POST-CLOSURE LIABILITY TRUST
FUND SIMULATION MODEL:
VOLUME III: MODEL DESCRIPTION
Office of Solid Waste
U.S.,.Environmental Protection Agency
May 1985
-------
(PAGE INTENTIONALLY LEFT BLANK)
-------
PREFACE
This report is the joint effort of two co-contractors, IGF Incorporated
and Battelle Pacific Northwest Laboratories. Battelle Pacific Northwest
Laboratories developed the Release Simulation Model, a model that predicts
facility releases. ICF Incorporated developed the overall Post-Closure
Liability Trust Fund Simulation Model, which uses the results of the Release
Simulation Model to assess the adequacy of the PCLTF.
This report is organized in three volumes, as follows.
Volume I, "Model Overview and Results," introduces the Post-Closure
Liability Trust Fund and presents background information on the RCRA program;
describes in general terms the PCLTF Simulation Model and how it is used;
presents Model results for five simulations; presents an analysis of the
revenue requirements under alternative Fund coverage policies; and presents a
discussion of necessary limitations and next steps.
Volume II, "Graphs and Tables of Model Results," presents the graphs and
tables that contain the Model results discussed in Volume I. Volume II is
separately bound so that the reader can conveniently peruse the Model output
when reading the Volume I report.
Volume III, "Appendices," contains four appendices that provide technical
detail on various aspects of the Model. Appendix A presents detail on
methods, data and assumptions used in the PCLTF Simulation Model. Appendix B
provides detail on the Release Simulation Model. Appendix C describes several
aspects of model implementation for both the PCLTF Simulation Model and
Release Simulation Model, including run characteristics, subroutines, data
files, random number generators, and statistical properties of the results.
Appendix D reproduces the set of options available to the Model user.
This study was conducted under the guidance of EPA's Office of Solid
Waste. Ms. Carole Ansheles, the Project Officer, and her associate Mr.
Richard Allan were in charge of this project. Mr. Peter Guerrero was the
former Project Officer.
-------
TABLE OF CONTENTS
Page
VOLUME lit: APPENDICES
Preface i
Appendix A: Detailed Model Description A-l
A. 1 Model Overview A-2
A.2 Estimation of Basic Model Units A-4
A.2.1 Facility Population A-4
A.2.1.1 Existing Facility Population A-4
A.2.1.2 New Facility Population A-12
A.2.2 Facility-Level Characterization A-12
A.2.2.1 Facility Milestones A-13
A.2.2,2 Facility Attributes A-28
A.2.3 Modeling of Releases A-53
A. 2.4 Actions A-63
A.2.4.1 Monitoring Actions A-64
A.2.4.2 Response Actions A-70
A.2.4.3 Post-Closure Care A-76
A.2.5 Claims A-80
A.2.5.1 Personal Injury Claims A-81
A.2.5.2 Real Property Damage Claims A-98
A.2.5.3 Economic Loss Claims A-102
A.2.5.4 Natural Resource Damage Claims A-103
A.2.6 Funding Sources ..;.... A-107
A.3 Relationships Among the Basic Model Units A-lll
A.3.1 Economic Relationships A-113
A.3.1.1 Relationship Between the Demand for Land
Disposal Capacity and the Facility
Population A-114
A.3.1.2 Relationships Among Fund Balance, Revenues,
and Spending A-122
-------
TABLE OF CONTENTS (continued)
Page
A.3.2 Regulatory Policy A-125
A.3.2.1 Permit Policy A-125
A.3.2.2 Action Policy A-128
A.3.3 Financial Relationships A-135
A.3.3.1 Background Rate of Bankruptcy A-138
A.3.3.2 Bankruptcies Due to Costs at Hazardous Waste
Facilities A-138
A.3.4 Cost Allocation Policy A-140
A.3.4.1 Qualification Policy A-140
A.3.4.2 Allocation of Costs to Funding Sources A-142
A.3.5 The Effect of Response Actions on Releases A-147
A.3.6 Legal Validity of Claims A-153
Appendix B: Release Simulation Model B-l
B.I Model Methodology B-l
B.1.1 User-Supplied Inputs and Default Values B-l
B.1.2 Internally Stored Inputs B-7
B.I.3 Relationships and Functions B-16
B.2 Outputs B-27
B.3 Discussion B-29
B.3.1 Low-Probability Acute Events B-29
B. 3.2 Land Treatment Facilities B-29
B.3.3 Injections Wells B-29
B.3.4 Corrective Actions B-30
-------
TABLE OF CONTENTS (continued)
Appendix C: Model Implementation C-l
C.I PCLTF Simulation Model C-l
C.I.I Run Characteristics C-l
C. 1.2 Pseudo-Random Number Generator C-l
C.I.3 Statistical Properties of the Results C-6
C. 1.4 Subroutines C-10
C. 1.5 Data Bases C-14
C.2 Release Model C-14
C.2.1 Run Characteristics C-14
C . 2.2 Pseudo-Random Number Generator C-15
C.2.3 Subroutines C-15
C.2.4 Data Bases C-17
Appendix D: User Options D-l
-------
(PAGE INTENTIONALLY LEFT BLANK)
-------
APPENDIX A
DETAILED MODEL DESCRIPTION
This appendix describes the PCLTF Simulation Model. The goal is to lay ~
open for examination all the assumptions, data, data analyses, and simulation
techniques that underly the Model. This material is not a "users' manual";
flow charts and program codes are not presented. Instead, this appendix
concentrates on presenting the substantive workings of the Model.
Recognizing that a detailed description of a model as large and complex as
the PCLTF Simulation Model would necessarily be quite lengthy, this appendix
is organized to enable the reader to refer selectively to individual topics of
interest. For example, an individual interested in examining the Model's
methods for calculating the costs of response actions can easily identify the
relevant material for review without reading the entire appendix. This
appendix is organized as follows:
A.I Model Overview provides a brief overview of the entire
Model. All readers should review this brief section to become
familiar with the overall flow of the Model.
* A.2 Estimation of Basic Model Units describes each of the
major building blocks of the Model. Each building block is
defined, and the data used and data analyses performed to
estimate each building block are described. The basic model
units are:
Facility Population;
-- Facility-Level Characterization;
-- Modeling of Releases;
Actions;
-- Claims; and
-- Funding Sources.
A.3 Relationships Among Basic Model Units describes how the
major building blocks relate to each other. The relationships
can be divided into six types:
Economic Relationships;
Regulatory Policy;
Financial Relationships;
-- Cost Allocation Policy; *
-- The Effect of Response Actions on Releases; and
-- The Legal Validity of Claims.
-------
A-2
A.I MODEL OVERVIEW
As displayed in Exhibit A-l, the PCLTF Model analysis proceeds as follows:
The estimate of the population of land disposal facilities over
time is based on economic relationships describing the supply of
and demand for land disposal of hazardous waste.
For each facility in the facility population, a facility-level
characterization is performed which estimates a series of
characteristics (such as facility size and design). Based on
these characteristics and data defining the modeling of releases,
the facility's potential releases are estimated.
The potential releases at a facility are an important driving
influence of the facility-level characterization. Given the
actions characterized at the facility based on the action data,
potential releases may be detected. Then, in conjunction with
regulatory policies and the effects of response actions on
releases, the detected releases lead to additional monitoring
actions, response actions, and claims. The costs of these
actions and claims are estimated using the action data and claim
data.
The costs of the actions and claims estimated at the facility
level are allocated to the available funding sources according to
specified cost allocation policies. The PCLTF is one of the
potential funding sources available to cover costs that arise at
the facility level. Various issues regarding the legal validity
of third-party claims also play a role.
* If costs are allocated to the owner/operator, the firm may be
more likely to go bankrupt. Using the financial relationships,
the viability of firms is assessed to reflect the influence of
actions and claims costs on firms. If a firm goes bankrupt, its
facilities cease accepting waste and it is no longer available to
cover costs.
An important feedback between funding sources and the
facility-level characterization exists. If no funding source is
available for a monitoring or response action, then the action is
not taken. This feedback is important because a response action
can prevent additional releases from occurring at a facility.
Consequently, not having a funding source for certain actions may
lead to additional numbers of releases.
The following two sections describe the basic units and relationships
included in the PCLTF Model. These sections are organized according to
Exhibit A-l, which lists the relevant section numbers in parentheses.
-------
A-3
EXHIBIT A-l
SIMPLIFIED REPRESENTATION OF THE PCLTF SIMULATION
MODEL ANALYSIS1
(A.2.1)
FACILITY
POPULATION
OVER TIME
(A. 2. 3)
MODELING OF
RELEASES
(A.2.4)
ACTION DATA
(A.2.5)
CLAIM DATA
(A.2.2)
FACILITY LEVEL CHARACTERIZATION:
Characteristics
Releases: Potential and Detected
Monitoring Actions
Response Actions
Claims
(A.2.6}
FUNDING
SOURCES
(A.3.1)
ECONOMIC
.RELATIONSHIPS
(A.3.2)
REGULATORY
POLICY
(A.3.5)
,THE EFFECT OF1
RESPONSE ACTIONS
.ON RELEASES.
(A.3-3)
FINANCIAL
i RELATIONSHIPS,
(A.3.6)
LEGAL VALIDITY
OF CLAIMS
(A.3-4)
COST
ALLOCATION
POLICY
1 Rectangles display basic model units. Ovals display the relationships
among the basic model units. Numbers in parentheses refer to the section in
Appendix A.
-------
A-4
A.2 ESTIMATION OF BASIC MODEL UNITS
A.2.1 Facility Population
In order to estimate PCLTF expenditures over time, it was necessary to
estimate the current and future population of land disposal facilities.
Despite the fact that there exist several surveys of hazardous waste disposal
facilities in the United States, no single data base exists that contains both
(1) a comprehensive listing of currently operating facilities and (2) the
characteristics of these facilities. Because this information is needed for
the Model, it had to be developed as part of this project. Three sources of
relevant information exist; these sources were analyzed to ensure that (1) all
relevant information was used and (2) the current set of facilities used in
the Model agrees closely with EPA's current best estimate of their land
disposal process distribution. This section describes the several steps
performed to accomplish these goals.
First, we developed a representation of the existing facility population,
i.e., currently-operating hazardous waste land disposal facilities throughout
the United States. Second, because the facility population will change as
facilities close and new facilities open, we simulated the addition of new
facilities as they are needed to meet the demand for the land disposal of
hazardous wastes.
The population of existing and future land disposal facilities considered
in the model includes those facilities with the following land disposal
processes:
surface impoundments (storage, treatment, and disposal);
landfills;
land treatment; and
injection wells.
An individual facility may have one or more of these four processes.
Facilities that handle hazardous waste but do not have at least one of these
processes were not considered in our analysis.
A.2.1.1 Existing Facility Population
To represent the existing facility population (i.e., the facility
population as of October 1, 1983, which is the beginning of year 1 of the
PCLTF Model), we constructed and analyzed a data base containing facility
information from the most recent data describing Interim Status facilities.
This newly-created data base is called the facility data file. The three data
sources employed in constructing the facility data file were:
Part A Permit Application Data (1979-1980), which
identify approximately 2,600 land disposal facilities;
-------
A-5
Telephone Verification Survey Data (1982) for
approximately 1,171 of the 2,600 facilities identified
in the Part A Data; and
National Mail Survey of RCRA-Regulated Hazardous Waste
Handlers Data (1983), which provide the best estimates
of the current aggregate facility population and
additional detailed information on a smallnumber of
land disposal facilities.
Each of these data sources has strengths and weaknesses. No one source
provides a fully comprehensive and verified listing of facilities that
currently operate one or more of the land disposal processes analyzed in the
Model.
To use the strengths of each source of information, we devised a procedure
to estimate the relationships among the data sources and EPA's current best
estimate of the size of the existing facility process population.1 These
estimated relationships were then used in the PCLTF Simulation Model along
with the Part A and TVS data to simulate the existing facility population.
First, the statistics themselves are described along with the procedure used
to estimate them. Then, the method used to simulate the existing facility
population is presented.
Relationships Among the Data Sources Describing the Existing
Facility Population
To estimate the existing facility population, we first created a data base
of facility-specific information using the Part A Data and the Telephone
Verification Survey (TVS) Data. We then analyzed this data base in
conjunction with the results of the Mail Survey to estimate the relationships
among the three sources. The relationships enable the Model to simulate a
comprehensive facility population using all the data that are currently
available.
The most accurate of the three data sources was assumed to be the Mail
Survey Data, followed by the TVS Data, and then the Part A Data. However,
only Part A Data are available for all 2,600 facilities that may have land
disposal processes. TVS Data are available for 1,171 of the 2,600 Part A
facilities and Mail Survey Data are available for only 672.2 This means
1 U.S. Environmental Protection Agency, "National Survey of Hazardous
Waste Generators and Treatment, Storage, and National Disposal Facilities
Regulated under RCRA in 1981: Preliminary Highlights of Findings," August 30,
1980.
2 While the Mail Survey responses were not random, indicating that these
data may be biased, the Survey's high response rate suggests that potential
bias may not be a serious problem. See footnote 1.
-------
A-6
that neither the Hail Survey Data nor the TVS Data alone are sufficient for
simulating the population of existing land disposal facilities.
To overcome this lack of TVS and Mail Survey data, the following
statistics were estimated to describe the relationship among the data sources:
Probability of observing a particular TVS response
given the observed Part A response; and
Probability of observing a particular Mail Survey
response given the observed TVS response.
Given that Part A Data are available for all the facilities and TVS data are
available for many, these statistics can be used to simulate what the Mail
Survey response would have been for each facility.
To estimate the relationship between the Part A and TVS data, we compared
the process data for the 1,171 facilities reporting both sets of data.
Exhibit A-2 displays the Bayesian probabilities estimated as a result of this
comparison.3 For example, 0.92 (or 92 percent) of the facilities reporting
at least one surface impoundment in the Part A Data also reported at least one
in the TVS Data. Conversely, 48 percent that did not report a surface
impoundment in the Part A Data did report one or more in the TVS Data.
Using these probabilities and the Part A and TVS data, we estimated the
frequency with which each of the 15 combinations of disposal processes is
anticipated to occur at the facilities. These probabilities, reported in
Exhibit A-3, were constructed as follows:
1) Using those facilities with Part A Data (but without TVS Data),
count the number of times each of the 15 facility configurations
occurs;
2) Using the Bayesian probabilities reported in Exhibit A-2,
estimate the probabilities with which each Part A Data-based
observation could result in each of the 15 configurations (e.g.,
the probability of an observed Part A configuration 1 resulting
in a TVS-adjusted configuration 1 is:
(0.92)(1.0-0.06)(1.0-0.04)(1.0-0.02) = 0.81);
3 Bayes' Theorem forms the basis of the Bayesian analysis performed
here. Bayes' Theorem is:
P(E/H) P(H)
p (H/E) -
P(E)
where the variables E and H represent Part A and TVS responses, respectively.
For an explanation of Bayes1 Theorm, see an introductory probability text such
as Probability with Statistical Applications by F. Mosteller, R. Rourke, and
G. Thomas, (Reading, Massachusetts: Addison-Wesley Publishing Company, 1970),
pp. 158-164.
-------
A-7
EXHIBIT A-2
BAYESIAN PROCESS PROBABILITIES DEVELOPED BY
COMPARING THE PART A DATA AND THE TELEPHONE
VERIFICATION SURVEY DATA1
Bayesian Probability of Having this Process
Part A Responseby Process ^.n the TVS Data Given the Part A Response
Surface Impoundment:
YES 0.92
NO 0.48
Landfill:
YES 0.70
NO 0.06
Land Treatment:
YES 0.70
NO 0.04
Injection Well:
YES 0.87
NO 0.02
1 Based on 1,171 observations.
-------
A-8
EXHIBIT A-3
FREQUENCY DISTRIBUTION OF FACILITY CONFIGURATIONS1
Facility , Probability of
Configuration' Probability of Occurrence Occurrence Adjusted
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
SI
Y
N
N
N
Y
Y
Y
Y
Y
Y
Y
N
N
N
N
LF
N
Y
N
N
Y
N
N
Y
Y
N
Y
Y
Y
Y
N
LT
N
N
Y
N
N
Y
N
Y
N
Y
Y
Y
N
Y
Y
IW
N
N
N
Y
N
N
Y
N
Y
Y
Y
N
Y
Y
Y
Based on TVS/Part A Analysis
.676
.093
.045
.041
.067 .
.025
.021
.012
.005
.005
.001
.005
.003
.001
.000
1.000
for Mail Survey Results
.673
.147
.049
.066
.032
.009
.012
.002
.001
.001
.OOO3
.002
.004
.000*
.002
1.000
1 Based on 1,171 TVS observations and 1,429 Part A observations.
2 SI = Surface Impoundment; LF = Landfill; LT = Land Treatment; IV =
Injection Well
3 Less than 0.0001.
* Less than 0.0005.
-------
A-9
3) Multiply the probabilities estimated in step 2 times the number
of observations from step 1 to estimate the expected number for
each configuration from those facilities with only Part A Data;
4) Count up the number of each configuration observed in the TVS
Data, and add to the results of step 3; and
5) Calculate the probability with which each configuration is
expected to occur by dividing the estimated number for each
configuration by the total number of facilities expected to have
at least one of the four disposal processes (approximately 200 of
the original 2,600 facilities are expected not to have any of the
four disposal processes).
This procedure produced the TVS/Part A distribution in Exhibit A-3 and the
expected number of each disposal process reported in the first row of Exhibit
A-4.
These estimates were then adjusted for the Mail Survey results using an
estimate of the relationship between the TVS Data and the Mail Survey Data.
The TVS/Mail Survey relationship was estimated based on the results of the
Part A/TVS analysis (described above) and EPA's estimate of the number of each
disposal process believed to exist today (see the second row of Exhibit A-4),
which is based on the Mail Survey results.
An eligibility rate was calculated for each process (see the fourth row of
Exhibit A-4). These rates were interpreted as the probability of actually
having each process (as indicated by the Mail Survey), given that it was
reported in the TVS (as indicated by the results of the Part A/TVS analysis).
These probabilities were used to estimate the expected likelihood that each of
the TVS-observed configurations could be a Mail Survey-adjusted configuration
(e.g., TVS configuration 5 has a (0.40 x 0.45 = 0.18) chance of being a Mail
Survey-adjusted configuration 5, a (0.40)(1.0-0.45) = 0.22 chance of being a
Mail Survey-adjusted configuration 1, a (1.0-0.40)(0.45) = 0.27 chance of
being a Mail Survey-adjusted configuration 2, and a (1.0-0.40)(1.0-0.45) =
0.33 chance of having no land disposal process after the Mail Survey
adjustment). These likelihoods were multiplied by the TVS/Part A
probabilities of occurrence and summed across the 15 configurations. The
total was 0.44, meaning that overall, there is a 0.44 (or 44 percent) chance
that a facility which reported having at least one land disposal process in
the TVS would actually have at least one process (as indicated by the Mail
Survey results).
The distribution across the 15 facility configurations was normalized
using this 0.44 estimate to produce the Mail Survey-adjusted probabilities
reported in Exhibit A-3. The results of these statistics (i.e., the Bayesian
probabilities in Exhibit A-2, the calculated eligibility rates in Exhibit A-4,
and the probabilities of the configurations in Exhibit A-3) are displayed on
an expected value basis in the fifth row of Exhibit A-4. These estimates
closely match the EPA estimates of the current number of each land disposal
process in existence today. The manner in which these statistics were used in
the PCLTF Model to simulate the existing facility population is described next.
-------
A-10
EXHIBIT A-4
ADJUSTMENT OF POPULATION ESTIMATE FOR
MAIL SURVEY RESULTS
1. Expected Value
Based on TVS
and Part A1
2. EPA Estimate2
3. EPA-Reported
Eligibility
Rate2
4. Calculated
Eligibility
Rate (2*1)
5. Mail Survey-
Adjusted
Expected Value
Surface
Impound- Land Injection
ments Landfills Treatment Wells Facilities
1,946
0.40
769
448
0.45
198
225
0.31
68
184
770 200 70 90
0.49 0.46 0.37 0.42
0.49
90
2,395
980
1 Calculated by applying the probabilities reported in Exhibit A-2 to
the facility data file to estimate the number of facilities, and then using
the TVS/Part A analysis probabilities reported in Exhibit A-3 to estimate the
number of each process.
2 U.S. Environmental Protection Agency, "National Survey of Hazardous
Waste Generators and Treatment, Storage, and Disposal Facilities Regulated
Under RCRA in 1981: Preliminary Highlights of Findings," August 30, 1983.
-------
A-ll
Simulating the Existing Facility Population
The results of the above analysis provided the relationships needed to
simulate probabilistically an existing facility population for the PCLTF
Model. On an expected value basis (i.e., on average), this facility
population closely resembles the existing facility population in the United "
States as estimated by EPA. It also reflects the current uncertainty
surrounding the true size of the facility population. The number of existing
Interim Status facilities simulated in each iteration of a model run will vary
due to the probabilistic simulation approach adopted. Because this variance
may influence the PCLTF balance, the Model results reflect the uncertainty
surrounding the size of the facility population.
We used the following procedure in simulating the existing facility
population:
1. For each of the 2,600 facilities in the facility data base, the
types of land disposal processes reported to be present and the
source of the data (either Part A or TVS) were identified.
2. These data were adjusted probabilistically as follows to reflect
the TVS results:
If TVS is the source of the data and a land disposal process
is reported, then the facility is assumed to have at least
one land disposal process.
If the data source is Part A, whether a land disposal
process exists is determined probabilistically using the
Bayesian probability of having a specific process in the TVS
Data, given the Part A Data. (This is done using a
pseudo-random number and comparing it to the associated
probability for each of the processes.) If one or more
processes is simulated to exist, the facility moves on to
step 3. If none are simulated to exist, the facility is
dropped from the iteration.
3. If one or more land disposal processes exist at the facility
according to step 2, then we determined probabilistically whether
the facility exists according to the Mail Survey results. Each
facility has a 0.44 chance of existing at this step.
4. If one or more land disposal processes exist at the facility
according to step 3, then the facility became part of the
simulated existing facility population for that iteration of the
Model run. The characteristics of that facility were estimated ,
based in part on the Part A Data available for that facility and
based in part on estimates of facility characteristics in the
overall population (based on analyses of the TVS and Mail Survey
data). The manner in which facility characteristics were
simulated (for both existing and new facilities) is discussed
below in Section A.2.2, Facility-Level Characterization.
-------
A-12
Using the above procedure for each facility, we determined the population
of existing facilities for the PCLTF Model. Because the adjustments and
simulations in this process are based on probabilities, the simulated number
of facilities (generated with the use of pseudo-random numbers) will vary.
The expected value (i.e., the average) of the number of facilities and
processes in the existing facility population is reported in row 5 of Exhibit
A-4.
A.2.1.2 New Facility Population
New facilities are added to the facility population over time as
additional capacity is required to meet the demand for land disposal of
hazardous wastes. In Simulation 1, the demand for land disposal of hazardous
waste was assumed to decrease by 2 percent per year for the first 50 years and
then to remain constant for the following 50 years. In addition, the average
capacity of new facilities was assumed to grow by 1 percent per year to
reflect increases in the size and efficiency of new land disposal processes
over time. Using these values results in approximately 185 new facilities
being added over the next 100 years within the United States.
It should be noted that the future demand for land disposal of hazardous
waste is very uncertain. Proposals are currently being considered to ban or
restrict the land disposal of hazardous waste in the near future. Whether
these proposals are enacted will influence the future demand for land
disposal, and consequently, the estimate of the future size of the facility
population. Due to this uncertainty, the Model was configured to allow a wide
variety of assumptions regarding the future demand and supply of land
disposal. The Model does, however, ensure that the assumptions used are
iternally consistent. Several economic relationships are used which constrain
supply and demand to be approximately equal over time. These relationships are
described in Section A.3.1, Economic Relationships.
A.2.2 Facility-Level Characterization
Each facility in the facility population was characterized in detail
within the Model. Because detailed facility-specific data were not available
for most facilities, the values for their .characteristics were simulated using
descriptors of the distribution of facility characteristics within the
facility population. For example, based on the 44 landfill responses to the
Mail Survey, the distribution of landfill size (in acres) in the existing
facility population was estimated. This distribution was then used to
simulate the sizes of existing landfills within the Model. The size
distribution of new landfills (those simulated to open in the' future) may, of
course, differ from the distribution for existing landfills. Consequently, a
size distribution for new landfills was assumed because new (i.e., future)
landfills do not yet exist.
The characteristics simulated at the facility level were divided into two
types: milestones and attributes. Milestones are activities that every
facility must perform at some point in time. For example, every facility must
eventually close (i.e., cease accepting hazardous waste for land disposal).
Milestones were simulated at the facility level by estimating when they
-------
A-13
occur. Attributes, on the other hand, describe the facility at a point in
time. For example, the size of a landfill is a facility attribute. The size
was simulated by estimating a particular value, e.g., 2 acres.
Both the milestones and attributes are important descriptors of the
facility-level analysis performed in the Model. Together they provide a
description of the level of resolution or detail inherent in the Model. In
reading the following sections on milestones and attributes, the reader will
be able to identify the major facility characteristics used to simulate the
performance of land disposal facilities.
A.2.2.1 Facility Milestones
The importance of facility milestones is that they identify points in time
during a facility's life at which important determinations or decisions are
made. In particular, the timing of these milestones will influence the
allocation of costs among potential funding sources (i.e., PCLTF, state funds,
Superfund, and the owner/operator). By simulating these key decision points,
the Model analysis can represent the dynamic aspects of a facility's life
cycle.
Below, we discuss seven facility milestones: facility begins handling
hazardous waste for land disposal; owner/operator applies for a final RCRA
permit; facility ceases accepting hazardous wastes for disposal; EPA
determines if the facility is eligible for PCLTF coverage; termination of all
business activity by the owner/operator; owner/operator stops performing
routine post-closure care and monitoring; and the end of the post-closure care
period.
Facility Begins Handling Hazardous Waste for Land Disposal
The year in which existing facilities opened (the year the facility began
accepting hazardous wastes for land disposal) was simulated based on Part A
arid Mail Survey data. The Mail Survey Data were used to derive a frequency
distribution of the open year (see Exhibit A-5). This distribution was used
to simulate an open year for those existing facilities which did not have Part
A Data describing their open year. Using this distribution, for example, a
facility has a 0.36 chance of having an open year between 1970 and 1979.
Both the Part A Data and Mail Survey Data were used to estimate a
relationship between the Part A Data and the Mail Survey responses based on
facilities that were included in both data sets (see Exhibit A-6). This
relationship was used to adjust, probabilistically, the Part A open year
responses to reflect the Mail Survey responses. For example, if a facility's
Part A open year response was 1965, it would have a 0.68 chance of having a
simulated open year between 1960 and 1969. It would also have a chance of
being opened between 1910 and 1919 (0.03), 1930 and 1939 (0.02), 1940 and 1949
(0.02), 1950 and 1959 (0.05), 1970 and 1979 (0.18), and 1980 and 1983 (0.02).
To determine the open year, a random number was drawn and compared to these
probabilities to identify the open year range. Then, a year was drawn from
the identified range. This approach enabled all the available data to be used
-------
A-14
EXHIBIT A-5
DISTRIBUTION OF OPEN YEARS REPORTED IN MAIL SURVEY DATA1
Range of Open Year Probability
1850-1899 . O.OO2
1900-1909 O.OO2
1910-1919 0.02
1920-1929 0.02
1930-1939 0.03
1940-1949 0.07
1950-1959 0.13
1960-1969 0.21
1970-1979 0.36
1980-1983 0.16
1.00
1 Based on 316 observations.
2 Less than 0.01.
-------
A-15
SO
UJ
O
O
W
C£
Q-
Z
LJ
O
LU
CD W
1 I
u w
LJ
EC
O
CO
o o
o <*> w
OS rO CM OS CV
O O O OS
0
ro Os f*- GO CO
ooo 1^0
oodddoodo
a- o
o
in fo oo os
o vo o o
J o
a- m
CM
in m
o in o o
o o o in o o o
ooooooooo
OOs^
O
d o
o
O CM
CM
VO
CV CO
O O O
O VO O
d o o d o" o o
m cv
ooomooooo
ooooooooo
O tsj
ooooooooo
ooooooooo
dddddo'ddd
Os
{SJOOOOOOOO
ooooooooo
oooooooo
do'ddddodd
CD
JJ
TO
Q
ID
t-
U5
0)
JC
n
o
0>
4J
O
£>
O
2
o
K
Q.
o oo it
cv m so os
u)
C
o
JJ
OJ
(D
Os
O
1
§
Os
o\
OS
1
o
o\
OS
CM
OS
I
O
CM
Os
ON
rO
O\
I
0
M
ON
OS
^
OS
1
o
a-
OS
Os
in
Os
1
0
in
ON
Os
SO
Os
r
O
VO
Os
OS
f-
Os
~
O
OS
tr>
CO
ON
1
o
CO
OS
CO
CM
C
o
a
CU
V)
as
CO
-------
A-16
and provided open year estimates which had less variance than if the Part A
Data had been excluded from consideration.
New facilities were simulated to begin operation as required to meet the
demand for land disposal. Consequently, the open year for a new facility is
simply the year in which it is simulated-to come on line.
Owner/Operator Applies for a Final RCRA Permit
Because all new facilities are required to open with a RCRA permit,
applying for a final permit is applicable only to owner/operators of existing
facilities.
The year in which an existing facility will apply for a final RCRA permit
was based on an assumed rate at which EPA will require and review permits at
existing facilities. Because EPA anticipates that all existing facilities
will be reviewed within the next five years, we simulated the following: 5
percent of the existing facilities will be requested to submit final permit
applications in each of the next two years, and 30 percent will be requested
to submit applications in each of the following three years. Each existing
facility has an equal chance of being requested to submit its application in
each of the five years. No relationship between other facility
characteristics (e.g., groundwater contamination) and the date of permit
application is assumed.
The determination of whether the facility is granted or denied a permit is
based on an assumed RCRA permitting policy and the simulated groundwater
quality at the facility. In the current simulations, evidence of groundwater
contamination was not assumed to result in a permit being denied. However,
alternative permitting policies may be examined; these are discussed in
Section A.3.2, Regulatory Policy.
Facility Ceases Accepting Hazardous Wastes for Disposal
The year in which a facility ceases accepting and disposing of hazardous
wastes is called the year of closure. Closure marks the beginning of two
important time periods:
Qualification Period: Period during which the
facility is monitored to determine if it can transfer
liabilities to the PCLTF at the end of this period; and
Post-Closure Period: The period during which the
owner/operator is required by his permit to perform
post-closure care. (Note: facilities may excavate and
decontaminate their land disposal processes at closure,
thereby avoiding any post-closure requirements.)
The closure year was determined by simulating a facility lifetime and
adding it to the year in which land disposal was simulated to have begun. The
facility lifetime was simulated by drawing from a lognormal distribution of
facility life developed from 174 responses to the Mail Survey. Exhibit A-7
-------
A-17
EXHIBIT A-7
FACILITY LIFETIMES
Range of
Expected Lifetime
(Years)
0-10
10-15
15-20
20-25
25-30
30-35
35-45
45-55
55-70
70-100
100 -
Observed Frequency1
18
17
18
11
17
10
20
15
11
16
21
174
Expected Frequency2
11.3
16.3
18.0
17.4
15.7
13.5
21.5
15.5
15.3
15.5
14.0
174.0
1 Based on 174 observations from Mail Survey.
* Estimated using lognormal distribution (3.49651, 0.7876, 0.0).
-------
A-18
compares the observed lifetime distribution (from the Mail Survey) to the
expected frequency of lifetime observations attained using a three-parameter
lognormal distribution with a mean of 45 years and a median of 33.* A
chi-square statistic of 12.2 was obtained from the difference between the two
distributions shown in Exhibit A-7, indicating that the hypothesis that the
observed distribution was drawn from the hypothesized lognormal distribution"
cannot be rejected at the 10 percent level of significance. Consequently,
because the continuous log normal distribution reflects the available
empirical data, it is used to model the closure year. Based on discussions
with EPA, it was believed that new facilities would have longer lifetimes than
indicated by the data used here to describe the lifetime of existing
facilities. Consequently, the lifetimes of new facilities were simulated by
drawing from the above distribution and then adding 15 years.
EPA Determines if the Facility is Eligible for PCLTF Coverage
At this milestone, EPA must determine whether closed facilities will
qualify for PCLTF coverage. In the current PCLTF statute, this time is five
years following closure. Prior to the five years after closure, the
facility's PCLTF qualification is said to be "pending." At the time when
qualification is determined, the facility becomes either "qualified" or
"unqualified."
The Model is configured to allow the timing of this qualification
determination to vary. The user specifies the number of years following
closure when the determination is initially made at all facilities. (The
value used must be greater than or equal to zero.) The same time period must
apply to all facilities.
The decision rules upon which a qualification determination is based must
reflect the CERCLA requirements for coverage.5 However, the regulations
interpreting the requirements have not yet been promulgated by EPA.
Consequently, the Model is designed to allow a variety of decision rules to be
examined. In the current simulations, any evidence of a release disqualifies
a facility from Fund coverage. Section A.3.4, Cost Allocation Policy,
describes in detail how the decision rules were constructed and applied.
Termination of All Business Activity by the Owner/Operator
At any point in time, an owner/operator may terminate all business
activity by declaring bankruptcy. At bankruptcy, the owner/operator's
facility (or facilities) cease accepting hazardous waste for disposal (if they
* The lognormal parameters are: v = 3.49651, o = 0.7876, and
minimum ~ 0. Note: the mean and median of the distribution are defined as:
mean = exp(v + l/2o2) -I- minimum; and
median = exp(y) + minimum.
5 See CERCLA, section 107(k).
-------
A-19
are in operation) and the owner/operator is no longer an available funding
source for costs. If an alternative funding source is not available to cover
ongoing actions, then the actions are terminated.
The rate at which bankruptcies occur is based on the financial
characteristics of the firms owning hazardous wastes disposal sites and on the
costs simulated to be incurred by these firms. In this section, we discuss
these financial characteristics. Section A.3.3, Financial Relationships,
discusses how bankruptcies are simulated using the financial characteristics
and the simulated costs.
Firms are classified according to several financial characteristics: net
worth, total assets, ownership status, and number of facilities owned.
Exhibit A-8 -presents the distribution of firms currently owning land disposal
facilities. The exhibit shows 61 types of facility owners. Each type is
defined by its probability of occurrence and its characteristics (net worth,
etc.). The net worth and total assets are used to simulate the firm's
likelihood of bankruptcy.6 A firm's ownership status may be government
owned or non-government owned. Government ownership may be at the federal,
state or municipal level. As shown in Exhibit A-8, government-owned
facilities do not have a net worth or total assets associated with them.
Because government-owned facilities are assumed to not go bankrupt, these
values are not required. Non-government ownership includes private ownership,
public ownership, and foreign ownership. Firm types 1 thru 58 in Exhibit A-8
indicate non-government ownership. Firm types 59 thru 61 indicate federal,
state and municipal government ownership, respectively. If the facility's
owner is the government, it does not go bankrupt.
Firm financial characteristics were estimated based on an analysis of
owners of active hazardous waste land disposal facilities with one or more of
the following processes: surface impoundments; waste piles; land treatment
units; or landfills. These facility types are not precisely the same as those
simulated in the PCLTF Model, which includes injection wells and does not
include waste piles.
The characteristics of the owner/operator for an existing facility are
simulated by randomly drawing a firm type from the distribution of owners
presented in Exhibit A-8. For example, 2.24 percent of the existing
facilities are owned by firms with total net worth between $0 and $10 million
and total assets between $2.5 and $2.9 million (firm type 5). Owner/operators
of new facilities are assumed to be larger than owner/operators of existing
facilities. For the purpose of modeling the owners of new facilities, the
distribution of existing facilities owned by particular firms was changed so
that no new facilities will have owners with less than $10 million in total
assets. The probabilities for all firms of greater than $10 million in total.
assets were increased to correct for this change. The net worth and total
6 The simulation of bankruptcy is discussed in Section A.3.3, Financial
Relationships.
-------
A-20
EXHIBIT A-8
DISTRIBUTION OF FIRM CHARACTERISTICS
Firm
Type
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Prob for
Existing
Facilities
.0184
.0281
.0281
.0105
,0224
.0416
.0058
.0058
.0058
.0261
.0019
.0252
.0058
.0068
.0019
.0058
.0048
.0010
.0068
.0019
.0310
.0116
.0029
.0058
.0164
.0058
.0029
.0068
.0174
.0010
.0425
.0271
.0232
.0029
Prob for
New Facil-
ities
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0312
.0072
.0085
.0024
.0072
.0060
.0000
.0084
.0024
.0385
.0144
.0036
.0072
.0204
.0072
.0036
.0085
.0216
.0012
.0528
.0337
.0288
.0036
Net Worth
(Millions)
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
0.-
10.-
10.-
10.-
10.-
10.-
10.-
10.-
10.-
10.-
10.-
10.-
10.-
100.-
100.-
100.-
100.-
100.-
10.
10.
10.
10.
10.
10.
10.
10.
10.
10.
10.
10.
10.
10.
10.
10.
10.
100.
100.
100.
100.
100.
100.
100.
100.
100.
100.
100.
100.
1000.
1000.
1000.
1000.
1000.
Total Assets
(Millions)
0.0-
0.5-
1.0-
2.0-
2.4-
3.0-
3.2-
3.0-
3.0-
5.1-
7.8-
10.2-
12.4-
21.2-
22.8-
75.8-
146.2-
7.5-
18.0-
16.3-
20.0-
25.4-
44.7-
35.4-
51.1-
58.0-
. 84.7-
64.7-
107.3-
11.5-
100.3-
151.1-
150.5-
367.4-
0.5
1.0
2.0
2.4
2.9
5.0
5.0
4.9
3.0
9.5
7.8
20.0
12.8
31.1
22.8
152.5
146.2
7.5
20.0
16.3
47.3
39.8
44.7
35.4
99.4
96.2
84.7
64.7
202.5
11.5
455.4
496.9
482.6
367.4
Number of-
Type of Facilities
Owner
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Owned
1
1
1
1
1
1
2
3
6
1
2
1
2
1
2
1
5
1
1
2
1
2
3
6
1
2
3
7
2
1
1
1
2
3
-------
A-21
EXHIBIT A-8 (continued)
DISTRIBUTION OF FIRM CHARACTERISTICS
Firm
Type
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
Prob for
Existing
Facilities
.0232
.0135
.0087
.0039
.0251
.0145
.0039
.0048
.0087
.0097
.0155
.0010
.0280
.0290
.0290
.0377
.0348
.0242
.0522
.0068
.0464
.0193
.0270
.0155
.0339
.0087
.0232
Prob for
New Facil-
ities
.0288
.0168
.0108
.0049
.0312
.0180
.0049
.0060
.0108
.0121
.0193
.0012
.0348
.0361
.0361
.0469
.0433
.0301
.0649
.0085
.0576
.0240
.0336
.0192
.0421
.0108
.0288
Net Worth
(Millions)
100.- 1000.
100.- 1000.
100.- 1000.
100.- 1000.
100.- 1000.
100.- 1000.
100.- 1000.
100.- 1000.
100.- 1000.
100.- 1000.
100.- 1000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
1000. -20000.
0.- 0,
0.- 0.
0.- 0.
Total Assets
(Millions)
502.6- 993.7
526.3- 787.0
661.7- 709.4
710.3- 710.3
1048.4- 3272.7
1111.1- 2117.4
1594.2- 1594.2
1636.0- 1636.0
1274.9- 1274.9
3908.9- 3908.9
1192.9- 1192.9
450.0- 450.0
1053,2-12417.0
1555.0-16016.0
1708.4-13729.4
3033.4-21632.8
2038.2-20614.0
2001.4-21961.7
2862.0-62288.6
6060.9- 6060.9
2048.0-27114.0
6077.0-19432.0
10616.0-41397.8
21615.0-21615.0
0.0- 0.0
0.0- 0.0
0.0- 0.0
Type of
Owner
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Private
Federal
State
Municipal
Number of
Facilities
Owned
1
2
3
4
2
3
4
5
9
10
16
1
.1
1
2
3
4
5
6
7
8
10
14
16
1
1
1
Source: EPA estimates.
-------
A-22
assets for the firm were simulated by drawing between the lower and upper
bounds of the particular ranges as shown in Exhibit A-8.
Owner/Operator Stops Performing Routine Post-Closure Care and
Monitoring
Post-closure care refers to the routine monitoring and care activities
undertaken after a facility ceases accepting waste for land disposal and is
closed. Each facility owner/operator is required to demonstrate financial
assurance for performing post-closure care during the post-closure period
(which is currently, by regulation, 30 years). However, in some cases, the
owner/operator may stop performing this care before the end of the 30-year
period because of insufficient funds. When the owner/operator stops
performing the care must be simulated in order to estimate potential PCLTF
expenditures because under certain assumptions, the PCLTF may pick up these
costs and because if the owner/operator stops too soon, it may affect the
facility's qualification for Fund coverage.
When an owner/operator stops performing post-closure care is a function of
the type of financial assurance mechanism the owner/operator used. There are
primarily three types of financial assurance mechanisms which the
owner/operator may use to demonstrate his ability to meet the costs of
post-closure care:7
Financial Ratio/Assets Tests: used by large firms to
demonstrate that they are financially able to cover the costs of
post-closure care. The owner/operator pays for the necessary
care actions during the post-closure period. If the owner/
operator terminates business, there are no funds set aside
specifically for the post-closure care activities. However,
firms that are able to pass the financial ratio/assets tests are
less likely than average to go out of business. Also, the costs
of post-closure care would be a liability for the company when it
went out of business. Therefore, a settlement to provide funds
for the remaining post-closure care costs may be obtained by EPA
if the owner/operator goes out of business before the end of the
post-closure period.
Trust Fund/Insurance: sets monies aside for use during the
post-closure period. During the post-closure period, the
available funds are used as needed. If the trust fund or
insurance is insufficient to cover costs for the entire period,
the responsibility to do so remains with the owner/operator. If
the owner/operator goes out of business prior to when the trust
fund/insurance runs out, then he is not available to cover these
costs. If, however, the owner/operator is still in business, he
may cover the costs of post-closure care until the end of the
post-closure period or the time he goes out of business. If the
For the financial assurance regulations, see 40 CFR 264, Subpart H.
-------
A-23
owner/operator goes out of business after the trust
fund/insurance runs out but before the end of the post-closure
period, a settlement to provide funds for the remainder of the
post-closure period may be obtained by EPA.
Letter of Credit/Surety Bond: a separate party (e.g., a
bank) agrees to guarantee that the owner/operator is able to
cover the costs of post-closure care. During the post-closure
period, the owner/operator pays the necessary care costs.
However, if the owner/operator goes out of business, the
guarantor covers the costs of post-closure care up to the limit
of his guarantee.
To simulate the performance of these financial assurance mechanisms, the
following estimates are required:
the frequency with which firms in different net worth
categories use each mechanism;
the frequency with which each of the financial
assurance mechanisms is used overall;
the size of the settlement that EPA could expect when
the firm goes out of business (expressed as a percentage
of the remaining post-closure care costs for which the
owner/operator is liable); and
the expected duration of trust funds and insurance.
To estimate the frequency with which firms in various net worth categories
use each mechanism, we started with data describing the frequency with which
users of each mechanism were found to fall into each net worth category.
Exhibit A-9 shows, for example, that 0.526 (or 52.6 percent) of those
facilities using trust funds or insurance were found to have net worths of
less than $10 million.
The second piece of data required is the overall frequency with which each
financial assurance mechanism was found to be used. Using Mail Survey
responses of approximately 225 land disposal facilities which reported the
mechanism used for post-closure care, the following distribution was obtained:
Financial Ratio/Corporate Guarantee: 63.3 percent;
Trust Fund/Insurance: 12.8 percent;
Surety Bond/Letter of Credit: 11.5 percent; and
Other: 12.4 percent.
After dropping the "other" category, the resulting percentages for the first
three categories are 72.2, 14.6, and 13.2, respectively.
