the data is not
particularly revealing of any one trend in ground-water usage. At the zip
code level, however, it is significant to note (as shown in Exhibit 4) that
nearly one-quarter (23.7 percent) of all zip codes use a high level ground
water and nearly one-half (45.6 percent) derive thirty percent or less of
their water from private ground-water sources.
-------
-3-
EXHIBIT 1
NUMBER OF HOUSEHOLDS PER COUNTY
Number of Households Frequency Percent of All Counties
0-5,000 1,101 35.0
5,000-10,000 775 24.7
10,000-15,000 364 11.6
15,000-20,000 210 6.7
20,000-25,000 119 3.8
25,000-30,000 92 2.9
30,000-35,000 60 1.9
35,000-40,000 53 1.7
40,000-45,000 36 1.2
45,000-50,000 34 1.1
50,000-55,000 23 0.7
Over 55,000 270 8.6
Total 3,137 100
NUMBER OF HOUSEHOLDS PER ZIP CODE
Number of Households Frequency Percent of All Zip Codes
< 1,000 22,337 63.8
1,000-2,000 3,905 11.2
2,000-3,000 1,776 5.1
3,000-4,000 1,054 3.0
4,000-5,000 864 2.5
5,000-10,000 2,680 7.7
10,000-15,000 1,439 4.1
15,000-20,000 607 1.7
20,000-25,000 263 0.8
25,000-30,000 77 0.2
30,000-35,000 26 0.1
Over 35,000 7 under 0.1
Total 35,035 100
-------
-4-
EXHIBIT 2
Mean Income
(OOOs)
$0-5
$5-10
$10-15
$15-20
$20-25
$25-30
Over $30,000
Total
COUNTY MEAN INCOME
Number of Counties Percent of All Counties
0
4
876
1,760
405
73
19
3,137
0
0.1
27.9
56.1
12.9
2.3
0.6
100
Mean Income
(OOOs)
$0-5
$5-10
$10-15
$15-20
$20-25
$25-30
Over $30
Total
ZIP CODE MEAN INCOME
Number of Zip Codes Percent of All Zip Codes
.4
.9
0.6
4.0
28.
40.
17.0
5.2
3.8
100
-------
-5-
EXHIBIT 3
ILLUSTRATION OF AGGREGATION EFFECT
County Mean Income Distribution
Zip Code Mean
Income Distribution
All Individuals
NATIONAL
MEAN
INCOME
-------
-6-
PART II: FRDS DATA
Data identifying public water supply systems supplement the census data
which identifies locations and usage levels of private wells. One good source
of public water supply (PWS) data is the Federal Reporting Data System
Interactive (FRDS/Interactive).
FRDS was developed for EPA's Office of Drinking Water (ODW) to provide
data on the size, characteristics, and compliance of Public Water Systems
(PWS). FRDS provides a relatively easy way of accessing, searching,
retrieving, and reporting data from the PWS files.
Five general types of PWS data are maintained in FRDS including: (1)
identification and statistical summary data for each PWS; (2) inventory data
(i.e., PWS capacity and source information); (3) violation data based on
non-compliance with EPA's standards; (4) data on variances and exemptions from
EPA's standards; and (5) enforcement action data. The database is updated
quarterly; the most recent complete fiscal year accessible is 1984. A version
for each completed FY is retained for historical analysis. Approximately
62,000 community PWS's are tracked, along with approximately 160,000
non-community PWS's. Community water supplies are those that operate year
round and that serve a residential population. Examples of non-community
PWS's include those used for recreation or for restaurants or hotels. The
majority of these 200,000 plus systems are active although those that are
non-active are also tracked.
An individual PWS is identified as having a source type of either: (1)
surface water, (2) purchased surface water, (3) ground water, or (4) purchased
ground water. The source for a particular PWS is based on EPA's regulatory
definitions which rank the different source types. That is, if any surface
water is used, regardless of the existence of other source types, the source
will be identified as surface water. Likewise, if no surface water is
utilized, the source will be categorized as ground water (if any amount is
present). Groundwater is ranked second, preceded by surface water.
Public water systems can be identified in FRDS by plant location based on
state, county, SMSA, zip code, latitude/longitude, and hydrologic cataloging
unit. Plant latitudes and longitudes were derived by identifying the
centroid of the plant zip code. Specific PWS water sources can, in some
cases, be identified by latitude/longitude. However, since approximately only
36-38% of the source latitude/longitudes are reported, there is not a full
inventory source. It may be reasonable to use the location of the plant as a
proxy for the source, but in cases where the actual source of the PWS is not
proximate to the plant, ground-water contamination near the plant may not
affect the water sources, and hence the water supplied by the plant.