-------
A-24
EXHIBIT A-9
FREQUENCY DISTRIBUTIONS OF OWNER/OPERATOR NET WORTH
BY FINANCIAL ASSURANCE MECHANISM USED
Financial Ratio/Assets Tests:
RANGE OF NET WORTH
(MILLIONS OF $)
0.1
10.0
100.0
1000.0
10.0
100.0
1000.0
5000.0
Trust Fund/Insurance:
RANGE OF NET WORTH
(MILLIONS OF $)
0.1 - 10.0
10.0 - 100.0
100.0 - 1000.0
1000.0 - 5000.0
Surety Bond/Letter of Credit:
RANGE OF NET WORTH
(MILLIONS OF $)
0.1
10.0
100.0
1000.0
10.0
100.0
1000.0
5000.0
PROBABILITY
0.000
0.038
0.308
0.654
PROBABILITY
0.526
0.211
0.105
0.158
PROBABILITY
0.526
0.211
0.105
0.158
SOURCE: ICF Memorandum to EPA, "Financial Data on Firms Owning Land Disposal
Facilities," July 29, 1983.
-------
A-25
Finally, the distributions of financial assurance mechanisms for each net
worth category were derived by applying Bayes' Theorem.8 For example, for
the category of net worth greater than $1,000 million, the probability of each
mechanism was calculated as follows:
Probability of Financial Ratio/Assets Tests =
(0.654)(0.722)/(0.516)9=0.915;
Probability of Trust Fund/Insurance =
(0.158)(0.146)/(0.156) = 0.045; and
Probability of Letter of Credit/Surety Bond =
(0.158)(0.132)/(0.516) = 0.040.
The results of these computations are displayed in Exhibit A-10. Given
the simulated net worth of each facility's owner, the financial assurance
mechanism was simulated by drawing from the appropriate distribution shown in
the exhibit.
Based on discussions with EPA, a settlement fraction of 0.10 was adopted
for each of the financial assurance mechanisms. The expected duration of the
trust funds (Exhibit A-ll) was assumed to follow an exponential distribution
and is based on discussions with EPA. There is considerable uncertainty
surrounding the duration that post-closure care trust fund/insurance may be
anticipated to last.
To simulate when an owner/operator stops performing post-closure care, the
data described above were used as follows:
using the distribution of financial assurance
mechanisms corresponding to the owner/operator's net
worth, assign a financial assurance mechanism to the
facility;
if financial assurance group 2 was assigned, simulate
the duration of the trust fund;
Bayes Theorem is:
P(H/E) = P(E/H)-P(H)
P(E)
where the variables E and H represent net worth and financial assurance
mechanisms, respectively.
9 The probability of a net worth greater than $1,000 million is derived
from these financial assurance data as:
(0.722*0.654 + 0.158*0.146 + 0.158*0.132) = 0.516.
-------
A-26
EXHIBIT A-IO
DISTRIBUTIONS OF FINANCIAL ASSURANCE MECHANISMS
BY NET WORTH CATEGORY1
Net Worth
(millions 1982 Financial Ratio/ Trust Fund/ Letter of Credit/
dollars) Assets Tests Insurance Surety Bond
less than 10 0.0 0.525 0.475
10-100 0.314 0.360 0.326
100-1000 0.884 0.060 0.056
greater than 1000 0.915 0.045 0.040
1 Rows sum to 1.0.
Source: See text.
-------
A-27
EXHIBIT A-ll
FREQUENCY DISTRIBUTION OF TRUST FUND DURATIONS
Duration (years) Probability
21 0.011
22 0.007
23 0.012
24 0.020
25 0.032
26 0.053
27 0.088
28 0.145
29 0.238
30 0.394
SOURCE: Based on discussions with EPA.
-------
A-28
the owner/operator performs care for:
at a minimum, the duration of the trust
fund/insurance (if he uses this mechanism), longer
if he is still in business when the trust
fund/insurance runs out;
-- the entire post-closure period (30 years) if he was
not simulated to terminate business prior to this
time; or
as long as he stays in business plus the settlement
percentage times the remaining time in the
post-closure period if he was simulated to terminate
business prior to the end of the post-closure period.
This procedure simulates the length of time the owner/operator is expected
to be able to cover post-closure care costs as currently required by
regulation. This estimate is made for each facility when it is first
considered by the Model, i.e., before releases and response actions are
simulated. In the majority of cases, this estimate for the facility remains
unchanged over time. However, in some cases, a situation may arise which
results in the owner/operator going bankrupt before the time simulated here.
In these cases, the new bankruptcy date is used to adjust the estimate of when
the owner/operator stops paying for post-closure care. -The method used to
simulate these special case bankruptcies is described in Section A.3.3,
Financial Relationships.
End of Post-Closure Care Period
As described above, the owner/operator's responsibility for covering
post-closure care costs only extends through the post-closure period. We
determined the end of the post-closure period by adding 30 years to the
closure year. The 30-year post-closure period may be adjusted on a
case-by-case basis. For modeling purposes, when a non-routine monitoring or
response action is ongoing at the end of the 30-year period, the post-closure
period is extended until the action is completed.
A.2.2.2 Facility Attributes
This section describes the facility-level attributes simulated in the
PCLTF Simulation Model. Each attribute is described in terms of its role in
the Model, the data analysis performed for estimating it, and the procedure
used to simulate it in the Model. Exhibit A-12 summarizes the attributes and
displays the values each attribute can have and the bases for the
facility-level estimates of the attributes. For example, the facility
location is defined in terms of the state, county, zip code, and groundwater
region in which the facility is simulated to be located. The simulation of
the location is based on the observed distribution of existing facilities
throughout the country.
-------
A-29
EXHIBIT A-12
ATTRIBUTES SIMULATED AT THE FACILITY LEVEL
Facility-Level
Attribute
Location
Disposal Processes:
Size and Design
Potential Releases
and Associated Dis-
tance to Off-Site
Detection Location
Detected Releases
Status of Monitor-
ing, Response, and
and Post-Closure
Actions
Status of Claims
Value of Attribute
State; county; zip code;
groundwater region.
Size in acres;
prototype design.1
Year after facility opening
in which release is expected
to occur; distance to off-
site detection location
(in feet).
True or false for each
release type.
Year action began;
duration of action;
cost of action.
Cost of claims brought.
Population Which May
Potentially Become
Exposed to Hazardous
Waste Constituents
Due to an Off-site
Release
Number of people.
Basis for Simulation
Distribution of existing
facility locations.
Distribution of sizes and
designs of existing
facilities; expected
distributions for new
facilities.
Release data; facility
location; process design.
Potential releases;
monitoring actions.
Detected releases; RCRA
policy; action data.
Detected releases;
location; population
which may potentially
become exposed; claim
data.
Distribution of ground-
water and surface water
use near disposal facili-
ties; number of people
drinking surface water and
groundwater supplied near
the facility location.
1 There are seven prototype designs incorporated into the Model.
Appendix B.
See
-------
A-30
Of note is that some simulated attributes form the basis for simulating
other attributes. For example, facility location and process design serve as
the bases for simulating potential releases. Furthermore, potential releases
form the basis (along with monitoring actions) for simulating detected
releases. Detected releases are important for simulating the actions
(monitoring and response) which must be taken at the facility. Clearly, the "
attributes simulated at the facility level are highly interrelated, and many
inter-relationships can be identified from Exhibit A-12. Each attribute is
described separately.
Location
The location of each facility is defined in terms of the following:
County;
State;
Zip code (5 digit); and
Groundwater region.
Each of these location definitions is used within the Model to reference a
data base or a simulated value. For example, the county location is used to
reference data on population density. The state location is used to reference
the simulated state legal liability regime.18 Exhibit A-13 summarizes the
uses of each location definition.
Data from the Part A Data Base were used to simulate the locations of both
existing and new facilities. For the 2,600 facilities in the Model's facility
data base, the state was identified as the first two characters of the
facility identification number (which corresponds to the standard U.S. Post
Office state abbreviation) reported in the Part A Data Base. The county
identification code was also requested in the Part A Data Base and in roost
cases, it was reported. However, 101 county codes were missing, and were
filled in by using the reported facility address (city and street) to identify
the appropriate county from standard state and county maps. Additionally,
errors in reported county codes (i.e., county codes which do not exist in the
reported state) were identified and corrected.
The facility zip code was reported in the Part A data in nearly all
cases. Missing zip codes were identified using the reported facility address
(city and street) and the standard post office zip code reference book. The
groundwater region was identified using the facility latitude and longitude
reported in the Part A data. Missing and invalid values were filled in using
the reported facility address.11
10 Legal liability regimes are used to simulate the legal validity of
claims. These regimes may vary from state to state due to differences in
state statutes and state common law. The legal liability regimes are
discussed in section A.3.6, Legal Validity of Claims.
11 Invalid latitude and longitude values were identified as those
outside the lower 48 states.
-------
A-31
EXHIBIT A-13
SUMMARY OF USE OF EACH LOCATION DEFINITION
Location Definition Used to Identify
County Population density; median housing value; housing
density; mean farm value; percentage of area used as
farmland.1
State Simulated state legal liability regime; simulated
state coverage regime;2 hazardous waste market
region.3
Zip Code (5 digit) Number of people who drink water supplied from
within the zip code.*
Groundwater Region Release matrix for simulating releases at the
facility level.5
1 Data obtained from Bureau of the Census, The County and City Data
Book, 1977, U.S. Department of Commerce.
2 State legal liability regimes are used to simulate the legal validity
of claims; see section A.3.6, Legal Validity of Claims. State coverage
regimes define the funding available at the state level to cover costs arising
at land disposal facilities; see section A.2.6, Funding Sources.
3 Each state is assigned to one of four regional hazardous waste
disposal markets. These markets are used to balance the demand for and supply
of hazardous waste disposal capacity; see section A.3.1, Economic
Relationships.
* Data obtained from the Drinking Water Survey, Office of Drinking
Water, U.S. Environmental Protection Agency, 1981.
5 The lower 48 states are divided into 12 groundwater regions. These
regions are used in the Release Simulation Model (see Appendix B) and to
aggregate the release results into release matrices; see section A.2.3,
Modeling of Releases.
-------
A-32
After this data gathering was performed, all the necessary location
parameters were available for all 2,600 facilities included in the facility
data base. These data formed the pool of facilities for simulating the
population of existing facilities (see section A.2.1, Facility Population).
If a facility was simulated to be part of the existing facility population,
its location identified from the Part A Data (and from the corrections and
additions made) was used.
The locations of new facilities were simulated by drawing randomly from
the locations of existing facilities, as indicated in the Part A Data. The
Model is designed to allow the user to specify the extent to which new
facilities are opened at "better" locations than existing facilities. A
location is defined to be better if it has: (1) fewer people using the
groundwater or surface water near the facility; (2) lower property values near
the facility; and/or (3) a lower likelihood of release. A location that is
better along these dimensions is preferred because it is less likely to lead
to costs and damages.
The first two location characteristics (people using water and property
value) are modeled for new facilities by modifying the manner in which new
facility locations are drawn. Rather than draw from the full range of
existing facility locations, the user can define a portion of the range that
may be used. For example, the user may specify that only the top one-third of
the existing facility locations (in terms of ground and surface water use) may
be selected as new facility locations. This is accomplished by ranking the
existing facility locations according to the desired characteristics (lowest
to highest) and only drawing from the top one-third of the ranked locations.
If two characteristics are specified (both water use and property value), then
the allowable set of locations for new facilities may be limited further. In
current simulations, only the top one-third of existing locations in terms of
ground and surface water use were specified as allowable candidates for new
facility locations. No limitations on property value were used.
The manner in which releases at new facilities were limited is discused in
section A.2.3, Modeling of Releases. The method of limiting releases at new
facilities is independent of the method of limiting water use and property
value described here.
Disposal Processes
Each facility is simulated to have one or more of the following four
processes:lz
surface impoundment (storage, treatment, disposal);
* landfill;
land treatment; and
injection well.
12 Facilities may also have other hazardous waste management processes,
such as storage in containers and treatment in tanks. However, these
processes are not simulated in the PCLTF Simulation Model.
-------
A-33
The processes at each facility must be simulated because releases occur at
individual processes, and the sizes of the processes influence the costs of
monitoring, response, and post-closure care actions. At each facility, the
presence, size, and design of each process is simulated as described below.
Processes Present. Facilities simulated to be part of the facility
population (either as existing facilities or new facilities generated within
the Model) will have one or more of the four disposal processes. Using the
most recent data describing the population of existing land disposal
facilities,13 a frequency distribution was developed to describe the chance
of each process being used at each facility. The data analysis performed to
estimate this distribution is described in section A.2.1, Facility Population.
The estimated distribution is shown in Exhibit A-3 in that section. The
disposal processes present at each facility are simulated by drawing from this
distribution. No distinction is made between existing and new facilities.
Process Size. The sizes of the disposal processes present at each
facility are expressed in terms of their surface area and are measured in
acres. These sizes are used to estimate the costs of monitoring, response,
and post-closure ca're actions taken at the facility.1*
Sizes for processes at existing facilities were simulated based on
frequency distributions developed from the Mail Survey Data. Exhibit A-14
displays the estimated distributions. For surface impoundments, the total
area of all surface impoundments at each of 135 facilities was used to
estimate the distribution. For example, Exhibit A-14 shows that 45.1 percent
of the facilities with surface impoundments have a total surface impoundment
area of between 0.1 and 1.0 acres. The landfill size distribution, based on
44 Mail Survey responses, was estimated by multiplying the facility's reported
average cell size by the total number of cells reported. The size
distribution for land treatment processes, based on only 30 values, was
developed directly from the Mail Survey responses, which reported the surface
area in acres.
The sizes of processes at existing facilities were simulated by drawing
from these distributions. First, the size range was determined by drawing a
random number and comparing it to its probabilities of falling into the
various possible size ranges. Then, a size was drawn from a uniform
disribution bounded by the end points of the chosen size range.
No data were available upon which to estimate sizes of processes at new
facilities. We assumed that new landfills, surface impoundments and land
treatment processes would be larger than existing ones and estimated the size
distributions of these processes by dropping the smallest size categories in
the distributions for existing facilities. These assumed distributions are
13 The data sources used were the Part A Data, the Telephone
Verification Survey Data, and the Mail Survey Data.
l* See section A.2.4, Actions, for a description of the costs of actions.
-------
A-34
EXHIBIT A-14
FREQUENCY DISTRIBUTIONS OF PROCESS SIZES
FOR EXISTING FACILITIES
Surface Impoundments:1 mean = 3.35 acres; median = 0.83 acres
Size Range
(acres) Probability
0.01-0.1
0.
.0
1-1.
1.0-3.0
3.0-7.0
7.0-9.0
9.0-10.0
10.0-26.0
26.0-100.0
0.143
0.451
0.181
0.128
0.007
0.030
0.045
0.015
1.000
Landfills:2 mean = 46.0 acres; median = 9.2 acres
Size Range
(acres) Probability
0.01-0.50
0.5-4.1
4.1-6.8
6.8-14.7
14.7-27.5
27.5-142.0
142.0-468.0
0.18
0.24
0.02
0.20
0.11
0.18
0.07
1.00
Land Treatment:3 mean = 36.5 acres; median = 10.0 acres
Size Range
(acres) Probability
0.1-1.0
1.0-7.0
7.0-10.0
10.0-13.0
13.0-15.0
15.0-60.0
60.0-100.0
100.0-188.0
188.0-680.0
0.100
0.300
0.100
0.033
0.067
0.300
0.033
0.034
0.033
1.000
1 Based on 135 observations from the Mail Survey.
2 Based on 44 observations from the Mail Survey.
3 Based on 30 observations from the Mail Survey.
-------
A-35
displayed in Exhibit A-15. The sizes of processes at new facilities were
drawn from these distributions.
Process Design. A process design refers to the type of cap and liner
the process employs and whether it has a leachate collection system. The
design of each process is used to simulate releases at the facility level.
The following seven prototypes designs were developed:15
* Design 1: unlined surface impoundment with a clay
cap following closure;
Design 2: clay-lined surface impoundment with a
clay cap following closure;
Design 3: synthetic membrane and clay cap,
synthetic membrane liner, leachate collection system,
bedding material under synthetic membrane liner;
Design 4: synthetic membrane and clay cap, two
synthetic liners, leak detection, and leachate
collection system;
Design 5: two synthetic membranes and clay cap,
synthetic membrane .and clay outer liner, clay inner
liner, and leachate collection system;
Design 6: unlined landfill with a clay cap
following closure; and
Design 7: clay-lined landfill with a clay cap
following closure.
Injection wells and land treatment processes were not assigned a design
because releases at injection wells and land treatment processes were not
simulated. Because releases may in fact occur at these processes, the Model
may underestimate total releases. The responses to the Mail Survey were used
to estimate the frequency with which these designs occur at existing surface
impoundments and existing landfills.
The design data reported in the Mail Survey for surface impoundments and
landfills were limited primarily to the liner materials used. Exhibit A-16
displays the data obtained and the prototype designs assumed to be
representative of the Mail Survey responses. For processes with a single
synthetic membrane liner, design 3 was used. For processes with either two
synthetic membrane liners or one synthetic membrane liner and one clay liner,
design types 4 and 5, respectively, were used. Using the observations
i5 The prototype designs are defined in detail in Appendix B, Release
Simulation Model.
-------
A-36
EXHIBIT A-15
FREQUENCY DISTRIBUTIONS OF PROCESS SIZES
FOR NEW FACILITIES
Surface Impoundments:1
Size Range
(acres)
0.01-0.1
0.1-1.0
1.0-3.0
3.0-7.0
7.0-9.0
9.0-10.0
10.0-26.0
26.0-100.0
Landfills:2
Size Range
(acres)
0.01-0.50
0.5-4.1
4.1-6.8
6.8-14.7
14.7-27.5
27.5-142.0
142.0-468.
Land Treatment:3
Size Range
(acres)
0.1-1.0
1.0-7.0
7.0-10.0
10.0-13.0
13.0-15.0
15.0-60.0
60.0-100.0
100.0-188.0
188.0-680.0
Probability
0.000
0.526
0.211
0.149
0.008
0.035
0.053
0.018
1.00
Probability
0.00
0.29
0.03
0.24
0.13
0.22
0.09
1.00
Probability
0.000
0.333
0.111
0.037
0.074
0.333
0.037
0.038
0.037
1.000
1 Estimated by eliminating smallest size category from distribution of
existing surface impoundment sizes.
2 Estimated by eliminating smallest size category from distribution of
existing landfill sizes.
3 Estimated by eliminating smallest size category from distribution of
existing land treatment sizes.
-------
A-37
EXHIBIT A-16
DESIGN TYPES
Surface Impoundments
Surface Impoundment
Liner Material(s) Assumed Design Type1 Responses (Percent)
Unlined 1 126
Clay-lined 2 82
Synthetic membrane 3 66
Two synthetic membranes 4 14
Synthetic membrane and clay 5 5
Other 2 28
321
Landfills
Landfill
Liner Material(s) Assumed Design Type1 Responses (Percent)
Unlined 6 37 47.4
Clay-lined 7 31 39.7
Synthetic membrane 3 3 3.8
Two synthetic membranes 4 1 1.3
Synthetic membrane and clay 5 0 0.0
Other 7 _6 7.8
78 100.0
Refers to prototype designs, see text.
-------
A-38
reported in Exhibit A-16, we assigned the following frequencies to the designs
at existing surface impoundments:
Design 1: 0.392;
Design 2: 0.342;
Design 3: 0.206;
Design 4: 0.044;
Design 5: 0.016;
Design 6: 0.0; and
Design 7: 0.0.
The observations reported in Exhibit A-16 for landfills were similarly
used to estimate a frequency distribution of designs, with the following
results:
Design
Design
Design
Design
Design
Design
Design
1:
2:
3:
4:
5:
6:
7:
0
0
0
0
0
0
0
-0;
.0;
.038;
.013;
.0;
,474;
.475.
and
The designs at existing landfills and surface impoundments were simulated by
drawing from these distributions.
No data were available for estimating design frequencies for new landfills
and surface impoundments. Prototype designs 3, 4 and 5 each meet existing EPA
design requirements for new facilities. Based on discussions with EPA, we
assumed that the frequencies of designs at new facilities would be:
Design 3: 0.340;
Design 4: 0.495; and
Des ign 5: 0.165.
The use of prototype designs and the lack of data upon which to
characterize the designs of new facility processes are clearly shortcomings of
the data currently used to run the Model. However, the results of the Release
Simulation Model indicated that the release potential is similar for designs
3, 4 and 5. Consequently, the Model results are not sensitive to the assumed
distribution of new facility designs (among designs 3, 4 and 5).
Releases: Potential and Detected
The purpose of this section is to identify the release information
maintained at the facility level. This information can be divided into two
parts:
descriptors of potential releases; and
identification of releases that have been detected.
-------
A-39
This information is used for a variety of purposes within the Model. In
particular, the information about detected releases identifies (along with
user-supplied action policies) the actions required at the facility. Not
discussed here are:
the definitions of the seven release types modeled
(see section A.2.3, Modeling of Releases);
the manner in which releases are simulated (see
section A.2.3, Modeling of Releases and Appendix B,
Release Simulation Model);
the effect that response actions have on releases (see
section A.3.5, Effect of Response Actions on Releases);
and
the manner in which monitoring and response actions
detect releases (see section A.2.4, Actions).
Potential Releases. Potential releases are the Model's estimate of
those releases expected to occur at the facility in the future. They are
determined by random draws from the release matrices provided by the Release
Simulation Model (see Section A.2.3 and Appendix B). The Model makes this
estimate when the facility is simulated to open, and then revises it as needed
when response actions are simulated to be taken which influence the potential
for future releases.
Potential releases at a facility are identified by the time they are
expected to occur and by their location. For example, the timing of one
release type may be identified as 30 years after opening, and the timing of a
second release type may be 50 years after opening. If a release is not
anticipated to occur at the facility, its potential time is set to the value
999.
The locations of releases can be either on site or off site. On-site
releases are assumed to be detected at the on-site monitoring wells. Off-site
releases are assumed to be detected at the potable well or naturally occurring
body of surface water closest to the facility. The distance to this off-site
point of detection (in feet) is simulated in the Release Simulation Model and
maintained at the facility level as a descriptor of potential releases.
Detected Releases. Detected releases are those releases that have been
simulated to occur and have been simulated to be detected. When the facility
opens, no releases have been detected. The Model updates its identification
of detected releases as releases are simulated to be detected. A release
remains in the detected state until all the facility's potential releases have
been detected and cleaned up. After the facility has been cleaned up, the
facility is returned to having no detected releases.
-------
A-40
Status of Monitoring, Response, and Post-Closure Care Actions
Actions are undertaken at a facility whenever a release is detected and at
the beginning of the post-closure period. Several facility-level parameters
are maintained for each action in order to describe its status over time.
Below, we present each of these parameters. A detailed description of the
actions and the methods used to estimate their costs is presented in Section
A.2.4, Actions.
The statuses of monitoring, response, and post-closure care actions are
summarized using the following parameters:
Year action began. The year after facility closure in which
the action began is recorded for each action. If the action was
never started or if the action was stopped, this parameter is set
to 999 or -999, respectively.
Capital investment. Each action has a capital investment
requirement. This includes planning for construction and
installing capital equipment (i.e., monitoring wells and wells
used for fluid removal and treatment). For each action, an
indicator is maintained as to whether the capital investment
required for that action has been completed. This indicator is
important because monitoring actions cannot detect releases until
the capital investment is completed. In addition, capital
investment occurs only once for each action at each facility.
When an action is undertaken for a second time, capital
investment is not required if this investment was previously
performed.
Action Completion. An indicator is used to denote if an
action has been performed for its full duration (i.e., has been
completed). (See Section A.2.4, Actions, for a description of
the durations of response actions.)
Annual Costs. The annual capital cost and annual operation
and maintenance (O&M) cost ,is maintained at the facility level
for each action.16 The annual capital cost is incurred while
capital investment is being performed. The O&M costs are
incurred for the duration of the action.
The above parameters are maintained for all actions. In addition, the
following information is maintained on response actions in order to simulate
their effect on releases:
16 The estimates for both capital and O&M costs are presented in
section A.2.4, Actions.
-------
A-41
Fraction of response action completed. When a response
action ends, its actual duration as a fraction of its required
duration is recorded. This fraction is used to assess whether
the response action is effective in preventing or delaying future
releases and in determining the required duration of the response
action if it is undertaken again.
Required duration. The duration for which a response action
must be performed may vary at the facility level. When a
response action is taken at a facility for the first time, its
required duration is assumed to be the period defined for the
action (see section A.2.4, Actions, for a description of the
durations of response actions). However, if a response action is
stopped before completion and subsequently restarted, its new
required duration is simulated to be less than its original
duration. In this manner, "credit" is given for having performed
the action for some period of time. The amount of credit given
is a function of the fraction of the action completed (the larger
the portion completed, the greater the credit). Because the
required duration of the response action varies depending on past
actions at the facility, the required duration is maintained at
the facility level.
Status of Third-Party Claims at the Facility
Claims are brought by third parties who believe they have been harmed by
an off-site release at a facility. The four claim types are:
Personal Injury. Medical costs, lost time due to
illness, lost time due to mortality, and medical monitoring
costs comprise personal injury claims.
Real Property. Real property claims reflect a
reduction in the value of housing and farmland near the
location of an off-site release.
Economic Loss. The capital cost of replacing a
contaminated source of drinking water was used to
approximate economic loss claims.
Natural Resource Damage Claims. The cost of restoring
small areas of contaminated surface water, the value of lost
recreation for one year, and the commercial value of
potential fish kills comprise the estimate of natural
resource damage.
The methods and data used to calculate these claims are described in section
A,2.5, Claims. This section describes the information on third-party claims
which is maintained at the facility level, and is divided into two parts:
cost information, and the uses of ground and surface water near the facility.
-------
A-42
Cost Information. The following three cost parameters are recorded at
the facility level for each of the four claim types:
Generated Claims: the claims (in 1982 dollars) that
have been brought by third parties as the result of
off-site releases.
Legally Valid Claims: that portion of generated
claims which is simulated to be legally valid under the
legal liability regime of the state in which the claims
were brought.17 The legally valid claims can be any
fraction of the generated claims.
Covered Claims: that portion of the legally valid
claims for which compensation is provided. If no
funding source is available to pay for a legally valid
claim, the claimant will not be compensated and the
covered claims will be zero.
This claims information is summarized during a Model run to identify the
overall fraction of generated claims which are simulated to be legally valid,
and the overall fraction of legally valid claims which are covered. For
example, in Simulation 1, approximately 18 percent of the total personal
injury claims were simulated not to be legally valid. Of the remaining
legally valid claims, funding sources were simulated to be available for
nearly 98 percent of the total claims.
The information on covered claims is also used at the facility level to
avoid double counting of claims costs. Two different off-site releases can
lead to third-party claims. When release type 5 (detectable concentration of
constituents off site) occurs, claims are estimated. If release type 6
subsequently occurs (toxic concentrations of constituents off-site), claims
are estimated again. However, both releases cause the same type of
harms.1* Therefore, to avoid double counting, the generated claims for
release type 6 are adjusted by subtracting the previously covered claims from
release type 5. Of note is the assumption that only covered claims need to be
subtracted, and not generated claims. This means that if a claimant was not
compensated for harms caused by release type 5, he would be able to make
another claim if release type 6 occurred subsequently.
Uses of Ground and Surface Water Near the Facility. To estimate
personal injury claims and economic loss claims, the uses of ground and
surface water near the facility must be simulated. Although a variety of uses
may be present, this analysis focuses on the number of people drinking the
17 The method used for simulating the legal validity of claims is
described in section A.3.6.
18 Release type 6, however, causes larger claims for real property
damage.
-------
A-43
water which originates from near the facility. This estimate is maintained at
the facility level and is used as a proxy for the number of people who
consider themselves harmed by the off-site release. Because there may be
other uses of the ground and surface water near the facility (e.g., industrial
and agricultural uses) which may be affected by the release, the estimates of
these claims may be biased downward.
Data describing the number of people drinking water originating from
nearby hazardous waste disposal facilities (referred to here as the drinking
water population) have not been assembled. Consequently, no single
authoritative data base was available upon which to estimate or simulate the
drinking water population at the facility level. To overcome this deficiency,
two EPA data sources were used jointly to simulate the necessary values: the
drinking water survey and the site visit survey. The following discussion
first identifies the data obtained from each survey. Then, the procedure used
to establish a relationship between the two data sources is described.
Finally, the method used to simulate the drinking water population using the
two data sources and the estimated relationship between them is presented.
The drinking water survey is compiled' annually by EPA's Office of Drinking
Water. Each year, all suppliers of drinking water must report a series of
information items to the appropriate state authorities,19 who then forward
the information to the federal EPA. Over 95 percent of all drinking water use
is reportedly included in this annual survey.
Considerable information is collected in the survey from each water supply
plant in the U.S. A water supply plant is the location where the drinking
water is treated (if necessary) and fed into the water distribution system.
Nearly all water treatment plants are adjacent to or very near the source of
the water used (i.e., the lake, stream, or wells from which the water is
obtained).20 Consequently, the data reported by the water supply plants
describing the number of people they serve were used in the PCLTF Simulation
Model as estimates of the number of people using water that originates from
the location of the water supply plant.
The most detailed data available to describe the location of the water
supply plants were their 5-digit zip codes. Because more than one water
supply plant may exist in a single 5-digit zip code, the data were aggregated
by zip code. The resulting data base describes the total number of people who
are served by water originating from each zip code. The people who are
reported as being served by the plant may not necessarily reside in the zip
code.
19 In states without authorized drinking water programs, the information
is reported to the appropriate regional EPA offices.
20 In the western United States, supply plants are occasionally far from
water sources. For example, water is pumped from Northern California to
supply plants in the southern portion of the state.
-------
A-44
The information on people served was reported by each water supply plant
in one of four water source categories:
1. surface water;
2. groundwater;
3. purchased surface water (i.e., surface water purchased
by the water supply plant); and
4. purchased groundwater (i.e., groundwater purchased by
the water supply plant).
If a single plant used more than one water source (for example, both surface
water and groundwater), the information was reported by the plant in the
category ranked highest on the above list (in this example, surface water).
However, because the data were aggregated within 5-digit zip codes for use in
the Model, a zip code could have values reported for more than one category
(because there can be more than one water supply plant in a zip code).
These data from the drinking water survey provide a comprehensive picture
of U.S. sources of drinking water. However, they do not identify the extent
to which water supplies may be influenced by releases from hazardous waste
land disposal facilities. In particular, 5-digit zip codes may be large
(especially in sparsely populated areas) and it is not realistic to assume
that a release from a hazardous waste land disposal facility would necessarily
harm the entire water supplies in the facility's zip code. A method was
required to assess the likelihood that releases would affect water supplies.
To address this need, the data collected by EPA in its site visit survey of
1982/1983 were used.
EPA identified 109 facilities for site visits. Of these, data were
reported on 63. A variety of data were collected in these visits, including
information on the proximity and use of ground and surface water. For ground-
water, data were collected on whether wells for the following uses were near
the facility:
single residential: wells owned and used by residents;
municipal: wells owned by a municipality and used to
supply water to the surrounding population; and
other: wells used for agriculture and industrial uses.
Information on the distance to groundwater wells was also reported in the
site visit survey. For each site, the closest well was identified as either
within 1 mile or farther away than 1 mile. However, much of these data were
missing or inconsistent and consequently were not used. Of those distances
reported, 65 percent were within 1 mile, and 35 percent exceeded 1 mile.
-------
A-45
For surface water, data were collected on whether the following uses of
surface water were within 1,000 feet, 1 mile, 5 miles or greater than 5 miles
of the facility:
drinking;
recreation (e.g., swimming and boating); and
commercial or industrial.
The data obtained from the site visits are summarized in Exhibit A-17. For
example, 26 of the 63 facilities reported single residential wells near the
facility. Similarly, 8 of the 63 facilities reported surface water within
1,000 feet being used for drinking.
These site visit data alone are not sufficient to estimate the drinking
water population required for claims estimates. The data are available for
only a small number of facilities, and they provide no indication of the
number of people whose drinking water may be affected by a release.
Consequently, a method was developed to combine the site visit survey data
with the drinking water survey data to provide a basis for simulating the
drinking water population.
To combine the two data sets, the following four steps were performed:
Step I: create mutually exclusive categories of
. groundwater use and surface water use based on the site
visit survey data. These categories, and the number
falling into each, are:
.-- Groundwater Use Well Type:
1. Single residential drinking: 22
2. Municipal drinking: 8
3. Other uses only (no drinking): 9
4. Single residential and municipal drinking: 4
5. No wells: 20
-- Surface Water Use:
1. Water present but not used: 6
2. Drinking: 2
3. Recreation: 29
4. Other uses only: 9
5. Drinking and recreation: 14
6. No surface water present: 3
Step 2: Create mutually exclusive categories of
responses to the drinking water survey. These
categories are:
-- Groundwater only (people served by groundwater or
purchased groundwater)
-- .Surface water only (people served by surface water
or purchased surface water)
-------
A-46
EXHIBIT A-17
SUMMARY OF SITE VISIT SURVEY DATA
ON GROUNDWATER AND SURFACE WATER USE
Groundwater Use1
Use
Single Residential
Municipal
Other
No Wells Reported
Number Reporting
26
12
9
20
67
Surface Water Use1
USE
Drinking
Recreation
Commercial/
Industrial
Not Used
~
TOTAL
1,000 feet
8
20
17
2
47
Distance
Occurring
1 mile
i
7
16
11
2
36
to Closest
Body of Sui
5 miles
1
7
5
2
15
Naturally
rface Water
5 miles
0
0
0
3
3
Total
16
43
33
9
101
1 More than one use can be reported by a single facility.
totals exceed the 63 responses reviewed.
Consequently,
Source: See text.
-------
A-47
Both ground and surface water
-- No response (no water supply plant in the zip code)
Step 3: For each response to the site survey,
identify the facility's 5-digit zip code. Using this
zip code, identify the category of response from the
drinking water survey. Of the 63 facilities in the site
survey: 22 were in zip codes with groundwater only; 5
were in surface water only; 20 were in both ground and
surface water; and 16 were in zip codes that had no
response on the drinking water survey.
Step 4: Using the results of steps 1 and 3, create
the following two sets of tables:
Well Types: given the drinking water survey
response category, identify the probability of
having each of the five possible groundwater uses.
Because distance data are not available, assume
these probabilities reflect usage patterns between
1,000 feet and 5 miles. Assume no usage within
1,000 feet and assume that usage at distances
exceeding 5 miles will not be affected. This set of
tables is displayed in Exhibit A-18.
Surface Water Use: given the drinking water
survey response category and the site survey
distance category, identify the probability of
having each of the six possible surface water uses.
This set of tables is displayed in Exhibit A-19.
The tables created in step 4 are a summary of the relationship between the
drinking water survey data and the site visit survey data. For example,
Exhibit A-18 shows that of those facilities in the groundwater category (of
the drinking water survey), 42 percent had well type 1 (single residential),
32 percent had well type 2, 11 percent had well type 3, and 15 percent had
well type 4. These probabilities were assumed to apply to distance categories
2 and 3 (1,000 feet to 5 miles). It was further assumed that there would be
no wells within 1,000 feet (consequently, there is a 100 percent chance of
well type 5, no wells, for this distance) and that wells in excess of 5 miles
would not be affected by releases.
Exhibit A-19 shows a similar relationship for surface water use. However,
distance information was available in this case. For example, of those
facilities in the groundwater category and in distance category 1, 83 percent
had surface water use 3 (recreational use) and 17 percent had surface water
use 4 (other uses only).
These two exhibits display considerable consistency between the two data
sets. For example, in Exhibit A-18, there are no entries for groundwater
drinking (well types 2 and 5) in the table for surface water responses to the
-------
A-48
EXHIBIT A-18
FREQUENCY DISTRIBUTIONS OF WELL TYPES1*
BY DRINKING WATER SURVEY RESPONSE AND DISTANCE2
TO OFF-SITE POINT OF DETECTION
Response to Drinking Water Survey Probability of Well Type by Distance
Groundwater*
Distance to Facility
Well type
Surface Water1*
Well type
1
2
3
4
5
Both Groundwater and Surface Water5
Well type
1
2
3
4
5
No Response*
Well type
1
2
3
4
5
0
0
0
0
1
1
.000
.000
.000
.000
.000
0
0
0
0
0
2
.420
.320
.110
.150
.000
Distance to
0
0
0
0
1
1
.000
.000
.000
.000
.000
0
0
0
0
0
2
.500
.000
.500
.000
.000
Distance to
0
0
0
0
1
1
.000
.000
.000
.000
.000
0
0
0
0
0
2
.720
.070
.140
.070
.000
Distance to
0
0
0
0
1
1
.000
.000
.000
.000
.000
0
0
0
0
0
2
.290
.140
.570
.000
.000
0
0
0
0
0
3
.420
.320
.110
.150
.000
0
0
0
0
1
4
.000
.000
.000
.000
.000
Facility
0
0
0
0
0
3
.500
.000
.500
.000
.000
0
0
0
0
1
4
.000
.000
.000
.000
.000
Facility
0
0
0
0
0
3
.720
.070
.140
.070
.000
0
0
0
0
1
4
.000
.000
.000
.000
.000
Facility
0
0
0
0
0
3
.290
.140
.480
.000
.000
0
0
0
0
1
4
.000
.000
.000
.000
.000
* Footnotes on following page.
-------
A-49
FOOTNOTES TO EXHIBIT A-18
See text for definition of well types.
Distances are defined as follows:
1. less than 1,000 feet
2. 1,000 feet to 5,280 feet
3. 5,280 feet to 26,400 feet
4. greater than 26,400 feet
Based on 22 observations from the site survey which also had a drinking
water supply survey response.
Based on five observations from the site survey which also had a
drinking water supply survey response.
Based on 20 observations from the site survey which also had a drinking
water supply survey response.
Based on 16 observations from the site survey which did not have a
drinking water supply survey response.
Source: See text.
-------
A-50
EXHIBIT A-19
FREQUENCY DISTRIBUTIONS OF SURFACE WATER USE1*
BY DRINKING WATER SURVEY RESPONSE AND DISTANCE2
Response to Drinking Water Survey Probability of Surface Water Use By Distance
Groundwater3 '
Distance to Facility
Surface Water Use
Surface Water"
1
2
Surface Water Use 3
4
5
6
Both Groundwater and Surface Water5
Surface Water Use
1
2
3
4
5
6
No Response*
Surface Water Use
1
2
3
4
5
6
I
0.000
0.000
0.830
0.170
0.000
0.000
2
0.126
0.000
0.500
0.000
0.374
0.000
Distance to
1
0.000
0 . 000
0.333
0.000
0.667
0.000
5
2
0.000
0.000
0.000
0.000
0.000
1.000
Distance to
1
0.241
0.120
0.482
0.120
0.037
0.000
2
0.000
0.000
0.000
0.000
1.000
0.000
Distance to
1
0.000
0.000
0.375
0.375
0.250
0.000
2
0.167
0.000
0.333
0.333
0.167
0.000
3 4
0.168 1.000
0.000 0.000
0.661 0.000
0.168 0.000
0.000 0.000
0.000 0.000
Facility
3 4
0.000 0.000
0.000 0.000
0.500 0.000
0.500 0.000
0.000 0.000
0.000 1.000
Facility
3 4
0.000 0.000
1.000 0.000
0.000 0.000
0.000 0.000
0.000 0.000
0.000 1.000
Facility
3 4
1.000 1.000
0.000 0.000
0.000 0.000
0.000 0.000
0.000 0.000
0.000 0.000
* Footnotes on following page.
-------
A-51
FOOTNOTES TO EXHIBIT A-19
See text for definition of surface water uses.
Distances grouped as follows:
1. less than 1,000 feet
2. 1,000 feet to 5,280 feet
3. 5,280 feet to 26,400 feet
4. greater than 26,400 feet
Based on 22 observations from the site survey which also had a drinking
water supply survey response.
Based on five observations from site survey which also had a drinking
water supply survey response.
Based on 20 observations from the site survey which also had a drinking
water supply survey response.
Based on 16 observations from the site survey which also had a drinking
water supply survey response.
Source: See text.
-------
A-52
drinking water survey. Some inconsistencies do exist. For example, in
Exhibit A-19, the first table showing the groundwater responses to the
drinking water survey shows an entry of 0.374 in use type 5 (drinking and
recreation) in distance category 2 (1,000 feet to 1 mile). However, no
surface water was reported used for drinking in this zip code (as is evidenced
by it being in the groundwater category). Consequently, these responses are'
inconsistent.
The inconsistencies may be due to a variety of factors, including:
The drinking water usage reported in the site survey
may not meet the definition of water supply used in the
drinking water survey (e.g., the water supply may not be
used year round).
The responses to the drinking water survey are
hierarchical. As discussed above, if both surface and
groundwater are supplied by a plant, the plant's report
is placed under the surface water category.
The inconsistencies reflect the fact that using the drinking water survey data
may result in an underestimate of the drinking water population because some
water uses near facilities may not be reported.
The tables displayed in Exhibits A-18 and A-19 were used to simulate the
drinking water population at each facility as follows:
Identify the drinking water survey response category for the
facility using the facility's zip code. The number of people
served by groundwater and surface water are recorded. (If there
is no response, the facility is in the no response category.)
Identify the distance to the off-site point of detection from
the release matrix (see the section above on potential and
detected releases).
Using the drinking water survey response category and the
distance, simulate the well type and surface water use at the
facility by drawing a random number for each and comparing it to
the appropriate column in the well type and surface water use
tables, respectively.
The drinking water population from groundwater usage is
estimated as follows for each well type:
-- Welt Type I: Single Residential. The number of people
within the distance to the off-site point of detection are
assumed to be affected. This is estimated using the
population density for the county and the distance, to the
off-site point of detection.
-------
A-53
-- Wei! Type 2: Municipal. The number of people served by
groundwater, as reported in the drinking water survey.
-- Well Type 3: Other Uses. Number of people set to zero.
-- Well Type 4: Single Residential and Municipal. Sum of
people calculated for well types 1 and 2.
-- Well Type 5: No Wells. Number of people set to zero.
The drinking water population from surface water usage is
estimated as follows for each use type:
-- Use Type I: Water Present but Not Used. Number of people
set to zero.
-- Use Type 2: Drinking. The number of people served by
surface water as reported in the drinking water survey.
-- Use Type 3: Recreation. Number of people set to zero.
-- Use Type 4: Other Uses. Number of people set to zero.
Use Type 5: Drinking and Recreation. Number of people
served by surface water as reported in the drinking water
survey.
-- Use Type 6: No Surface Water Present. Number of people
set to zero.
The total drinking water population is estimated as the sum of
the ground and surface water populations.
The result of this method is an estimate of the number of people using
surface and groundwater that originates from near the facility. The approach
starts with the drinking water survey data, and adjusts them to reflect the
results of the site visit survey data. It should be recognized that this
simulated drinking water population is generally not affected by releases at
the facility. Only when releases are simulated to migrate off site does this
population play a role in claims. (In the Simulation 1 Model run, releases
migrated off site at fewer than 15 percent of the facilities.)
A.2.3 Modeling of Releases
Seven release types were modeled based on release matrices produced by the
Release Simulation Model developed by Battelle Pacific Northwest Laboratories
as part of this PCLTF assessment project. The Release Simulation Model was
implemented separately from the PCLTF Simulation Model, primarily for
computing efficiency. The same existing facility population which forms the
basis for the PCLTF Simulation Model was used in the Release Simulation
Model. A detailed description of the Release Simulation Model is provided in
Appendix B. Below, we define the seven release types modeled, and then
-------
A-54
describe the release matrices and how they were used in the PCLTF Simulation
Model.
Definition of the Releases21
A release from a hazardous waste disposal facility can be defined in a
variety of ways. For example, any migration of waste outside of containment
may be considered a release. Alternately, a release may have occurred only if
a certain critical rate of migration outside of containment is achieved. For
the purposes of this analysis, six releases were defined based on two
characteristics: location and concentration. A seventh release type was
defined in terms of location only. Release types 1 through 6, displayed in
Exhibit A-20, were formed by combinations of two locations and four
concentrations. Each location and concentration is described below.
On-Site Monitoring Well. The on-site monitoring well is the
first location where the owner/operator may find evidence of a
release into groundwater. For facilities designed with single
liners, this location was assumed to be at the perimeter of the
waste and at the depth of the water table (i.e., the uppermost
aquifer). For facilities with double-liner leak detection
systems, this location was assumed to be between the two liners.
In the discussions below, this location is referred to as
"on-site."
» Closest Off-Site Potable Well or Naturally Occurring Body of
Surface Water. The off-site location where individuals (in the
general population) could first become exposed to the hazardous
waste migrating out of containment into groundwater is the
potable well or body of surface water closest to the facility
boundary. This location represents the point of first potential
off-site exposure. The distance to this location, referred to
here as "off-site," may vary considerably. Some facilities are
within 1000 feet of groundwater wells used to supply drinking
water; others are miles away from any potable well or surface
water. Distributions describing the distances to this off-site
location likely to be observed at land disposal facilities were
developed based on data from EPA's surface impoundment assessment
survey.22
21 This section is essentially identical to the definition of the
release types presented in Chapter 2 of Volume I. The reader interested in a
more detailed definition of the release types should refer to Appendix B.
22 Separate distributions describing distances to the off-site location
were developed for different parts of the 48 states analyzed in the Release
Simulation Model. These distributions were estimated based on "Surface
Impoundments and Their Effects on Groundwater Quality in the United States --
A Preliminary Survey," U.S. Environmental Protection Agency, 510/9-78-005,
June 1978.
-------
A-55
EXHIBIT A-20
SIX RELEASE TYPES DEFINED IN TERMS OF
LOCATION AND CONCENTRATION
Release Type Location
1 On-Site Monitoring Well
2 On-Site Monitoring Well
3 On-Site Monitoring Well
Closest Off-Site Potable Well
or Naturally Occurring Body of
Surface Water
Closest Off-Site Potable Well
or Naturally Occurring Body of
Surface Water
Closest Off-Site Potable Well
or Naturally Occurring Body of
Surface Water
Concentration
Change in Indicator
Parameters
Detectable Concentration of
Constituents
Toxic Concentration of
Constituents
Detectable through Taste
and Odor
Detectable Concentration of
Constituents
Toxic Concentration of
Constituents
-------
A-56
Change in Indicator Parameter. The concentration of
hazardous waste which would be required in order to result in a
change or increase in an indicator parameter in groundwater was
defined for each EPA waste code. The indicator parameters used
were total organic carbon (TOG) and total disolved solids
(TDS).23 The trigger TDS concentration was defined as 100
percent higher than the background TDS level.2* The trigger
concentration for TOC was defined as 10 ppm. (The TOG trigger
does not vary throughout the country because most groundwater is
very low in organic content.) This concentration is referred to
here as "indicator parameter" or IP.
Detectable Concentration of Constituents. The concentration
of priority pollutant constituents which would lead to their
identification during routine analysis was defined as the
detectable concentration. For organic constituents, a 10 Vg/1
detectable concentration was used based on the expected
sensitivity of standard gas chromatography/mass spectrometry
analysis techniques.25 For inorganic constituents analyzed
through indirectly-coupled plasma emission spectrometry, a
detection limit of 20 ug/1 is used.26
Toxic Concentration of Constituents. The toxic concentration
was defined as the criteria level set for each of(the 129
priority pollutants. If the pollutant is carcinogenic, this
concentration implies an increased risk of cancer of one in
one-million based on a 70-year period of exposure. Toxicity
levels for non-priority pollutants were not defined.
23 The indicator parameters pH and total organic halides were not used.
pH values were not simulated because there were very few data about the large
influence of local soil conditions. Total organic halides were reflected in
the use of TOC. TDS was used here as a substitute for conductivity, a
commonly used and easily measured indicator parameter. The conductivity of
wastes and natural waters are not well catalogued. The TDS for each waste
code was estimated using the solubility of its inorganic constituents. The
TOC for each waste code was estimated as the sum of the carbon portions of
each water-soluble organic constituent.
2* Background TDS levels were identified for different areas of the
country from "Surface Impoundments and Their Effects...," op. cit.
25 The analysis techniques are described in "Guidelines Establishing
Test Procedures for the Analysis of Pollutants; Proposed Regulations,"
Federal Register, December 3, 1979, pp. 69464-69575.
26 Ibid.
-------
A-57
Detectable through Taste and Odor. Taste and odor threshold
concentrations were identified for the hazardous waste
constituents.27 Taste thresholds were utilized where
available. In cases when only odor threshold values for airborne
concentrations were available, the concentrations in water needed
to yield the airborne vapor concentration thresholds were
identified.
The seventh release type is different from the six discussed above in that
it does not have a concentration as part of its definition. It was defined as
the overflow of contaminated leachate from the facility. This release occurs
when the cap on a closed land disposal process allows more water to infiltrate
into the facility than the facility can hold or pass through. This is often
referred to as the "bathtub" effect because it resembles a bathtub
overflowing. . It could be caused by excessive erosion of a clay cap or the
failure of a synthetic membrane cap.
These seven release types were chosen in order to model the process of
detecting and responding to releases. Each release type could be detected by
one or more monitoring actions which may be undertaken at the facility.
Additionally, the detection of each release type may lead to a particular set
of monitoring actions, response actions, or claims.
For example, routine monitoring actions at the facility may lead to the
discovery of release type 1, IP on site. As the result of this discovery,
more detailed sampling and analysis may be required at the facility for a
certain length of time. This more detailed analysis may be capable of
detecting release type 2, detectable concentration on site, and release type
3, toxic concentration on site, if and where they occur. The discovery of one
or both of these releases may lead to another set of actions, which may
include a corrective action and/or monitoring off site, for example. If
on-site monitoring fails to detect a release, the appearance of taste or odor
at the off-site location could be the first indication of a release.
Relationships defining how release detection leads to monitoring and
response actions were adopted; these relationships correspond to current RCRA
regulations. These relationships are discussed in Section A.3.2, Regulatory
Policy. Here it is important to note that the seven release types defined
above form the lowest level of detail achievable in the Model's analysis of
releases. Actions at the facility level were simulated as the result of the
simulated detection of these releases. This level of detail enables the Model
to be used for a variety of policy analyses. For example, the implications of
alternative RCRA policies which identify those monitoring and response actions
required due to the detection of releases can be assessed.
The next section describes how the release matrices were used to simulate
releases at the facility level.
27 F.A. Fazzalori, Compilation of Odor and Taste Threshold Values Data,
American Society for Testing Materials, 1978.
-------
A-58
Release Matrices
The Release Simulation Model produced 24 release matrices which were used
to simulate releases at the facility level in the PCLTF Simulation Model. A
total of 24 matrices were developed by:
dividing the country into four hydrogeologic regions;
and
analyzing each region six times, one time for each of
six process designs.
The choice of the four hydrogeologic regions was based on an analysis of
twelve smaller regions. This analysis indicated strong similarities in the
release estimates across some regions. Regions with similar release estimates
were combined, resulting in the four regions displayed in Exhibit A-21.
The six design types analyzed represent the seven prototype designs
developed as part of the Release Simulation Model (see Exhibit A-22). The
prototype designs represent possible configurations of layers at different
land disposal processes. Design types 1 through 7 are all observed at
existing facilities. Design types 3, 4 and 5 satisfy EPA design requirements
for new processes and are consequently assigned to new facilities. Design
types 4 and 5 were combined into a single design category because their
release estimates were very similar. Therefore, a total of six design
categories were used.
An illustrative release matrix is shown in Exhibit A-23. Each of the 24
release matrices has approximately 1,500 rows (Exhibit A-23 only shows the
first 10 rows). Each row is an independent estimate of when each of the seven
release types may occur (measured in years since the process began accepting
wastes). Consequently, a process (e.g., a landfill) located in a given
hydrogeologic region has an equal chance of having releases as described by
each row.
Also reported in each row of the release matrices is the simulated
distance to the off-site point of detection which underlies the estimated
times to the off-site releases. Because this simulated distance plays an
important role in both the estimates of the times to the off-site releases and
the estimates of the costs of claims and cleanup, the distance must be
reported by the Release Simulation Model for use in the PCLTF Simulation
Model. By using the simulated distance in conjunction with the estimates of
release times, consistency is maintained in the simulation of both releases
and costs.
To simulate releases at the facility level, the following steps are
performed:
identify the hydrogeologic region in which the facility
is located;
-------
A-59
EXHIBIT A-21
FOUR HYDROGEOLOGIC REGIONS USED IN THE
SIMULATION OF RELEASES
\J
-------
A-60
EXHIBIT A-22
PROTOTYPE PROCESS DESIGNS
USED IN PCLTF SIMULATION MODEL1
Prototype
Design
1
2
3
Process
Surface Impoundment
Surface Impoundment
Surface Impoundment:
Cover Liner
Clay Unlined
Clay Clay
Synthetic membrane: Synthetic membrane:
Landfill
Surface Impoundment;
Landfill
Surface Impoundment;
Landfill
Landfill
Landfill
clay
Synthetic membrane;
clay
Two synthetic mem-
branes; clay
Clay
Clay
bedding material
Two synthetic mem-
branes
Synthetic membrane;
clay
Unlined
Clay
1 These designs were developed for use in the Release Simulation Model.
See Appendix B.
-------
A-61
EXHIBIT A-23
ILLUSTRATIVE RELEASE MATRIX1
Time to Each Release Type
Distance to
Off-Site Point of
Iteration
Number
1
2
3
4
5
6
7
8
9
10
(Years
1
999*
50
10
20
999
999
40
90
30
999
2
25
45
8
40
999
110
70
50
999
40
Since
3
30
999
8
15
999
999
999
90
999
38
Facility Opening)
4
999
999
15
50
999
999
999
100
999
50
5
50
999
13
60
999
999
80
60
999
50
6
60
999
13
70
999
999
999
60
999
46
7
999
999
999
999
999
999
999
999
999
999
Detection
(feet)
2,000
1,000
800
4,000
600
8,000
10,000
5,000
9,000
2,000
1 Values are for illustrative purposes only.
2 999 means release was not observed din first 125 years.
-------
A-62
for each land disposal process (i.e., surface
impoundment, landfill),28 do the following:
identify its simulated design;
based on the simulated design and the facility's
hydrogeologic region, identify the appropriate
release matrix;
randomly draw a row from the release matrix;29
the releases at the facility are taken to be the first
occurrence of the releases simulated at each of the
disposal processes; and
the simulated distance for the facility is the minimum
distance simulated at each of the disposal processes.
This procedure results in a timing of each release type being assigned to
each facility. Additionally, a distance to the off-site point of detection is
assigned which is consistent with the simulated release times. Facilities
with multiple disposal processes require more than one draw from the release
matrices. Because more than one process design may be present at a facility
with more than one process, a facility may draw from up to two different
release matrices or may draw up to two times from a single matrix (once each
for landfill and surface impoundment).
Releases at new facilities are simulated in a slightly different manner.
Because new land disposal processes must satisfy EPA's design and siting
requirements (including public participation), it is expected that the
locations of new facilities are likely to be "better" than the locations of
existing facilities; better in the sense that releases will be less likely to
occur. Therefore, for the purpose of modeling releases, the Model has been
designed to allow the user to identify the extent to which the new locations
are better.
To model better locations, the user first chooses one of the seven release
types as a basis for selecting new facility locations. In current
simulations, release type 2, detectable concentration of constituents, was
adopted. Then the user identifies the fraction of existing facility locations
which is expected to approximate the distribution of new facilities. For
example, in current simulations, it was assumed that new facilities would
perform (in terms of the chosen release type 2) as well as be in the top
28 Releases are not simulated for injection wells and land treatment
processes.
29 To reduce the variance across runs, the draws from the release
matrices are always made in the same random order. Also, draws from the
matrices, are made without replacement.
-------
A-63
one-third of all existing facilities. Therefore, only the top one-third of
all existing facility locations (in terms of the time it takes for release
type 2 to occur) is available as a basis for simulating new facility
locations. Releases at new facility locations are drawn randomly from the top
fraction (one-third in current simulations) of release times at existing
facility locations.30
A.2.4 Actions
Actions refer to a set of activities undertaken at hazardous waste
disposal facilities. The PCLTF Simulation Model was designed to incorporate
those actions necessary to model the process of release detection and
response. This section describes the characteristics of each of the actions
included in the Model.
The material in this section is presented using the action definitions and
characteristics adopted in the current simulations. These values correspond
to the actions required under current EPA regulation,31 as well as several
additional actions not currently covered by regulation. Although these values
have not been varied in the Model runs performed to date, it should be noted
that any set of actions can be analyzed using the PCLTF Model, so long as it
can be defined using the action characteristics built into the Model. By
defining actions in terms of their characteristics and allowing the user to
vary any or all of the characteristics as desired, the Model remains very
flexible. This flexibility is very valuable because the Model is capable of
examining the implications of altering a characteristic of an action or of
analyzing an alternative set of actions.
Not discussed here is the definition of when and under what circumstances
the actions are simulated to be taken at facilities. Section A.3.2,
Regulatory Policy, describes the manner in which the Model user defines how
actions are simulated, i.e., the "action policy." The action policy adopted
in the current simulations corresponds to current EPA regulations.
The actions are divided into three types: monitoring actions, response
actions, and post-closure care actions. All the actions are characterized in
terms of the following general properties:
* duration;
ability to detect releases;
capital costs; and
operation and maintenance (O&M) costs.
30 If a fraction of 1.0 (i.e., 100 percent) were used, the distribution
of releases at new facility locations would be drawn from the full
distribution at existing facility locations, meaning that both new and
existing facilities would have the same distribution of releases.
31 See 40 CFR 246 Subpart F for current monitoring and corrective
action requirements.
-------
A-64
The duration is the length of time (in years) that it takes to complete the
action. The ability to detect releases refers to whether the action can
detect one or more of the seven release types when they are simulated to occur
at the facility. Capital costs pertain to one-time costs for the planning and
construction of equipment (i.e., the placement and construction of monitoring
wells) at each facility. Capital costs are spread over a capital investment
period, which is the time required for planning and construction to take
place. During this period, the action is unable to detect releases. O&M
costs are the annual costs of the action, and may include equipment
maintenance and costs for the sampling and analysis of groundwater.
In addition to these characteristics, response actions are also defined in
terms of their ability to clean up contaminated groundwater and to prevent and
delay releases. The following sections describe each action and its
properties.
A.2.4.1 Monitoring Actions
The PCLTF Simulation Model considers five monitoring actions. The
characteristics of each are summarized in Exhibit A-24. The exhibit displays
the releases each monitoring action can detect, its capital investment period,
and its required duration. Each action is discussed separately.
Routine Monitoring for indicator Parameters at On-Site Monitoring
, Wells
Routine monitoring for indicator parameters is analogous to detection
monitoring required at all land disposal facilities.32 This action is
capable of detecting release type 1, a change in indicator parameters, and
release type 7, the bathtub effect.33 It is modeled without a capital
investment period because the action is assumed to be required in
perpetuity.311 (This means that the action can continue so long as someone
is available and required to pay for it (e.g., the PCLTF). This duration is a
characteristic of the action, not the policy of who must undertake the
action.) Periodically, the monitoring wells must be refurbished or replaced,
and this capital cost was amortized over the life of the wells and added to
the O&M cost.35 The range of capital costs for constructing four wells is
displayed in Exhibit A-25. Also displayed is the range of annual sampling and
32
See 40 CFR 264, subpart F, Detection Monitoring Program.
33 If an individual is performing this action, it is assumed that when
he visits the facility, he will be able to observe the occurrence of release
type 7.
34 The action stops when a more extensive action, such as monitoring for
constituents, is begun. However, when the more extensive action or actions
are completed, this action resumes.
38 A well life of 25 years was assumed.
-------
EXHIBIT A-2<4
CHARACTERISTICS OF MONITORING ACTIONS
Action
Routine Monitoring For Indicator
Parameters at On-Site Monitoring Wells
Monitoring for Waste Constituents at
On-Site Monitoring Wells
Plume Delineation and Tracking On Site
Monitoring for Waste Constituents at
Off-Site Monitoring Wells
Plume Delineation and Tracking Off Site
Release Type 1/
1 2 3 it 5 6
D* 2/
7
D
Capital
Investment
Pe r i od
(Years)
(NA) t|/
Duration
( Yea rs )
Ongoing in
perpetui ty
D* D* D*
D 1/D D
ODD
(NA)
I
I
Lifetime or 30
yea rs
30 years
30 years
30 years
I/ Release Types:
I. Change in Indicator Parameters at the On-Site Monitoring Well
2. Detectable Concentrations of Constituents at the On-Site Monitoring Well
3. Toxic Concentrations of Constituents at the On-Site Monitoring Well
t|. Taste and Odor Off Site
5. Detectable Concentrations of Constituents Off Site
6. Toxic Concentrations of Constituents Off Site
7. The "Bathtub Effect." If any of the five monitoring actions are being performed, then
this release type may be observed, and consequently, by definition, detected.
2/ D* indicates that monitoring action can detect release only if monitoring wells are placed
correctly. See text.
3/ D indicates that monitoring action can detect release.
M/ Not Applicable.
5/ Plume delineation and tracking is not modeled as a separate action in the current simulations.
Instead, it is considered as part of the fluid removal and treatment actions. Characteristics
shown are illustrative values which might be used if plume delineation were treated as a separate
action.
-------
A-66
EXHIBIT A-25
MONITORING ACTION COSTS1
Lower Upper
Bound Bound
Range of capital costs for construction of $25,000 $50,000
4 monitoring wells2
Range of annual sampling and analysis costs for $ 1,000 § 2,000
monitoring for indicator parameters, 4 wells3
Range of annual sampling and analysis costs for $ 1,750 $ 3,000
monitoring constituents, 4 wells'*
Range of annual sampling costs for plume $ 3,500 $ 6,000
delineation and tracking5
1 All costs reflect a range of boring depths of 25 to 150 feet. Costs
are much higher for depths greater than 150 feet.
2 Costs include:
well construction;
plan development; and
* report preparation.
3 Costs include:
annual sampling of water quality parameters (chloride, iron,
manganese, phenols, sodium, sulfate); and
semi-annual sampling of indicators of contamination (pH,
specific conductance, total organic carbon, total organic halogen),
4 Costs include:
annual sampling of water quality parameters;
semi-annual sampling of indicators of contamination; and
analysis for the presence of hazardous waste constituents.
5 ICF estimate. Assumed to be twice the costs of annual sampling and
analysis of constituents because more samples are taken and analyzed for plume
delineation and tracking. Because plume delineation and tracking are not
modeled as a separate action in the current simulations, these costs are
treated as part of fluid removal and treatment. See Exhibit A-28.
Source: Geraghty & Miller, Inc., "Development of Groundwater Monitoring
Requirements and Costs for Current RCRA Regulatory Requirements,"
working paper, January 1982.
-------
A-67
analysis costs. Costs are simulated each time the action is initiated by
randomly drawing from a uniform distribution bounded by the ranges shown in
Exhibit A-25.
The ability of this monitoring action to detect changes in indicator
parameters on site also depends on the accurate placement of the on-site
monitoring wells. Only if the wells are placed correctly (i.e., at the
correct depth and location relative to plume movement) will this action be
able to detect releases. (Exhibit A-24 displays this requirement for
correctly placed wells with an asterisk.) Because some hydrologies are
exceedingly complex, it is likely that some number of facilities may have
monitoring systems that were improperly designed or installed. Additionally,
a system which was initially placed correctly may fail because of subsequent
changes in groundwater flow caused by nearby construction, pumping, or
grading. However, the frequency with which monitoring wells are incorrectly
placed (thus making them unable to detect releases) is not well known and no
basis exists for quantifying it.
To model this phenomenon, the Model user supplies an estimate of the
probability with which the on-site monitoring wells will fail to detect a
release when it occurs (i.e., the chance that the wells are placed
incorrectly). In the current simulations, this probability was assumed to be
25 percent.36 The placement of each facility's on-site wells is
characterized by drawing a random number and comparing it to this
probability. The placement of the on-site wells is determined only once for
each facility. .
Monitoring for Hazardous Waste Constituents at On-Site Monitoring
Wefls
Monitoring for hazardous waste constituents is analogous to compliance
monitoring and is required at land disposal facilities when detection
monitoring indicates a release has occurred.37 No capital expenditure
period is used for this action because the same on-site wells used for
monitoring indicator parameters are assumed to be used. Consequently, the
capital costs are amortized over the well life and added to the O&M costs.
The range of O&M costs used in the current simulations is displayed in Exhibit
A-25.
Because the same on-site monitoring wells are assumed to be used, the
ability to detect hazardous waste constituents with this action depends on the
correct placement of the on-site monitoring wells. If the wells were
simulated to be placed correctly at the facility (the probability of which is
defined by the user as described above), then this monitoring action can
36 A lower value (e.g., 5 percent) means that more releases are caught
before going off site. Consequently, more opportunity exists to take response
actions to prevent off-site releases from occurring.
37 See 40 CFR 264, subpart F, Compliance Monitoring Program.
-------
A-68
detect changes in indicator parameters on site (release type 1), detectable
concentrations of constituents on site (release type 2), and toxic
concentrations of constituents on site (release type 3).38 The bathtub
effect (release type 7) is assumed to be observable if this action is ongoing.
The duration of this action is simulated as the maximum of 30 years or the
operating lifetime of the facility. This is consistent with current EPA
regulations.
Plume Delineation and Tracking On Site
The current simulations do not model either on- or off-site plume
delineation as separate actions; rather, they are modeled as part of on- and
off-site fluid removal and treatment, respectively. It is possible, however,
to model plume delineation separately by specifying the appropriate inputs.
Plume delineation and tracking on site can be performed to identify the extent
of groundwater contamination and to track the movement of the plume. This
action can be used in the Model to simulate whether: (1) the plume of concern
has already migrated off site; and (2) the contamination will migrate off site
in the future, if it has not already done so. This action may be modeled as a
response to the detection of hazardous waste constituents on site. Because it
required whenever a corrective action is taken, plume delineation and tracking
on site is combined with fluid removal and treatment on site to form a single
action.39
Exhibit A-26 displays a series of capital costs for plume delineation and
tracking as a function of plume size*0 and depth of well borings which might
be used if plume delineation and tracking were trated as separate actions.
Although some planning is required to determine the placement of the wells,
the capital investment period or assumed to be one year. The capital costs
38 If the toxic concentration is less than the detectable concentration,
then this action can detect the toxic release only once the detectable
release has occurred.
39 See 40 CFR 264, subpart F, Corrective Action Program. In the Model,
plume delineation and tracking can be modeled separately from an overall
corrective action (which would include fluid removal) so that alternative
policies defining when this action should be undertaken could be examined.
For example, by defining this action separately, it was possible to model a
policy requiring plume delineation before making a determination on the need
for fluid removal. The information on the potential for off-site migration of
the plume (supplied by this action) can be utilized in the simulated decision
regarding the need for fluid removal. Currently, however, plume delineation
and tracking are not treated as an action separate from fluid removal and
treatment.
*° We assumed the plume size is the sum of facility's process sizes plus
an additional 15 percent to account for administrative buildings and space
between the facility processes.
-------
EXHIBIT A-26
PLUME DELINEATION AND TRACKING CAPITAL COSTS J./
(Thousands of 1982 Dollars)
Depth
(feet)
25
50
75
100
200
Plume Size (square feetl
500.OOP I.250.0QQ 2.OOP.000 5.000.000 8.000.000 20.000.000
15.0
22.5
26.0
31.0
60.0
15.0
22.5
26.0
3U. 0
60.0
32.5
11.0
55.0
67.5
120.0
37.5
50.0
65.0
82.5
138.0
10.0
56.0
73.0
91 .0
153.0
16.0
65.0
85.0
104.0
.187.5
59.0
80.0
I0<4
126.0
225.0
0
80.0
108.0
138.0
170.0
300.0
!/ Costs include:
o groundwater quality assessment plan;
o test borings;
o conversion or test borings .to monitoring welts;
o test boring abandonment;
o groundwater sampling;
o groundwater sample analyses; and
o consultant fee.
Assumptions include:
o I boring/acre; and
o borings spaced 100 feet apart around perimeter of plume.
i
Ch
SOURCE: Geraghty & Miller, Inc., "Cost Estimates for Containment of Plumes of Contaminated
Groundwater," working paper, March 1982, p. 11.
-------
A-70
are assumed to be borne during this time. The annual sampling and analysis
costs were assumed to be two times the annual sampling and analysis costs for
monitoring constituents (see Exhibit A-25). This assumption was made because
there are more wells and samples required for plume delineation and tracking
than for monitoring constituents. Because plume delineation is modeled as
part of fluid removal and treatment in the current simulations, its costs
(both capital and O&M) are reflected in the costs of fluid removal and
treatment.
This action can detect indicator parameters on site (release type 1),
detectable and toxic concentrations of constituents on site (release types 2
and 3), and the bathtub effect (release type 7).
Monitoring for Hazardous Waste Constituents Off Site
This monitoring action is analogous to on-site monitoring for hazardous
waste constituents, but takes place at an off-site location. This action can
detect constituents off site (release types 5 and 6) and the bathtub effect
(release type 7).
We assumed a capital investment period of one year for this action. This
action's capital costs are the construction costs for four monitoring wells
which are placed off site. These costs, displayed in Exhibit A-25, are
assumed to be borne when the action is taken. The sampling costs are the same
for monitoring constituents on site (see Exhibit A-25).
Piume Delineation and Tracking Off Site
Plume delineation and tracking off site is analogous to plume delineation
and tracking on site, but takes place at an off-site location. Again, it is
not modeled as a separate action but is instead simulated as a part of fluid
removal and treatment. If treated as a separate action, its capital costs (as
a function of plume size*1 and depth of well borings) and its sampling and
analysis costs would be the same as for plume delineation and tracking on site
(see Exhibits A-25 and A-26). This action can detect constituents off site
(release types 5 and 6) and the bathtub effect (release type 7). In addition,
this action can identify whether a release will exceed toxic concentrations
off site, if it has not already done so.
A.2.4.2 Response Actions
Response actions are performed for the purpose of cleaning up releases
which have occurred and preventing or delaying future releases. Additionally,
response actions may be defined as having the ability to detect releases when
** Plume size is estimated as the area of a rectangle with a length
equal to the distance to the off-site point of detection and a width equal to
the facility width times a dispersion factor. This dispersion factor serves
to increase the simulated width of the plume in proportion to the length of
the plume.
-------
A-71
they occur. Because the response actions are the only available means for
cleaning up contaminated groundwater and preventing releases, they play an
important role in the PCLTF Simulation Model. The way in which these actions
are simulated to influence groundwater contamination is described in section
A.3.5, The Effect of Response Actions on Releases. This section presents the
characteristics of the response actions themselves.
The PCLTF Simulation Model considers three response actions: cap repair
(surface sealing), fluid removal and treatment on site, and fluid removal and
treatment off site. The current simulation values for each action are
summarized in Exhibit A-24. Each action is discussed separately.
Cap Repair (Surface Sealing)
Surface sealing refers to the repairing of the site cover. This may
include placing recompacted clay on the site, improving drainage and grading,
and taking other measures to reduce infiltration into the site. As shown in
Exhibit A-27, surface sealing is assumed to prevent or delay the bathtub
effect (release type 7).
Because surface sealing is assumed to take one year to complete, it has no
OSeM costs. The capital costs for surface sealing are a function of the size
of the cap. Based on available estimates,"2 we assumed the following range
for costs per acre (in 1982 dollars):
* lower bound: $ 6,500
* median value: $20,000
upper bound: $40,000
Given the sizes (in acres) of the facility's processes, the costs for surface
sealing were simulated by: (1) drawing a random number to determine whether
the cost is above or below the median, (2) drawing randomly from the range of
the median and the appropriate bound, and (3) multiplying the simulated cost
per acre by the facility's size.
Fluid Removal and Treatment On Site
Fluid removal and treatment on site is analogous to a corrective
action.1*3 It is assumed to be able to clean up groundwater contaminated due
It 2
The surface sealing estimates identified were:
$6,500 to $32,620 per acre: Geraghty & Miller, op. cit;
$4,000 to $20,000 per acre: IR&T, op. cit; and
$27,000 to $37,500 per acre: SCS Engineers, op. cit.
*J Corrective action is required-at land disposal facilities when
compliance monitoring demonstrates certain levels of groundwater
contamination. In the current simulations, this action includes plume
delineation and tracking on site (see the description of monitoring actions)
See 40 CFR 264, subpart F, Corrective Action Program.
-------
EXHIBIT A-27
CHARACTERISTICS OF RESPONSE ACTIONS
Action
Surface Sea I ing
Fluid Removal and Treatment On Site
Fluid Removal and Treatment Off Site
Release Type I/
123156 7
Capita 1
Investment
Period Duration
(Years) (Years)
D 2//P 3/1 1
D/P D/P D/P P P P
P D/P D/P
5 100
5 100
jy Release Types:
I. Indicator Parameters On Site
2. Detectable Constituent Concentrations On Site
3. Toxic Constituent Concentrations On Site
k. Taste and Odor Off Site
5. Detectable Constituents Off Site
6. Toxic Constituents Off Site
7. The "Bathtub Effect"
2/ D indicates that response action can detect release.
3/ P indicates that response action can prevent or delay release.
-------
A-73
to release types 1, 2, and 3. Further, because cleaning up contaminated
groundwater on site will prevent it from migrating off site, this action can
prevent (or delay) the occurrence of release types 1 through 6 (if they have
not already occurred). If releases have already occurred off site, then this
action was assumed to have no influence on the off-site release.
The capital investment period for fluid removal and treatment was assumed
to be five years because time is required to locate the contaminated plume and
set up the well system for fluid removal and treatment. Although part of the
plume may be delineated within a year, it may take a few years before the
contaminated plume is fully identified.
It was assumed to take 100 years to complete this action fully. There is
considerable uncertainty surrounding the time needed to clean up a site
fully. This estimate reflects the fact that if the contaminant source is not
removed, fluid removal and treatment will be required continuously. If the
action were stopped, plume growth might be expected to resume immediately.
Duration also plays an important role in the relationship used to define the
influence which response actions have on releases (see section A.3.5, The
Effect of Response Actions on Releases).
The costs of fluid removal and treatment were calculated based on the
output of a stochastic simulation model.4* This cost model simulated the
costs of fluid removal and treatment under several different conditions.
Based on plume dimensions and hydrogeologic characteristics, appropriate
response actions were modeled. The model chose the most efficient removal
option available and then simulated either one or more of the following fluid
treatment methods:
* No treatment;
Carbon adsorption;
Ion exchange; and
* Reverse osmosis.
The result was a 1,000-observation data file with information on fluid removal
and treatment costs as well as other key variables.
The cost model operates under the assumption that the source of the
contamination may be removed, which would reduce the necessary duration of
groundwater removal. The PCLTF Model, however, does not incorporate the cost
of this removal nor is such removal currently required by regulation.
Consequently, the duration of the fluid removal and treatment actions has been
set to 100 years to reflect the ongoing need for fluid removal and treatment
which arises if the contaminant source is not removed.
** Geraghty & Miller, Inc., "Stochastic Model of Corrective Action Costs
at Hazardous Waste Management Facilities," January, 1984.
-------
A-74
Because the PCLTF Model simulates plume dimensions, a method has been
developed to estimate cleanup costs as a function of plume size. However,
other factors create additional variation in costs after accounting for plume
size. Consequently, the method incorporates this uncertainty.
Based on plume size (i.e., plume length times plume width), the. 1,000
observations in the cost-model data file were sorted into eight groups.
Within each group, the mean plume size along with the distribution of both
capital and O&M costs were estimated. The results are displayed in Exhibit
A-28.
Using the information in Exhibit A-28, the Model simulates the cost of
on-site fluid removal and treatment in the following way.
STEP ONE: A random number is drawn and assigned to each facility for
the duration of the iteration.
STEP TWO: Plume size is estimated as the sum of the size of all
processes at the facility plus 15 percent.
STEP THREE: Using estimated plume size and the random number from step
one, costs are estimated from Exhibit A-28. Linear interpolation is used both
horizontally and vertically to determine costs. If the plume's dimension is
greater than 5,102,554 square feet, linear extrapolation is used to determine
costs; however, the maximum cost used is twice the maximum given in Exhibit
A-28. O&M costs and capital costs are estimated using the same random number
to ensure consistency between the two types of costs. Further, given that the
random number is maintained at the facility level, cost estimations for a
given facility are consistent with one another throughout the duration of the
iteration.
Fluid Removal and Treatment Off Site
Fluid removal and treatment off site is analogous to fluid removal and
treatment on site, except that the action is undertaken off site. This action
both cleans up and can prevent off-site release types 5 and 6. In addition,
it can prevent taste and odor problems from occurring (release type 4). In
the current simulations, this action includes plume delineation and tracking.
The capital and O&M costs for this action are based on the same cost data used
for on-site fluid removal and treatment (see Exhibit A-28). Like on-site
fluid removal and treatment, duration is set at 100 years. The following cost
estimation process was process used.
STEP ONE: Total plume size is computed by adding on-site plume size
(i.e., the sum of the facility's process sizes plus 15 percent) to the
off-site plume size (i.e., a rectangle with a length equal to the distance to
the off-site point of detection and a width equal to the facility width times
one plus a simulated dispersion factor).
STEP TWO: Using the method described for on-site fluid removal and
treatment, cleanup costs are estimated for a plume with dimensions equal to
-------
EXHIBIT A-28
PROBABILIT(ES OF ON-SITE AND OFF-SITE FLUID REMOVAL AND TREATMENT COSTS
(Thousands of 1982 Dollars)
Cum it I a t i ve
ProbabiIity
Capital Costs J./
.0000
. 1875
.5000
.8125
1.0000
O&M Costs 2/
.0000
. 1875
.5000
.8125
I.0000
PIu.me Size t square feet)
1.495
173.6
237.9
291. 1
387.7
434. 1
22.7
30.5
39.2
53.3
59.7
20. 198
202.7
262.5
326.2
418.7
1*90.2
31.5
38. 1
48. 8
59.6
70.3
89^ 102
210.7
276.6
351.1
440.2
603.5
314.9
43.5
55.6
74.2
1 1 1.1
279.992
242.5
314.5
422.8
683.5
1,293.8
36.0
47.8
75.4
1 18.8
241.9
538.638
262.3
448.0
686.4
1, 167.6
3, 108.4
48.0
73.0
121 .9
226.3
464. 1
987.773
314.8
438.9
675.2
2,452.8
3,990.4
53. 1
76.0
121.9
338.5
649-3
2.428.574
4(7.0
697.3
1,111.2
2,452.8
4,465.9
76.4
1 13.2
210.7
338.5
767.0
5. 102.554
417.0
697.3
1, 197.9
6,548.0
9,697.7
76.4
1 18.0
243.3
1,299. 1
2,565.6
I/ Capital costs include:
o Plume delineat ion
o System design engineering
o We 11/drain installation
o Construction engineering
o Surface infrastructure
o Fluid treatment
o Capital replacement
2/ O&M costs include:
o We I I/drain system operation
o Surface infrastructure
o Fluid treatment
o System monitoring
>
-4
Source: IGF analysis of cost model simulation results. See Geraghty & Miller, Inc., "Stochastic
Model of Corrective Action Costs at Hazardous Waste Management Facilities," January, 1984.
-------
A-76
those determined in step one. The same random number used for estimating
on-site cleanup is used again to ensure consistency.
STEP THREE: On-site cleanup costs are estimated in a manner identical
to that described for the fluid removal and treatment on-site action.
STEP FOUR: The on-site cleanup cost is subtracted from the total
cleanup cost to yield the cost of off-site fluid removal and treatment. This
approach ensures that only the marginal cost of expanding cleanup from on site
to off site is measured. This is an important consideration because under
current regulation, it is extremely likely that an off-site cleanup will
always be taken after or at the same time as an on-site cleanup.
A.2.4.3 Post-Closure Care
Post-closure care refers to the routine activities that are undertaken
after the facility ceases accepting waste for land disposal and is closed.
Based on an analysis by IR&T,"5 we identified seven components of
post-closure care:
1. Maintain Benchmarks;
2. Maintain Soil Integrity of the Cap;
3. Security;
4. Fence Maintenance;
5. Gas Monitoring;
6. Leachate Monitoring; and
7. Leachate Collection and Removal.
i
Each activity is described using two factors: (1) the cost of the
operation (e.g., the cost of maintaining the soil integrity of the cap) and
(2) the proportion of facilities required to undertake the activity. The cost
estimates for these operations were derived from the IR&T analysis and are
described in the footnotes to Exhibit A-29. Because current regulations
covering post-closure care indicate that all landfills, surface impoundments,
and land treatment facilities (at which waste remains following closure) would
be required to perform the same post-closure care activities, we utilized the
same cost estimates for each of the seven post-closure care requirements for
all these process types.
It was assumed that all facilities will be required to perform the first
four operations (see Exhibit A-29), with cap maintenance being the most costly
of these four items (for facilities over 16 acres in size). Fence maintenance
*5 International Research and Technology Corporation (IR&T), An
Analysis of Methods of Ensuring Continuity of Operation for Hazardous Waste
Disposal, draft report submitted to EPA under Contract No. 68-01-5794
(undated). Because monitoring for indicator parameters is required in
perpetuity (see 40 CFR 264, subpart F, Detection Monitoring Program), it is
implicitly part of post-closure care. However, this action is modeled
separately in the PCLTF Simulation Model.
-------
EXHIBIT A-29
SUMMARY OF POST-CLOSURE CARE ACTIVITIES AND COSTS*
(1982 do Ilars)
Component
of Care
I. Ma intain
Benchmarks
2. Ma intain Cap
3. Security
i*. Fence Mainte-
nance
5. Gas Monitoring
6. Leachate
Moni toring
7. Leachate
Col lection and
Remove I
Probab i I i ty Lower
Requi red Bound
i
ng
ind
on
An
1.0
1.0
1.0
1.0
0.05
1.0
1.0
f o 1 1 ow i ng
Analysis
$60
60/acre
2/acre
355/(acre). 57
?7 591
1,030
M77
2 IO/( acre) .
page.
of Methods of Ensuring Cont
Med ian
Est imate
$89
1 la/acre
6/acre
473/(acre). 57
1,182
1,115
U77
830/(acre) .
inui t.v of Operation
Upper
Bound
$118
176/acre
12/acre
59 (/(acre). 57
(.773
(.800
177
l,450/(acre).
for Hazardous Waste
Source
IR&T, p. A-37
IR&T, p. A-37
IR&T, p. A-37
IR&T, p. A-37 I/
IR&T, p. A-21 3/
<«/
57
Disposal Sites.
^
draft report submitted to EPA under Contract No. 68-01-5794 (undated);- and "Cost of Groundwater
Sampling and Analysis," Draft EPA Memo, May 20, 1982.
-------
A-78
FOOTNOTES TO EXHIBIT 29
Computed assuming facility is square so that its perimeter (the length
of fencing) is equal to: P = A(A).5; where A equals the area of the
treatment process (e.g., landfill). IR&T's best estimate of the cost of-
fence maintenance is 5 percent of $29.20 per meter per year (where $29.20
per meter is the initial cost of the fence).
IR&T referenced Arthur D. Little, Inc. as stating that approximately 5
percent of all facilities will likely need gas monitoring (IR&T, p. A-21).
IR&T estimated the costs for gas monitoring during operation. Assuming
monitoring is decreased during the post-closure period, its costs will be
less. Using the IR&T assumptions of $100 to $300 per sample/analysis»
five analyses per site, and reducing the frequency to once per year, the
cost is $500 to $1,500 per year (1980 $).
IR&T assumed leachate monitoring would be required at a subset of those
facilities with leachate collection systems and stated that leachate
monitoring will cost approximately the same as groundwater monitoring (p.
A-20). However, IR&T's cost estimates of leachate and groundwater
monitoring were made for monitoring during operation. Therefore, the
monitoring costs from the EPA Memo were used. The costs were estimated as
follows: $752 per year for analysis, and $278 to $1,030 per year for
sampling, yielding $1,030 to $1,782 per year total. These cost estimates
are consistent with IR&T's cost estimates for minimum analyses. (See pp.
A-12 to A-16, IRT report.)
These cost estimates were derived from the IR&T data, but are slightly
lower than the IR&T cost estimates. This portion of the post-closure care
cost estimate may be refined to take advantage of information on leachate
generation which is estimated in the Release Simulation Model; however,
IR&T treated these costs as independent of leachate quantity. These cost
estimates were derived as follows:
Number of cells per site:
" 10 cells/15 acre site; 10 years of operation (IR&T, p. A-23)
-- 30 cells/150 acre site; 10 years of operation (IR&T, p. A-24)
20 cells/30 acre site; 20 years of operation (IR&T, p. A-23)
60 cells/300 acre site; 20 years of operation (IR&T, p. A-23)
These data can be fitted to two curves:
(1) number of cells = 2.75 (acre).*77: 10 years of operation
(2) number of cells =3.95 (acre).477: 20 years of operation.
These two curves can be averaged together as:
number of cells = 3.35 (acres).1*77
-------
A-79
FOOTNOTES TO EXHIBIT 29
(continued)
(fn. 5 continued)
One pump required per cell (IR&T, p. A-23)
Annual operation and maintenance cost per pump of approximately 2
percent of initial capital cost (IR&T, p. A-23)
Approximately 5 percent of pumps replaced annually (IR&T, p. A-34)
at a replacement cost of $3,000 each (IR&T, p. A-23).
Total annual cost equals:
(0.02+0.05)(3000)(3.35/(acre).*77) = $735/(acre).47T (1980 dollars)
A range of uncertainty of plus and minus 75 percent (IR&T, p.
A-37), and an upper bound of $36,000 (IR&T, p. A-38) were used.
Comparison with IR&T's leachate collection and removal estimate of $4,500
+ 44.50/acre (p. A-37) yields (1980 $):
SIZE IN ACRES
30 100 150 300
IR&T ESTIMATE 5,800 8,950 11,175 17,850
REVISED Upper 6,200 11,000 13,400 18,600
ESTIMATE Median 3,500 6,300 7,600 10,600
Lower 900 1,600 1,900 2,700
-------
A-80
was assumed to be required by all facilities, although no data were available
upon which to base this assumption. The gas monitoring frequency was taken
from the IR&T study. The leachate-related frequencies reflect the assumption
that leachate will be required to be monitored and collected at all facilities
with leachate collection systems. These leachate frequencies are very
uncertain at this time.
The costs of each activity are determined at each facility as follows:
to simulate whether the activity is undertaken, draw a
random number and compare it to the probability of the
activity being required;
for activities undertaken, draw a random number to
determine if the cost will be above or below the median
cost estimate;
draw from the range of the median estimate and the
upper or lower bound to determine the activity's cost;
and
" if appropriate, multiply the cost times the area of
the faculty's processes (in acres) raised to the
appropriate power.*6
The sum of the costs of all seven activities is used as the estimate of the
facility's post-closure care cost. Summaries of the individual cost
components are not maintained by the Model.
A.2.5 Claims
Claims refer to compensation requested by third parties for harms they
believe they have suffered as the result of off-site releases of hazardous
waste. This section presents the definitions of the claims and the methods
used to estimate their size (in dollars). The definition of when and under
what circumstances claims are simulated to be brought is discussed separately
in section A.3.2, Regulatory Policy.
Claims are divided into the following four categories:
Personal injury;
Real property;
Economic loss; and
Natural resource damage,
The division of claims into these categories is important, primarily for
simulating their legal validity under state statutory and common law.*7 The
*' A minimum of one acre is used for the size of the facilities if the
simulated facility size is less than one acre.
kl The legal validity of .claims is discussed in section A.3.6.
-------
A-81
above categories are somewhat artificial for the purpose of calculating the
total dollar value of claimed damages. For example, if a farmer's irrigation
wells were contaminated, the cost of providing an alternative supply of water
suitable for irrigation could be placed in the economic loss category.
However, the loss of the existing irrigation wells could also be valued in
terms of its effect on the fair market value of the farmer's land.
Presumably, the land would decrease in value due to the loss of the safe use
of the wells. A decrease in property value however, falls under the real
property damage category, not under the economic loss category. Therefore,
the farmer could seek recovery under either claim category for the damage to
his wells.
For this analysis, we have assumed that the following harms will fall
under each of the categories:
Personal injury: out-of-pocket medical expenses; lost
time due to illness and mortality; costs of monitoring
individuals' health conditions;
Real property: reductions in property value;
Economic loss: costs of alternative supplies of
potable water; and
Natural resource damage: lost recreation value; value
of fish killed or contaminated; cost of dredging
contaminated stream bed or lake bottom.