FRDS provides information characterizing the water supply industry as a
whole (Exhibit 5) in terms of the number of systems and the size of the
population served (ranging from very small systems serving 25-500 people to
very large systems serving over 100,000 people); the type of water source
(ground and surface waters only); the ownership of the systems (public vs.
private); and the total population served by each system-size category. It is
-------
-7-
EXHIBIT 4
USE OF PRIVATE WELLS
Percent of Housing
Units in Zip Code Using Number of Percent of
Private Ground Water Zip Codes Zip Codes
Low 0-10% 9,372 26.8
10-20% 3,392 9.7
20-30% 3,212 9.2
Total 15,976 45.7
Moderate 30-40% 3,146 9.0
40-50% 2,918 8.3
50-60% 2,585 7.4
60-70% 2,089 6.0
Total 10,738 30.7
High 70-80% 1,868 5.3
80-90% 2,375 6.8
90-100% 4,078 11.6
Total 8,321 23.7
-------
-8-
o\
00
ON
CO
lA
CO O
ONiA
LAM
CM ro
r-
CM
O
ro
CM
O
O
CD
CD 03
o r-
o r-
<- O CM
CD •
> o
o o
LA CM
a-ro
CM
— VO
ro a-
CM
O
a-
>
o.
CL
I O ON
00 CO
oo
o *
• o
LAO
I
I
— o
o o
00
OlA
IAJ
r
i
— o r-
OO ON
O O VO
LAO
CM LA
I
— O ro
00 VO
OO t—
OLA —"
— CM
r—vo
vovo
roa-
P-CM
CMa-
ON a-
— a-
COON
CO J"
00
vO
VO
CO
CM
OON
LA —
VOVO
r- co
a- CM
lA
CM
ro
ON
ON
IT
LJ
LA
I- 2-3-
— co
CO Ld
— X >
X I- U.
X ~— '
I O
— o
oo
o -
-o
LA —
co
co
o
ro lA
— ON
I
— O
OO
ro O
f-
00
ro LA —
COON
r— ro
ONro
lA —
lAro
a-
O
lA
N
C£
LJ
I-
o
<
cr
<
x
U
— O —
00 CM
LAro J
I
— O ro
OO O
OLA a-
—"CM" vo"
o o
i o c\j
— O CM
o -
LA — VO
— o
COLA
ONa-
co —
ro O
a-W
IAIA
LAVO
03 ro
VO ON
O O
CM CM
CO CM
ONVO
CMO
COO
— ro
ON CM
ro
CM
CO
a-
a-
ro
CM
NO
I
— O
00
— LA
VO
ro
CM
co"
CO
I O vo
LAO
CM — ON
00 lA
r-a-
lACM
LAO
LAro
— LA
«"—"
voa-
LA 00
rOlA
a- ON
CO
ro
LA
VO
CO
o
c
o
O CD
a. c/3
c*_
O
U)
i- E
O CD
E
3 >>
ZOT
(U
a
._ 4)
o o
C £-
•*- 3
!- O
O. (/)
CD
-o o
C CO
3 IK
O S-
<- 3
O CO
a
•—
.c
^
CD
C
a
0
O -P
— 03
— >
n -^
3 1-
CL Q-
CU
4J C
03 O
E —
— u
X 03-0
O — CD
(_ 3 >
a a i-
ao CD
in
c
o
__
__
._
^
„_
VI
Q
o:
o
!_
3
O
-------
-9-
clear from this data that the majority of Americans are served by very large
systems but that a significant fraction of the population is served by the
several thousand, very small systems. It is also significant to note that
over four times as many people are served by ground-water sources as by
surface water.
PART 3: OTHER DATA SOURCES
A variety of other data sources were used in connection with the issue
papers contained in this package. The two most important are described here.
First, in order to obtain information on active RCRA treatment, storage, and
disposal facilities, we consulted the Hazardous Waste Data Management System
(HWDMS). This data base, maintained by EPA, contains information on the
location of these facilities, including addresses, zip codes, and geographic
coordinates. This information permits a matching of data from HWDMS to other
data sources. For example, based on zip codes, we were able to determine how
many private well users were located near hazardous waste management
facilities. The second key data source was the data base created in
conjunction with the Hazard Ranking System (HRS). Maintained by the Mitre
Corporation, the HRS data base contains information on all hazardous waste
sites that have been evaluated under the Superfund program. We limited our
analysis to those sites that are on the National Priorities List. For these
sites, we obtained data on longitudes and latitudes which enabled us to match
CERCLA sites with data from our other sources.
------- |