Each claim type is described separately below.
A.2.5.1 Personal Injury Claims
Personal injury claims are simulated by estimating the number of people
who consider themselves to have been harmed by the off-site release of
hazardous waste into groundwater, and by estimating the amount of compensation
these people may request. Of particular importance is that this method is
based on an assessment of those who may consider themselves to have been
harmed, as opposed to those who were actually harmed. This distinction is
important because the number of harmed people will likely be much smaller than
the number of people who consider themselves harmed. Estimating personal
injury claims based on harms alone could result in an underestimate.
The number of people who consider themselves to be harmed are, at a
maximum, all those who came into contact with the ground or surface water
contaminated by the off-site release. This number of people was calculated at
the facility level by simulating:
the usage of the ground and surface water near the
facility; and
the number of people using the ground and surface
water as a source of drinking water.
-------
A-82
These estimates are based on data from: EPA's site visit survey (1982);
EPA's drinking water survey (1981); and the County and City Data Book (1977).
Section A.2.2.2, Facility Attributes, describes the analysis performed at the
facility level to simulate the number of people using water from near the
facility for drinking (i.e., the drinking water population). This estimate is
used as the basis for simulating personal injury claims at facilities with
off-site releases. However, because these claims will arise in the future and
because the compensation requested may vary by the age of the claimant, two
adjustments to the drinking water population are required:
divide the population into age groups; and
adjust the number of people in each age group to
reflect births, deaths, and aging over time.
Exhibit A-30 displays the data used to perform these adjustments.
Age categories were constructed at ten-year intervals from ages 0 to 99.
The fraction of the U.S. population in each age group in 1980 was used to
divide the simulated drinking water population into age groups. Because the
age distribution of the population near facilities will vary from the age
distribution of total U.S. population, this division into age groups will not
be correct for individual facilities. However, the procedure will produce
results representative of the mean across all facilities.
To reflect changes in population over time, the birth rate and probability
of survival were used to estimate how the distribution of ages will vary over
time. A matrix was developed that represents the size of each age category
over time relative to its size in 1980. Exhibit A-31 displays this matrix.
The age distribution in 1980 is taken as the distribution for the total U.S.
population (see Exhibit A-30). To estimate the age distribution in 1990, the
following steps were performed:
* Multiply each age category by one minus its
probability of not surviving to obtain the size of the
next older age category in the next time period (age
category 2 in 1990 = age category 1 in 1980 times one
minus the probability of not surviving from age category
1 to 2: 0.1430 = 0.1459 x (1.0 - 0.0198)).
The size of age category 1 in the next time period
equals the sum of the births in this time period (0.1606
= (0.1740 x 0.147) -f (0.1803 x 0.577) + (0.1392 x 0.212)
+ (0.1005 x 0.011)).
Once the age distribution is calculated for 1990, the same procedure is used
to estimate the age distribution for 2000, and so on.
Of note is that the matrix displayed in Exhibit A-31 is comprised of
values reflecting the size of each age category relative to the 1980
population. Consequently, if personal injury claims are brought in year 2010,
the size and age distribution of the drinking water population is estimated by
-------
A-83
EXHIBIT A-30
1980 U.S. POPULATION DISTRIBUTION
Age Category
(years)
1. 0-9
2. 10-19
3. 20-29
4. 30-39
5. 40-49
6. 50-59
7. 60-69
8. 70-79
9. 80-89
10. 90-99
Fraction of 1980 Birth
U.S. Population1 Rate2
0.1459 0.
0.1740 147.
0.1803 577.
0.1392 212.
0.1005 11.
0.1030 0.
0.0830 0.
0.0512 0.
0.0197 0.
0.0032 0.
1.0000
Probability of Not
Surviving to the
Next Age Group
0.0198
0.0070
0.0129
0.0159
- 0.0374
0.0892
0.1904
0.3827
0.6850
1.0000
1 Bureau of the Census, U.S. Department of Commerce, "Estimates of the
Populations of the United States, by Age, Sex and Race: 1980 to 1982."
Current Population Reports, Population Estimates and Projections Series P-25,
No. 929, May 1983, Table 4, pp. 22-23.
2 Births per 1,000 people (male and female) over the 10-year period,
U.S. Department of Health and Human Services, "Health United States 1981,"
DHHS Publication No. (PHS) 82-1232, December 1981, p. 108.
-------
EXHIBIT A-31
AGE DISTRIBUTION OF THE POPULATION OVER TIME
Age
Ca tego rv
I
2
3
t4
5
6
7
8
9
10
Age
(Years)
0-9
10-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
2000
2010
2020
0. 1459
0. 1740
0. 1803
0. 1392
0. 1005
0. 1030
0.0830
0.0512
0.0197
0.0032
0. 1606
0. 1130
0. 1728
0. 1780
0. 1370
0.0967
0.0938
0.0674
0.0316
0.0062
0. 1604
0. 1574
0. 1420
0. 1706
0. 1751
0. 1319
0.0881
0.0760
0.0416
0.0100
0. 1431
0. 1572
0. 1563
0. 1402
0. 1678
0. 1685
0. 1201
0.0713
0.0469
0.013!
0. 1445
0. 1402
0. 1561
0. 1543
0. 1379
0. 1616
0. 1535
0.0972
O.O'l'lO
0.0148
0. IU5I
0. 1417
0. 1393
0. 1541
0. 1519
0. 1328
0. 1472
0. 1243
0.0600
0.0139
Age
Category
I
2
3
4
5
6
7
8
9
10
Age
(Years)
0-9
10-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
2040
0. 1355
0. 1422
0 . 1 907
0.1375
0. 1516
0. 1462
0. 1209
0. 1191
0.0767
0.0189
0. 1327
0. 1328
0. 1412
0. 1389
0.1353
0. 1460
0.1331
0.0979
0.0735
0.0242
0.1319
0. 1301
0. 1319
0. 1394
0. 1367
0. 1302
0. 1329
0. 1078
0.0604
0.0231
0. 1263
0. 1293
0. 1292
0. 1302
0. 1372
. 0. 1316
0. 1 186
0. 1076
0.0665
0.0190
0. 1226
0. 1238
0. 1284
0. 1275
0. 1281
0. 1320
0. 1 198
0.0960
0.0664
0.0210
0. 1207
0. 1201
0. 1229
0. 1267
0.1255
0. 1233
0. 1202
0.0970
0.0593
0.0209
I
O3
F-
Age
Category
I
2
3
4
5
6
7
8
9
10
Age
(Yea rs)
0-9
10-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
2100
21 10
0. 1168
0. 1183
0. 1193
0. 1213
0. 1247
0. 1208
0. 1 123
0.0974
0 . 0599
0.0187
0.
0.
0.
0.
0.
132
145
175
177
194
0. 1201
O.I 100
0.0910
0.0601
0.0189
0. 1 109
0. 1 1 10
0. 1 137
0. 1 160
0. 1 159
0. 1 149
0. 1094
0.0891
0.0561
0.0189
0. 1078
0. 1087
0.1102
0. 1 122
0. 1 141
0. 1 1 15
0. 1047
0.0885
0.0550
0.0177
0. 1046
0. 1056
0. 1079
0. 1088
0. 1 105
0. 1099
0. 1016
0.0848
0.0547
0.0173
0. 1020
0. 1025
0. 1049
0. 1065
0. 1071
0. 1063
0. 1001
0.0822
0.0523
0.0172
0.0993
0 . 1 000
0. 1018
0. 1035
0. 1048
0. 1031
0.0968
0.0810
0.0508
0.0165
SOURCE: See text.
-------
A-85
multiplying the 1980 estimate for the facility by the age distribution
estimated for 2010. For example, if 1,000 people were simulated to be
drinking the water near the facility in I960, the estimate for 2010 would be
143.1 people in age category 1, 157.2 people in age category 2, etc., for a
total of 1,184.5 people.
This population (divided into age categories) becomes the basis for
estimating personal injury claims. For each age group, an estimate is made of
the average claim per person" which would be brought given that the individuals
in the age group consider themselves to have been harmed by the off-site
release. The average claim per person is then multiplied by the number of
people in each age group. The costs for each age group are summed to produce
an estimate of the total personal injury claim for the drinking water
population.
To estimate the average claim per person in each age group, we divided the
claims into four components:
Lost time due to disability;
Lost time due to mortality;
* Medical costs related to illness; and
Medical monitoring costs.
The first three components (lost time due to illness, lost time due to
mortality, and medical costs) are driven by the illness present in the age
group. To estimate the harms caused by the off-site release, one could
estimate the impact that exposure had on the presence of disease and illness.
The personal injury caused by the release would be the lost time and medical
costs of. the extra cases of disease and disease-related deaths caused by the
release.
This method, which is based on the incremental cases of disease, does not
reflect the fact that the incremental cases caused by the release cannot be
distinguished from the cases of the same or similar disease which would have
occurred even in the absence of a release.** The people with diseases which
would have occurred anyway (referred to as background cases of disease) will
consider their diseases to have been caused by the release. Because it is not
possible to determine whether an individual's disease was caused by the
release (background cases and incremental cases are indistinguishable), all
individuals with diseases that could be related to the release are potential
claimants. Because background cases will exceed incremental cases by several
magnitudes,1*9 background cases become the primary factor to consider when
48 There are several exceptions to this. For example, a case of liver
angiocarcinoma, if correctly diagnosed, would be almost unambiguously be
associated with exposure to vinyl chloride.
49 This will be true if exposures are in the range experienced to date
as a consequence of the majority of releases at hazardous waste sites.
Exceptions include situations involving direct contact and ingestion of high
concentrations of hazardous substances over prolonged periods of time.
-------
A-86
estimating personal injury claims. Consequently, the costs of lost time due
to illness and mortality and medical costs due to illness are based on
estimates of background cases only.50
Choice of Diseases. To implement this method based on background cases
of illness, a set of appropriate illnesses must be identified that will
potentially lead to claims. The background incidence of the chosen illnesses
forms the pool of potential claimants. Each individual with one of the
illnesses may claim for lost time due to the disability caused by the illness
and medical costs associated with treating the illness. Additionally, if an
individual dies from one of the illnesses, his or her heirs may claim for the
losses due to premature death. Clearly, the choice of illnesses will have a
significant influence on the claims estimated using this background cases
method. Each illness chosen must be potentially causable by exposure to
hazardous substances. Additionally, once the illness is diagnosed, it must
not be possible to prove that the cause of the illness was something other
than the exposure to hazardous substances. The illnesses were identified by
their three-digit International Classification of Diseases codes adapted for
use in the United States (ICDA codes). The broad categories of the chosen
diseases are listed in Exhibit A-32, along with their ICDA "codes and the
corresponding National Medical Care Utilization and .Expenditure Survey
(NMCUES) codes. As described below, the NMCUES data were used to estimate
values for simulating potential claims associated with the selected
illnesses.51 Because these broad categories of diseases are potentially
over inclusive, numerous diseases were eliminated at a more detailed level of
identification, the four-digit ICDA code level. The illnesses excluded from
the broad three-digit ICDA code categories are displayed in Exhibit A-33 along
with their four-digit ICDA codes.
Of note is that these illnesses were chosen for purposes of assessing the
claims that may arise under the PCLTF. Because the PCLTF accepts the
58 To estimate incremental cases, a-risk assessment would be performed.
This was not undertaken in this analysis because: (1) unavailable
site-specific data are necessary to model human exposures with reasonable
accuracy; (2) causal and behavioral models needed to predict incremental
incidence from exposure are relatively primitive and very imprecise; and (3)
such a complex analysis was beyond the resources available to this project.
51 U.S. Department of Health and Human Services, National Center for
Health Statistics, National Medical Care Utilization and Expenditure Survey,
Washington, 1983. The data base contains detailed information about
approximately 6,000 families (17,000 individuals). The respondents were
interviewed five times over a period of fourteen months during 1980 and early
1981. They provided extensive information on their medical conditions, health
insurance coverage, health care utilization, and expenditures. They also
provided extensive demographic information on individual sources of income,
ethnicity, age, sex, and employment status. The unique value of NMCUES lies
in its compilation of data on illnesses suffered, health care charges
incurred, and sources of payments for the health care changes.
-------
EXHIBIT A-32
ILLNESSES USED TO ESTIMATE PERSONAL INJURY CLAIMS
jmgjES Condit ion
SWCUES Code
ICDA Code
Malignant neoplasm oE lip, oral cavity and pharynx OB
Malignant neoplasm of digestive organs and peritoneum 09
Malignant neoplasm of respiratory and intrathoracic organs 10
Malignant neoplasm of bone, connective tissue, skin and breast 11
Malignant neoplasm of genitourinary organs 12
Malignant neoplasm of other and unspecified sites 13
Malignant neoplasm of lymphatic and haemopoietic tissue 14
Benign neoplasm 15
Carcinoma in situ 16
Other and unspecified neoplasm 17
Endocrine and metabolic diseases, immunity disorders 18
Diseases of blood and blood-forming organs 20
Mental disorders 21
Diseases of the nervous system 22
Disorders of the- eye and adnexa 23
Diseases of the ear and mastoid process 24
Diseases of pulmonary circulation and other forms of heart disease 28
Diseases of the upper respiratory tract 31
Other diseases of the respiratory system 32
Diseases of other parts of the digestive system 34
Diseases of urinary systems 35
Diseases of male genital organs 36
Diseases of female genital organs 37
Abortion 38
Diseases of skin and subcutaneous tissue 42
Congenital anomalies . 44
Certain conditions originating in the perinatal period 45
Signs, symptons and ill-defined conditions 46
Poisonings and other toxic effects 53
140-149
150, 151, 153-155, 157-159
160-162, 171
170, 172-174
180, 183-185, 187-189
190, 191, 193, 195-199
200-202, 204, 205, 208
211, 214-217, 219-223, 225, 226, 228, 229
230, 234
235, 236, 238, 239, 600
242-244, 246, 251-256, 258, 259, 270, 273, 275, 277, 279
280-282, 285-289
296, 300, 301, 303, 306-309, 312-316, 390, 905
331-337, 345-347, 349-358, 382
360, 364, 368, 370, 373, 377, 379, 437, 930, 117
322, 384, 386-388
415-417, 429, 459, 492
472, 477, 478, 491, 493
382, 388, 465, 496, 511, 515
530, 535-537, 555, 556, 558, 562, 564, 570, 571, 573, 575-577, 596
437, 581-586, 593, 596, 598, 599
601, 602, 605, 606, 608
620, 621, 623-626, 628, 629, 614-617
632, 634, 637
692, 693, 695, 696, 698, 701, 702, 704-706, 709, 787, 995
742-748, 750-753, 756, 757, 759
760-779
780-791, 793, 795-799
473, 987, 9B9
>
00
-------
A-88
EXHIBIT A-33
FOUR-DIGIT ICDA CODES EXCLUDED FROM
DISABILITY AND MEDICAL COST ANALYSIS
Illness ICDA Code
Toxic diffuse goitre ' 242.0
Sickle-cell trait 282.5
Sickle-cell anemia 282.6
Acute posthaemorrhagic anemia 285.1
Lesion of ulnar nerve 254.2
Lesion of radial nerve, other 354.3, 354.8
Hordeolum and other deep inflammation of eyelid 373.1
Other infective dermatitis of eyelid 373.3
Nystagmus and other irregular eye movements 379.5
Otosclerosis involving oval window, nonobliterative 387.1
Simple chronic bronchitis 491.0
Oesophagitis 530.1
Ulcer of oesophagus 530.2
Alcoholic gastritis 535.3
Postgastric surgery syndromes 564.2
Vomiting following gastrointestinal surgery 564.3
Other digestive system postoperative functional disorders 564.4
Megacolon, other than hirschsprung's 564.7
Alcoholic fatty liver 571.0
Acute alcoholic hepatitis 571.1
Alcoholic cirrhosis of liver 571.2
Alcholic liver damage, unspecified 571.3
Acute cholecystitis 575.0
Hydrops of gallbladder 575.3
Perforation of gallbladder 575.4
Fistula of gallbladder . 575.5
Cholesterolosis of gallbladder 575.6
Postcholecystectomy syndrome 576.0
Cholangitis 576.1
Perforation of bile duct 576.3
Fistula of bile duct 576.4
Acute pancreatitis 577.0
Acute prostatitis 601.0
Chronic prostatitis 601.1
Abscess of prostate 601.2
Prostatocyst it is 601.3
Prostatitis in diseases classified elsewhere 601.4
Seminal vesiculitis 608.0
Spermatocele 608.1
Torsion of testis 608.2
Other inflammatory disorders of male genital organs 608.4
Stricture or atresia of vagina 623.2
Tight hymenal ring 623.3
Old vaginal laceration 623.4
-------
A-89
EXHIBIT A-33 (continued)
FOUR-DIGIT 1CDA CODES EXCLUDED FROM
DISABILITY AND MEDICAL COST ANALYSIS
Illness
Vaginal haematoma
Polyp of vagina
Hypertrophy of clitoris
Hypertrophy of labia
Old laceration of scarring vulva
Maematoma of vulva
Polyp of labia and vulva
Excessive or frequent menstruation
Puberty bleeding
Postcoital bleeding
Associated with anovulation
Of pituitary-hypothalamic origin
Of tubal origin
Of uterine origin
Of cervical or vaginal origin
Ac.ute salpingitis and oophoritis
Lupus erythematosus
Lichenification and lichens implex chronicus
Unspecified
Prickly heat
Coma and stupor
Syncope and collapse
Hyperhidrosis
Other general symptoms
Functional and undiagnosed cardiac murmurs
Other abnormal heart sounds
Gangrene
Shock without mention of trauma
Venom poisoning
Abnormal glucose tolerance-test
Excessive blood level of alcohol
Bacteraemia, unspecified
Viraemia, unspecified
Nonspecific reaction to tuberculin test
False positive serological test for syphilis
ICDA Code
.6
.7
.2
623,
623.
624.
624.3
624.4
624.5
624.6
626.2
626.3
626.7
628.0
628.1
628.2
628.3
628.4
614.0
695.4
698.3
698.9
705.1
780.0
780.2
780.8
780.9
785
785
785
785
789
790
790
790
790.8
795.5
795.6
-------
A-90
liability of owner/operators established under CERCLA or any other state or
federal law, the illness categories are fairly broad. A program designed
specifically to compensate individuals harmed due to exposure to uncontrolled
releases of hazardous substances may be designed more narrowly. Fewer
illnesses may be designated as eligible for compensation. Also, diagnostic
criteria for those illnesses may be established to help ensure that only those
individuals who were harmed are compensated. These measures would narrow the
applicability of the program, and would, if successfully implemented, result
in expenditures less than the claims estimated with the background cases
method used here.
Several types of claims might arise under the PCLTF. For this study,
individuals who have been harmed due to one of the selected illnesses are
simulated to seek compensation for the following losses, as applicable:
time lost due to disability;
time lost due to premature death;
medical charges; and
* medical monitoring charges.
Each is discussed below.
Time Lost Due to Disability. Individuals who suffer from one of the
selected diseases may become temporarily or permanently disabled, resulting in
lost work days, lost time for homemaking, lost leisure time, or reduced
enjoyment of leisure. The amount that people would claim for these losses (in
dollars) is uncertain. Clearly, the severity and duration of the disability
are important factors. Individuals fully disabled for extended periods of
time would be expected to claim for larger amounts than individuals partially
disabled for short periods.
Ideally, to simulate claims we would like to estimate the value that
disabled individuals place on the time lost due to their disability. This
value would be the amount of money necessary to make the individual "whole;"
i.e., the amount needed to return the individual to his or her pre-disability
condition. When the disability is irreversible, there is no amount of money
capable of making the individual whole in the physical sense. Consequently,
"wholeness" must be interpreted more broadly so that an appropriate level of
compensation may be established.
There is no established procedure for evaluating the total compensation
(in dollars) suitable for various disabilities. Every individual will have
his or her own unique circumstances that would affect compensation levels. As
a lower bound, we focused on one fairly uncontroversial component of losses
due to disability: lost wages. Currently there are several government
programs that routinely compensate disabled individuals a portion of their
pre-disability wages (e.g., workers' compensation programs and the social
security disability program). Therefore, for purposes of simulating claims,
the amount of wages lost due to disability was used.
To estimate lost wages we first estimated lost work days from the NMCUES
data. These data include information on lost work days by ICDA code and by
-------
A-91
age category for individuals age 14 and older. The following procedure was
used to estimate work days lost per day per person due to one of the selected
illnesses:
Step 1: We calculated the total lost work days for the
entire population (over age 14) by counting all
lost work days reported in the survey that listed a
chosen disease as the condition causing the lost
work days.
Step 2: The distribution of the total lost workdays across
the age categories was estimated using the broad
classifications of diseases.sa
Step 3: The percentages of lost workdays by age from step
two were multiplied by the aggregate lost workdays
of step one to estimate lost work days by age.
Step 4: For each age group, the results from step three,
lost work days by age, were divided by the employed
population for that age group in 1980.S3 This
yielded an estimate of lost work days per person in
1980.
Exhibit A-34 displays the estimated average number of lost work days per
person associated with the chosen diseases by age category.
To value one day of lost work we used one-fifth of the median weekly wage
per person by age, shown in Exhibit A-34.5* To simulate claims we applied
these estimates of lost work days and wages to all persons exposed to the
hazardous wastes, not just those who had been working before the illness
occurred. This will overstate wage losses due to these illnesses. However,
it reflects the claims that may be brought by individuals who do not work for
wages (e.g., those who work inside the home). Consequently, because these
individuals would be affected by these diseases in the same manner as those
52 It was not possible to estimate the distribution of lost work days
across the age categories after screening out the four-digit ICDA code
illnesses in Exhibit A-33. Consequently, the distribution by age is based on
a broader definition of diseases than is the total number of lost work days.
53 1980 employed population figures taken from U.S. Department of Labor,
Bureau of Labor Statistics, Employment and Earnings, January 1981.
511 1982 median weekly earnings for full time wage and salary workers,
Unpublished Bureau of Labor Statistics data.
-------
EXHIBIT A-34
DATA USED TO ESTIMATE PERSONAL INJURY CLAIMS I/
AGE
0-9
10-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
Work Loss
Days
(per person.
per year)
0.0
0.55
1.22
0.95
1.20
1.76
2. 17
0
0
0
Va.lue per
Work Loss Day
(1982 S)
0
13
55
68
71
69
65
0
0
0
Fraction of
Deaths Due
to Relevant
Diseases
.7796
.1867
. 1705
.3089
.4228
.4805
.4685
.4121
.3569
.2764
Rema i n i rig
Years
of Life
70
61
51
42
33
24
(7
11
3
0
Med ica 1
Costs
(per person,
per year)
( 1982 S)
145
82
142
226
240
378
686
1, 139
1 , 1 40
767
Med ica 1
Moni tor ing
Costs
(per person,
per year}
(1982 S)
160
160
160
160
160
(60
160
160
160
(60
1
VO
K)
For sources, see text.
-------
A-93
people in the workforce, this method provides a reasonable estimate of this
component of the claims.55
In simulating future claims, we must recognize that real wage rates are
projected'to increase over time. To account for the changes in wage over
time, the median wages shown in Exhibit A-34 are increased using rates based"
on economic projections of real wages over time.56
Time Lost Due to Premature Death. Some individuals may die from one of
the selected illnesses. Survivors may submit a claim for compensation for the
death. In estimating the value of such a claim, we did not attempt to value a
life. Rather, our objective was to estimate the likely value of a successful
claim. As is the case with disability, no amount of money can compensate for
the premature death of an individual. Again, as a lower bound, we focused on
the lost potential wages due to the individual's premature death. Clearly,
this loss is only one of the losses resulting from the death.
To estimate these lost wages, we first estimated the fraction of all
deaths caused by the diseases of interest. These estimates, also shown in
Exhibit A-34, were made for each age group; for example, 17.05 percent of all
deaths of individuals between 20 and 29 were due to one of the diseases shown
in Exhibit A-32. These values were estimated by: (1) calculating the number
of deaths in each age category using the ICDA codes for the selected
illnesses; and (2) dividing the number of deaths due to selected diseases by
the total number of deaths in the given age category.57
The number of premature deaths caused by the chosen diseases is simulated
by multiplying the percentages of deaths due to selected diseases by the death
rates found in Exhibit A-30. For example, if 17.05 percent of all deaths of
individuals between 20 and 29 are due to one of the selected diseases, and if
55 The economists' appropriate measure of such costs is consumer
surplus, an extremely difficult concept to implement. We use the methods
discussed here because they are most likely to be used in the process of
actually assigning loss.
56 Projected increases in real-wages are taken from House Document
98-199, 1984 Annual Report of the Board of Trustees of the Federal Hospital
Insurance Trust Fund, Scenario II-B, 98th Congress, 2nd session, 1984. An
average annual increase of 1.15 percent is used for the first ten years; 1.5
percent per year is used for the next ninety years (which is the long-term
projected rate of increase); and the rate is zero thereafter.
57 U.S. Department of Human Health and Services, National Center for
Health Statistics, Vital Statistics of the United States. 1979, Volume II -
Mortality. Table 1-25, 1984. NMCUES data do not report mortality rates.
Because of the need to use a data source other than NMCUES, the ICDA codes
used to estimate mortality rates are not identical to those used to estimate
medical costs and work loss days. However, the variation in use of codes is
believed to have a negligible effect on the estimated death rates.
-------
A-94
1.29 percent of the people in this age group do not survive to the next age
group (30 to 39) then: .1705 x .0129 = .0022 of the population aged 20 to 29
is estimated to die from one of the selected diseases in any given decade.
The time lost as the result of these deaths is estimated as the expected years
of life remaining for each age category, as shown in Exhibit A-34.58
The above procedure produces an estimate of years of life lost as the
result of the premature deaths caused by the background cases of the chosen
diseases. To simulate claims, the loss per year is valued at 52 times the
median weekly wage for each age group (the wage data used for the disability
claims are also used here, see above). For example, an individual who dies at
age 30 (and thus has an expected remaining life of 42 years) receives 10 years
of compensation at the rate of $68 per work day, ten years at $71, ten years
at $69, ten years at $65, and 2 years at zero compensation. In simulating
claims, these values are also increased as appropriate to reflect projected
increases in real wage rates. These values clearly underestimate the actual
costs of premature deaths. However, they are reasonable lower bounds to
potential claims.
Medical Costs Related to Selected Illnesses. A major source of claims
under the PCLTF could be reimbursement for medical costs associated with the
selected illnesses. To estimate annual per capita medical costs, we began
with an estimate of total medical charges in 1980, developed from the NMCUES
data. We used a process similar to the process used to estimate lost work
days, with an additional step.
Step 1: Medical charges were estimated for the entire
population for the selected conditions listed in
Exhibit A-32 (and screened for those in Exhibit
A-33).
Step 2: The charges estimated in step 1 were adjusted
upwards to account for undercounting of government
charges for certain illnesses.59 This
undercounting of government charges occurred for a
number of reasons; for example, many individuals in
the NMCUES survey were unaware of the extent of the
Medicare and Medicaid costs of their illnesses.
58 For the ages up to 79 the following publication was used to computed
expected remaining life: U.S. Department of Health and Human Services,
National Center for Health Statistics, Vital Statistics of the United States,
1979. Volume II - Mortality. Table 6-3, 1984. For age categories 80-89 and
over 90 the death rates (found in Exhibit A-30) were used to estimate
remaining life.
59 Upwards adjustments based on ICF analysis of NMCUES data and
estimates prepared by Dr. Thomas Hodgson of the National Center for Health
Statistics, based on National Health Accounts data, compiled by the Health
Care Financing Administration.
-------
A-95
This procedure produced an adjusted estimate of the
total aggregate medical charges for the entire
population.
Step 3: The distribution of medical charges across the age
categories was estimated using the broad
classification of diseases.60
Step 4: The percentage of charges by age from step 3 were
multiplied by the total charges of step 2 to
estimate total medical charges by age.
Step 5: In each age group, the total medical charges were
divided by 1980 population figures for that age
group to calculate per capita medical charges by
age group.*1
These five steps produce estimates of medical charges in 1980 (the year of the
NMCUES Survey). Because the PCLTF Model is based on 1982 dollars, the NMCUES
charges were adjusted to reflect 1982 dollars and 1982 health care
expenditures.62
In simulating future medical cost claims we must recognize that health
care costs and expenditures do not remain constant in real terms.
Historically, the prices for health care services have increased more rapidly
than the prices of goods in general. Also, health care expenditures have
increased as the age mix of the population changed and as the intensity in the
use of health care services has increased. Potential future changes in the
age mix of the population are simulated using the population dynamics matrix
(see Exhibit A-31). However, changes in prices and intensity had to be
incorporated.
"Intensity" is the amount of medical services used to treat given
conditions. It is believed that intensity has historically increased at a
60 It was not possible to estimate the distribution of medical charges
by the age categories after screening out the four-digit ICDA code illnesses
in Exhibit A-33. Consequently, the distribution by age is based on a broader
definition of diseases than is the estimate of the total medical charges.
61 Resident population of the United States by age as of 4/1/80, taken
from "Estimates of the Population of the United States, By Age, Sex, and
Race: 1980-1982," Current Population Reports, Population Estimates and
Projection Series P-25, No. 929, May 1983, Table 4, pp. 22-23.
s2 Adjustments based on data from Statistical Abstract of the United
States, 1984 edition, p. 103. Health care expenditures increased 30.77
percent from 1980 to 1982, taking into account population growth, changes in
intensity of use, and changes in price. We assumed that the costs in all age
categories increased at the same rate.
-------
rate of approximately one percent per year, and is projected to continue to
increase at the same rate.63 The model thus allows for a one percent
increase in intensity each year. Increases in real prices for medical
services likewise contribute to growing medical charges over time. Prices for
medical services have historically risen an average of 1.1 percent per year
faster than inf lation. 6I* This figure is used as the long-term projected
increase in medical costs (i.e., from year 41 to year 100). The annual
increases in prices for medical services are projected to vary in the near
term, beginning at 2.0 percent and falling to 1.3 percent by model year
31. 6S After year 100, we assume a zero percent real increase in prices.
Medical Monitoring. The final component of the personal injury claim
is medical monitoring costs. The purpose of this monitoring is to keep watch
on a person's health over time to determine if the person contracts a disease
due to exposure to the environmental pollutants. Symptoms may or may not
exist, and consequently, an examination must be performed which.includes
non-routine tests and x-rays, such as tests of liver and kidney functions and
chest x-rays.
The inclusion of the monitoring costs in our analysis of personal injury
claims is warranted because:
past settlements for harms due to occupational exposure
to toxic substances have included medical monitoring
costs; and
claims currently pending for harms due to exposure to
environmental pollutants include medical monitoring.
Consequently, it is likely that these costs will be claimed by people who
consider themselves to have been put at risk by their exposure to the off-site
release. Of note is that this monitoring is provided for all individuals, not
just those with evidence of illness.
The cost of medical monitoring is likely to range from $70 to $250 per
person per year. The lower .estimate is for a routine exam in a clinic (which
is the type of exam often provided by employers to their employees who work
with toxic substances). The higher estimate reflects the total cost of tests
and examination by a private physician in an urban area. The average of these
two estimates, $160, was used (see Exhibit A-34).
63 House Document 98-199, 1984 Annual Report of the Board of Trustees
of the Federal Hospital Insurance Trust Fund. Scenario II-B, 1984, p. 53.
6* Based on a comparison of the Consumer Price Index (CPI) and the
Medical Care Price Index for 1940 to 1983. See Economic Report of the
President. 1984. p. 279.
65 House Document 98-199, op. cit., p. 53.
-------
A-97
Using the costs displayed in Exhibit A-34 and the population over time
displayed in Exhibit A-31, the total personal injury claim was computed using
the following assumptions:
personal injury claims are requested as lump sum
payments by all individuals who believe themselves to
have been harmed;
the lump sum payments reflect the future expected costs
for each individual for the remainder of his life;
only 75 percent of those eligible for medical
monitoring make claims; and
the lump sum payment is computed by taking the net
present value of costs anticipated in the future.
For example, to estimate the personal injury claims from an off-site release
in 2010, the following calculations are performed:
Identify the drinking water population in 1980 (this
is a facility.-level attribute, assume 1,000 people for
this example).
Calculate the number of people in each age group- in
2010. For example, from Exhibit A-31, there are 1,000 x
0.1678 = 167.8 people in age group 5 in 2010.
Calculate the stream of direct medical costs, the
value of lost time due to disability and premature
death, and medical monitoring costs for each age group
over time.
» Discount the cost streams for each age group to a net
present value using the real discount rate employed in
the Model.
Sum the net present values for each age group to
estimate the total personal injury claim.
Although the approach used here produces reasonable estimates of potential
personal injury claims, there are still a variety of data problems and
uncertainties. In particular, the simulation of the number of people who
consider themselves to have been exposed is based on very limited data from
site surveys and the drinking water survey. No adjustments are made for
changes in water usage near facilities over time. Additionally, several of
the estimates of direct medical costs are based on gross disease categories
and are very uncertain. The background incidence of disease and the costs of
diseases are not adjusted over time. Finally, cost components for pain and
suffering are not included. Claims for pain and suffering may be large, and
if they are awarded, the estimates reported here may be too low.
-------
A-98
A.2.5.2 Real Property Damage Claims
Real property damage claims may include the following:
diminished market value;
impaired use or interrupted enjoyment;
destruction or contamination; and
restoration of property.
Diminished market values may result from any detected gradual release to
groundwater, on site or off site. On-site releases may cause reductions in
surrounding property values because people may perceive some increased risk
presented by the site. This proposition is somewhat speculative and we have
not identified any documented cases of reductions in property value caused
solely by on-site releases. Therefore, for the current simulations, we do not
model real property damages due to on-site releases.
Off-site releases may reduce the value of property near the site. If the
release comes in contact with people's property, the costs of restoring the
property to its original condition could fall under the category of real
property damage claims. In the development of the PCLTF Simulation Model, we
did not estimate this type of real property damage because it is too site
specific to be considered in our general approach and may be unlikely to occur
as the result of the releases into groundwater considered in the Model. Also
excluded was consideration of damages to groundwater that is not currently
used as a water supply. (This unused groundwater can be considered to be real
property.) The costs of replacing contaminated groundwater which was in use
are included under economic loss claims described below. The exclusion of
damages to groundwater currently not in use biases our results downward.
Additional research may be useful for estimating the importance of this bias.
The real property damage caused by off-site releases is limited to a
simulated reduction in property value. Conceptually, one could envision that
all the property within a given distance of a site which is found to be
releasing hazardous waste off site may be affected. Within this given
distance, property values may drop by an average of some percent relative to
what they were prior to the discovery of the off-site release. The actual
percentage reduction will be a function of the perceived risk of the release.
A catastrophic release could result in permanent evacuation with property
values going to zero (e.g., the inner rings at the abandoned Love Canal). A
less serious release could cause very modest (and perhaps temporary) changes
in market values. Based on this general understanding that off-site releases
affect property values, the parameters necessary to model this phenomenon are
as follows:
distance from the site that is affected;
pre-release property values;
type of off-site release (i.e., toxic or detectable);
-------
A-99
whether property values are affected by the releases;
and
percentage by which property values fall due to the
discovery of the release for both types of off-site
releases.
Each of these parameters is discussed below in turn.
Any property closer to the site than the place where the release is
detected is likely to experience some decline in property values.
Additionally, some properties farther away from the site than the detection
point may also be affected. For off-site releases, we assume the affected
area to be within one-and-one-half times the distance to the point of
detection. (The distance to the point of detection is the distance discussed
in Section A.2.3, Modeling of Releases.)
Clearly, the choice of one-and-one-half times the distance to the point of
detection of off-site releases is somewhat speculative. Additional research
would be useful to identify more precisely the expected market reaction to
information regarding a release of hazardous waste. Nevertheless, the value
is reasonable in that it is greater than one (which could be considered a
lower bound), is not excessively large (the size of the area increases faster
than the distance chosen to define the area),66 and utilizes the distance to
the potable well which is assumed to be the point of discovery.
The pre-release property values are required as a base from which to
estimate the reduction in property values. The County and City Data Book,
1977 (Data Book)67 was used to obtain median housing values and average
farm values (per acre) on a county basis for the entire country.
Additionally, the Data Book reports the percentage of area in each county that
is farmland. Assuming that the county is homogeneous, the value of any given
area can be computed as:
Value per unit area = (1.0-farm percentage)(housing density
per unit area)(housing value) + (farm
percentage)(farm value per unit area).
66 The size of the affected area increases more than twice as fast as
the radius chosen to define the area. Therefore, a small increase in the
radial distance can mean a much larger area. Algebraically this can be shown
as follows: let initial radius = R; initial area = TT R2; increase in
radius « AR; and increase in area = ir(R+AR)2-irR2 = ir(2RAR+AR2).
The proportional increase in the radius is AR/R. The proportional
increase in the area is more than twice as large and is equal to
ir(2RAR+AR2)/itR2 * 2AR/R+(AR/R)2.
57 County and City Data Book_, 1977 and Data Book, A Statistical
Abstract Supplement, U.S. Department of Commerce, 1977.
-------
A-100
This valuation technique omits several important considerations:
" counties are not homogeneous;
commercial or industrial facilities will be within many
affected areas; and
land values near the site may be well below the county
median and average, possibly as a consequence of the
site being there, or possibly the low land values were
an impetus for locating the site there initially.
Additional parameters describing the initial housing and farmland values
relative to the county median and average were used to provide a partial
correction for these considerations. For the current simulations, we assumed
that housing and farmland values near facilities will each be 50 percent of
the county-wide values.68 Although this does not incorporate industrial or
commercial facilities, it does recognize the fact that land near a facility is
likely to be of lower value than the average price of land in a county. If
the industrial use of land is not severely restricted by the discovery of an
off-site release (e.g., the industry itself may be the owner of the site, and
it may utilize numerous hazardous substances), then the omission of industrial
facilities may not introduce a significant bias in our estimates.
The percentages by which property values fall due to the discovery of
on-site and off-site releases are crucial parameters about which we have
uncovered very little data. Toxic constituents off site are assumed to cause
an average of a 30 percent decline in the pre-release value of the property
surrounding the site. Detectable constituents off site are assumed to cause
an average of a 15 percent decline in the pre-release value of the property
surrounding the site. The assumptions of 15 and 30 percent, although highly
speculative, were chosen to indicate that while the land need not lose all of
its value, a substantial reduction is possible. The reductions in property
values need not be uniform within the affected area. The percentage of value
reduction is meant to represent an average for the affected area.
Finally, the frequency with which property values are affected by releases
may be specified. The frequency of real property damage resulting from
off-site releases is set to 1.0, meaning that each release off-site results in
the reductions in property values outlined above. Exhibit A-35 summarizes the
values used to calculate real property damages.
6* Different fractions may be chosen for housing and farm values. The
choice of 50 percent is speculative; additional research would be valuable in
refining this assumption.
-------
A-101
EXHIBIT A-35
SUMMARY OF ASSUMED PARAMETERS USED TO
ESTIMATE REAL PROPERTY DAMAGES
Initial housing values near the site expressed as 0.50
fraction of median county values (0-1)
Initial farmland values near the site expressed 0.50
as a fraction of average county values (0-1)
Fraction of property value lost due to detectable 0.15
release off site
Fraction of property value lost due to detectable 0.30
and toxic release off site
Frequency of real property claims due to detect- 1.00
able release off site
Frequency of real property claims due to detect- 1.00
able and toxic release off site
-------
A-102
The real property claim is calculated using the following three equations:
VALUE1 = (1.0 - FARMPT) x HOSDEN x MEDHOS x HOSVAL (1)
where:
FARMPT = percentage of land w-ithin county which is
farmland
HOSDEN = housing density per unit area within county
MEDHOS = median value of houses within county
HOSVAL = housing values near the site expressed as a
fraction of the county median (assumed to be 0.5)
VALUE2 = FARMPT x MEDFAR x FARVAL (2)
where:
FARMPT = percentage of land within county which is
farmland
MEDFAR = average farm value per unit area
FARVAL = farm values near site expressed as a fraction of
the county average (assumed to be 0.5)
Real Property claim = RPDEV x AREA x (VALUE1 + VALUE2) (3)
where:
RPDEV = property devaluation due to release off site
(assumed to be 0.15 for detectable releases off
site and 0.30 for toxic releases off site).
AREA = affected area as defined above.69
A.2.5.3 Economic Loss Claims
Economic loss claims may include lost productive capacity of income-
producing property and the replacement of a contaminated water supply. We
have concentrated only on the water supply component of economic loss claims
because it is the only aspect about which data are available. Consequently,
the estimates developed may be biased downward.
69 Affected area is. in the shape of a "doughnut," with the circumference
defining the boundary of the affected property and the hole being the
facility. The facility size is adjusted by 15 percent to account for the area
between processes and administrative buildings.
-------
A-103
The capital and O&M costs of providing an alternate supply of water are
presented in Exhibit A-36. These costs were developed based on the following
assumptions:7B
Tapping a new source would require a raw water transmission main
of between 5 and 10 miles, with an average of 7.5 miles.
Alternative water source meets all drinking water standards.
Costs associated with acquiring new land or easements for the
transmission main are not included due to the site-specific
nature of these costs; however, these costs could be considerable.
The existing treatment and distribution facilities would be
used with the new source without major operating changes.
* For surface water sources:
-- Raw water intake facilities would be installed.
-- Raw water low-lift pumping at a total dynamic head of 50 feet
(average) would be required.
For groundwater sources:
New wells at a maximum capacity of 1 mgd each would be
installed.
The costs of a geologic study were not included because of
the site-specific nature of these studies; however, these
costs could be considerable.
We assumed that individuals will not be compensated for O&M costs, so that
the capital costs displayed in the exhibit are used as the estimate of the
economic loss claim. The drinking water population simulated at the facility
level in 1980 and adjusted for changes in population over time (see personal
injury claims) was used to estimate the size of the water supply required.
A.2.5.4 Natural Resource Damage Claims
Natural resource claims may include compensation for the destruction of
publicly-owned groundwater, surface water, natural resources, and recreation
70 U.S. Environmental Protection Agency, Technology and Costs for the
Removal of Fluoride from Potable WaterSupplies. Final Draft, July 1983, pp,
58-59.
-------
A-104
EXHIBIT A-36
COSTS OF ALTERNATIVE WATER SUPPLY
(thousands of 1982 dollars)
Population Served Capital Costs O&M Costs
50-99 300.0 9.3
100-499 360.0 11.2
500-999 540.0 17.9
1,000-2,499 910.0 34.1
2,500-4,999 1,585.0 52.8
5,000-9,000 2,120.0 87.1
10,000-99,999 9,135.0 269.0
100,000-999,999 45,265.0 1,760.0
greater than 1,000,000 . 292,115.0 15,900.0
SOURCE: U.S. Environmental Protection Agency, Technologies and Costs for the
Removal of Fluoride from Potable Water Supplies, Final Draft, July,
1983, pp. 58-78.
-------
A-105
areas.71 These costs will be highly site specific, depending on the
proximity of the release to areas controlled or owned by states or the federal
government.
The largest costs in this category would likely be the costs of cleaning
up contaminated groundwater outside the property boundaries of the facility."
These costs could be considered to be natural resource damages because
groundwater is a natural resource and in most states, the use of groundwater
is controlled by the state. Consequently, the state could request that
response actions be taken to restore this resource.
In developing the Model, we assumed that groundwater cleanup actions would
be required even if the state did not request them. Other parties could
require that these actions be initiated (e.g., EPA, property owners) so it was
assumed that these actions would be undertaken as a response to the detection
of off-site releases. Their costs are counted under the category of response
costs, and including them again under natural resource damages would be double
counting. Therefore, the natural resource damages do not include the costs of
cleaning up groundwater.
Similarly, the costs of providing an alternative water supply are not
counted here because they are included in economic loss claims. Therefore,
natural resource damages are limited to the following three components:
cost of restoring areas of contaminated surface
water;
value of lost recreation; and
« cost of fish kills.
To estimate the costs of restoring areas of contaminated surface water, we
assumed that the primary cost would be the dredging and disposing of the
contaminated material in the river bed or lake bottom. Dredging and disposal
costs72 are as follows:
Removal of contaminated river $3.00 per cubic yard
bed or lake bottom (less than
40-50 feet deep)
71 CERCLA section lll(b): "... claims resulting from a release or
threat of release of a hazardous substance ... may be asserted against the
Fund under this title for injury to, or destruction or loss of, natural
resources, including cost for damage assessment: Provided, however, that any
such claim may be asserted only by the President, as trustee, for natural
resources over which the United States has sovereign rights, ...', or by any
State for natural resources within the boundary of that State belonging to,
managed by, controlled by, or appertaining to the State."
72
Personal communication with Army Corps of Engineers.
-------
A-106
Transport costs of contaminated $55.00 per cubic meter
water to a land disposal facility ($42.00 per cubic yard)
Disposal costs of contaminated $100.00 per cubic meter
material ($76.50 per cubic yard)
Using these values, we assumed the contaminated area would be on the order of
one acre and that the bottom material would have to be dredged to a depth of 3
feet. The cost of dredging is therefore estimated as:
cost = (4,840 cubic yards of material) x (3.00 4- 42.00 +
76.50 dollars per cubic yard),
or approximately $588,000. This estimate is clearly speculative because the
size of the area is assumed to be one acre. If the area is larger, or if more
activities would be performed, this cost would be much higher.
Estimates of the value of recreation include:73
* General recreation: $1.60 to $4.80 per person per day
(1982 dollars).
« Specialized recreation: $6.50 to $19.00 per person
per day (1982 dollars).
Using the median value of $3.20 per day for general recreation, we estimated
the recreational loss per release using the following assumptions:
affected area will be unusable for a year;
* the area would have been used by 50 people per
weekend; and
weekend use accounts for half of total use.
Using the above assumptions, we calculated recreational loss as follows:
(50 people/weekend) x (52 weekends/year) x 2 (to reflect use during the week)
= 5,200 person days. Multiplying the number of person days by $3.20 yields
$16,640 (1982 dollars).
Although this number is very small, one must recognize that the recreation
loss as the result of the types of releases considered in the Model is likely
to be very localized. Because the affected area is likely to be small, it is
likely that there will be numerous substitute recreation areas available at
little or no incremental cost to the user. The value of the lost recreation
could therefore be very small.
73 Ibid.
-------
A-107
The cost of fish kills was estimated by investigating past cases where
surface water areas of less than an acre were affected. Exhibit A-37 presents
the cases on which we based our analysis. Based on these cases, we assumed
3,000 fish would be killed per release into surface water. Given that the
value per fish is generally below $1.00J7fc we assumed a value of $0.50.
Therefore, the fish kill component of the natural resource damage is $1,500.'
The total natural resource damage per off-site release is assumed to be
the sum of the three estimates ($588,000 + $16,640 + $1,500) or approximately
$606,000. This value is used in the Simulation 1 run of the Model.
As is clearly evident from the above discussion, the natural resource
damage estimate is based on speculative assumptions. Additional research may
yield improved estimates of the likely range of these costs.
A.2.6 Funding Sources
The potential funding sources available to cover the costs of actions and
claims are the owner/operators, PCLTF, Superfund, and state funds. Below, we
describe each funding source. The manner in which costs are allocated to the
funding sources is discussed in section A.3.4, Cost Allocation Policy.
Owner/Operator
For the owner/operator to be available as a funding source, one of the
following two conditions must be met:
the owner/operator has an active financial assurance
mechanism (such as insurance or a trust fund) dedicated
to covering the type of cost being allocated; or
the owner/operator is still in business and
consequently pays the cost of the action or claim.
These characteristics are simulated in the facility-level characterization
(see section A.2.2). It should be noted that financial assurance is currently
only required for routine monitoring and care during the post-closure period.
Consequently, most costs would not be covered by a financial assurance
mechanism, and owner/operators must remain in business in order to cover costs.
PCLTF
The PCLTF will cover costs at a facility only if the facility is qualified
for PCLTF coverage. Under the current PCLTF statute, the Fund will cover all
costs for actions and claims following the end of the post-closure period.
Prior to this point, only third-party claims are covered at qualified
facilities.
7* American Fisheries Society, "Monetary Values of Freshwater Fish and
Fish-Kill Counting Guidelines," Publication No. 13 (155N00097-0638), 1982.
-------
A-108
EXHIBIT A-37
DATA ON FISH KILLS
Cause
Pesticides
Fertilizers
Manure Drainage
Food
Chemicals
Petroleum
Metals
Other
# of Cases
9
4
2
1
4
3
2
1
# of Fish Killed
38,068
15,554
3,200
150
7,862
2,269
9,050
100
Average Kill/Case
4,229
3,888
1,600
150
1,968
756
4,525
100
TOTAL
26
76,263
2,933
SOURCE: U.S. EPA, "Fish Kills Caused by Pollution in 1976."
-------
A-109
Superfund
The Superfund may be specified as a funding source for action and claims
costs. In current simulations, Superfund is designated to cover non-routine
response costs which are not covered by owner/operators at facilities not
covered by the PCLTF. Additionally, the Superfund may cover the costs of
natural resource damages.
State Funds
The availability of state funds to cover costs for actions and claims is
modeled using two sets of prototype state coverage regimes. One set of
regimes describes state coverage for claims and response actions. The second
describes its coverage for monitoring and post-closure care actions. Each is
discussed below.
State Coverage Regimes for Third-Party Claims and Response Costs. For
facilities that qualify to transfer their liability to the PCLTF, the PCLTF
assumes the costs of legally valid claims and response actions. If a facility
does not qualify to transfer its liability to the PCLTF, then a state fund
might be available to pay for certain or all of these two types of costs. To
estimate the likelihood that such state funds will be available, the laws of
all 50 states, plus the District of Columbia, Puerto Rico, and the Virgin
Islands, were surveyed from the following sources:
* Peter Guerrero and David Colbert, "State Post Closure
Approaches," Office of Solid Waste, U.S. Environmental
Protection Agency, May 10, 1982;
ICF Incorporated, Review ofState Liability
Provisions, og. cit.;
ICF Incorporated, Review ofState Closure and Post-
Closure Financial Assurance Provisions, report to U.S.
Environmental Protection Agency, June 1982; and
G. Cohen and D. Derkics, "Financial Responsibility for
Hazardous Waste Sites," 9 Capital University Law Review
509, 1980.
From this review, we developed six state coverage regimes for third-party
claims and response actions (see Exhibit A-37). The regimes divide
third-party damage claims into personal injuries and other damages.
The research revealed that most states do not have a fund that covers
third-party damages or response actions (Regime 1). Slightly less than
one-fourth of the states were found to have a fund that would cover response
actions only (Regime 2). Only three states covered any third-party claims:
one state covered all third-party damage claims, but not response costs
(Regime 6), and two states had a fund that covered response costs and third-
party claims for damages other than personal injuries (Regime 3). Exhibit
A-38 displays the resulting frequencies for each state coverage regime. These
-------
A-110
EXHIBIT A-38
STATE COVERAGE REGIMES FOR THIRD-PARTY CLAIMS
AND RESPONSE COSTS
Damage/Response
Cost Coverage
Regime
1.
2.
3.
4.
5.
6.
Coverage
No funding for third-party claims or
response actions
Funding for response actions only (when
federal funds and the owner/operator do
not pay)
Funding for response actions and third-
party claims other than personal injury
Funding for all response actions and
third-party claims
Funding for third-party claims other
than personal injury
Funding for all third-party claims
Frequency
0.71
0.23
0.04
0.00
0.00
0.02
SOURCE: ICF Incorporated estimates.
-------
A-lll
frequencies were used to probabilistically assign the regimes to the states in
each iteration of a Model run. The potential for new state funds being
established over time was not examined.
State Coverage Regimes for Routine Monitoring Actions and Post-Closure
Care. Several states have instituted mechanisms for using state funds to
pay for monitoring and post-closure care under certain conditions at closed
hazardous waste sites. Based on a survey of state post-closure provisions as
of June 1, 1982:75
41 states have no provisions for covering monitoring
and post-closure care costs; and
12 states have provisions authorizing the use of state
funds for monitoring and post-closure care costs under
certain conditions. Of the 12 states with such
provisions, 2 would provide funds even if federal funds
were available, 2 would not, and the outcome is unclear
for the remaining 8.
Based on these figures, we structured three state monitoring and
post-closure care regimes, as shown in Exhibit A-39. Each state is
probabilistically assigned a state coverage regime using the frequency shown
in the exhibit.
A.3 RELATIONSHIPS AMONG THE BASIC MODEL UNITS
This section presents the major relationships incorporated into the
Post-Closure Liability Trust Fund Simulation Model. The term relationship is
used here to describe the manner in which two or more factors are simulated to
interact. For example, Fund expenditures and revenues jointly determine the
Fund balance. Consequently, there is a relationship among these three
quantities (expenditures, revenues, and balance). This section describes how
this and other relationships are modeled.
This section is organized as follows:
Economic Relationships:
Relationship between the demand for land disposal capacity
and the facility population. The method used to simulate the
demand for land disposal capacity is presented. The manner
in which facilities are added to achieve a balance between
supply and demand is described.
-- Relationship among Fund expenditures, Fund revenues, and Fund
balance. Expenditures and revenues jointly determine the
75 ICF Incorporated, Review of State Closure and Post-Closure Financial
Assurance Provisions, op. cit.
-------
A-112
EXHIBIT A-39
STATE MONITORING AND POST-CLOSURE CARE REGIMES
State Monitoring and Frequency
Coverage Regime Post-Closure Care Coverage of Occurrence
1 None 0.78
2 That portion not covered by PCLTF 0.11
3 100% 0.11 .
SOURCE: ICF Incorporated estimates.
-------
A-113
balance. Also, the (unobligated) balance influences the tax
revenues received by the Fund. The interaction of these
quantities and the roles of inflation and interest rates are
described.
Regulatory Policy:
Permit policy. EPA's assumed permit policy and its impact on
the facility population is described.
-- Action policy. The action policy identifies the relationship
between the detection of releases and the undertaking .of
actions. The method used to define and simulate this
relationship is described.
Financial Relationships. The relationships among owner/operator
financial characteristics, owner/operator expenditures, and the
likelihood of owner/operator bankruptcy are described.
Cost Allocation Policy:
Qualification policy. The relationship between the
qualification status of a facility (i.e., whether the
facility qualifies for PCLTF coverage) and the facility's
characteristics is defined in terms of a qualification policy.
Relationship between funding sources, costs, and facilities.
The allocation of costs among the available funding sources
depends on the cost type and the facility's characteristics.
The procedure used to define the relationships is presented.
The Effect of Response Actions on Releases. The relationship
between response actions and releases is presented.
Legal Validity of Claims. The relationship between claims which
are brought by third parties and the claims for which
compensation may be obtained is described.
A.3.1 Economic Relationships
This section describes economic relationships among basic Model units.
These relationships are divided into two groups:
Relationships among market forces which control the supply of
and demand for land disposal capacity and its alternatives; and
Relationships within the structure of the Fund. Here, the two
primary components of the Fund balance -- expenditures and
revenues -- are analyzed. Inflation and interest rates are also
discussed.
Each is discussed below in turn.
-------
A-114
A.3.1.1 Relationship Between the Demand for Land Disposal
Capacity and the Facility Population
The interaction of the supply of and demand for disposal capacity plays a
critical role in Model simulations. This is true for at least three reasons:
The Model, as currently configured, simulates the opening of
new disposal facilities in response to unmet demand. This has a
direct impact on PCLTF expenditures because each facility has the
potential to qualify for Fund coverage and thus generate costs to
the Fund.
The demand for disposal is also the tax base of the Fund. If
the demand for disposal slackens (i.e., the volume of disposed
wastes decreases), the tax base shrinks and, if the tax rate
remains unchanged, Fund revenues fall.
The fact that the Model incorporates market forces allows the
user to simulate different policies that affect the price of
disposal (e.g., a change in the tax rate) as well as the demand
for disposal capacity (e.g., a ban on the land disposal of
certain wastes). Because there is considerable uncertainty about
the nature of these economic relationships, the Model allows the
use of a wide range of economic assumptions. In doing so, the
Model imposes an internal consistency which is described below.
This section presents the methods used to model both the supply of and the
demand for disposal capacity. First, demand and the factors which affect it
are discussed. Second, the Model's treatment of land disposal capacity is
discussed.
The Demand for Land Disposal Capacity
The level of demand is one factor influencing the rate at which new
facilities are added to the facility population; consequently, it influences
the level of PCLTF expenditures. Furthermore, by modeling this demand, the
implication for the PCLTF of various assumptions about the responsiveness of
demand to changes in the prices of land disposal and substitute methods of
waste management can be assessed.
The total demand for land disposal is simulated to be affected by three
factors, each of which can either increase or decrease the volume of wastes.
These factors are:
growth of industrial output and changes in technology;
price of land disposal; and
price of substitutes for land disposal.
Each factor is discussed in turn and then the method for combining them is
presented. Finally, the effect of inflation on demand is discussed.
-------
A-115
Industrial Output and Technology. The first component of demand
depends on growth in the level of industrial output and changes in
technology. If the prices of land disposal and its substitutes stay fixed and
the technical factors of waste generation are held constant, the volume of
waste would be expected to grow in rough proportion to the growth of
industrial output. With improvements in technology, the volume of waste may "
be expected to decline. Therefore, the Model calculates waste volumes using
rates of growth and/or decline supplied by the user. To provide this
information to the Model, the user specifies the annual rate of change for
specified years (e.g., 1983, 1990, 2000, 2010, 2020, 2050, and 2083). For
years between these dates, the Model interpolates a rate of change.
Because the influence of prices is accounted for elsewhere (see below),
these assumptions about rates of growth or decline need only reflect expected
changes in technology over time. Consequently, the rate of decline of 2
percent per year (used in Simulation 1) could be interpreted as a reduction in
the amount of waste being produced (at constant real prices), despite expected
growth in output due to changes in technology.
Price for Land Disposal. The second determinant of the demand for land
disposal capacity is the price that is attached to it. The Model incorporates
two facets of this market force. The first is a scenario of the projected
price for land disposal in each year of the Model run. As with industrial
growth, the user inputs prices for seven points in time in the Model
simulation. Interpolation is used to determine the price in any of the
intervening years. The second is the price responsiveness, or elasticity, of
demand for disposal capacity. Elasticity is measured in the traditional form:
Percent change in quantity of waste that is land disposed
Elasticity =
Percent change in price of disposal
Assume, for example, that the elasticity of demand is -0.5. This implies that
if the price of land disposal were increased by 10 percent, there would be a
drop of 5 percent in the quantity of land disposal capacity demanded.
Taken together, the projected prices and the elasticity of demand allow
the Model to calculate the effect of changes in land disposal prices on total
demand. In current simulations, the price of disposal is held constant at an
arbitrary $100 per ton and the elasticity of demand is set equal to zero. The
effect of these assumptions is to make the demand for land disposal
independent of its price.
Prices for Substitutes for Land Disposal. In a manner analogous to
that discussed for the price of land disposal, demand can also be affected by
the price of substitutes for land disposal. In short, if an alternative to
land disposal such as incineration or recycling were to change in price, we
would expect the demand for land disposal to change somewhat in response. The
magnitude of this change is described by the cross price elasticity of demand
which takes the traditional form:
-------
A-116
Cross price Percent change in quantity of waste that is land disposed
Elasticity =
Percent change in price of non-land disposal option
As above, the user inputs to the Model a projected price scenario for the
substitute. As with the price and elasticity of land disposal, the price
projection and cross price elasticity for the substitute disposal method in
current simulations have been set equal to a constant $100 per ton and 0.0,
respectively.
Total Demand for Disposal. The three forces described above (the
growth of output and changes in technology, the price and elasticity of demand
for land disposal, and the price and elasticity for substitutes for land
disposal) are combined in the Model to produce an estimate of demand for land
disposal in each year of the Model run. Because the factors contributing to
total demand are relative measures (i.e., they are calculated relative to the
previous year's volume or price), the initial volume of waste is required. In
the current simulations, this value is 64 million wet-weight tons"of waste
disposed in 1983 (taken from the preliminary results of the Mail Survey).
Once the initial value is set, total demand is calculated for each year in the
Model run in the following way:
Step One: The effect of the rate of growth or decline is
calculated. The volume of wastes from the previous
year is multiplied by one plus the rate supplied by
the user. For year one, the previous year's waste is
simply the initial level supplied by the user.
Step Two: The result obtained in step one is adjusted for the
effects of price changes. This is done in two parts:
Step Two-A: The effect of a change in the price
of land disposal is calculated. First,
the proportional change in the real
price is computed. This price change
includes shifts in the tax rate which
may result from a policy decision or
because inflation causes the rate to
drop in real terms. The proportional
change in price is multiplied by the
price elasticity for disposal to yield
the proportional change in the volume
of wastes that are land disposed. The
volume of wastes from step one is then
multiplied by this proportion to
determine the change attributable to
the prices of land disposal.
-------
A-117
Step Two-B: The effect of a change in the price
of the substitute is calculated. The
proportional change in the real price
of the substitute is computed and then
multiplied by the elasticity to yield
the proportional change in the volume
of land- disposed waste. The result is
then multiplied by the volume of wastes
determined in step one and yields the
change attributable to a change in the
price of a substitute.
Step Three: The effects of these three economic forces are
combined to estimate the total volume of waste
capacity demanded. This is done by simply summing up
the results of steps one and two. Depending on the
magnitude and direction of the individual changes, the
total volume of waste may go up or down.
Step Four: The demand for land disposal capacity is then
apportioned among four separate economic regions. The
Model divides the country into four economic regions
that are treated as isolated markets when analyzing
the supply of and demand for disposal capacity.
Exhibit A-40 is a map of the United States showing the
regions'used in the current simulations. The user
specifies which states are in each of the regions and
the proportions of the national waste stream being
land disposed in each. With these proportions, the
total volume determined in step three is divided among
the regions. Using both the initial volume of waste
and the regional proportions from the current
simulations, Exhibit A-41 displays the amount of waste
simulated to be disposed of in each region in the
first year of the Model run. These waste levels may
change during the Model run, but the proportion of
waste in each region will remain fixed.
Inflation. All calculations of demand are made in real dollars. Thus,
the price projection scenarios are based on 1982 dollars, as are the
elasticities. Of note is the fact that the tax rate is not in constant
dollars, but is set at a fixed $2.13. With inflation, the rate will drop in
real terms and must be adjusted back to 1982 dollars each time the annual
calculation of total demand is done. The user supplies information about the
inflation rate prior to the simulation. The user specifies the inflation
rates for seven years (e.g., 1983, 1985, 1990, 2000, 2020, 2050 and 2083).
For years between these dates, the Model interpolates to determine the annual
inflation rates. These inflation rates are used as required to adjust between
1982 dollars and dollars in the other years of the analysis.
-------
A-118
EXHIBIT A-40
MAP OF ECONOMIC REGIONS USED IN THE MODEL
-------
EXHIBIT A-UI
LAND DISPOSAL OF WASTES IN FIRST YEAR OF MODEL RUN
REGION
I - Northeast
2 - Southeast
3 - Midwest
k - West
TOTAL
PROPORTION OF WASTE
VOLUME OF WASTE
(Hi I I ions of Tons!
MEAN NUMBER OF FACILITIES
(Standard Error of Estimate)
AVERAGE FACILITY CAPACITY J./
(Thousands of Tons per Year)
.31
.27
.30
.12
19.8
17.3
19.2
7.7
272
393
190
127
(2.1 )
(2.7>
(2.0>
(2.0)
109
66
152
91
1.00
982 (3.9)
98
J./ Includes adjustment for excess capacity. See text.
Source: ICF Incorporated estimates based on Simulation I.
vo
-------
A-120
The Supply of Land Disposal Capacity
The supply of land disposal capacity plays an important role in the Model
results. If there is unmet demand within one of the economic regions, the
Model simulates the opening of new facilities. These facilities, if qualified
for Fund coverage, may have a direct impact on Fund expenditures -and, if
unqualified, may generate costs to other funding sources.
The supply of land disposal capacity is assumed to be perfectly inelastic
(i.e., elasticity = 0.0). This implies that supply is not affected by prices
of disposal. Instead, supply is simulated to be provided (i.e., facilities
are simulated to open) whenever there is unmet demand. This approach is
consistent with the exogenous specification of the prices of disposal (and its
substitutes) over time by the user. The Model assumes that demand, at the
user-supplied prices, will be met.
In this section, we first discuss how the supply of land disposal capacity
in each region is estimated. We then turn our attention to the means by which
new capacity is added.
Capacity Estimates for Each Economic Region. To estimate the available
capacity in each economic region, the Model goes through the following steps
at the beginning of each year:
Step One: The average capacity of facilities in each region in
year 1 is determined. The total volume of wastes (e.g., 64
million tons in 1983) is divided among the four economic
regions based on proportions supplied by the user. This
volume is then divided by the number of facilities to yield
the average volume per facility. Finally, this number is
increased to reflect an excess capacity factor supplied by
the user. Based on Simulation 1, Exhibit A-41 shows the
mean number of facilities in each of the four regions in
year 1. For example, in region 1 there are 272
facilities. The exhibit also shows that about 19.8 million
tons of waste were disposed in region 1. Thus, on average,
each faciliity received 73,000 tons per year. Given the
excess capacity factor of 50 percent (used in the current
simulations), the Model estimates that the average capacity
per facility in region 1 in the initial year is 109,000
tons per year (i.e., 73,000 x 1.5). The estimate of the
initial average capacity in the current simulations is
shown in Exhibit A-41.
Step Two: In each year of analysis, the Model assesses the number
of "facility capacity units" in the region. At the outset
of the Model run, the number of facility capacity units in
each region is set equal to the number of existing
facilities in the region. Thereafter, when a new facility
opens in a given region, the number of capacity units for
that region is incremented by an amount which depends both
on the year and on the growth rate of average capacity as
-------
A-121
supplied by the user. This growth rate is used to
determine the capacity units of the average facility
opening in any given year with the following formula:
year
Capacity Units - (1 + Growth Rate)
where:
Growth Rate = growth in average facilty capacity
(user input)
Year = Model year of facility opening
For example, in the current runs of the Model, the growth
rate is 1 percent. Thus, for facilities opening in year 5,
capacity is assumed to be (1.01)5 times the average year
1 capacity of facilities in the particular region.
Whenever a facility closes, the Model subtracts from the
region total the number of capacity units associated with
the facility which is closing.
Step Three: For each of the four regions, total available capacity
is calculated. The total number of capacity units (based
on the number of facilities in the region and their opening
dates) determined in step two is multiplied by the average
capacity determined in step one. The result is the total
available capacity in the region.
New Facility Opening: Once the total supply and demand for capacity
have been calculated for a given region, a determination of whether to add new
facilities is made. This occurs in the following way:
Step One: Within each economic region, total available capacity is
compared to total demand. Total capacity may drop in any
year if one or more facilities in the region have closed.
Total demand will change based on industrial growth and price
changes. If enough capacity is available to meet the demand
for land disposal, nothing further is done and the Model
moves on to evaluate capacity in the remaining economic
regions.
Step Two: If there is not enough capacity to handle .the volume of
wastes in the region, the Model will simulate the opening of
a new facility. The capacity of this facility will be equal
to the year 1 average for that region plus the growth in
average capacity that has taken place since. The regional
count of capacity will be adjusted to reflect the new
facility and will again be compared to the total demand in
the region. If the demand is met by the addition ofvthe
facility, the Model moves on to evaluate the remaining
regions. If supply is still less than demand, more
-------
A-122
facilities are added until the condition no longer exists.
The number of facilities that can be added in a given region
in any one year is currently limited to 10 percent of the
facilities existing in that region in the year.
Also of note is that as currently configured, the Model does not simulate
the closing of facilities in response to the growth of excess capacity. For
example, if waste volumes were assumed to decline due to regulation, new
facilities would not be added (as appropriate), but existing facilities would
not be simulated to close prematurely in response to the decline.
A.3.1.2 Relationships Among Fund Balance, Revenues,
and Spending
This section discusses the economic relationships that exist within the
structure of the PCLTF Model. These relationships are an important
determinant of the fiscal adequacy of the Fund. We first discuss how Fund
expenditures are simulated on a fiscal year basis. Next is an analysis of
Fund revenues, which requires the examination of both calendar and fiscal
years. Finally, the interaction of the two is presented.
Fund Expenditures
Monitoring, Response, and Claims Costs. The costs associated with land
disposal, as simulated by the Model, are discussed in section A.2.4, Actions,
and section A.2.5, Claims. As these costs arise, different funding sources
are responsible for meeting them. (See section A.2.6, Funding Sources). The
ways that the costs are distributed are discussed in section A.3.4, Cost
Allocation Policy. In this section, we discuss only those costs allocated to
the PCLTF.
Depending on the number of years since closure and the facility's
qualification status, the PCLTF may be responsible for:
capital and recurring costs for response actions;
liabilities arising from third-party claims; and
routine monitoring and care costs.
At the end of each fiscal year, the amount of money spent on these costs by
the Fund is totaled. This is by far the largest Fund expenditure.
Administrative Expenses. The costs of operating and administering the
Fund are also borne by the Fund. The Model simulates these costs based on two
user-supplied parameters:
fixed administrative expense (e.g., $1 million per
year); and
proportion of fund expenditures used for administration
(e.g., 10 percent of monitoring, response, and claims
costs).
-------
A-123
At the end of each fiscal year, the Model computes administrative costs based
on the two methods outlined above. It chooses the cost estimate that is the
higher of the two and uses only that one to simulate administrative expenses.
In current simulations, the fixed-expense component is set equal to zero,
while the proportional component is set at 10 percent.
Fund Revenues
Tax Base. The tax base is defined as the waste being disposed of in
landfills, surface impoundments, injection wells, and land treatment
facilities. Under the current statute, this waste is taxed on a dry-weight
basis to fund the PCLTF. The total volume of waste from the Model's demand
estimate forms the basis for the tax base. Total demand consists of all
wastes, while the tax is levied only on the dry-weight portion. The user
supplies a fraction of total waste which is dry weight so that the tax base
can be calculated. In current simulations, this fraction is 28.4
percent.76 This means that of the initial 64 million tons of wastes
disposed, only about 18.2 million tons are taxable.
Tax Rates. Currently set at $2.13 per dry-weight ton, the tax is the
primary method of raising revenue for the PCLTF. The user supplies the
starting tax rate and determines if it will be indexed to inflation. At the
end of each year, the appropriate rate is multiplied by the tax base to yield
total revenues.
Fund Ceiling. Based on the statutory requirements of CERCLA, the Fund
has a ceiling of $200 million, defined in terms of unobligated Fund balance.
The unobligated Fund balance is the total fund balance minus any fund
obligations (this concept is discussed further below). When the unobligated
Fund balance reaches $200 million, the tax is no longer collected. When the
balance drops below $200 million, the tax is again collected. In Simulation 1
(which is meant to reflect current policy), the unobligated balance first
exceeds the ceiling in the sixth year.
Unlike Fund expenditures which are calculated on a fiscal year basis,
taxes are collected on a calendar year basis. This is primarily because of
the statutory provision for suspending the tax on January 1 of any given year
if three months earlier, on September 30, the unobligated balance exceeds the
ceiling. To make revenues comparable to spending, the assumption is made that
revenues flow into the Fund at a steady rate over the calendar year. This
assumption permits the Model simply to take three-quarters of the revenues
collected in, say, calendar year 1985 and add one-quarter of the revenues from
calendar 1984, to compute the total revenues for fiscal year 1985.
76 This was computed by assuming that the dry-weight portion equals:
100 percent of landfill wastes (equal to 3 percent of the total), 50'percent
of land treatment wastes (equal to 1.4 percent of the total), 50 percent of
surface impoundment wastes (equal to 38 percent of the total), and 10 percent
of injection well wastes (equal to 57 percent of the total).
-------
A-124
Fund Balance
In addition to the revenues and expenses described above, the Fund balance
is affected by the borrowing and lending of Fund monies. These practices are
described below and the method of determining the Fund balance is discussed.
Finally, the definition of Fund obligations (used to calculate the unobligated
Fund balance) is presented.
Interest Payments and Revenues. Interest payments and revenues may
result from the financial management of the Fund. The Secretary of the
Treasury may borrow money from the Treasury as deemed necessary. The
Secretary is also permitted to invest such portions of the Fund which, in his
judgment, are not required to meet current obligations. Investment must be in
U.S. Treasury bonds or notes.
The model calculates the interest payments and revenues in the following
manner:
Step One: The tax revenues and fund expenditures are.
calculated on a quarterly basis over the fiscal year;
Step Two: If the quarterly balance is negative, money is
borrowed at the user-supplied interest rate in order
to bring the Fund into balance;
Step Three: If the quarterly balance is positive, the monies
are invested at the user-supplied interest rate; and
Step Four: Net interest, which is expressed in nominal
dollars, is calculated as the difference between
interest payments and earnings.
In the same way that the user specifies economic assumptions about growth
and inflation, an interest rate scenario is supplied. The following
assumptions were used in the current simulations:
1983 1985 1986 1990 1995 2033 2083
Inflation 3.1 5.3 5.2 4.6 4.0 4.0 4.0
Interest Rate 8.0 10.0 9.2 5.7 5.1 5.1 5.1
The data for 1985 reflect current estimates. From 1990 onward, the difference
between the interest rate and inflation is 1.1 percent. This difference is
based on the observed relationship from 1960 to 1982 between the Consumer
Price Index and the average Treasury security interest rate (averaged over the
following maturities: 3 months, 6 months, 3 years, and 10 years).
Fund Balance. To determine the annual Fund balance, the Model simply
sums (algebraically) interest payments; interest revenues; costs of
monitoring, response, and claims activity; and tax revenues. As noted before,
values of the Fund balance are expressed in inflated (i.e., nominal) dollars.
-------
A-125
Obligated Fund Balance. Monies in the PCLTF may be either obligated or
unobligated, depending on their commitment to specific projects. The
unobligated Fund balance is equal to the total Fund balance minus the amount
of obligated funds. Thus, a definition of obligated funds is required.
The Model permits the user to define obligated funds in one of three ways
(Policy 3 is used in the current simulations)'.
Obligated Fund Policy IConcept of obligated funds
not used. Obligated funds equal zero.
Obligated Fund Policy 2--Obligated funds are set
equal to the post-closure care and recurring response
costs for the current year multiplied by a
user-specified number of years.
Obligated Fund Policy 3--Obligated funds are set
equal to the average of the previous three years of
total PCLTF expenditures.
A.3.2 Regulatory Policy
This section presents the two major sets of regulatory policies which play
important roles in the PCLTF Simulation Model:
Permit Policy: defining the rate at which EPA
requires and reviews permit applications from existing
land disposal facilities and the bases used to make the
permitting decision;
Action Policy: defining the situations leading to
the undertaking of monitoring and response actions.
Regulatory policies not discussed here include:
Financial assurance requirements: discussed in
section A.3.4, Cost Allocation Policy;
PCLTF qualification requirements: discussed in
section A.3.4, Cost Allocation Policy; and
Allowable facility designs: discussed in section
A.2.2, Facility-Level Characterization.
A.3.2.1 Permit Policy
Permit policy refers to the manner .in which EPA requests and reviews final
permit applications from existing land disposal facilities. (All new
facilities are assumed to begin operation with permits.) This policy must be
modeled because:
-------
A-126
Facilities which are not granted final permits will be
required to cease accepting hazardous waste. The need
for new land disposal capacity will be influenced by the
reductions in capacity caused by these closures.
" Only facilities with final permits are eligible to be
covered by the PCLTF. 7 7
Because of these two roles of permits, the modeling of EPA's permit process
must identify both when permit applications will be reviewed and whether
permits will be granted to each of the facilities. This analysis was
simplified by assuming that each existing facility will be reviewed once for
the purpose of granting a permit. If a permit is granted, the facility is
thereafter treated as a permitted facility. Otherwise, if a permit is denied,
the facility will never obtain a final permit.
The permitting policy is defined by three characteristics: (1) rate; (2)
information; and (3) decision rule. Each characteristic is defined in turn,
and the assumptions used for each in the current simulations are presented.
Rate of Permit Review. The rate of permit review is defined by three
variables: two time periods and one fraction. In the current simulations,
the values are 2 years, 5 years, and 0.10, and are interpreted as follows:
Within the first time period (2 years from the
starting year of the Model), permit applications are
reviewed at the defined fraction of facilities, 0.10
(i.e., 10 percent of all the existing facilities); and
* Between the first time period and the second time
period (2 years to 5 years), the permit applications of
the remaining fraction of facilities (1.0-0.1 =0.9;
i.e., 90 percent of the existing facilities) are
reviewed.
This assumed rate of permit review is used to assign a year to each
facility which identifies the time of its permit review. This year is stored
at the facility level (as described in section A.2.2.1, Facility Milestones).
The year for each facility is simulated'as follows:
A random number is drawn and compared to the defined
fraction. If the random number is less than or equal to
the fraction (0.10 in the current simulations), then the
first time period is used. Otherwise, the second time
period is used.
77 This is the current Agency interpretation of Section 107(k)(l) of
CERCLA, which requires that the disposal facility "has received a permit under
Subtitle C of the Solid Waste Disposal Act" in order to qualify for Fund
coverage.
-------
A-127
Using the appropriate time period, the facility's
permit review date is simulated by drawing randomly from
a uniform distribution within the time period (i.e., the
time is drawn from the range of 1 to 2, or the range 3
to 5).
* The facility's permit review date is stored at the
facility level.
Of note is that the facility's permit review date is simulated
independently from other facility characteristics. The possibility of
reviewing permit applications from certain known problem facilities first is
not examined. If a facility closes prior to having its permit application
reviewed, it does not obtain a permit.
Information Supplied by the Facility for Permit Review. In the permit
review process, the facility's owner/operator (0/0) must provide information
to the Agency in order for the Agency to make its permit determination.
Although a variety of data may be required, the modeling approach has been
limited to the consideration of releases only.78
Information regarding releases is defined in terms of the seven release
types included in the Model. At the simulated time of permit review, one or
more of the releases may be simulated to have occurred, and the occurrence of
the release(s) may or may not be simulated to be known. The occurrences of
releases are simulated as described in section A.2.3, Modeling of Releases.
At the time of permit review, releases that have occurred may be detected via
two mechanisms: routine monitoring or site assessment.
As described in section A.2.2.2, Facility Attributes, and section A.2.4,
Actions, routine monitoring may detect releases. However, this action may not
be fully reliable, so that a release may go undetected. This failure to
detect a release when it has in fact occurred is incorporated into the Model,
as described in section A.2.4. For purposes of modeling the information
available for permit review, it is assumed that the information obtained from
the routine monitoring action is supplied to the Agency.
In addition to the routine monitoring data, the Model includes an option
to require site assessments at the time of permit review. Although not used
in the current simulations, this option was designed to simulate the impacts
on permitting and Fund adequacy if more information were required at the time
of permitting. The site assessment requirement is defined in terms of which
of the seven releases the assessment would be able to detect if, in fact, they
have been simulated to occur. For example, a site assessment requirement of
release type 2, detectable constituents on site, would be interpreted as a
requirement to undertake a study of the site which would reliably reveal
71 Other information required during permitting may include data
concerning: liner/waste compatibility; financial assurance; liability
insurance; waste management practices, and other data.
-------
A-128
whether release type 2 had already occurred. This information obtained
through the site assessment would be in addition to the data uncovered through
routine monitoring.
The final result of the information step in the permit review simulation
is an indication of whether or not each of the seven release types has already
occurred.
Permitting Decision Rule. The purpose of the permitting decision rule
is to allow the Model user to specify the conditions under which a permit will
be denied to a facility. The permit decision is based only on the release
information for the facility, as described above. Consequently, the decision
rule must describe whether an individual release or combinations of releases
will result in a permit being denied.
Based on discussions with EPA, it was assumed for the current simulations
that the detection of releases would not result in the denial of a permit.
.(This assumption was used for modeling purposes only and does not necessarily
reflect current or future Agency policy.) Consequently, all facilities which
desire final RCRA permits may receive them. However, EPA believes that some
facility owner/operators may prefer to close rather than receive final
permits. In the current simulations, it was assumed that a random 25 percent
of existing facilities would prefer not to continue operation and would
consequently close without obtaining final permits.
The user may specify a permitting decision rule which would deny permits
on the basis of the evidence of releases, or combinations of releases. Using
such a rule would selectively close down those facilities with grotmdwater
cont aminat ion.
A.3.2.2 Action Policy
The purpose of the action policy is to define the actions (i.e.,
monitoring actions and response actions) which must be undertaken in certain
situations. For example, when routine detection monitoring discovers a change
in indicator parameters in groundwater, compliance monitoring is required.
The action policy is used to specify this compliance monitoring requirement
and the other action requirements.
To model the actions taken at land disposal facilities, a method was
required which would identify those actions that should be undertaken in all
situations which could arise. Because, with seven release types and various
other factors, there could be tens or hundreds of different situations arising
over time, it was not feasible to list every situation that could arise and
identify the set of actions required in each. At a minimum, such a method
would be unwieldy and difficult to modify to reflect alternative policies.
As an alternative to listing all the situations that could possibly arise,
a rule-based approach was used. In this method, the user supplies a limited
set of rules which describe how actions are to be turned on and off. As
situations arise, the Model applies these rules and chooses actions as
indicated. The advantages of this method include:
-------
A-129
a small set of inputs can describe the actions that
should be taken in an unlimited number of situations; and
the rules can be changed easily by the Model user to
test alternative action policies and the implications
the alternative policies have for costs and releases at
facilities.
Such an approach, however, does require extra care on the part of the Model
user to ensure that an internally consistent set of rules has been specified.
For the purposes of this presentation, the action rules have been divided
into the following four categories:
* Knowledge-to-Action Rules identify which actions
are taken in response to obtaining knowledge about the
existence of a release. This set of rules is a major
driving component of the overall Model because it
specifies the relationship between release detection and
action.
Substitute Rules identify which action is to be
undertaken as a substitute when an action specified in
the knowledge-to-action rules cannot be undertaken.
Reset Rules identify which action is to be
initiated when an action is terminated.
Dominance Rules identify the relationship among the
various actions.
These four sets of rules define the relationships necessary to model actions
at facilities. The first set, knowledge-to-action rules, is used to reflect
regulatory policies. These rules may be changed to test alternative
policies. The other three sets of rules reflect various relationships among
the different actions. These relationships may be changed to reflect various
assumptions; however, their current structure is believed to represent
adequately the relationship among the actions modeled. Therefore, even if the
knowledge-to-action rules were changed substantially to test various policies,
the other rules could remain unchanged. Each set of rules is discussed below
in turn.
Exhibit A-42 displays the knowledge-to-action rules used in the current
simulations. These rules are organized in a matrix with 16 rows and 17
columns (not counting the rule number column). Each row is an individual
rule. The columns are defined as follows:
Columns i and 2 identify the name and number of each release
type which can lead to actions being taken. For example, row 1
is for release type 1, indicator parameters on site.
-------
EXHIBIT A-M2
KNOWLEDGE-TO-ACTION RULES USED IN THE CURRENT SIMULATIONS
Rule »
1
2
3
k
5
6
7
8
9
10
1 1
12
13
11
15
16
Columns:
Release Name
IP ON
DET ON
DET ON
DET ON
DET ON
TOX ON
TOX ON
TOX ON
TOX ON
T/0 OFF
DET OFF
DET OFF
DET OFF
DET OFF
TOX OFF
BATHTUB
1
Re lease
Type
1
2
2
2
2
3
3
3
3
U
5
5
5
5
6
7
2
Re lease
Cond it ion
0
0
1
-1
3
0
1
-1
3
0
0
2
-2
H
0
0
3
Monitoring Actions
1 2 3 l| 5
-1 1000
-1 - 1 1 0
-1 - 110
-1 - 110
-1 - 1 10
-1 - 110
-1 - I.I 0
-1 - 110
-1 - 110
-1 -1 100
000-1 1
000-1 1
000-1 1
000-1 1
00000
00000
H 5 6 7 8
Response
Act ton
1 2 3
000
1 1 0
1 1 0
1 1 0
1 1 0
000
000
000
000
000
1 1 1
1 1 1
1 1 1
1 1 1
000
1 0 0
9 10 II
Ca re
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
12
Exposure
0
0
0
0
0
0
0
0
0
0
0
0
0
0
-1
0
13
Cla ims
1 2 3 l(
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0110
0110
0110
1 1 1 1
0000
IU 15 16 17
-------
A-131
Column 3 identifies the release condition. The release
condition is a code that defines the extent to which knowledge
about the current and future status of the release has been
obtained. For example, based on the performance of plume
delineation and tracking on site, it may be known that the plume:
will migrate off site over time;
will not migrate off site over time; or
has already migrated off site.
The choice of actions to undertake may vary depending upon this
knowledge. For example, if plume delineation and tracking
demonstrates that the plume will not migrate off site, then fluid
removal and treatment may not be necessary. Alternatively, if
the plume will migrate, or has already migrated off site, then
fluid removal and treatment may be desired. This release
condition code allows the user to specify action policies that
vary by condition. There are seven possible values for the
condition code:
0: no information;
1: release has not yet migrated off site, but it will;
-1: release will not migrate off site;
2: release has not yet become toxic off site, but it
will;
-2: release will not become toxic off site;
3: release has already migrated off site; and
A: release is already toxic off site.
Columns 4 through 17 identify the actions to be taken.
Within each column, a -1, 0, or 1 can be entered to signify
stopping the action, no effect on the action, or starting the
action, respectively. The actions are:
-- Monitoring Actions:
1. routine monitoring for indicator parameters at
on-site monitoring wells
2. monitoring for hazardous waste constituents at
on-site monitoring wells
3. plume delineation and tracking on site
4. monitoring for hazardous waste constituents off site
5. plume delineation and tracking off site
Response Actions:
1. cap repair
2. fluid removal and treatment on site
3. fluid removal and treatment off site
-- Post-Closure Care
-------
A-132
Exposure: defined as the use of groundwater near the
facility for drinking. This action can only be assigned
a 0 or -1 (the -1 meaning that people stop using the
water for drinking)
Claims:
1. personal injury claims
2. real property claims
3. economic loss claims
4. natural resource damage claims
These claims can only be assigned a 0 or a 1 (the 1
meaning that the claim is brought).
Using these column definitions, each rule (i.e., each row of the matrix in
Exhibit A-42) can be interpreted. For example, rule 1 means that in response
to detecting response type 1, monitoring action 1 (routine monitoring for
indicator parameters on site) is stopped and monitoring action 2 (monitoring
for constituents, i.e., compliance monitoring) is begun. Rule 1 also implies
that there is no impact on any of the other actions or claims as the result of
detecting release type 1.
Rule 2 implies that in response to discovering detectable concentrations
of constituents on site, plume delineation and tracking on site, cap repair,
fluid removal and treatment on site, and monitoring for constitutents off site
are undertaken. Also, monitoring actions 1 and 2 are terminated.
As the matrix shows, rules were not varied using the condition codes.
Also, rules 6 through 9 are redundant because no additional actions are taken
when a release is found to be toxic relative to when it is only detectable.
(Because a release must be detectable before it can be found to be toxic,
rules 2 through 5 must be in force if any of the rules 6 through 9 apply.
Therefore, rules 6 through 9 are not necessary under the set of assumptions
presented in the exhibit. However, they are included in the exhibit to show
the level of resolution that could be attained if desired.)
Several important observations can be made regarding these knowledge-to-
action rules:
Claims are only simulated to be brought in response to off-site
releases. No claims are simulated for release types 1, 2, 3, 4,
or 7.
Corrective action (i.e., plume delineation and tracking and
fluid removal and treatment) is assumed to be taken in response
to finding detectable concentrations of constituents on site,
regardless of the potential for off-site migration. EPA believes
this rule is most consistent with the current corrective action
regulations. Alternative rules would permit an analysis of the
cost implications of not requiring corrective actions in cases
when releases are not anticipated to migrate off site.
-------
A-133
Only monitoring is taken in response to the detection of taste
and odor off site. This monitoring action would subsequently
lead to other actions, including corrective actions, given that
the release in fact has migrated off site.
Up to 24 different rules can be identified by the user. Whenever a
release is detected, the condition of the release is evaluated, and the
appropriate rule is found by matching the release number and release code.
Once the rule is found, the action changes (-1, 0, +1) are identified and
performed. If an action which is not ongoing is to be terminated (i.e., has a
value of -1), then the -1 has no effect. If an action which is already being
performed is to be initiated (i.e., has a value of +1), then the starting date
for the action is updated to reflect the later request for the action to
begin.79 If a claim is brought for a second time, only the incremental
value of the second claim which is in excess of the compensated portion of all
previous claims is counted. In this manner, care is taken to ensure against
double counting of the costs of actions and claims.
These knowledge-to-action rules are the primary means via which actions
are modeled. In addition, the substitute, reset, and dominance rules play a
role in special circumstances.
The substitute rules are used to specify a substitute action when a
desired action cannot be undertaken. For example, based on knowledge-to-
action rule 1 (see Exhibit A-42), monitoring action 1 may be terminated and
monitoring action 2 may be initiated. However, if a funding source is not
available for monitoring action 2, it is not initiated. Were this the end of
the story, the result would be no ongoing monitoring actions at the facility
(because monitoring action 1 was terminated and 2 could not start). To
incorporate the fact that alternative actions may be contemplated when funding
is not available, the substitute rules were developed.
The substitute rules identify one substitute for each action, as displayed
in Exhibit A-43. For example, the substitute for monitoring action 2 is
monitoring action 1. Therefore, if monitoring action 2 cannot be initiated
due to a lack of funding, then its substitute, monitoring action 1, is
initiated. The realism of the choice of actions is enhanced through the use
of these substitute rules. As described below, before the substitute action
is initiated, it must also conform to the dominance rules.
The reset rules are similar to the substitute rules. When an action is
terminated after having been performed for a period of time, the actions at
79 Updating the starting date has the effect of increasing the length of
time the action is performed. For example, an action may begin in year 10 and
may be required to be performed for 30 years. If in year 20 the action is
again specified to be started under a separate rule (even though it is already
ongoing), the start time is updated to year 20. Because the 30 year duration
still applies, the action must be performed until year 50, or for a total of
50 - 10 = 40 years.
-------
A-134
EXHIBIT A-43
SUBSTITUTE ACTIONS1
Substitute Actions
Monitoring Actions
Response Actions
Post-Closure Care
1
2
3
4
5
'l
2
3
Response
Monitoring Actions2 Actions3 Post-Closure
12345 123 Care
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
o .
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0 0
1 indicates action which is a substitute.
Monitoring Actions:
1: Monitoring for indicator parameters at the on-site monitoring
well.
2: Monitoring for hazardous waste constituents at the on-site
monitoring well.
3: Plume delineation and tracking on site.
4: Monitoring for hazardous waste constituents off site.
5: Plume delineation and tracking off site.
i
Response Actions:
1: Surface sealing.
2: Fluid removal and treatment on site.
3: Fluid removal and treatment off site.
-------
A-135
the facility are reset, or returned, to a specified condition. For example,
when an owner/operator completes the necessary compliance monitoring at a
facility, he reverts to detection monitoring. This principle is applied
through the use of the reset rules, which are displayed in Exhibit A-44. As
the exhibit shows, it was assumed that monitoring action 1 was the reset
action for all the other actions.80
When actions are specified with the substitute or reset rules, they are
not taken automatically. To ensure against taking unnecessary actions, the
concept of action dominance is introduced. If action A dominates action B,
and if action B is requested as a substitute or reset action, it is only
initiated if action A is not already ongoing. For example, when fluid removal
and treatment on site is completed, monitoring action 1 is identified as a
reset action. However, if monitoring action 2 is ongoing (which could occur),
then monitoring action 1 need not be initiated. Using the concept of
dominance, the heirarchy of actions is defined so that only the necessary
substitute and reset actions are initiated. Exhibit A-45 displays the
dominance rules. For example, the first row implies that monitoring action 1
is dominated by monitoring actions 2 and 3 and response action 2.
These four sets of action rules are an enormously powerful tool for
assessing alternative action policies. Of primary importance is the
knowledge-to-action rules which perform the bulk of the work. The substitute,
reset, and dominance rules enhance the,realism of the analysis and allow the
full dynamic nature of the actions to be reflected in the Model.
A.3.3 Financial Relationships
This section describes the financial relationships used to assess an
owner/operator's ability to cover expenditures at his facilities. The
relationships are an important driving force in the Model because they
influence the rate at which owner/operators go bankrupt, which renders them
unable to cover costs at facilities. The relationships are described in the
following two sections:
Background rate of bankruptcy. These data describe
the rate at which owner/operators terminate business for
reasons other than expenditures at owned facilities.
Bankruptcies due to costs at hazardous waste
facilities. For firms not simulated to fail due to
background causes, the relationship between revenues and
expenditures is used to simulate the rate at which
owner/operators choose to terminate business due to the
expenditures at owned facilities.
80 No reset action was specified for monitoring action 1 because it is
assumed to be required in perpetuity, unless terminated by a
knowledge-to-action rule or the lack of funding.
-------
A-136
EXHIBIT A-44
RESET ACTIONS1
Reset Actions
Monitoring Action
Response Action
Monitoring
1
2
3
4
5
1
2
3
1
0
1
1
1
1
0
1
1
2
0
0
0
0
0
0
0
0
3
0
0
0
0
0
0
0
0
Actions2
4
0
0
0
0
0
0
0
0
5
0
0
0
0
0
0
0
0
Response Actions
1
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
3
0
0
0
0
0
0
0
0
1 indicates action which is the reset action.
Monitoring Actions:
1: Monitoring for indicator parameters at the on-site monitoring
well.
2: Monitoring for hazardous waste constituents at the on-site
monitoring well.
3: .Plume delineation and tracking on site.
4: Monitoring for hazardous waste constituents off site.
5: Plume delineation and tracking off site.
Response Actions:
1: Surface sealing.
2: Fluid removal and treatment on site.
3: Fluid removal and treatment off site.
-------
A-137
EXHIBIT A-45
DOMINANCE RULES1
Ongoing Actions
Substitute or
Reset Action
Monitoring Actions
Response Actions
Post-Closure Care
1
2
3
4
5
1
2
3
Response
Monitoring Actions2 Actions' Post-Closure
12345 123 Care
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
-1
0
0
0
0
0
0
0
-1
-1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
-1
0
0
0
0
0
0
0
0
0
0
0
0
-1
-1.
-1
0
0
0
0
0
0
0
0
-1
-1
0
0
0
0 0
-1 indicates that the ongoing action dominates the substitute
or reset action.
Monitoring Actions:
1: Monitoring for indicator parameters'at the on-site monitoring
well.
2: Monitoring for hazardous waste constituents at the on-site
monitoring well.
3: Plume delineation and tracking on site.
4: Monitoring for hazardous waste constituents off site.
5: Plume delineation and tracking off site.
Response Actions:
1: Surface sealing.
2: Fluid removal and treatment on site.
3: Fluid removal and treatment off site.
-------
A-138
A.3.3.1 Background Rate of Bankruptcy
Owner/operators may terminate all business activity at any time during a
facility's life cycle. The background rate at which this may occur (i.e., due
to causes not associated with the land disposal of hazardous waste) has been
estimated for the current population of owner/operators.81 This background
rate is important because it defines the minimum rate at which owner/operators
will declare bankruptcy.
An owner/operator's net worth provides an indication of the
owner/operator's financial strength and is used to assess firm viability. The
Model user can specify two categories of simulated net worth that determine
the rate at which firms declare bankruptcy. Estimated background bankruptcy
rates of 0.42 percent and 0.26 percent for firms whose net worth is less than
$10 million and greater than $10 million, respectively, are used for the
simulations described in this report. These rates are for bankruptcies due to
reasons other than expenditures for actions at facilities and are indicative
of the minimum rate at which firms owning land disposal facilities will go
bankrupt.
A.3.3.2 Bankruptcies Due to Costs at Hazardous Waste Facilities
To assess PCLTF adequacy, it was necessary to go beyond the background
bankruptcy rates to simulate how owner/operators may cover the costs of
monitoring, response, and claims arising at hazardous waste facilities. For
example, because the owner/operator's share of corrective action costs may be
very large, this cost may itself force the owner/operator into bankruptcy.
Neglecting these cases of owner/operator bankruptcy would result in an
overestimate of the owner/operator's ability to cover costs, and possibly, an
underestimate of PCLTF expenditures. Thus, after the chance of routine
business failure has been simulated, the Model determines if bankruptcy is
likely to result from the>costs of a firm's land disposal operations.
This section presents the method used to simulate bankruptcies
attributable to land disposal activities. The key assumption underlying this
method is that firms will cease operations if they perceive that expenses will
exceed revenues for some period of time (e.g., over the next 10 years). The
viability of the owner/operator of each facility is evaluated at the beginning
of each year. Facilities that are owned by the government (federal, state or
municipal) are assumed always to have a viable owner (consequently, they are
not evaluated for possible bankruptcy). The following procedure is used to
assess the viability of each facility's owner/operator.
Step 1: Select the appropriate return on assets for the firm. The
Model user provides a return on assets for each of the two
net worth categories used for simulating background business
failures. In the simulations described in this report, a
return on assets of 6 percent was assumed for all firms.
81 The firm population on which the distribution is based includes firms
other than those owning land disposal facilities.
-------
A-139
Step 2: Using the firm's simulated total assets*2 and the return
on assets from Step 1, estimate the firm's expected annual
revenue. The expected annual revenue is the return on
assets times the total assets.
Step 3: Simulate the owner/operator's current expenditures for
capital investments, O&M costs, and claims of its
facilities. Because firms may own more than one facility,
the sum total of all the expenditures for the firm must be
simulated. For example, if the owner/operator of facility A
owns a total of four facilities, the total firm expenditures
will include the expenditures from facility A plus simulated
expenditures from three other facilities. The expenditures
for the three other facilities are simulated by drawing from
the distribution of costs at other facilities owned by
non-government viable owners.
Step 4: Adjust expenditures to reflect taxes and market
constraints. Capital expenditures are adjusted downward to
reflect the investment tax credit and are then depreciated
over time.*3 Next, all costs (including depreciation of
capital costs, O&M expenses, and claims costs) are adjusted
downward to account for:
The degree to which costs can be passed on to customers;
and"1
The marginal corporate tax rate.
>s
Step 5: Compute the present value of both revenues and expenditures
over the next ten years. A zero percent discount rate was
used to calculate these present values. This low value is
warranted because if the firm declares bankruptcy, the
owners may lose control of the company's assets and may be
unable to make profits in the foreseeable future.
Consequently, the opportunity cost of the assets (as seen by
the firm owners) may be essentially zero.
82 An owner/operator's total assets are simulated as described in
Section A.2.2.1, Facility Milestones.
82 An investment tax credit of 10 percent is used for all the
simulations described in this report.
8* For the purposes of this analysis, it has been assumed that it is not
possible for facility owner/operators to pass on any of their costs to their
customers. This is a "worst case" because it maximizes the number of
bankruptcies.
85 The marginal corporate tax rates used are 0.40 for firms with a net
worth of less than $2.5 million and 0.46 for all other firms.
-------
A-140
Step 6: Determine if the firm chooses to cease operations by
comparing the present values using a "rational quit" rule.
Under this rule, a firm will go out of business if, on a
present-value basis, expenses are simulated to exceed
revenues over the next ten years.
The outcome of Steps 1 through 6 for a particular facility is that the
facility's owner/operator may go bankrupt and consequently no longer be able
to cover costs at the facility. When this occurs, the costs of ongoing
actions and claims are reallocated to the alternative funding sources (state
funds, PCLTF, Superfund), as available.
A.3.4 Cost Allocation Policy
This section describes the relationships used to allocate costs among
available funding sources. Costs arise at facilities, and using the various
relationships described here, they are assigned to one of the funding sources
described in Section A.2.6, Funding Sources. This section is divided into two
parts:
Qualification Policy identifies the rules used to
simulate whether facilities qualify for Fund coverage;
and
Allocation of Costs to Funding Sources identifies
the relationship between cost allocation and facility
characteristics.
Each is discussed in turn.
A.3.4.1 Qualification Policy
The qualification policy is used to model Fund coverage. A series of
rules is specified which define those conditions that must be met in order for
a facility to be covered by the PCLTF. If a facility meets the specified
conditions, it is said to be "qualified for Fund coverage," or just qualified.
At any point in time, a facility's qualification status may be:
pending, i.e., yet to be determined;
qualified; or
unqualified.
Once a facility becomes qualified or unqualified, its qualification status may
no longer change. The relationships discussed in this section are used to
model the decision regarding qualification.
Section 107(k) of CERCLA identifies three basic requirements to qualify
for Fund coverage:
a final RCRA permit has been received;
-------
A-141
the owner/operator has not violated permit
requirements that would affect the performance of the
facility following closure; and
based on up to five years of monitoring after closure,
there is no substantial likelihood of future release
from containment or other risk to public health or
welfare.
The regulations interpreting these statutory requirements have not yet been
promulgated by EPA. For purposes of modeling the qualification process, we
developed several simple rules the user can use to specify a qualification
policy. As described below, these rules have three parts which specify: the
timing of the qualification decision; the roles which releases and responses
to releases play; and the role which financial assurance plays. Of note is
that the rules do not incorporate various types of permit violations, such as
inadequate closure, improper waste handling, or other factors.
The first part of the rule specifies the timing of the qualification
decision. In the current simulations, all permitted facilities are assumed to
have this determination made five years after closure, as specified in CERCLA
section 107(k).*6 This time can be altered by the Model user by supplying
an alternative value. The qualification determination period does not vary
across facilities, except as specified in the second part of the qualification
rule.
The second part of the rule is used to reflect the CERCLA requirement of
"no substantial likelihood of release." Based on discussions with EPA, this
requirement is modeled by using the simulated detection of releases at
facilities at the time of the qualification decision. Because EPA feels it is
unlikely that analyses predicting the likelihood of future releases will be
required to meet the CERCLA requirement, the Agency, has chosen to model the
qualification decision based on the evidence of past releases. This
information would be available at the time the decision is made.
Consequently, the second part of the rule specifies the extent to which the
detection of any of the seven release types influences the qualification
decision.
In the current simulations, the detection of any release prior to the
qualification decision was assumed to be sufficient to disqualify the facility
from Fund coverage. In this manner, only those facilities that had not had
any release up to that point were accepted into the Fund.
Because there is uncertainty regarding the appropriate interpretation of
the "no substantial likelihood" requirement, several alternatives were built
into this part of the qualification rule. For each of the seven release
types, the user may identify one of four alternative influences on facility
qualification:
86
Non-permitted facilities are disqualified when their permit is denied.
-------
A-142
The release has no impact on qualification, i.e., if the
release was detected, the facility may still qualify for Fund
coverage;
If the release is detected and all appropriate response actions
have been taken, the facility's qualification determination is
delayed until the response actions are completed. If all the
necessary response actions are completed, the facility will
qualify for Fund coverage at the time of completion. If the
owner/operator is unable to complete the actions as required, the
facility is disqualified.
If the release is detected and all the appropriate response
actions are completed prior to the qualification determination,
then the facility may qualify for coverage. If, at the time of
the decision, the actions are ongoing, the facility is
disqualified.
* If the release is detected, the facility is disqualified.
In the current simulations, the fourth alternative is used for all the
releases: disqualification due to detection. The other alternatives are less
stringent and would allow wider Fund coverage than that currently considered.
These alternatives may be used to assess the implications of alternative
qualification requirements on potential Fund expenditures.
The final part of the qualification rule defines the role in the
qualification decision of the owner/operator's obligation to perform
post-closure care. The Model user may define whether defaulting on
post-closure care obligations prior to the qualification determination will
result in disqualification. In the current simulations, it was assumed that
if the owner/operator failed to provide post-closure care at least until the
time the qualification decision is made, then the facility would be
disqualified. The user also has the option of omitting this requirement.
Based on these rules, the qualification status of each permitted facility
is simulated. The facility's qualification status is stored at the facility
level and is used as a facility characteristic for allocating costs. The
method used to perform the allocation is described next.
A.3.4.2 Allocation of Costs to Funding Sources
When a cost is simulated to arise at a land disposal facility, the Model
looks for a funding source to cover the cost. When performing this cost
allocation, certain rules specify which funding sources may be used and the
order in which they are used. The order is particularly important because if
the funding source specified as first (e.g., the owner/operator) is able to
pay the entire cost, then the second and third funding sourcs (e.g., PCLTF and
state funds) would not be called on to cover the cost.
The potential funding sources (discussed in section A.2.6) for actions and
claims are:
-------
A-143
Owner/operator financial assurance mechanism;
Owner/operator;
State fund;
Superfund;
PCLTF; and
Gap (i.e., the costs go unfunded).
These funding sources need not cover all types of actions and claims. For
example, in Simulation 1, financial assurance covers only routine monitoring
and care.
The allocation policy is defined in three steps. First, allocation
orders are defined. Exhibit A-46 shows the orders used for Simulation 1.
For example, order number 4 (the fourth row in the exhibit) gives the
financial assurance mechanism first responsibility. Should it fail to cover
the cost, the model looks to the owner/operator. If the owner/operator is out
of business or the cost exceeds his ability to pay, the model looks to the
state fund if one is simulated to exist. Absent a state fund, the cost would
go uncovered, and consequently goes to the "gap." In each of the orders
(i.e., rows) in Exhibit A-46, a zero indicates that the funding source is not
used. The choice of allocation order is discussed below.
The method of modeling the adequacy of each funding source is described
elsewhere in this appendix. The viability of financial assurance mechanisms
is described in section A.2.2.1, Facility Milestones. Owner/operator
viability is discussed in section A.3.3, Financial Relationships. The
simulation of state funds and the Superfund is described in section A.2.6,
Funding Sources. Finally, the determination of PCLTF coverage is described in
Section A.3.4.1, Qualification Policy.
The second part of an allocation policy involves reflecting the cost
sharing provisions of CERCLA. The Model user specifies the fraction of monies
spent by Superfund that is provided by the states.87 The state share of
these costs may vary depending on the type of cost and ownership of the
facility. Exhibit A-47 displays the allocation of the state share of
Superfund spending used in Simulation 1. The states are simulated to cover 10
percent of the non-claims costs paid by Superfund at private facilities and 50
percent of these costs at state and municipal facilities. The state share of
Superfund spending is paid regardless of the simulated state coverage
regime.88
The third part of an allocation policy involves applying the orders and
state share of Superfund allocations developed in the previous steps to
particular costs at different stages of a facility's lifetime. An allocation
order is specified for each cost type over time. Exhibit A-48 shows the
87 See CERCLA section 104(c).
88 See Section A.2.6 for a description of State coverage regimes for
third-party claims and response costs.
-------
A-144
EXHIBIT A-46
ALLOCATION ORDERS
OWNER/OPERATOR STATE
ORDER # FINANCIAL ASSURANCE OWNER/OPERATOR FUND PCLTF SUPERFUND GAP
1
2
3
4
5
6
7
8
9
0
0
0
1
0
0
0
0
1
1
1
0
2
1
0
0
1
2
0
0
1
3
2
0
0
0
0
0
0
0
0
0
0
1
2
0
0
2
0
0
0
1
0
0
0
2
3
2
4
3
2
0
0
3
-------
A-145
EXHIBIT A-47
STATE SHARE OF SUPERFUND ALLOCATIONS
FACILITY OWNER
COST TYPE
MONITORING:
Indicates Parameters On Site
Constituent On Site
Plume Del. On Site
Constituent Off Site
Plume Del Off Site
RESPONSE:
Surface Sealing
Fluid Removal On Site
Fluid Removal Off Site
POST-CLOSURE CARE
CLAIMS:
Personal Injury
Real Property Damage
Economic Loss
Natural Resource Damage
Private Municipal State Federal Other
0.1
0.1
0.1
0.1
0.1
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.1
0.1
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
. 0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-------
EXHIBIT A-t»8
COST ALLOCATION POLICY USED IN SIMULATION I
MONITORING ACTION TYPE:
I. Indicator parameters monitoring on site.
2. Constituent monitoring on site.
3. Plume delineation on site.
H-*
.p-
-------
A-147
policy for Simulation 1. Using this information in conjunction with the
allocation orders from Exhibit A-46, an allocation policy is completely
specified. Thus, the allocation policy can be described as follows (row
numbers refer to Exhibit A-48):
Row 1: The owner/operator alone is responsible for monitoring and
response costs during the facility's operation (allocation
orders 1 and 9). The owner/operator is also initially
responsible for natural resource claims although if the
owner/operator is unable to provide the necessary funds, the
Superfund is simulated to cover these costs.
Row 2: Following closure, a facility's PCLTF status is pending, and
allocation orders 2, 4 and 5 are used. For routine
monitoring and care, the owner/operator's financial assurance
mechanism has the first responsibility to cover costs
(allocation order 9).
Row 3: For qualified facilities closed less than 30 years, the PCLTF
will cover the costs for third-party claims. Responsibility
for monitoring, response, and care actions is not affected
(i.e., the allocation orders for these costs remain
unchanged).
Row 4: Following the post-closure period (30 years), the PCLTF
covers all costs at qualified facilities (allocation order 7).
.Row.5: For unqualified facilities, no costs are covered by the
PCLTF. The responsibility for cost coverage remains as if
the facility's qualification status was pending.
Row 6: After the end of the post-closure period (30 years), the
owner/operator is no longer responsible for certain costs.
Consequently, the allocation order 4 (in row 5) is replaced
with 3, and order 2 is replaced with 6. Depending on the
type of facility owner, the State will pay a share of
Superfund expenditures (see exhibit A-47).
The use of different allocation orders for different costs permits the
user a great deal of flexibility in simulating different policies. One of the
Model's strengths here is that it allows very dissimilar policies to be
analyzed. For example, by varying the allocation rules, the user might assign
all costs to the states. Likewise, a policy that holds owner/operators
responsible for more costs over a longer period of time can be simulated.
A.3.5 The Effect of Response Actions on Releases
Various actions are simulated by the PCLTF Model to be taken at hazardous
waste land disposal facilities in response to the detection of releases. As
described in Section A.2.4, Actions, these actions have costs and are able to
detect releases. In addition, the three response actions (cap repair, fluid
removal and treatment on site, and fluid removal and treatment off site) can
-------
A-148
clean up contaminated groundwater and can influence whether and when releases
will occur in the future. This ability to influence releases is a benefit of
taking the actions which must be considered in the facility-level modeling in
the PCLTF Simulation Model. Were this characteristic of response actions
omitted, the Model would be biased because the costs of performing the actions
would be included, but the benefits of the actions would not be considered. -
Therefore, a relationship describing the effects of response actions on
releases is a necessary component of the PCLTF Model.
To model the effectiveness of response actions, one would prefer to
characterize and simulate the physical processes that govern the manner in
which the response action takes place. For example, a series of differential
equations could be used to describe how the concentration of constituents in
groundwater changes over time and space as fluid is withdrawn through a group
of wells. The solution to these equations would identify the length of time
(or amount of fluid) required to reduce groundwater contamination to a given
level.
Unfortunately, the physical processes underlying the effectiveness of
response actions are only partially understood. Very little empirical data
are available which describe the performance of well fields used to clean up
groundwater contaminated with hazardous substances. EPA is currently
undertaking research to improve its understanding of the effectiveness of the
methods available for cleaning up groundwater. For purposes of the PCLTF
Simulation Model, an alternative approach was required.
To simulate response actions in the Model, a parametric approach was
adopted. A parametric modeling approach does not, simulate the underlying
physical processes of a phenomenon, but instead uses a simplified expression
believed to be capable of exhibiting the major properties of the phenomenon.
The simplified expression is defined using one or more parameters that can be
easily modified to reflect various assumptions regarding the properties of the
phenomenon.
The expression used in the PCLTF Model to model the effectiveness of
releases uses the following components:
required duration of the action at the facility;
completed portion of the action at the facility;
a user-supplied parameter describing the effectiveness
of response actions at all facilities.
The required duration is initially defined for each facility using the
duration characteristic of the action (see section A.2.4). For example, in
the current formulation of the Model, fluid removal and treatment on site
requires 100 years to complete. This required duration is adjusted at the
facility level if the action is resumed after being partially completed
previously. The facility is given credit for partially completed actions,
thereby reducing the required duration when the action is initiated for a
second.time (see section A.2.2.2, Facility Attributes).
-------
A-149
The completed portion of the action is also a facility-level
characteristic. When a response action is stopped at a facility, the portion
of the action completed is calculated. Stored at the facility level is the
fraction of response action completed (referred to here as the "fraction
completed"), which is the portion completed (defined in terms of time) divided
by the required duration. This fraction completed is used along with the
user-supplied parameter as the basis for simulating the effectiveness of
response actions.
We have hypothesized that the following properties should be reflected in
the relationship:
the effectiveness of the response action is an
increasing function of the fraction of the response
action completed;
if the response action is fully completed (i.e.,
fraction completed equals 1.0), then it is fully
effective;
if the response action is not initiated (i.e.,
fraction completed equals 0.0), then it has no effect;
and
the user-supplied .parameters should be capable of
scaling the effectiveness of the response action between
the bounds of zero effectiveness and fully effective to
reflect alternative assumptions regarding the shape of
'the function between these bounds.
These properties are reasonable because the longer the response action is
performed, the more successful it is likely to be. At the limit, when it is
performed for its full duration, then the action is completely successful.
Clearly, with the use of this method, the assumed action duration becomes an
important input to the Model. Unfortunately, there is considerable
uncertainty surrounding the necessary durations of response actions because of
the incomplete understanding of the physical processes that govern the
actions. The durations chosen as input are, however, based on the roost recent
analyses of these actions. In developing the Model, we felt that allowing the
user to specify a duration was preferred to specifying assumptions regarding
complex physical relationships. Using a duration simplifies the Model and
makes it easier for the user to control the assumptions used in the analysis.
The relationship that forms the basis for simulating the effectiveness of
response actions is as follows:
where:
SCALE
Effectiveness = (Fraction completed)
Effectiveness = a measure of the effectiveness of the
response action;.
-------
A-150
Fraction Completed = portion of the action completed divided by
the action's required duration; and
SCALE = user-supplied scaling factor.
Exhibit A-49 displays the effectiveness as a function of the fraction
completed for various values of SCALE. When SCALE - 1, the effectiveness is a
linear function of the fraction completed, as the exhibit shows. If the
action is half completed (fraction completed = 0.5), then the effectiveness is
also 0.5. When SCALE is greater than 1, the effectiveness for a given
fraction completed drops. For example, when SCALE = 3, the effectiveness of a
half-completed action is only 0.125.
The boundary conditions for effectiveness are satisfied, as displayed in
Exhibit A-49. However, the value of SCALE allows the user to specify how the
effectiveness behaves between the boundaries. A higher value for SCALE
results in less effectiveness for a given fraction completed.
This basic relationship describing the effectiveness of response actions
is used to simulate whether releases are prevented or delayed. If a response
action is effective, it prevents a release from occurring. If it does not
prevent the release, it may delay it. Calculations are made using the
effectiveness estimate to simulate prevention and delay in two situations:
a release is anticipated to occur while a response
action is ongoing; and
a response action is terminated.
Each situation is discussed below.
Release Anticipated While Response Action is Ongoing
When a release is anticipated to occur at a facility in a given year, all
the response actions ongoing at the facility in that year are evaluated in
terms of their effectiveness in preventing or delaying the release. The
effectiveness of each response action is computed using the relationship
described above. (Only those response actions defined as being capable of
influencing the release are considered. This is a characteristic of each
response action, as described in section A.2.4, Actions, and as displayed in
Exhibit A-27.) The maximum effectiveness across the relevant response actions
is then used.
For example, release type 6 (toxic consituents off site) may be
anticipated to occur 40 years after the facility opens. However, in that
year, two response actions may be ongoing (due to other releases detected
earlier): fluid removal and treatment on site; and fluid removal and
treatment off site. The first action may be one-quarter complete (25 out of
100 required years) and the second may be 10 percent complete (10 out of 100
required years). If SCALE is defined by the user to be 3, then the
effectiveness of the two actions are 0.016 and 0.001, respectively (i.e.,
0.016 = (0.25)3). The larger effectiveness, 0.016, is used.
-------
A-151
EXHIBIT A-49
GRAPH OF EFFECTIVENESS AS A FUNCTION OF FRACTION
COMPLETED FOR VARIOUS VALUES OF SCALE1
Effect iveness
1.0
0.8
0.6
0.4
0.2
0.0
SCALE = 2"
SCALE = 1
0.0 0.2 0.4 0.6 0.8
Fraction Completed
SCALE = 3
1.0
1 Effectiveness = (Fraction Completed)SCALE.
-------
A-152
This effectiveness is used as follows:
A random number is drawn between zero and one. If the
random number if less than the effectiveness (0.016),
then the release is prevented.
If the release is not prevented, it will be delayed.
The amount of the delay is estimated by multiplying the
effectiveness by the time it took the release to occur.
In this example, the delay equals one year (0.016 x 40 =
0.6, which is rounded up to 1.0). This expression
reflects the notions that: (1) the delay should
increase with the effectiveness of the action and (2)
the delay may be greater for those actions which took a
long time to occur initially.
The result of this procedure is either that the release is no longer
anticipated to occur (i.e., is prevented) or is delayed. If the release is
delayed, the procedure is repeated when it is next anticipated to occur.89
If the response actions are still ongoing at that time, their effectiveness
will be higher and prevention will be more likely.
A Response Action Terminated
When a response action is terminated, the influence the response action
had on future releases is evaluated. This evaluation cannot be performed
until the action is terminated because the influence depends on the fraction
of the action completed. The effectiveness of the response action is
calculated as described above. This value is used to evaluate whether the
response action prevents releases as follows:
A random number between zero and one is drawn. If the
random number is less than the effectiveness, then the
action prevents releases.
If the action prevents releases, then all the releases
influenced by response actions are prevented.
If the action does not prevent releases, then all the
releases influenced by the response actions will be
delayed.
The effectiveness estimate is used as the basis for simulating the delay
of future releases. However, the estimate must be adjusted to reflect the
proximity (in time) of the response action to the expected future release. If
the release is expected far in the future, then the response action may have
little influence on the underlying processes leading to the releases.
89 If the delay is less than one-half year, then the release occurs in
the original year and is not delayed.
-------
A-153
Alternatively, if the release were anticipated shortly after the termination
of the response action, then the action was likely to have been in progress
during the time when the release was occurring. Consequently, the response
action may be expected to be effective in delaying the release.
To make the adjustment required for estimating delays, a new quantity was
calculated: fraction of release remaining. This quantity reflects the
portion of time which has passed relative to the total time it takes for the
release to occur. For example, if the release was expected to take 50 years
to occur, and if 40 years have passed, then the fraction of release remaining
is 0.2, ((50-40){50). To estimate the delay, the effectiveness is reduced
by this fraction, and the result is raised to the SCALE power, and then
multiplied by the time it takes for the release to occur. If the
effectiveness were 0.75 and SCALE is equal to 3, the delay in this example
would be:
Delay ~ (0.75 - 0.20)J x 50 = 8.3 years.
Exhibit A-50 displays estimated delays for various values of SCALE,
fraction completed, and fraction of release remaining. The exhibit shows the
following:
The greater the fraction completed, the higher the
effectiveness and the longer the delays.
Higher SCALE values have a substantial impact on
delays. In order to produce long delays, the response
action must be substantially complete, and the release
must be fairly near if SCALE equals 3 or 5.
The lower the fraction of release remaining, the
greater the delay.
A.3.6 Legal Validity of Claims
Before third-party claims (personal injury, real property, economic
loss)90 can be allocated to the PCLTF or other funding sources, it is
necessary to assess whether the claims are legally valid. Legal validity is a
necessary condition for compensation. If a claim is not considered legally
valid, the claimant has no right to compensation. The analysis described in
this section provides the link between claims that are brought and claims that
are allocated to the funding sources for compensation.
The legal validity of claims varies from state to state, depending on
state statutes and accepted legal theories. To simulate the legal validity of
90 Natural resource damage claims are assumed always to be legally valid
due to the statutory liability which CERCLA specifically establishes.
Consequently, in the current simulations, natural resource damage claims are
not subject to this legal validity requirement.
-------
A-154
EXHIBIT A-50
DELAYS IN RELEASE DUE TO RESPONSE ACTIONS FOR
VARIOUS VALUES OF SCALE, FRACTION COMPLETED, AND
FRACTION OF RELEASE REMAINING
SCALE
1
1
1
1
3
3
3
3
3
3
5
5
5
5
5
5
FRACTION
OF ACTION
COMPLETED
0.60
0.90
0.60
0.90
0.60
0.90
0.95
0.60
0.90
0.95
0.60
0.90
0.99
0.60
0.90
0.99
FRACTION OF
RELEASE
REMAINING
0.50
0.50
0.20
0.20
*
0.50
0.50
0.50
0.20
0.20
0.20
0.50
0.50
0.50
0.20
0.20
0.20
TIME FOR
RELEASE
TO OCCUR
(YEARS)
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
EFFECTIVENESS1
0.60
0.90
0.60
0.90
0.216
0.729
0 . 857
0.216
0.729
0.857
0.078
0.590
0.951
0.078
0.590
0.951
DELAY2
(YEARS)
5.0
20.0
20.0
35.0
0.0
0.6
2.3
0.0
7.4
14.2
0.0
0.0
0.9
0.0
0.5
11.9
1 Calculated as: (Fraction of action completed)
2 Calculated as: (Effectiveness - Fraction of Release Remaining)
x (Time for Release to Occur).
SCALE
-------
A-155
claims in a manner reflecting the differences across states, we developed
seven representative state liability regimes based on a review of existing
state statutes and legal theories. Each representative liability regime is
associated with a set of probabilities which reflects the likelihood that each
of the third-party claim types will be legally valid. The seven regimes
chosen represent the wide variety of- legal liability regimes currently
observed across the country.
We then estimated the frequency with which each liability regime exists
today. We used the following data sources to develop the state liability
regimes and to estimate their frequencies:
* ICF Incorporated, Review of State Liability
Provisions, report to U.S. Environmental Protection
Agency, June 1982;
CERCLA Section 301(e) Report: Injuries and Damages
from Hazardous Wastes -- Analysis and Improvement of
Legal Remedies, U.S. GPO, 1982;
Congressional Research Service, Six Case Studies of
Compensation for Toxic Substances Pollution, prepared
for the Senate Committee on Environment and Public
Works, 96th Cong.,.2nd S., U.S. GPO, 1980;
W. Prosser, Law of Torts, West Publishing, 1971;
Note, "Strict Liability for Generators, Transporters,
and Disposers of Hazardous Waste," 64 Minnesota Law
Review 949, 1980;
Ginsberg & Weiss, "Common Law Liability for Toxic
Torts: A Phantom Remedy," 9 Hofstra Law Review 859,
1981; and
J. Trauberman,- "Statutory Reform of 'Toxic Torts':
Relieving Legal, Scientific, and Economic Burdens on the
Chemical Victim," Environmental Law Institute, 1983.
Below, we first describe the seven representative state liability regimes.
Then, we present our estimates of their frequencies. Finally, the procedure
used in the Model to simulate the legal validity of claims is described.
Representative Liability Regimes
In order for a claim to be legally valid, it must be recoverable under an
applicable liability rule. There are two sources of liability rules that
apply to claims arising from hazardous waste disposal facilities: state
statutes and state legal theories. There are two distinct types of statutory
-------
A-156
liability rules.91 There are four types of applicable legal theories:
strict liability, trespass, nuisance, and negligence.
To develop the representative liability regimes, we first assumed that
injured persons may plead in the alternative, that is, they may seek
compensation under a statute and one or more legal theories simultaneously.
For example, a claim for personal injury could be based on strict liability,
nuisance, and negligence; and although the claim based on strict liability
might fail, recovery might be available under a nuisance or negligence
theory. In general, the more liability rules that are applicable to a given
claim, the greater the probability of recovery.
Second, we assumed that when claimants plead in the alternative, the
likelihood of a claim being legally valid under one statute or legal theory is
independent of the other statutes or legal theories being used and their
likelihoods of legal validity. This allowed us to assume independent
probabilities of legal validity for each statute and legal theory. These
probabilities, which may vary by the type of release leading to the claim, are
presented in Exhibit A-51.
Finally, the legal regimes were constructed by choosing combinations of
the statutes and legal theories that may be available as the bases for seeking
recovery (see Exhibit A-52). The probabilities of legal validity for each
available basis for recovery were used to estimate an overall' expected
probability of legal validity for the regime. For example, if only the
theories of trespass and negligence can be used as bases for seeking recovery
from a detectable (but not toxic) release, the probability of legal validity
(based in the values in Exhibit A-51) would be:
(0.40) + (1.0-0.40) x (0.20) = 0.52
Exhibit A-53 lists the probabilities that each of the three claim types will
be legally valid under each of seven state regimes by release type. Each
regime is discussed in turn.
(1) Statutory Liability I. A state statute is assumed to apply to
claims for personal injuries and real property damage. Recovery under the
statute for these damages is available under all conditions, i.e., the
probability of recovery is 100 percent. A negligence action is assumed to be
the only liability theory that can be applied for recovering an economic
loss. This assumption was made so that this legal regime would conform
closely to known state laws.
(2) Statutory Liability II. A state statute is assumed to apply to
claims for real property damage and economic loss; all such claims were
assumed to be legally valid. The statute does not apply to personal injury.
91 ICF Incorporated, Review of State Liability Provisions, report to
U.S. Environmental Protection Agency, June 1982.
-------
A-157
EXHIBIT A-51
ASSUMED PROBABILITY OF LEGAL VALIDITY OF CLAIMS
BROUGHT UNDER STATUTES AND LEGAL THEORIES
Probability of Legal Validity
(Reported in Percent)
Basis for
Seeking Recovery
State Statute
Strict Liability
Trespass
Nuisance
Negligence
Claims Due to
Detectable Releases
100
60
40
30
20
Claims Due to
Toxic Releases
100
67
50
30
30
i
Source: ICF estimates.
-------
A-158
EXHIBIT A-52
STATUTES AND LEGAL THEORIES AVAILABLE AS BASES FOR RECOVERY
FOR EACH LIABILITY REGIME IDENTIFIED BY CLAIM TYPE1
Liability Regime
1. Statutory Liability I
2. Statutory Liability II
3. Strict Liability I
4. Strict Liability II
5. Trespass
6. Nuisance
7. Negligence
Personal Injury
Claims
Real Property
Damage Claims
Economic Loss
Claims
STAT
TRES,
NEGL
STRL,
NUIS,
STRL,
NUIS,
TRES,
NUIS,
NEGL
NUIS,
TRES,
NEGL
TRES,
NEGL
NEGL
NEGL
STAT
STAT
STRL,
NUIS,
STRL,
NUIS,
TRES ,
NUIS,
NEGL
TRES,
NEGL
TRES,
NEGL
NEGL
NEGL
NEGL
STAT
TRES,
NEGL
STRL,
NUIS,
TRES,
NUIS,
NEGL
NUIS,
TRES,
NEGL
NEGL
NEGL
1 STAT
STRL
TRES
NUIS
NEGL
State statute available as a basis for recovery.
Strict liability available as a basis for recovery.
Trespass available as a basis for recovery.
Nuisance available as a basis for recovery.
Negligence available as a basis for recovery.
-------
A-159
EXHIBIT A-53
PROBABILITY OF CLAIMS BEING LEGALLY VALID
FOR EACH STATE LEGAL REGIME
BY RELEASE TYPE
(Reported in Percent)
Statutory
Liability I
Statutory
Liability II
Strict
Liability I
Strict
Liability II
Trespass
Nuisance
Negligence
Claims Due to
Detectable Constituent
Personal
Injury
100
66
86
86
52
44
20
Real
Property
Damage
100
100
86
86
52
44
20
Economic
Loss
20
100
66.
86
52
44
20
Claims Due to
Toxic Constituents
Personal
Injury
100
79
92
-92
65
51
30
Real
Property
Damage
100
100
92
92
65
51
30
Economic
Loss
30
100
79
'
92
65
51
30
Source: ICF Incorporated estimates.
-------
A-160
Claimants can seek to recover for personal injuries under the legal theories
of trespass, nuisance, and negligence.
(3) Strict Liability I. All four legal theories (strict liability,
trespass, nuisance, and negligence) are assumed to apply to claims for
personal injuries and real property damages. Claims for economic loss cannot
be recovered under a strict liability theory, but can be recovered under the
other three legal theories. Therefore, the probability of recovering for
economic loss is slightly lower than for personal injury and real property
damage.
(4) Strict Liability II. All four legal theories are assumed to apply
to all damage claims. The probability of recovery is thus the same for
personal injury, real property damage, and economic loss.
(5) Trespass. The legal theories of trespass and negligence are assumed
to apply to all damage claims. Recovery is assumed to be unavailable under
any state statute, strict liability theory, or a theory of nuisance.
(6) Nuisance. Only the legal theories of nuisance and negligence are
assumed to apply to all damage claims.
(7) Negligence. All damage claims can be based on a negligence theory
only.
It should be noted that our analysis does not include an assessment of the
strategic behavior likely to be undertaken by harmed parties seeking relief
for damages (harmed parties may overstate their losses in a strategic attempt
to obtain favorable judgments). Thus, the percentages given in Exhibit A-53
relate to claims for the actual harms which people believed they have
suffered, and not to any overstated claims that may be requested by
individuals or groups seeking recovery. Because the PCLTF Model attempts to
estimate claims for the actual harms people believe they have suffered, the
question of strategic behavior on the part of harmed parties is implicitly
assumed away. No apparent bias is introduced by this approach.
State Regime Frequencies
Our data sources revealed that two states have statutory liability rules
roughly similar to the first state liability regime, Statutory Liabilty I, and
two other states have rules similar to the second regime, Statutory Liability
II. About two thirds of all states have liability rules similar to one of the
strict liability regimes. A review of reported cases and the legal literature
indicated that nuisance actions were more common than trespass actions.
Furthermore, negligence bases for trespass and nuisance claims are included in
those categories. On this basis, we estimated the frequencies with which the
legal regimes are anticipated to occur among the states as:
Statutory Liability I: 0.04;
Statutory Liability II: 0.04;
Strict Liability I: 0.30;
Strict Liability II: 0.35;
-------
A-161
Trespass: 0.05;
Nuisance: 0.15; and
* Negligence: 0.07.
Although the distribution of legal regimes is likely to change over times we
have made no attempt to predict future changes in these frequencies.
Simulating Legally Valid Claims
Using the state regimes and their frequencies, we simulated the legal
validity of third-party claims for the PCLTF Model in the following manner:
1. For each of the lower 48 states, assign one of the seven
liability regimes using the frequency distribution described
above. The Model is designed so that the user can specify
liability regimes for particular states as he or she so
desires. In the absence of user-supplied inputs, each state is
randomly assigned a regime for each iteration.
2. When any third-party claim arises, identify the release type
leading to the claim and the state in which the release occured.
3. Based on the liability regime in the identified state and the
release type, identify the probability of legal validity for
each claim type (i.e., the probability from Exhibit A-53).
4. Simulate whether each claim is legally valid by drawing a random
number and comparing this number to the probability of legal
validity. If the claim is legally valid, it is then allocated
to one of the funding sources. If it is not, then the quantity
is kept track of separately and is not allocated. The Model
output summarizes the extent to which each claim type is found
to be legally valid.
-------
APPENDIX B
RELEASE SIMULATION MODEL
This appendix describes the Release Simulation Model. This model
simulates the movement of water into and out of chemical waste disposal sites
by characterizing the site and its immediate environment in terms of those
properties that affect hydrologic transport. A Monte Carlo simulation
technique is employed to describe predicted weather patterns. Subsequent
precipitation is followed through runoff, evapotranspiration, and infiltration
using a simple water balance approach. Leakage into the waste is assumed to
dissolve contaminants to their point of solubility. The resultant leachate is
then tracked as it progresses through the unsaturated zone and into the
aquifer. Transport velocity and plume dimensions are estimated using an area
source three-dimensional transport equation that models dispersion in two
dimensions (i.e., horizontal plane) and assumes a uniform concentration in the
vertical dimension. Releases are defined as seven discreet incidents where
detectable or toxic levels of contaminants arrive at a monitoring or potable
well, or overflow across the land surface.
B.I MODEL METHODOLOGY
The design of the failure/exposure prediction elements of the Release
Simulation Model is illustrated in Exhibit B-l. In essence, the functional
components rely on user-supplied data, data held within captive data banks,
programmed analytical constructs for predicting continuous events (e.g.,
leachate movement), and Monte Carlo simulations to predict probabilistic
events (e.g., rainfall patterns). The individual algorithms employed, data
sources, and assumptions made in developing each structural component of the
model are discussed in the following sections.
B.I.I User-Supplied Inputs and Default Values
At the beginning of a simulation, the user supplies a baseline set of
input information. This baseline information is supplemented by
internally-generated values to create the required data set outlined in
Exhibit 8-2. In addition, default values have been programmed into the model
which can be utilized whenever the user does not supply a value. Some inputs
are self-explanatory; those that are not self-explanatory are discussed in
this section with respect to their use and default values, when appropriate.
1. Type of Site
Each site is characterized by:
Number of layers;
Thickness of each layer (in cm);
Field capacity of each layer (in cm);
-------
B-2
EXHIBIT B-l
RELATIONSHIP OF COMPONENTS IN RELEASE SIMULATION MODEL
USER
SUPPLIED
INPUTS
WEATHER
SIMULATION X^
>>. i
NATIONAL BASELINE . WASTE
^/^
LOCATION S
REFERENCE
i
BASt
SITE
DATA
FILE
1
WATER
MOVEMENT
SIMULATION
BREACH
TABLE
-------
B-3
EXHIBIT B-2
USER-SUPPLIED INPUTS FOR RELEASE SIMULATION MODEL
USER INPUTS NEEDED AT THE START OF A RUN:
Number of sites
* Number of Monte Carlo replications
Base year (i.e., the time origin for the run)
* Starting "random" number seed(s)
VARIABLES THAT THE USER CAN RESET AT THE START OF A RUN:
Probability that a synthetic liner fails during a given month
(Default value based on expected life of 25 years)
Probability that a synthetic liner fails at time zero (Default
value: 0.25)
Probability that a synthetic cover fails during a given month
(Default value based on expected life of 50 years)
Probability that a synthetic cover fails at time zero (Default
value: 0.05)
Amount of water that defines a breach when the monitoring is done
inside the site
FOR EACH SITE OR FACILITY, THE USER MUST DEFINE THE FOLLOWING:
* Site type (if the user picks one of the standard site types, then
the thickness, field capacity,1 saturation, starting water
content,2 probabilities of synthetic liner failure, number of
layers, type of layers (e.g. clay, sand, etc.), drainage, and Type 1
breach flags are obtained from an internal data base; if the user does
not pick one of these standard types, he must supply his own values).
Latitude and longitude
Number of wastes in the site and an EPA waste code for each waste
Date opened and closed
Date when post-closure care ceases
Date on which study ends for this site
1 Field capacity refers to the amount of moisture a medium will hold
after the layer has been drained.
2 Starting water content refers to the initial amount of moisture in the
layer.
-------
B-4
EXHIBIT B-2 (continued)
USER-SUPPLIED INPUTS FOR RELEASE SIMULATION MODEL
Type of waste package (drum or sludge)
Width
Length
Distance to the monitoring well (which will be set to 1 meter if
monitoring is to be done at the site boundary)
THE COMPUTER WILL SUPPLY ALL OTHER VARIABLES. THE USER, HOWEVER, CAN OVERRULE
THE COMPUTER*AND SUPPLY ALTERNATIVE VALUES FOR THE FOLLOWING:
Seepage velocity range -- i.e., the velocity range of the aquifer
(speed at which the groundwater is moving)
Distance to drinking water well and/or surface water
Use of the Blaney-Criddle Method over the Thornthwaite Method to
characterize evapotranspiration (if the user chooses this option, he
must also supply all the parameters needed by the Blaney-Criddle
Method)
Runoff coefficients -- i.e., the fraction of precipitation that
runs off the surface rather than infiltrates into the soil
Erosion
Effective porosity range -- i.e., the range of values defining
pore or void space in the soil; this is the space between soil
particles through which soil moisture can flow
Range of depth to groundwater
Field capacity of the topsoil
Amount of water that defines a breach at a monitoring location
when the monitoring well is at the perimeter
Random number seeds for the climate calculations
Total dissolved solids and total organic carbon content of the
uppermost aquifer
Thickness of the aquifer
-------
B-5
" Saturation for each layer (in cm) -- i.e., amount of
water that can be received by a layer before it
overflows or backs up into the layer above;
Probability that synthetic liners/caps will fail in
any given month; and
Probability that synthetic liner/cap is punctured at ,.
time zero.
The first four of these characteristics can either be supplied by the user
or obtained from an internal data base. To use the latter option, the user
must identify the site as one of seven standard designs. An identifier for
each layer is assigned to designate the material type: waste-1, sand-2,
clay-3, synthetic (e.g., plastic, asphalt, concrete)-4, soil-5. A separate
identifier is used to denote the presence of a leachate collection system that
would prevent the upper soil layers from becoming saturated.
The seven design types are summarized in Exhibit B-3. These prototype
designs are defined by the material and thickness of the layers underlying the
site. The designs are representative of the series of layers for surface
impoundments and landfills.
As the footnotes to Exhibit B-3 explain, unlined surface impoundments and
landfills are modeled with underlying clay layers with a large minimum
permeability of 10 7 m/sec. This permeability is large enough so that in
general, the amount of water leaving the site is controlled by the amount of
moisture available from the infiltration of rainfall, rather than the liner's
permeability. Consequently, although clay layers are used in modeling the
unlined prototype designs, the layers play virtually no role in containing the
wastes in the facilities.
Prototype designs 3, 4 and 5 were developed to represent landfills.
However, because releases at facilities with these designs may be less
influenced by the additional hydraulic pressure from the standing water in
surface impoundments during its operation, these designs were also used to
represent surface impoundments.
Latitude and longitude "are used to specify the location of the site. This
allows data to be drawn from the National Data Base for the appropriate region
of the country and thus ensures the proper consideration of site-specific
features.
2. Waste Codes
The waste in each site is identified by EPA waste codes. Although the
code designation provides no input that is directly applicable to subsequent
analyses, it does identify a discrete waste type. This waste type is then
used to specify toxic constituents and their properties that will affect the
waste type's transport and fate in the environment.
-------
EXHIBIT B-3
MATERIALS USED IN SEVEN PROTOTYPE DESIGNS
PROTOTYPE
LAYERS AND THEIR THICKNESS IN CM I Top to Bottom)
DESIGN PROCESSES MODELED WITH DESIGN
1 Un lined Surface Impoundment
2 Clay- Lined
3 Landfill,
(4 Landfill,
5 Landfill,
Surface Impoundment
Surface Impoundment
Surface Impoundment
Surface Impoundment
6 Unl ined Landfi II
7 Clay- Lined
1 A 91 .5 cm thick
Landfi 1 1
1
Soi 1
61.0
Soi 1
61.0
Soi 1
61.0
Soi t
61.0
Soi 1
61.0
Soi 1
61.0
Soi 1
61 .0
clay layer with hydraulic
2
Sand
30.5
Sand
30.5
Sand
30.5
Sand
30.5
3
Clay
91.5
Sand
30.5
Sand
30.5
3
3
Clay
61.0
3
Clay
61.0
Plastic
0.005
Plastic
0.005
Plastic
0.005
3
Clay
61.0
Clay
61 .0
conductivity of
H
Waste
91.5
Waste
91.5
Sand
15.2
3
Clay
61.0
Sand
30.5
Waste
600.0
Waste
600.0
5 6 7 8 9 10 1 1
1
2
Clay
91.5
3
Clay Waste Sand Plastic Sand
61.0 600.0 30.5 0.20 15.2
Waste Sand Plastic Sand Plastic
600.0 30.5 0.005 30,5 0.20
3 33
Plastic Clay Waste Sand Clay Plastic Clay
0.005 30.5 600.0 30.5 61.0 0.008 15.2
i*
5
Clay
91.5
10-7 m/sec was used as the bottom layer. Due to the standing
03
I
water in the surface impoundment during operation, the effective hydraulic conductivity of this layer was modeled as
2x10-7 m/sec, which yields a minimum flow of 51.8 cm of water per month (assuming that much water is available to flow
through the facility).
Hydraulic conductivity of the layer assumed to be 10-8 m/sec. Due to standing water during operation, the effective
hydraulic conductivity is 2x10-8 m/sec, which yields a minimum flow of 5.18 cm of water per month (assuming that much
water is available to flow through the facility).
Hydraulic conductivity of the layer assumed to exceed 10-9 m/sec.
A 91.5 cm thick clay layer with hydraulic conductivity of 10-7 m/sec was used as the bottom layer. This yields a minimum
flow of 25.9 cm of water per month (assuming that much water is available to flow through the facility).
Hydraulic conductivity of the layer assumed to be 10-8 m/sec. This yields a minimum flow of 2.59 cm of water per month
(assuming that much water is available to flow through the site).
-------
B-7
3. Opening and ClosureDates
Opening and closure dates are specified to indicate: (1) when the face is
first opened and, therefore, able to collect precipitation; (2) how long the
face is open; and (3) when the cap is emplaced. Leachate migration is tracked
from the site's opening date. The date when post-closure care is terminated
is equated with the time when leachate collection is terminated and erosion is
allowed to proceed unabated. This assumption may be changed by the user to
reflect both the continued maintenance of the cap after the period of
owner/operator responsibility and the possibility that leachate collection may
cease when the cap is emplaced.
Leachate generation and migration will begin when the site opens, as will
the aging process for any synthetic liners. Infiltration waters will
accumulate at a faster rate while the site is in operation because of the lack
of a low-permeability cap. After closure, the cap will be in place and
infiltration will be reduced. Because leachate migration can be substantial
during operation, it must be tracked from the opening date.
4. Length of Run
The length of run is specified by the user upon activating the model.
This is the period of time following facility opening (i.e., following when
the facility begins accepting waste for disposal) that the model will check
for breaches.
5. Probability of Liner Failure
The probability of liner failure refers to the density function employed
in determining when a synthetic membrane liner is likely to fail because of
natural degradation. This function is separate from the probability of
failure resulting from improper installation or puncture during site operation
(i.e., failure at time zero). Liner failure addresses the limits placed on
the life of a properly-installed membrane as a result of chemical degradation
and natural forces. The user may specify a value for this probability or may
rely on the following default values: an average life of 25 years for
synthetic membrane liners and 50 years for synthetic membrane covers. The
probability of sudden breaches in soil liners is assumed to be zero.
B.I.2 Internally Stored Inputs
In addition to inputs supplied by the user, the Release Simulation Model
requires waste- and geography-specific inputs in order to conduct the
analysis. These inputs are stored in a data base internal to the Model.
1. Waste-Specific Data
The EPA waste identifier codes specify the waste type or source. For the
analysis here, information is required on the constituents of wastes and their
properties with respect to their transport and fate in the environment.
-------
B-8
For each waste code, toxic constitutents were identified for the purpose
of modeling breaches. Constituents were selected by reviewing EPA regulatory
documents1 and a draft report by MITRE.2 These documents list the most
hazardous constituents for each waste code.
For each of the identified toxic constituents, the following data were
compiled: solubility, drinking water or priority pollutant criteria levels,
detection limit, octanol-water partition coefficient (organics),3 and Kd
values (inorganics).'* With respect to toxicity criteria based on
carcinogenesis, the 10 6 risk level value was used.5 These physical
constants were derived from the "Background Document-Listing of Hazardous
Wastes," ojg. cit., and its reference material.6 Only those constituents
with primary drinking water standards or designated priority pollutant
criteria were considered for the purpose of determining a toxic breach.
Average detection limits of 10 vg/1 for organics and 20 ug/1 for
inorganics were assumed, based on EPA guidelines.7 These detection limits
assume that monitoring programs will use gas chromatography/mass spectrometry
(GC-MS) analyses for organics and inductively coupled plasma emission
spectrometry (ICP) for inorganics.
In addition, data are entered on the taste and odor threshold, the total
dissolved solids (TDS),.and total organic carbon (TOC) levels associated with
all constituents in each waste. These quantities denote when organoleptic
1 U.S. Environmental Protection Agency, "Background Document-Listing of
Hazardous Wastes," November 14, 1980.
2 MITRE Corporation, "Composition of Selected Hazardous Waste Streams,"
prepared under Contract 68-01-6092.
3 An octanol-water partition coefficient represents the degree to which
a chemical will distribute itself between water and octanol when shaken in a
mixture of the two. It is used as a measure of a chemical's affinity for
water as opposed to organic fluids.
11 "Kd" is a parameter used to represent the distribution of a chemical
between water and soil particles.
5 The 10 6 risk level is defined as the concentration at which
exposure over a lifetime (70 years) would result in one additional case of
cancer in 106 people.
6 Dawson, G. W., C. J. English, and S. E. Petty, "Physical-Chemical
Properties of Hazardous Waste Constitutents," U.S. EPA, March 5, 1980.
7 "Guidelines Establishing Test Procedures for the Analysis of
Pollutants; Proposed Regulations." Federal Register, 69464-69575, December
3, 1979.
-------
B-9
properties or an indicator parameter would lead to the detection of leakage in
potable or monitoring wells, respectively.
The detection of a breach is based on the arrival of mobile species at the
point of detection. Therefore, the most important constituent(s) in each
waste will be the "early arrival" contaminant(s), i.e., those constituents
that arrive first at the monitoring or potable well or a surface water in
detectable or toxic concentrations. Recognizing this, a representative
chemical(s) is identified for each waste code for the purposes of subsequent
transport modeling. This identification was accomplished by comparing the
solubility, detection limit, and toxicity criteria for the various chemicals
with data on their potential interactions with soil (e.g., attenuation). For
consideration of breaches defined by the arrival of detectable levels, the
ratio of the solubility (S) to the detection limit (DL) was calculated to
indicate the amount of dilution and attenuation necessary for each, constituent
below which leachate will not produce a detectable level at the point of
observation (it is assumed that all constituents are leached from the site at
their water solubility). DL may constitute the indicator parameter level, the
chemical analysis level, or the taste and odor threshold.
Similarly, for breaches defined by toxicity considerations, the ratio of
the solubility (S) to the toxicity criteria (TC) was derived to indicate the
level of dilution and attenuation necessary to prevent a toxic level from
occurring at the point of observation. For organic chemicals, the
octanol-water partition coefficient (P) is converted into a Koc8 using the
Karichoff relation (log Koc = 0.02 + 0.94 log P).9 This relation is
subsequently employed to calculate a Kd (soil-water partition coefficient) for
organic contaminants, assuming an average soil organic content of 2 percent.
Kd values are used directly in subsequent modeling algorithms to designate
soil attenuation. A Kd value is also specified to indicate soil attenuation
for inorganic constituents.
For each waste code, the values (Log (S/DL) - Log (Kd)) and (Log (S/TC) -
Log (Kd)) were calculated for each constitutent. The constituents yielding
the highest values were designated as "early arrival" contaminants for use in
subsequent analyses because they will move through the environment most
rapidly. Therefore, the one or two constituents with the highest values are
used to designate the properties of a given waste type. For a site receiving
more than one waste, the values of each constituent are compared and the
constituents that move most rapidly through the environment are used for
subsequent modeling. Thus, transport and concentration that result in a
8 "Koc" is the distribution coefficient adjusted to the organic carbon
content of the soil.
Koc = concentration of organic carbon in soil
concentration in water
9 See tf. B. Neely, Chemicals in the Environment, Marcel Dekker, New
York.
-------
B-10
breach will be based on up to four contaminants (one each for indicator
parameters, taste and odor, chemical detection, and toxicity). In essence,
this process reduces the modeling effort to one of following the "early
arrival" toxic constituents.
No degradation is accounted for in the transport model. It is assumed
that groundwater is abiotic (i.e., has no active biological community as a
result of filtration of bacteria by the soil and potential toxicity of
leachate) and, therefore, does not subject contaminants to biodegradation.
Photodegradation (decomposition sponsored by the presence of light, e.g.,
sunlight) is also eliminated in groundwater systems. Hydrolysis and chemical
oxidation may still occur, but because the worst-case contaminants are of
primary interest, constitutents with significant degradation rates are not
retained as indicator chemicals.
In addition, wastes are characterized by the number of compound classes
present among inorganic, polar organic, and nonpolar organic chemicals. This
is done to assist in predicting liner failure. Synthetic membrane liners are
often compatible with only one of these three classes of chemicals.
Therefore, synthetic membrane liner life can be expected to be reduced if more
than one class of chemicals is present. Finally, a flag is added to any waste
containing carcinogenic constituents. Thus, when a potable well or surface
water is contaminated at toxic levels, the model output indicates whether
carcinogens are involved.
2. Geography-Specific Data
The National Data Base houses site-specific data on environmental factors,
including pertinent hydrogeological, geological, and meterological data
registered by longitude and latitude. When a site is identified by its
geographic coordinates, the National Data Base can supply the appropriate
environmental data. Specific parameters included in this data base are listed
in Exhibit B-4.
Temperature data are used to determine losses of water from a site as a
result of evapotranspiration. Historic monthly temperature pattern data have
been obtained from the National Climatic Center for all first-order weather
stations based on 40 years of records. For each site, the closest weather
station is used to provide an estimate of temperature for any given set of
coordinates in the lower 48 states. Historic monthly precipitation data were
obtained in the same manner to determine infiltration.
The length of time a breach requires to reach the underlying aquifer will
depend in part on the depth of the aquifer, because it takes longer for
leachate to travel to deeper water tables. Hence, data from United States
Geological Survey and American Water Wells Association publications and
resources maps have been integrated into a single data set, which will provide
a range of depths to a potable aquifer for any set of coordinates.
-------
B-ll
EXHIBIT B-4
INPUTS INCLUDED IN THE NATIONAL DATA BASE
Temperature Profile (°C) - density curve drawn from historic data
Precipitation Profile (cm/mo) - density curve drawn from historic data
Depth to Groundwater (cm)
Hydraulic Conductivity (cm/sec)
Ratio Gradient/Porosity (dimensionless)
Distance to Drinking Water Well or Surface Water (cm)
Overburden Field Capacity (cm/cm)
Erosion Loss Rate (cm/mo)
Runoff Coefficient (dimensionless)
Aquifer Water Quality (mg/1)
Aquifer Thickness (cm)
-------
B-12
Subsequent transport of the plume in the aquifer will depend on the
aquifer's hydraulic conductivity, porosity, and gradient. To create a
comparable data base for these aquifer properties, Water Atlas10 data on the
aquifer type in each of twelve major groundwater regions of the lower 48
states (see Exhibit B-5) were integrated with data from Fugro, Inc.,11
Heath,12 and the EPA surface impoundment survey.13 Correlations were made
with typical permeability, porosity, and slope information for the soil/rock
types to estimate a velocity range for each aquifer.
The National Data Base is both general and broad in scope. More specific
data have been found for some aquifer properties. In particular, EPA's
surface impoundment survey provides more specific values for depth to
groundwater for some 1,200 data points shown in Exhibit B-6. These values are
used when a site lies within the same 7.5 minute grid as one of the data
points. Finally, if the user has site-specific data, they may be used in
place of the values normally provided by the National Data Base.
Overburden field capacity is used to determine how much moisture the soil
will hold before the water moves to the next layer. Starting with the
assumption that overburden for a particular facility is taken from surface
soils near that site, soil groups from the United States Department of
Agriculture's generalized soil map were matched with representative soils and
relevant taxonomic data were reviewed to determine the one-third bar moisture
content.1" For highly sandy soils, the fifteenth-bar value was employed.
This value is roughly comparable to field capacity, since it is a measure of
moisture content at reduced pressure. The resulting designation of field
capacity values is provided in Exhibit B-7.
Erosion loss rates are used to determine stresses put on the facility
cap. State-by-state values for average anticipated erosion were extracted
10 Geraghty, Miller, Van Der Leeden, and Troise, "Water Atlas of the
United States," Water Information Center Publication, Post Washington, New
York.
11 Fugro, Inc., "Survey of Potentially High Yield Aquifers Within the
Contiguous United States," Project No. 75-073-EG. Prepared for
Battelle-Northwest, 1975.
12 Heath, R.C., "Classification of Groundwater Systems of the United
States," Groundwater. Vol. 20, No. 4, July-August, 1982.
13 U.S. Environmental Protection Agency, "Surface Impoundments and Their
Effects on Groundwater Quality in the United States -- A Preliminary Survey,"
EPA 510/9-78-005, June 1978.
lh Soil Taxonomy, Agriculture Handbook 436. Washington, D.C.: U.S.
Department of Agriculture, 1975.
-------
B-13
EXHIBIT B-5
* GROUNDYVATER REGIONS OF THE
UNITED STATES, EXCEPT ALASKA AND HAWAII
-------
B-14
EXHIBIT B-6
LOCATION OF SPECIFIC DEPTH TO
GROUNDWATER VALUES
01
o
g
-------
B-15
EXHIBIT B-7
FIELD CAPACITY VALUES FOR SURFACE SOILS
(percent dry weight of soil)
-------
B-16
from the National Resource Inventory (NRI)15 and added to the National Data
Base. These values, provided in Exhibit B-8, represent the lowest reported
value for cultivated hay fields.
Runoff characteristics were used in determining infiltration at a site.
Runoff coefficients were derived on the basis of soil type, antecedent
rainfall, and season. First, four soil groups (shown for the U.S. in Exhibit
B-9) were developed from the USDA generalized soil map16 by matching soil
series names and soil groups.17 The antecedent moisture condition was then
calculated by dividing the average monthly rainfall value by six to determine
a five-day average value. The appropriate rainfall runoff fraction was
derived18 using the median value from rainfall events in the range 0 to 8
inches. The runoff coefficient differs between the growing and dormant
seasons for each soil group. The following values are based on conditions
expected in pasture land kept in good condition with an average slope of 2-5
percent:
Runoff Coefficient for Hydrologic Soil Group
5-Day Antecedent Rainfall
(inches) A B C D
Dormant Growing
(10/1 to 3/31) (4/1 to 9/30)
0.5 1.4 0.0 0.01 0.10 0.15
0.5-1.1 1.4-2.1 0.01 0.15 0.35 0.45
1.1 2.1 0.15 0.45 0.65 0.75
The National Data Base also includes a cross-reference file identifying
the FIPS code and latitude and longitude for the county seat for all
counties. This file was necessary to register the erosion loss data
identified by FIPS code; it was retained as part of the data base to
facilitate the addition of new data not registered by longitude and latitude
and to locate sites when only the county or FIPS code is given.
B.I.3 Relationships and Functions
The failure prediction provided in the Release Simulation Model is carried
out through a number of relations and functions as provided by the elements or
submodels entitled Weather Simulation, Baseline Analysis, Site Data File, and
Water Movement Simulation. These submodels are illustrated in Exhibit B-l.
15 United States Department of Agriculture.
16 Soil Taxonomy, op. cit., Figure 41.
17 See V. T. Chow, ed., Handbook of App1led Hydrology. New York:
McGraw-Hill, 1964, Table 21-8.
19
Ibid., Figure 21-6.
-------
B-17
EXHIBIT B-8
ANNUAL EROSION LOSS (tons/acres)
-------
B-18
EXHIBIT B-9
HYDROLOGIC SOIL GROUPS
-------
B-19
1. Weather Simulation
The weather simulation component utilizes a random draw or "Monte Carlo"
technique. The historic distribution of temperature and precipitation are
recreated in a density function of probable occurrences for each month of the
year. A first-order linear congruential pseudo-random number generator is
then employed to select a value for each month of the simulation. This
process is repeated for each of the primary weather stations in the U.S.; the
same weather pattern is simulated for all sites in the proximity of any given
weather station. The entire weather pattern for a simulation can be retained
by putting the starting draw number in memory, from which an identical series
can be generated by the program package. The output from the weather
simulation is ultimately employed to calculate monthly precipitation and
evapotranspiration losses at a site. Because the density function is
recreated from historic data, the potential for sustained wet or dry periods,
droughts, etc. is maintained in proportion to past occurrences. This prevents
the draw from making each month's weather totally independent from the
previous month.
2. Baseline Analysis
The baseline analysis element assembles all pertinent inputs for a given
site. In addition to data supplied by the user and data drawn from the waste
and national data bases, the baseline element provides the starting water
content of a facility. Water content is determined by assuming that all
infiltration during the operational period enters the system (i.e., no plants
are present to sponsor evapotranspiration, and evaporation is insignificant).
Water is then distributed throughout the site's soil layers in a manner
described in the section below on water movement simulation.
3. Site Data File
The site data file stores the integrated and calculated data inputs from
the baseline analysis for a given site.
4. Water Movement Simulation
The water movement simulation submodule is the heart of the breach
development mechanism. This unit models the infiltration of precipitation and
the subsequent migration of leachate to the point of detection. Because of
the random nature of initiating and contributing events, the simulation relies
on Monte Carlo techniques. This process begins by accounting for
uncertainties in seepage velocities, and dispersion values by drawing values
from density curves. These values are used for the rest of the simulation.
Having established the initial conditions at a site by either:
reading values in from the data base,
letting the user define the value, or
making a stochastic determination (as described above),
-------
B-20
the program simulates water movement over time. It begins this process with
the output from the weather simulation. With this information, the program
can account for the water balance in the surface layer for each month, using
the following methodology:
1. If post-closure care is no longer active at the site,
the thickness of the cover layer is reduced each month
by an amount indicating monthly erosion loss. Erosion
loss is based on either user-specified values (when
available), generalized data for each of the seven
prototype designs or, lacking either of these, the
state average value contained in the National Data Base.
2. The amount of raw infiltration (precipitation runoff)
is calculated using a runoff coefficient from the
National Data Base.
3. The potential evapotranspiration is calculated using
either the Thornthwaite19 or Blaney-Criddle method.
The choice of method is dependent on the site's
geographic location. If the soil cover has eroded,
evapotranspiration is considered to be negligible
because of the resulting lack of vegetation.
4. The amount of water in the cover layer is reduced if
the potential evapotranspiration exceeds the raw
infiltration. This calculation is made using a
function of the form shown in Exhibit B-10.
If there is a net movement of water into the cover layer beyond the
layer's field capacity, then water seeps into the layers below. Like the
cover material, the underlying layers exhibit a certain capacity to hold
water. In other words, once the field capacity of any layer is reached, it
passes any excess water it receives to the layer below. This process is
depicted in Exhibit B-ll. This "water balance" approach is analogous to
viewing each layer as a bucket with a finite capacity to hold water. The
inflows and outflows of water are calculated at .each time step. When the
bucket is full, it overflows into the next bucket (the underlying layer).
This water balance approach assumes that the rate at which the layers can
accept water is large enough so that within a time step (i.e., a month), a
layer can accept all the water "offered" to it. Because this is not
necessarily a valid assumption for either the clay layers or synthetic layers,
the approach is modified for these layers. For clay liners, the submodule
estimates the total amount of water the clay can accept in a one-month
19 Thornthwaite, C. V. and J. R. Mather, "instructions and Tables for
Computing Potential Evapotranspiration and the Water Balance," Publications
in Climatology. Laboratory of Climatology, Drexel Institute of Technology,
Vol. 10, No. 3, 1957, pp. 185-311.
-------
B-21
EXHIBIT B-IO
RELATION OF EXCESS EVAPOTRANSPIRATION
TO INFILTRATION
t , i . I , i , . I
s
aanxsiow iios
Source: Thornthwaite and Mather, 1957.
-------
B-22
EXHIBIT B-ll
CONCEPTUAL MODEL OF WATER BALANCE
CALCULATION
EVAPOTRANSPIRATION
I I
PRECIPITATION
RUNOFF
OVERBURDEN
LINER
WASTE
SAND
LINER
SOIL
J INFLOW 1 « PRECIPITATION - EVAPOTRANSPIRATION - RUNOFF
tINFLOW 2 =
INFLOW 1 -
FIELD CAPACITY 1
-------
B-23
period. First, an estimate is made for the hydraulic conductivity as a
function of water content (Exhibit B-12). Field experience suggests that for
clays employed as liners, the volumetric wetness remains at a minimum level
(e.g., 0.3), with the result that conductivity will not drop below 10 s
m/s. The driving force downward (head gradient) is.assumed to be the distance
between the midpoint of the two layers. If the maximum amount of water the
clay can accept in a month is less than the amount "offered," then the smaller
amount is accepted into the clay layer and the rest is redistributed in the
layers above the clay.28
If a synthetic layer is intact, it causes the same redistribution of
water, except that no water is allowed to pass through the synthetic layer.
If an active leachate collection system is present in a layer, all water in
excess of field capacity that is not passed by the next lower layer is assumed
to be removed by the collection system.
At the end of a simulated year's time, the submodel has an estimate of the
volume of water, if any, which has passed through the site for every month of
the simulated year. A running total of the total volume of leachate to date
is also kept, as is a running total for the volume of water collected in the
leachate collection system. If the total amount of water is less than the
"critical value" (defined below), then the calculation goes on to the next
year, with no calculation made for travel time to the monitoring well. In
other words, until the layers between the waste layer and water table are
saturated with contaminants, no significant movement of waste into the aquifer
will occur. If this critical value has been exceeded, however, the waste
enters the groundwater system, and dilution effects and travel time to both
the monitoring well and the drinking water well are calculated.
The hydrologic transport model is composed of two discrete sub-models:
* Landfill Leachate Source Model in which the
landfill site is assumed to be a two-dimensional
leachate source to the unsaturated soil column beneath
the site, and approximate dimensions of the site are
obtained from the data on site type. The leachate
discharge (concentration and flux) is assumed to be
uniformly distributed over the site at any instant in
time. However, the discharge varies with time according
to the moisture infiltration of the site.
» Saturated Zone Transport Model in which the
leachate plume from the unsaturated zone acts as a
two-dimensional horizontal area source on the water
table surface of the saturated zone. It is assumed that
the water flux from the unsaturated zone is
2 B
The hydraulic conductivity of the underlying clay layer for prototype
designs 1, 2, 6, and 7 is fixed at a value greater than 10 9 m/s. See
Exhibit B-3 for details.
-------
B-24
EXHIBIT B-12
MODIFIED SOIL HYDRAULIC CONDUCTIVITY
RELATION EMPLOYED IN RELEASE SIMULATION MODEL
u
o>
t/l
,-4
,-6
10
-3
10
-1C
10
-1
10
-14
10
-16
0.1
VOLUMETRIC WETNESS (M3/M3)
0.2 0.3 0.4 0.5 0.6
-------
B-25
insignificant compared to the water flux in the
saturated zone; thus, only contaminant particles are
released to the saturated zone. The leachate plume
moves through the saturated zone according to a constant
value seepage velocity and dispersion coefficient
influenced by geochemical retardation of flow and the
dispersion model's attenuation in the system.
The transport equation is given by:
3C Vx 3c Dx 92c
St R 3x R 9x
where C - C(x,y,z,t) = leachate concentration
at position (x,y,z) at time t
Vx = seepage velocity of water (x-direction)
Dx = dispersion coefficient of water (x-direction)
R = geochemical retardation coefficient of the
leachate.
This relationship is expressed in a simple three-dimensional hydrologic
transport model that monitors four fronts: (1) the detectable concentration
of the first arrival toxic constituent; (2) the toxic concentration of the
first arrival toxic constituent; (3) the taste and odor concentration of the
first arrival constituent (not just toxic constituents); and (4) the indicator
parameter (TDS for inorganics, TOC for organics) concentrations for the entire
waste. Initial concentrations are assumed to be at the constituent's
solubility level. Dilution and attenuation are accounted for during movement
through the soil column and the aquifer. The initial point of detection is
assumed to be the monitoring well, which is located at the downflow perimeter
of the site. The model employs a Gaussian three-dimensional analytical
solution which considers head drop, hydraulic conductivity, porosity, and
dispersion coefficients.
Two failure modes are considered for synthetic covers and liners. These
are: (1) failure resulting from improper installation and (2) failure
resulting from "aging." Failure resulting from improper installation will be
calculated using the user-supplied probability of improper installation.
Based on this probability, failure will then occur at randomly selected sites,
beginning at the opening date for liners and the closing date for covers.
Default values are 0.25 for liners and 0.05 for covers. These values were
based on limited data for actual landfill failures and should be revised over
time to reflect improved data and new construction practices.
Failure resulting from aging is determined by Monte Carlo simulation using
a probability distribution curve that varies over time. In other words, the
probability of rupture increases with the age of the liner or cover. However,
current sources -- including recent EPA publications describing the behavior
-------
B-26
and performance of synthetic liners --do not contain sufficient data to
predict liner life accurately.21 An average lifetime of 25 years for
synthetic liners and 50 years for synthetic covers is assumed by the model
unless overridden by user-supplied values. These figures are based on
expected values for pond liner life reported by Lee (1974)22 and on the
20-25 year guarantee that many synthetic membrane manufacturers offer for
their products. Liners are expected to age faster than covers because of
their interactions with wastes. The failure prediction is achieved through a
random draw from a bimodal distribution based on the mean average life for the
membrane.
Certain conditions can cause more rapid aging of liners and covers. It
has been observed that waste incompatibility will lead to a more rapid aging
of liners. To account for this, waste constituents are classified as polar
organics, non-polar organics, or inorganics. If more than one of these waste
types is present at a site, it is assumed that the liner will age at twice the
normal rate. Covers exposed to sunlight will age more rapidly than those
protected by soil cover. If erosion has removed soil over a cover, it is
assumed that the cover will then age at twice the normal rate.
A potential failure mode of clay covers is intrusion by plants and
animals. Based on work being conducted by Battelle Pacific Northwest
Laboratories and the results of Hakonson and Gladney,23 it was concluded
that the effect of plant root intrusion on permeability would probably be
small, but that animal intrusion may have a significant effect if the cover is
shallow enough to be in the burrowing zone of animals. Therefore, it is
assumed that animal intrusion will occur when erosion reduces the total
thickness of the cover and overburden to one meter or less. It is also
assumed that intrusion will affect 10 percent of the site area. Using data
presented by Moore and Ali24 and assuming that the effects of animal
21 Haxo, H. E., "Durability of Liner Materials for Hazardous Waste
Disposal Facilities," Proceedings of the Seventh Annual Research Symposium on
Land Disposal: Hazardous Waste. EPA-600/9-81-0026, EPA, Cincinnati, Ohio,
1981, pp. 140-156; Haxo, H. E., "Effects of Liner Materials on Long-Term
Exposure in Waste Environments," Proceedings of the Eighth Annual Research
Symposium of Land Disposal: Hazardous Waste, EPA-600/9-82-002, EPA,
Cincinnati, Ohio, 1982, pp. 191-211; and Matrecon, Inc., Lining of Waste
Impoundment and Disposal Facilities. SW-870, EPA, Washington, D.C., 1980.
22 Lee, J., "Selecting Membrane Pond Liners," Pollution Engineering, 6,
1, 1974, pp. 33-40.
23 Hakonson, T. E. and E. S. Gladney, Biological Intrusion of Low-Level
Waste Trench Covers, LA-UR-81-2972, Los Alamos National Laboratory, New
Mexico, 1981.
24 Moore, C. A. and E. M. Ali, "Permeability of Cracked Clay Liners,"
Proceedings of the Eighth Annual Symposium on Land Disposal of Hazardous
Wastes. 1982.
-------
B-27
intrusion are similar to cracking, the chance of cover failure will be
increased by a factor of two following animal intrusion.
Research being performed at Texas A&M University has shown that organic
leachates may have a significant effect on the permeability of clay liners.
Based on results presented by Anderson, Brown, and Green,25 the model
increases the permeability of clay liners by two orders of magnitude if
organic wastes are present in leachate.
B.2 OUTPUTS
The output from each release simulation is the timing and type of releases
for a given site. Replicate runs are made to generate a matrix of when each
of seven release types26 are estimated to occur in each of the replicated
release simulations. In addition, the simulated distance to the off-site
point of detection is reported in this matrix. Exhibit B-13 displays &
release matrix for 20 repetitions of the release simulation.
The release matrix can be developed at any level of aggregation. For
example, each site could be analyzed individually, thereby creating a separate
matrix for each site. Similarly, a group of sites could be run together,
resulting in a single (although potentially larger) matrix for the entire
group. The performance of each site in the group would be described by the
'same breach and timing tables.
25 Anderson, D., K. W. Brown, and J. Green, "Effects of Organic Fluids
on the Permeability of Clay Soil Liners," Proceedings of the Eighth Annual
Symposium onLand Disposal of Hazardous Wastes, 1982.
26 The seven release types (discussed more fully in Volume I of this
report) are as follows:
Release Type 1 - Indicator parameters arrive at monitoring well
Release Type 2 - Detection level for specific contaminants arrives at
monitoring well
Release Type 3 - Toxic level for specific contaminants arrives at
monitoring well
Release Type 4 - Taste and odor threshold arrives at potable
well/surface water
Release Type 5 - Detection level for specific contaminants arrives at
potable well/surface water
Release Type 6 - Toxic level for specific contaminants arrives at
potable well/surface water
Release Type 7 - Infiltration backs up and overflows
-------
B-28
EXHIBIT B-13
SAMPLE RELEASE MATRIX1
Time to Each Release Type
Distance to
Off-Site Point of
Iteration
Number
1
2
3
4
5
6
7
8
9
10
(Years
1
9992
50
10
20
999
999
40
90
30
999
2
25
45
8
40
999
110
70
50
999
40
Since
3
30
999
8
15
999
999
999
90
999
38
Facility Opening)
4
999
999
15
50
999
999
999
100
999
50
5
50
999
13
60
999
999
80
60
999
50
6
60
999
13
70
999
999
999
60
999
46
7
999
999
999
999
999
999
999
999
999
999
Detection
(feet)
2
1
4
8
10
5
9
2
,000
,000
800
,000
600
,000
,000
,000
,000
,000
1 Values are for illustrative purposes only.
2 999 means release was not observed in first 125 years.
-------
B-29
The advantage of analyzing groups of sites is that it greatly reduces the
required number of computations. For example, if 1000 repetitions are desired
to specify a release matrix, then individually analyzing 2000 sites would
require 1000 x 2000 = 2 million repetitions. However, the 2000 sites could be
divided into 10 groups of reasonably similar sites, with 200 sites in each
group. The analysis of each group still requires 1000 repetitions, or
1000/200 = 5 repetitions per site. Thus, the total number of repetitions
required is reduced from 2 million to 10,000.
Analyzing sites by groups is only appropriate for sites that are similar.
For example, separate release matrices would be desirable for sites that
differ by design and groundwater region. Analyzing the various sets of tables
would result in estimates of the expected influences on site performance of
the various factors used to distinguish groups. For the runs made in this
effort, 24 release matrices were created. (A total of 24 matrices were
created by analyzing each of six site designs within each of four geographic
regions covering the 48 states.)
B.3 DISCUSSION
B.3.1 Low-Probability Acute Events
V
It should be noted that the model does not account for acute disruptive
events such as earthquakes and extreme storms. A review of damage cases
showed that these events are probably very minor causes of discharges from
hazardous waste facilities.27 In light of the large number of breaches .
predicted for "normal" conditions, the omission of acute events has not had a
significant effect on the results.
B.3.2 Land Treatment Facilities
No special provisions have been made for the entry of land treatment
facilities in the model. It is assumed that when closed, these facilities
will be treated as landfills if hazardous residuals remain. If residuals do
not remain, no significant breach scenarios are evident. Hence, the landfill
model will suffice as long as the user provides information on the site
design, the moisture content at closing, and the nature of the waste.
B.3.3 Injection Wells
The Release Simulation Model is not designed to handle injection wells.
If injection wells are to be considered for qualification to enter the PCLTF,
an approach must be selected to predict the post-closure breach scenarios.
Two possibilities are of major interest: (1) the well is constructed
improperly and allows leakage into a potable aquifer during operation, which
results in the contamination of a drinking well, and (2) the receiving aquifer
27 Fred C. Hart Associates, Assessment of Hazardous Waste Mismanagement
Damage Case Histories, Draft, prepared for EPA under Contract 68-01-6474,
1981.
-------
B-30
is improperly characterized, allowing contaminants to reach a point of water
use after closure.
In the first case, the probability of improper construction would be a
user-supplied variable. Aquifer data employed in the landfill model would
then be utilized to predict the time of travel for toxic levels to a point of
use. However, deep wells usually reside in aquifers well below the surface
aquifer for which data exist in the model. Very little is known about these
deep aquifers.
In the second case, two relations must be derived. The first deals with
the likelihood that the aquifer is improperly characterized. This is likely
to vary directly with depth, i.e., the deeper the receiving aquifer, the more
likely that it is not well understood. The second relation required would
estimate the likelihood that a point of potable use is ultimately
contaminated. This probability will vary inversely with depth, i.e., the
deeper the receiving aquifer, the less likely that a potable source will be
affected. The selection of reasonable algorithms to describe these relations
would be controversial.
B.3.4 Corrective Actions
No special provisions have been made for analyzing the performance of
corrective actions. The model output currently describes the first occurrence
of the seven release types analyzed. If a corrective action is undertaken,
the descriptions of the remaining releases are no longer valid.
Although the Release Simulation Model has not been used to analyze the
performance of corrective actions, it could be modified to do so. For
example, it could be assumed that once corrective actions are specified, the
site can reenter the analysis. As currently envisioned, corrective action
could consist of two elements: cap/liner repair and withdrawal and
containment of contaminated groundwater. When cap/liner repair is undertaken,
the site could be reentered into the simulation as a new site with a design
that reflects the repair work.
Withdrawal wells would require slightly different treatment. The model is
too simple to evaluate the performance of well fields designed to withdraw
contaminated water for treatment. However, the effects of withdrawal would be
similar to the activation of a leachate control system, i.e., termination of
all uncontrolled contaminated flows through the liner during the period of
pumping. The user could specify how long pumping will be performed, with the
default value possibly being set as a period equivalent to the time it took
for the breach to be discovered in the first place. This is somewhat
analogous to the RCRA regulations that set the compliance period equal to the
period of operation of a given site. However, the validity of this assumption
is clearly uncertain. There is potentially a hysteresis effect in such
systems; desorption may well take much longer than the original adsorption
onto soils, depending on pumping rates and the contaminants involved. A
second area of concern is the residual contamination in the unsaturated zone,
which would effectively accelerate movement of subsequent breaches compared to
how the model would otherwise simulate the post-remedial action period. This
effect, while probably small, would constitute an inaccuracy.
-------
APPENDIX C
MODEL IMPLEMENTATION
This appendix describes how the PCLTF Simulation Model and the Release "
Simulation Model were implemented. The run characteristics, subroutines, data
files, and random number generators are described separately for each model.
Additionally, the statistical properties of the results of the PCLTF
Simulation Model are discussed.
C.I PCLTF SIMULATION MODEL
C.1.1 Run Characteristics
The PCLTF Simulation Model was run on an IBM 4341. Model execution
required approximately 7 megabytes of virtual storage. Disk storage
requirements were approximately 320 kilobytes for the input files, 1,200
kilobytes for the databases, and 3,900 kilobytes for the model routines.
Depending on the number of iterations and the number of years, the model
output is approximately 3 megabytes. A Base Case model run with 29 iterations
and a facility scaling factor of 1 for 100 years required approximately 16,000
CPU seconds to complete execution.
C.I.2 Pseudo-Random Number Generator
The PCLTF Simulation Model, like all stochastic models, relies on a
routine to generate random numbers for simulating probabilistic events.
Exhibit C-l displays the routine, called a pseudo-random number generator (or
random number generator), utilized in the PCLTF Simulation Model.
The random number generator shown in Exhibit C-l employs the
multiplicative congruential technique. The multiplier (equal to 65,541) and
the modulus1 (equal to 2,147,483,647), determine the statistical properties
of the number generated.
Random number generators of the type used here operate in cycles. The
starting seed for the routine is an input parameter which allows the model
user to control where the random number generator cycle begins. The cycle has
a finite period associated with the chosen multiplier and modulus. The period
for the values used here cannot exceed 1/4 of the modulus, or 536,870,912
random draws. After this number of draws, the values will begin to repeat in
the same order. Although the random number generator simulates a random
selection of values, the actual process is predictable (consequently, the term
"pseudo-random" is often used).
1 This modulus, 231-1, represents the largest number held by four
bytes, which is the FORTRAN default size for integer type variables.
-------
C-2
EXHIBIT C-l
THE PCLTF SIMULATION MODEL RANDOM NUMBER GENERATOR
SUBROUTINE RANDU (ISEED, RANDOM)
IRAND - ISEED * 65541
IF (IRAND) 5, 15, 15
5 IRAND = IRAND + 2147483647 + 1
15 RANDOM = IRAND
RANDOM = RANDOM * .4656613E-9
' ISEED = IRAND
RETURN
END
-------
C-3
The PCLTF Simulation Model employs the predictability of the routine to
its own advantage. Given a particular starting seed, the random number
generator will always produce an identical set of random numbers. This
property allows us to reduce total variance across model runs by choosing
identical starting seeds. Within runs the routine simulates random selection,
but across runs the model minimizes the variance due to randomness.
In order to qualify as a statistically valid random number generator, a
routine must satisfy at least two criteria. First, the routine should
generate the desired distribution across the selected interval; in this case a
uniform distribution between zero and one. Second, the routine should display
little or no correlation between successive random draws. Any statistically
discernible pattern between draws belies randomness.
Exhibits C-2 and C-3 contain the results of tests used to measure how well
the random number generator produces results that are uniformly distributed
between zero and one. The starting seed for the test was equal to 31,553 (the
starting seed used in PCLTF Simulation Model runs). The test routine first
draws ten random values using the random number generator, then it draws
another 10,000 random values and counts how many of these values fall within a
given range of the ten initial random values. The ranges shown in Exhibit C-2
are +0.01 and + 0.001. For example, out of the 10,000 random numbers
drawn, 197 were between 0.18634 and 0.20634 (the first row in Exhibit C-2).
Given a uniform distribution, we would expect to observe 200 (0.02 x 10,000)
values within 0.01 of the initial values.
The chi-square statistic was used to measure how closely the observed
values fit the results expected from a uniform distribution.2 The
chi-square values for the two sets of results in Exhibit C-2 are 3,66 and
3.35. With 9 degrees of freedom, the chi-square statistic at 99 percent
confidence is 21.7, and with 90 percent confidence is 14.7. Therefore, the
null hypothesis that the observed values were drawn from a uniform
distribution cannot be rejected.3
Exhibit C-3 displays the observed distribution for 50,150 random values
generated by the test routine. Here, we would expect approximately 2,508
(0.05 x 50,150) observations to fall within each of the 20 possible ranges of
values. The chi-square value for the distribution in Exhibit C-3 is 19.84.
Again, we cannot" reject the null hypothesis that the observed values were
drawn from a uniform distribution because the chi-square statistics with 19
degrees of freedom are larger than the observed chi-square values (chi-square
2 For a description of the chi-square statistics, see Wonnacott and
Wonnacott, Introductory Statistics for Business and Economics, (John Wiley &
Sons: New York, 1977).
3 In fact, the small chi-square values indicate that the observed
distribution deviates from a uniform distribution by an amount which is
consistent with the influences of random factors alone.
-------
C-4
EXHIBIT C-2
DISTRIBUTION OF RESULTS FOR THE RANDOM
NUMBER GENERATOR WITH RANGE = 0.01 AND 0.001*
Initial Random Value
0.19634
0.36418
0.73328
0.22834
0.95133
0.80463
0.26317
0.51596
0.58032
0.90421
Number Observed plus
or minus 0.01 of the
Initial Random Value
197
192
208
201
198
190
207
196
192
219
Initial Random Value
Number Observed plus
or minus 0.001 of the
Initial Random Value
0.83518
0.62162
0.33663
0.82586
0.84297
0.78309
0.75681
0.99074
0.98717
0.10324
19
19
24
25
18
19
21
23
17
20
1 Based on two sets of 10,000 random draws. The first set used.the
starting seed 31,553, and the second set followed the first set in sequence.
-------
C-5
EXHIBIT C-3
OBSERVED DISTRIBUTION OF VALUES FROM THE RANDOM
NUMBER GENERATOR ALONG THE INTERVAL FROM 0 TO I1
Range
0.0 to 0.05
0.05 to 0.10
0.10 to 0.15
0.15 to 0.20
0.20 to 0.25
0.25 to 0.30
0.30 to 0.35
0.35 to 0.40
0.40 to 0.45
0.45 to 0.50
0.50 to 0.55
0.55 to 0.60
0.60 to 0.65
0.65 to 0.70
0.70 to 0.75
0.75 to 0.80
0.80 to 0.85
0.85 to 0.90
0.90 to 0.95
0.95 to 1.00
# Observations
2,479
2,566
2,513
2,410
2,517
2,441
2,529
2,537
2,478
2,428
2,549
2,507
2,450
2,426
2,545
2,616
2,533
2,499
2,556
2,471
Based on 50,150 random draws with a starting seed of 31,553.
-------
C-6
statistics with 19 degrees of freedom are 36.2 at 99 percent confidence and
27.2 at 90 percent confidence).
The Mean Square Successive Difference Test (MSSD) is used to examine
correlation between successive random draws.* Based on 100 randomly drawn
values, a MSSD test statistic of 2.027 was observed. This value is compared
to the expected value of 2.0 (which would indicate no correlation between
successive random draws) using a Z statistic. The observed Z statistic of
0.27 is sufficiently small so that the null hypothesis of independence between
draws cannot be rejected.
The possibility of higher order correlation among draws from this random
number generator cannot be ruled out. The extent to which higher order
correlation in the random numbers would influence the model results is not
known. In general, however, based on the inability to reject the null
hypothesis discussed above, the method used here appears to be acceptable in
terms of both uniformity across the desired interval and independence across
successive observations.
C.I.3 Statistical Properties of the Results
The statistical properties of the results influence the manner in which
the Model is used. Specifically, the choice of the following run parameters
is affected:
Number of iterations: the number of repetitions used
to create the distributions of results;
Facility scaling factor: the portion of the total
facility population used as a basis for estimating Fund
performance; and
The starting seed used to initiate the random number
sequence generated by the random number generator.
Each is discussed below.
The number of iterations determines the sample size upon which model
results are computed. The more iterations, the closer a run's results
approximate the underlying population. More iterations, however, increase
execution time and costs. Consequently, one wishes to identify the fewest
number of iterations necessary to produce statistically useful results.
Exhibit C-4 displays a graph of how different numbers of iterations affect
four test statistics: the 10th and 90th percentiles, the median, and the mean
of the total Fund balance in year 75. Total Fund balance is a useful output
variable to use here because it is a primary indicator of Fund performance.
Year 75 was adopted for two reasons: (1) variance between iterations becomes
Wannacott and Wannacott, op. cit., pp. 488-490.
-------
C-7
EXHIBIT C-4
GRAPH OF ESTIMATES OF FOUR TEST STATISTICS AS A
FUNCTION OF THE NUMBER OF ITERATIONS PERFORMED
7000
6000
3000
TOTAL BALANCE
IN YEAR 75 4000
{Millions of Dollars)
3000 >
2000
1000
0 >
-1000
-2000
-3000
-4000 4-
-sooo
-6000
MEDIAN
STANDARD ERROfl
MEAN
STANDARD ERROR
TENTH PEHCENTIL8
NUMBER OF ITERATIONS
-------
C-8
more significant over time (and consequently would be more sensitive to the
choice of the number of iterations); and (2) year 75 represents a critical
time period for the Fund (when the mean and median balances begin to drop
rapidly in Simulation 1).
To assess the influence which the choice of the number of iterations
performed has on the Model results, we first test to see if there is a
statistical difference between a given estimate of the mean and the actual
population mean. Exhibit C-5 displays the mean and standard error of the mean
for the total Fund balance in year 75 for 9, 29, and 59 iterations.5 We
assume the mean and standard deviation for 59 iterations represent those of
the population.6 The estimate of the mean based on 9 iterations is not
statistically different from the 59-iteration estimate at the 99 percent
confidence level (t = 1.36). These results suggest that performing 9
iterations is valid at least for obtaining estimates of the mean. Likewise,
the median and ninetieth percentile estimates obtained from 9 iterations are
reasonably representative of the 59-iteration estimates. However, as seen in
Exhibits C-4 and C-5, the 9-iteration estimates of the tenth pecentile is much
less representative of the 59-iteration estimate.
The estimate of the mean based on 29 iterations is also not statistically
different from the 59-iteration estimate (t = .14). (In fact, the observed
difference is expected as the result of random factors with a probability
greater than 0.90.) Clearly, the performance of 29 iterations is sufficient
for estimating the mean Fund balance. Also, the tenth percentile, median, and
ninetieth percentile estimates based on 29 iterations are reasonable
representations of the 59-iteration values. Overall, it appears that 29
iterations are sufficient for examining the adequacy of the PCLTF with the
statistics presented here.
The facility scaling factor involves an analogous trade-off between costs
of performing Model runs and the representativeness of the results. Given a
facility scaling factor of n, the model processes every nth facility and
multiplies the total costs which arise by n. This procedure cuts execution
time and costs, but sacrifices statistical confidence in the results. To
evaluate the scaling factors, we examine whether the mean total Fund balance
5 The statistics describing the Fund balance were derived from model
runs done for the second draft report. Although some of the input parameters
and the modeling of firm bankruptcies has changed, there is no reason to
expect that the statistical properties of the model have changed.
Consequently, the values reported in this appendix for the Fund balance do not
correspond to the results reported in Volumes I and II. However, the
conclusions regarding the model's statistical properties described in this
appendix are considered to be valid.
6 The use of the 59-iteration results is reasonable because the rate of
change of the estimates in terms of the number of iterations is very small at
this level. In other words, the results are converging by the time 59
iterations have been performed.
-------
C-17
C.2.4 Data Bases
The following data bases are used in the Release Model:
WASTE. ID: correlates EPA waste code to chemical
constituents in the waste.
WASTE. PT: pointer locates data for a given waste.
* CODE. FILE: contains all pertinent properties for chemical
constituents.
LONGLATN.DAT: is the geographic data file holding the
various geohydrologic and soil information for a site.
CARD. IMG: contains failure probability and timing
information such as opening and closing dates.
PARTM.DAT: contains the weather data for the primary
weather stations partitioned by region.
LOCATION.DAT: contains site location data partitioned into
regions.
* WASTELIST.DAT: contains data on the number of wastes and
their codes for each site.
SI I E.DAT: is the output file of accumulated data for each
site in the Release Model.
-------
-------
C-9
EXHIBIT C-5
VALUES OF TEST STATISTICS FOR EVALUATING
THE NUMBER OF ITERATIONS
(Estimates are for the Total Fund Balance in
Year 75 in Millions of Dollars)
Number of Iterations
Test Statistic
Mean
Standard Error of Mean
Tenth Percentile Minus I1
Tenth Percentile
Tenth Percentile Plus lz
Median Minus I1
Median
Median Plus I2
Ninetieth Percentile Minus I1
Ninetieth Percentile
Ninetieth Percentile Plus I2
9
3,559
765
-33
387
2,846
3,787
5,091
5,697
6,179
29
343
1,473
-14,773
-5,122
-4,867
2,697
2,846
3,391
5,512
5,544
5,697
59
573
837
-8,770
-5,122
-4,867
2,691
2,846
2,973
5,465
5,383
5,328
1 "Minus l" refers to the value for the iteration ranked 1 below the
specified quantile.
2 "Plus l" refers to the value for the iteration ranked 1 above the
specified quantile.
3 Cannot be estimated with 9 iterations.
-------
C-10
for year 75 estimated with scaling factors greater than one is statistically
different from the population mean.
The mean total Fund balance in year 75 for a run of 29 iterations with a
scaling factor of 3 is 1,740, and the standard error of the mean is 1,055
(both in millions of dollars). This estimate of the mean is not statistically
different from the population mean at the 99 percent confidence level (t =
.46). This performance is somewhat better than the result obtained with 9
iterations and a scale factor of 1 (t = 1.36, see above) which has
approximately the same run cost. Likewise, the scale factor of 3 estimates of
the tenth percentile (-6,895 million),' and ninetieth percentile (5,750
million), are more representative than the 9-iteration estimates. The median
estimate (4,151 million) is not better. Performing 29 iterations with a scale
factor of 3 may be preferred to performing 9 iterations with a scale factor of
1.
The procedure for evaluating the influence of the starting seed on the
results is the same as described above. We postulate a null hypothesis that
there is no difference in the mean total Fund balance in year 75 for runs with
different starting seeds. The mean total balance in year 75 for a run with 29
iterations and starting seed 31,553 is 991 (in millions of dollars). The
corresponding result for a run with 29 iterations and starting seed 55,599 is
343. If the null hypothesis is true, then the expected difference between
these independent estimates is 0. The observed t statistic is 0.37, and
therefore we cannot reject the null hypothesis. We conclude that the starting
seed has no statistically discernable effect on the estimate of the mean. The
estimates of the tenth percentile (-5122 vs. -8770), median (2846 vs. 1964)
and ninetieth percentile (5544 vs. 5191) are also very similar. Consequently,
the model results are independent of the choice of starting seed, as is
desired.
C.I.4 Subroutines
The following routines read and write information from and to disk and
initialize certain variables:
$RLMAT: sets up the release matrices for use during each
iteration.
END$IT: writes the results of each iteration to binary
files and updates the sums and the sums of squares.
END$RN: outputs information at the end of a run.
IN$ITR: initializes variables for a current iteration.
IN$IYR: initializes variables for the current year.
IN$RUN: initialize scenario parameters for a run.
-------
Oil
IN$STA: organizes and completes the state data input.' If a
regime for a given state is known, it will be provided in the
input file and the model uses that specification. Otherwise, the
model generates the missing regime based on the appropriate input
probabilities.
RD$CTY: reads county data (population, housing and farm
values)
RD$EXC: reads user-supplied inputs.
RD$FAC: reads in the facility data and stores the data in -
arrays organized by economic market region.
RDSSCN: reads the scenario data file and stores the values
in appropriate common areas.
RD$FRM: reads firm data.
TABLE 1 through TABLE 13: write out statistics for creating
summary tables of model results.
WR$EXC, WR$SCN and WRSFRM: write the user-supplied input to
a file for verification following the run.
The following routines simulate the size of the facility population:
$KEEP: selects existing facility for the current iteration.
$WSTES: calculates the total waste and taxable waste in the
next calendar year based on the price of disposal, the price of
disposal substitutes, and the cross elasticities between disposal
and substitute prices.
ADD$FC: determines if any new facilities are needed and
adds enough new facilities to meet the demand for the current
model year.
GET$FC: describes which facilities to use from the existing
facility file, and then initializes these facilities for the
current iteration.
The following routines perform the facility-level characterization:
IN$FAC: assigns the following facility characteristics:
open year, state pointer, closing year, permit status, process
types, size and design by process type, status flags, and exposed
population.
$PRCHR: assigns a size and design type to each facility
process based on the appropriate distributions.
-------
C-12
IN$PHY: initializes the potential releases at a facility.
The routine obtains information from the Battelle release
matrices.
$PERMT: determines the permit status of an existing
facility.
N$FIN: initializes financial characteristics (i.e., net
worth, total assets, financial assurance mechanism) of a facility.
$CLOSE: identifies the release, detection, monitoring, and
response states of a facility at closure.
CH$SUB: determines if the capital and O&M costs for actions
are successfully allocated.
$SUB: identifies the substitute actions for a given action
or actions which cannot be undertaken due to lack of funding.
$FLGCH: controls the analyses of the changes occurring at
facilities over time.
$ACTVE: identifies any ongoing activities at a facility.
$QLPER: determines a facility's PCLTF qualification status
at the end of the qualification period.
DETECT*!: determines if a release as detectable this year.
KN$ACT: determines type of action to be implemented given
the detection of a release.
RS$END: determines if the response action ending affects
future releases.
PHY$PV: checks if ongoing response action prevents releases.
$PNTST: determines the expected future state of groundwater
contamination at a facility based on knowledge obtained from
monitoring actions.
$PRVKN: determines whether detection of a previous release
is needed to detect a new release.
$DOM: determines whether actions are dominated by another
act ion.
$RESET: determines any new activities that should be
ongoing given an array of activities that have ended.
-------
C-13
* $RSTRT: reinitializes the releases at a facility once all
response actions have been completed at the facility and the
facility is in a "cleaned-up" state.
$ADJST: adjusts the facility closing year and the variables
dependent on the closing year when bankruptcy is forced.
MN$CST: determines monitoring (capital and O&M) costs.
CR$CST: determines care costs by estimating the costs of
each post-closure care activity.
RS$CST: determines capital and O&M response costs. The
three possible response actions are: surface sealing, fluid
removal and treatment on-site, and fluid removal and treatment
off-site,
$CHCLM: determines what type of third-party claims are
brought given that a release is detected.
CM$CST: estimates claims (personal injury real property,
economic loss, and natural resources).
RP$CST: estimates real property losses due to off-site
releases.
* $PERIN: estimates personal injury claims.
The following routines perform cost allocation and balance the Fund:
* $ARULE: determines an allocation rule based on the current
qualification status of the facility and the number of years
after the facility closed.
$BLNCE: balances for the Fund in each year.
* $DIVDE: divides claims, monitoring, care or response costs
among five agencies (financial assurance, owner/operator, state,
federal, and gap).
*, $ELIM: deallocates current costs as a facility is being
eliminated. The routine gets called at closure if the facility
is allowed to decontaminate.
$POLCH: makes policy changes in response to the current
Fund balance or trends in Fund balance.
$REALO: reallocates current costs when PCLTF qualification
status changes, financial assurance runs out, owner/operator is
no longer solvent, or post-closure period ends.
$ROLL: rolls over costs generated in a given year.
-------
C-14
CH$BNK: evaluates viability of all firms at the beginning
of each model year.
F$BANK: evaluate effects of firm's bankruptcy on a
facility's operational status and ability to cover costs.
A variety of general purpose statistical and model control routines are
also used. (These routines are not listed.-)
C.I.5 Data Bases
The following data files are used in the PCLTF Simulation Model:
COUNTY: contains the county data on property values,
housing density, and population.
FACILITY DATABASE: contains the facility data file. For
each facility, this file includes: the EPA facility
identification number, the county code, the facility zip code,
the latitude and longitude, the groundwater region, economic
region, disposal process codes, drinking water survey data, and
the facility open year.
PC$LEG:' contains the data describing the legal regimes.
PC$FIRM: contains the data on firms.
PC$SCEN: contains the scenario data file. This file
includes user-specified inputs for cost estimates and facility
characteristics.
Run File: contains the model control inputs such as: the
number of iterations, the number of years, the facility scaling
factor, and the starting seed. ' In addition, this file includes
user-specified inputs for economic/financial data, PCLTF/RCRA
policies, sensitivity factors, and other run parameters that
define the policy simulation and sensitivity analysis runs.
RUNMAT01-RUNMAT24: are the release matrices. There are 24
release matrices; one matrix for each of 4 regions and 6 design
types.
C.2 RELEASE MODEL
C.2.1 Run Characteristics
The Release Model was run independently from the rest of the PCLTF model
on a Vax 11/780 (by Battelle Pacific Northwest Laboratories). It required
25,088 bytes of memory for the input files, 15,360,000 bytes for the data
bases, and 154,624 bytes for the predictive routines. The performance of
19,717 'iterations to test four site types in four representative regions
required 106,242 CPU seconds.
-------
C-15
C.2.2 Pseudo-Random Number Generator
The random number generator used in the Release Model was based on a first
order linear congruential pseudo-random number generator described in Pollard,
J.M., Handbook of Numerical and Statistical Techniques, Cambridge University
Press, New York, 1977, pp. 236-239. The basic formula is
Wn = K-(W ) MOD P. The constants Kj = 470001 and P = 999563 were
based on recommendations of J.G. Skellan in "Studies in Statistical Ecology,"
Biometrika, 39, 346-362. In essence, the next random number is set as the
remainder (MOD) after dividing K (W ) by P. Hence, the process requires
a "seed" number for W to initiate the process. When the same seed is
employed, the random number sequence can be duplicated.
A test run was made to evaluate the distribution of the numbers generated
by the algorithm. A series of 10,000 numbers was generated and analyzed
through determination of a standard chi-square (X2) value. The chi-square
statistic was found to be 7.84, indicating a level of 99 percent confidence
that the series was indeed uniformly distributed between zero and one.
C.2.3 Subroutines
The Release Model consists of a series of subroutines divided between two
major routines as described below.
Baseline Analysis. This major routine sets up the data file for each
facility, and consists of the following subroutines:
RRAN1: generates random numbers for the Monte Carlo
simulations.
* WASTE: performs the comparisons to determine which
constituents will be identified as "first arrivers" for a given
waste.
SORTWA: sorts the wastes in ascending order.
FILLUP: loads the information on the layers for the
selected prototypical site design.
LSITE: lists soil property data on a given facility and
sends it to the printer.
FINDCL: selects the location of the nearest primary weather
station and retrieves the desired climate data.
Water Balance Routine. This major routine constitutes the predictive
element of the model and creates subsequent release matrices. The following
subroutines are used:
-------
C-16
BREACH: inventories breaches for a site.
FORM: develops dispersion data required by the contaminant
transport model.
DI FUSE: calculates the concentration level for contaminants
at points of interest.
BACKUP: redistributes water levels back through the upper
layers.
LSITE: writes out the data for each layer in a site.
DECAY: determines the decay constant versus the accumulated
potential water loss as required by CURENT.
CURENT: updates the water inventory for the layer based on
the accumulated potential water loss.
START: sets an initial value for accumulated potential
water loss for a layer when the field capacity has not been
reached.
RAINFL: makes a draw for the precipitation pattern for a
site.
TEMPR: makes a draw for the temperature pattern for a site.
TBETA: regenerates the density distribution pattern for the
climate feature of interest using the output from GAMMA.
GAMMA: generates the gamma function for input to TBETA.
ANORML: creates a normal distribution for a set of data.
UNIFRM: makes a draw from a uniform distribution.
PET!: performs a Thornthwaite calculation to determine
evapotranspirat ion.
PET2: performs the Blaney-Criddle calculation to determine
evapotranspiration.
FSLOPE: finds the slope required for the PET1 calculation.
FINDHC: determines the gradient across a clay layer.
ERF: calculates the error function for the transport model
FORM.
-------
APPENDIX D
USER OPTIONS
This appendix presents the options which the user has in running the PCLTF
Simulation Model. The values presented here were used in Simulation 1.
-------
-------
THE FOLLOWING TABLE CONTAINS PREVIOUS KNOWLEDGE CONDITIONS,
WHERE: F = NOT NEEDED
T = REQUIRED
CHANGE IN
KNOWLEDGE STATE 1
LABEL: PREKNO(7,7)
EXISTING KNOWLEDGE STATE
23456
1
2
3
4
5
6
7
IP ON-SITE
DET ON-SITE
TOX ON-SITE
T/0 OFF-SITE
DET OFF-SITE
TOX OFF-SITE
BATHTUB
F
F
F
F
F
F
F
F
F
T
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
T
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
THE FOLLOWING TABLE INDICATES THE SUBSTITUTE ACTION FOR ACTIONS WHICH
CANNOT BE IMPLEMENTED DUE TO A LACK OF FUNDING
THE ACTIONS ARE:
1. IP-ON
2. CONST-ON
3. PD/PT-ON
«*. CONST-OFF
5. PD/PT-OFF
TABLE SUBMAT(10,10):
6. SURFACE SEALING
7. FLR/T-ON
8. FLR/T-OFF
9. CARE
10. EXPOSURE
ACTION CANNOT
IMPLEMENT
01
02
03
01
05
06
07
08
09
10
SUBSTITUTE ACTION
345678
10
0
1
0
0
0
0
0
0
0
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
THE FOLLOWING TABLE INDICATES THE NEW ACTIONS TO BE TAKEN AT THE
NORMAL COMPLETION OF AN ACTION
THE ACTIONS ARE:
1.
2.
3.
M.
IP-ON
CONST-ON
PO/PT-ON
CONST-OFF
5. PD/PT-OFF
6. SURFACE SEALING
7. FLR/T-ON
8. FLR/T-OFF
9. CARE
10. EXPOSURE
INPUT A 1 IF NEW ACTION SHOULD BE STARTED DUE TO OLD ACTION ENDING
ACTION ENDING
01
02
03
OU
TABLE RESET(10,10): 05
06
07
08
09
10
NEW ACTION
1567
10
0
1
1
1
1
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
b
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
-------
OTHER DATA
PERCENTAGE OF WASTE GENERATED
IN EACH REGION
NUMBER OF STATES
REGION
1231
DISTWR(I) 0.31 0.27 0.30 0.12
NSTATE 51
FIPS
02
01
05
01
06
08
09
11
10
12
13
66
15
19
16
17
18
20
21
22
25
21
23
26
27
29
28
30
37
38
31
33
31
: CHARACTERISTICS:
NAME
ALASKA
ALABAMA
ARKANSAS
ARIZONA
CALIFORNIA
COLORADO
CONNECTICUT
DISTRICT OF COLUMBIA
DELAWARE
FLORIDA
GEORGIA
GUAM
HAWAI 1
IOWA
IDAHO
ILLINOIS
1 NO 1 ANA
KANSAS
KENTUCKY
LOUS 1 ANA
MASSACHUSETTS
MARYLAND
MAINE
M 1 CM 1 CAN
MINNESOTA
MISSOURI
MISSISSIPPI
MONTANA
NORTH CAROLINA
NORTH DAKOTA
NEBRASKA
NEW HAMPSHIRE
NEW JERSEY
STATE(55
.6)
ABBR. REG.
d
AK
AL
AR
AZ
CA
CO
CT
DC
DE
FL
GA
GU
HI
IA
ID
IL
IN
KS
KY
LA
MA
MO
ME
Ml
MN
MO
MS
MT
NC
ND
NC
NH
NJ
-D
1
2
2
1
1
1
1
1
1
2
2
1
1
3
1
3
3
3
2
2
1
1
1
3
3
3
2
3
2
1
3
1
1
CLAIM MMC
LEGREG COV. COV.
(1-7) (1-8) (1-3>
-------
=** RUN EXPLANATION **==============
RUN NAME: >BASE CASE
DATE: >MM/DD/YY
RUN DESCRIPTION: >BASE CASE (SCALE FACTOR =1 ;NUMBER OF ITERATIONS =29)
OPERATOR:
======================#* MODEL CONTROL *#==========================
NUMBER OF ITERATIONS(1-50): 29<
STARTING SEED(MUST BE ODD): 31553<
NUMBER OF YEARS(1-100): 100<
FACILITY SCALING FACTOR(>=1): 1<
READ BINARY SCENARIO DATA(T/F): F<
WRITE SCENARIO DATA B(NARY(T/F): F<
READ BINARY FACILITY DATA(T/F): T<
WRITE FACILITY DATA BINARY(T/F): F<
READ BINARY COUNTY DATA(T/F): T<
WRITE COUNTY DATA BINARY(T/F): F<
READ BINARY INITIALIZED FACILITY DATA(T/F): F<
WRITE INITIALIZED FACILITY DATA BINARY(T/F): F<
STOP BEFORE YEAR 1(T/F): F<
-------
-------=,.-=-------==----#* ECONOMIC/FINANCIAL DATA **================
ENTER VALUES FOR EACH YEAR:
1983 1985 1986 1990 1995 2033 2083
ANNUAL INFLATION RATE(%)==> 3.1 5.3 5.2 1,6 U.O H.O «4.0
ANNUAL INTEREST RATE(%)> 8.0 10.0 9.2 5.7 5.1 5.1 5.1
1983 1985 1986 1993 2032 2033 2083
AVERAGE S/TON WASTE DlSPOSAL(1983$)==> 100. 100. 100. TOO. 100. 100. 100.
AVERAGE S/TON SUBSTITUTION
FOR DISPOSAL(1983$)==> 100. 100. 100. 100. 100. 100. 100.
ANNUAL RATE OF GROWTH IN OUTPUT(%)==> -2.0 -2.0 -2.0 -2.0 -2.0 0.0 0.0
ANNUAL RATE OF GROWTH OF AVERAGE
FACILITY DISPOSAL CAPACITY(%)==> 1.0 1.0 1.0 1.0 1.0 1.0 t.O
FIRST FISCAL YEAR OF MODEL RUN==> 198U<
STARTING TOTAL PCLTF BALANCE
(MILLIONS OF FIRST YEAR $)==> 0.0<
STARTING UNOBLIGATED PCLTF BALANCE
(MILLIONS OF FIRST YEAR $)==> 0.0<
TAX RATE($/DRY WEIGHT TON)==> 2.13<
FUND CEILING
(MILLIONS OF FIRST YEAR $)==> 200.<
FIRST YEAR WASTE DISPOSAL
(MILLIONS OF TONS)==> 64.<
FRACTION OF WASTE DISPOSAL WHICH
IS TAXABLE(%)==> 28.«4<
DEFINITION OF OBLIGATED FUNDS(1-3)==> 3<
1 = NO OBLIGATED FUNDS
2 = ONGOING MONITORING, MMC, AND
RESPONSE COSTS FOR A
SPECIFIED NUMBER OF YEARS
3 = AVERAGE OF FUNDS EXPENDED IN
PREVIOUS 10 YEARS
SPECIFIED NUMBER OF YEARS FOR
OBLIGATED FUNDS POLICY = 2 ==> 0<
MINIMUM ADMINISTRATIVE COSTS
CHARGED TO THE FUND EACH
YEAR (1000S OF 1983$)==> 0.0<
FRACTION OF FUND EXPENDITURES
USED TO CALCULATE ADMINISTRATIVE
COSTS (0-1.00>==> O.K
INDEX TAX RATE TO INFLATION(T/F) ==> F<
INDEX FUND CEILING TO INFLATION(T/F)==> F<
PRICE ELASTICITY OF DEMAND FOR
WASTE DISPOSAL(-1.5 TO -0.5)==> 0.0<
CROSS PRICE ELASTICITY OF DEMAND
FOR WASTE DISPOSAL(0.5 TO 1.5)==> 0.0<
-------
=** FINANCIAL DATA FOR BANKRUPTCY ANALYSIS **=
CUTOFF FOR NET WORTH CATEGORIES
(MILLIONS OF $)
BACKGROUND BANKRUPTCY FAILURE RATE ==>
AVERAGE RETURN ON ASSESTS ==>
DISCOUNT RATE FOR BANKRUPTCY ANALYSIS==>
INVESTMENT TAX CREDIT ==>
NUMBER OF YEARS FOR:
REVENUE AND EXPENDITURE PROJECTIONS ==>
DEPRECIATION OF CAPITAL EXPENDITURES >
10.0<
NET WORTH
$10 MILLION
OR LESS
0.00
-------
==========================«* PCLTF/RCRA POLICIES **====================
POST-CLOSURE PERIOD (YRS.) 30<
FINANCIAL ASSURANCE PERIOD (YRS.) 30<
PCLTF QUALIFICATION PERIOD (YRS.) 5<
POST-CLOSURE PERIOD RULES: THE POST-CLOSURE
PERIOD MAY BE EXTENDED UNTIL THE COMPLETION
OF AN ONGOING MONITORING OR RESPONSE ACTION.
IDENTIFY ACTIONS WHICH EXTEND THE POST-CLOSURE
PERIOD BY ENTERING A "T" OR "F".
ACTION TYPE:
1. IP MONITORING ON-SITE F<
2. CONST. MONITORING ON-SITE T<
3. PLUME DELINEATION ON-SITE T<
i». CONST. MONITORING OFF-SITE F<
5. PLUME DELINEATION OFF-SITE F<
6. SURFACE SEALING T<
7. FLUID REMOVAL ON-SITE T<
8. FLUID REMOVAL OFF-SITE F<
PCLTF QUALIFICATION RULES: ENTER 0, 1, 2, OR 3 FOR EACH RELEASE TYPE
0 = NO INFLUENCE ON QUALIFICATION
1 = DELAY IN QUALIFICATION IF RESPONSE TO RELEASE NOT COMPLETED
BEFORE END OF QUALIFICATION PERIOD
2 = DISQUALIFICATION IF "RESPONSE TO RELEASE NOT COMPLETED BEFORE
END OF QUALIFICATION PERIOD
3 = AUTOMATIC DISQUALIFICATION IF RELEASE OCCURS BEFORE END
OF QUALIFICATION PERIOD
RELEASE TYPE:
1. IP ON 3<
2. DET. CONST. ON 3<
3. TOX. CONST. ON 3<
14. T/0 OFF 3<
5. OET. CONST. OFF 3<
6. TOX. CONST. OFF 3<
7. BATHTUB 3<
IF DEFAULTING ON FINANCIAL ASSURANCE PRIOR TO THE END OF THE
QUALIFICATION PERIOD RESULTS IN DISQUALIFICATION,
ENTER 3, OTHERWISE ENTER 0==> 3<
-------
FINANCIAL ASSURANCE COVERAGE:
ACTION TYPE:
IDENTIFY FOR EACH ACTION OR CLAIM
WHETHER FINANCIAL ASSURANCE IS
REQUIRED BY THE 0/0 (ENTER T OR F)
1. IP MONI TOR ING ON
2. CONST. MONITORING ON
3. PLUME DELINEATION ON
U. CONST. MONITORING OFF
5. PLUME DELINEATION OFF
6. SURFACE SEALING
7. FLUID REMOVAL ON
8. FLUID REMOVAL OFF
9. POST-CLOSURE CARE
CLAIM TYPE:
1. PERSONAL INJURY
2. REAL PROPERTY
3. ECONOMIC LOSS
H. NATURAL RESOURCE DAMAGE
T<
T<
F<
F<
F<
F<
F<
F<
T<
F<
F<
F<
F<
FINANCIAL ASSURANCE TRUST FUNDS:
1.
2.
3.
l».
5.
6.
7.
8.
9.
10.
PROBABILITY
0.01K
0.007<
0.012<
0.020<
0.032<
0.053<
0.088<
0.1U5<
0.238<
0.39U<
TRUST FUNDS MAY LAST VARYING
AMOUNTS OF TIME. ENTER BELOW THE
DISTRIBUTION OF DURATIONS:
DURATION IN YRS
21.
22.
23,
2U.
25.
26,
27,
28
29.
30.<
-------
COST ALLOCATION:
COST ALLOCATION POLICIES ARE DEFINED IN TWO STAGES
1. DEFINE THE OPTIONS FOR THE ORDER OF ALLOCATION
2. DEFINE ALLOCATION RULES WHICH IDENTIFY THOSE
ALLOCATION OPTIONS TO BE USED FOR EACH COST
ORDER OF ALLOCATION OPTIONS: DEFINE THE ORDER OF ALLOCATION.
NUMBER OF OPTIONS
1 - 9 ) ====;
0/0 FINANCIAL
OPTION If ASSURANCE 0/0
1
2
3
U
5
6
7
8
9
0
0
0
1
0
0
0
0
1
1
1
0
2
1
0
0
1
2
STATE
0
0
1
3
2
0
0
0
0
9<
PCLTF SUPERFUND GAP
0
0
0
0
0
0
1
2
0
0
2
0
0
0
1
0
0
0
2
3
2
H
3
2
0
0
3
-------
ALLOCATION RULES: INSERT THE ALLOCATION OPTION (1-5) TO BE USED
FOR EACH COST TYPE
NUMBER OF RULES USED (1 - 29)==> 28<
MONITORING ACTION TYPE:
1. IP MONITORING ON
2. CONST. MONITORING ON
3. PLUME DELINEATION ON
«t. CONST. MONITORING OFF
5. PLUME DELINEATION OFF
RESPONSE ACTION TYPE:
1. SURFACE SEALING
2. FLUID REMOVAL ON
3. FLUID REMOVAL OFF
POST-CLOSURE CARE
CLAIM TYPE:
1. PERSONAL INJURY
2. REAL PROPERTY
3. ECONOMIC LOSS
H. NATURAL RESOURCE DAMAGE
FACILITY STATUS:
E$N = EXISTING (1) OR NEW (2)
P$N = PERMITTED (1), DENIED PERMIT {2) OR NOT YET DETERMINED (3)
QS = QUALIFICATION STATUS (1-U)
1 NOT YET DETERMINED
2 QUALIFIED
3 UNQUALIFIED
It DELAYED
NAME DEFINITIONS:
OPER = DURING OPERATION
Q P = QUALIFICATION STATUS PENDING
Q = QUALIFIED
U Q = UNQUALIFIED
QDEL = QUALIFICATION DETERMINATION DELAYED
-------
ALLOCATION RULE MATRIX ~=>
\ TIME
AFTER |E$N P$N|QS MONITORING ACT| RESP.
NAME* | CLOSURE M -2 1-211-UI 1 2 3 M 5 1 2 3
** EXISTING FACILITIES -- PERMIT TO BE DETERMINED
_______ _______ i _ _ _
OPER -99- -1
1(3 1 11111 111
** EXISTING FACILITIES -- PERMIT DENIED
OPER
Q P
Q P
Q
Q
UQ
UQ
QDEL
QDEL
** EXIS
OPER
Q P
Q P
Q
Q
UQ
UQ
QDEL
QDEL
** NEW
OPER
Q P
Q P
Q
Q
UQ
UQ
QDEL
QDEL
COLUMNS:
-99- -1 1
0-291 1
30-999! 1
0-29! 1
30-9991 1
0- 291 1
30-9991 1
0- 29! 1
30-9991 1
j
2
2
2
2
2
2
2
2
2
5TING FACILITIES
-99- -1| 1
0-291 1
30-999! 1
0-29! 1
30-999! 1
0-29! 1
30-999! 1
0-29! 1
30-9991 1
FACILITIES
-99- -1
0- 29
30-999
0- 29
30-999
O- 29
30-999
0- 29
30-999
2
2
2
2
2
2
2
2
2
1
1
1
1
1
1
1
1
1
- PI
1
1
1
1
1
2
2
3
3
4
4
--
1
1
1
2
2
3
3
I*
M
:RMI
1
i
i
2
2
3
3
k
U
11111
»4 4 2 2 2
33666
li M 2 2 2
77777
H H 2 2 2
33666
14 l» 2 2 2
33666
>ERMIT GRANTED
11111
*4 «4 2 2 2
33666
li H 2 2 2
77777
14 14 2 2 2
33666
H it 2 2 2
33666
GRANTED
11111
If H 2 2 2
33666
l» 14 2 2 2
77777
H U 2 2 2
33666
U 14 2 2 2
33666
1 1 1
222
666
222
222
666
222
666
1 1 1
222
666
222
777
222
666
222
666
1 1 1
222
666
222
777
222
666
222
666
CARE I CLAIMS
1 2 3 U
1 9992
1
«4
3
4
14
3
14
3
1
M
3
>4
7
U
3
14
3
____
1
<4
3
14
7
H
3
l|
3
9992
5552
5552
7777
5552
5552
5552
5552
9992
5552
5552
7777
7777
5552
5552
5552
5552
9992
5552
5552
7777
7777
5552
5552
5552
5552
.
-------
STATE SHARE OF SUPERFUND EXPENDITURES
BY TYPE OF-FACILITY OWNER
PRIVATE MUNICIPAL STATE
FEDERAL
OTHER
MONITORING
INDICATING PARAMETERS
CONSTITUENTS ON-SITE
PLUME DEL/TRK ON-SITE
CONSTITUENTS OFF-SITE
PLUME DEL/TRK OFF-SITE
RESPONSE
SURFACE SEALING
FLUID REM/TRT ON-SITE
FLUID REM/TRT OFF-SITE
POST-CLOSURE CARE
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.50
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
CLAIMS
PERSONAL INJURY
REAL PROPERTY
ECONOMIC LOSS
NATURAL RESOURCE
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
-------
KNOWLEDGE TO ACTION MATRIX:
THIS MATRIX DEFINES THOSE ACTIONS AND CLAIMS WHICH ARE UNDERTAKEN
IN RESPONSE TO THE DETECTION OF RELEASES. THE MATRIX HAS THREE
TYPES OF ENTRIES:
1. RELEASE CODE: EACH OF THE SEVEN RELEASE TYPES CAN BE DETECTED:
RELEASE TYPE:
1. IP ON
2. DET. CONST. ON
3. TOX. CONST. ON
4. T/0 OFF
5. DET. CONST. OFF
6. TOX. CONST. OFF
7. BATHTUB
2. SPECIAL CONDITION CODE:
CODE
1
-1
2
-2
3
THE ACTIONS TAKEN IN RESPONSE TO
DETECTED RELEASES MAY DEPEND ON
KNOWING THAT SPECIAL CONDITIONS
EXIST. THERE ARE 6 SPECIAL CONDITIONS
IN THE MODEL:
CONDITION
RELEASE HAS NOT YET GONE OFF SITE, BUT IT WILL.
RELEASE WILL NOT GO OFF SITE.
RELEASE HAS NOT YET BECOME TOXIC OFF SITE, BUT IT WILL.
RELEASE WILL NOT BECOME TOXIC OFF SITE.
RELEASE IS ALREADY OFF SITE.
RELEASE IS ALREADY TOXIC OFF SITE.
IF NO SPECIAL CONDITIONS ARE DESIRED, THEN A CONDITION CODE
OF 0 (ZERO) IS USED.
3. ACTION CODE:
FOR EACH RELEASE CODE/CONDITION CODE COMBINATION,
ACTION CODES DEFINE WHICH ACTIONS/CLAIMS ARE
AFFECTED. THE THREE ACTION CODES ARE:
ACTION CODE CONDITION
0 NO EFFECT
1 INITIATE ACTION (IF ACTION IS ALREADY ONGOING,
START DATE IS UPDATED)
-1 STOP ACTION (IF ACTION IS NOT ONGOING, THERE IS
NO EFFECT)
-------
THERE ARE EIGHT ACTIONS WHICH CAN HAVE ACTION CODES OF -1, 0, 1:
MONITORING ACTION TYPE: RESPONSE ACTION TYPE:
1. IP MONITORING ON
2. CONST. MONITORING ON
3. PLUME DELINEATION ON
I*. CONST. MONITORING OFF
5. PLUME DELINEATION OFF
1. SURFACE SEALING
2. FLUID REMOVAL ON
3. FLUID REMOVAL OFF
POST-CLOSURE CARE CAN BE GIVEN AN ACTION CODE OF -1 OR 0.
EXPOSURE CAN BE GIVEN AN ACTION CODE OF -1 OR 0.
EACH OF THE FOUR CLAIM TYPES CAN BE GIVEN ACTION COOES OF 0 OR 1:
CLAIM TYPE:
1. PERSONAL INJURY
2. REAL PROPERTY
3. ECONOMIC LOSS
4. NATURAL RESOURCE DAMAGE
UP TO 21* DIFFERENT KNOWLEDGE TO ACTION RULES CAN BE SPECIFIED.
INPUT THE NUMBER OF RULES USED ==> 16<
INPUT EACH OF THE RULES BELOW ==>
NAME
IP ON
DET ON
DET ON
DET ON
DET ON
TOX ON
TOX ON
TOX ON
TOX ON
T/0 OFF
DET OFF
DET OFF
DET OFF
DET OFF
TOX OFF
BATHTUB
Pf)l IIMKIQ
REL
- 1
_____
2
2
2
2
_____
3
3
3
3
4
5
5
5
5
6
7
COND
0
0
1
-1
3
0
1
-1
3
0
0
2
-2
H
0
0
MONITORING ACT
1 2 3 4 5
-11000
-1-1010
-1-1010
-1-1010
-1-1010
-1-1010
-1-1010
-1-1010
-1-1 01 0
00010
000-1 0
000-1 0
000-1 0
000-1 0
00000
00000
RESP.
1 2 3
000
1 10
1 1 0
1 1 0
1 1 0
000
000
000
000
000
1 1 1
1 1 1
1 1 1
1 1 1
000
1 0 0
CARE
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
EXP
0
0
0
0
0
0
0
0
0
0
0
0
0
0
-1
0
CLAIMS
1234
0 0 0 0|
0000
0000
0000
0000
0000
0000
0000
0000
___________
0000
___________
0110
0110
0110
0110
1111
0000
1 _ _ _ _
-------
SITE ASSESSMENT INFORMATION:
IN ORDER FOR AN EXISTING FACILITY TO
CLOSE OR RECEIVE A FINAL PERMIT IT
MAY BE REQUIRED TO PERFORM A SITE
ASSESSMENT. THE SITE ASSESSMENT IS
DEFINED IN TERMS OF THE EVIDENCE IT
SUPPLIES. IDENTIFY THE EVIDENCE
REQUIRED IN TERMS OF THE SEVEN RELEASE
TYPES. (ENTER "T" OR "r")
RELEASE TYPE:
1. IP ON
2. DET. CONST. ON
3. TOX. CONST. ON
14. T/0 OFF
5- DET. CONST. OFF
6. TOX. CONST. OFF
7. BATHTUB
AT CLOSURE
F<
F<
F<
F<
F<
F<
F<
FOR PERMIT
F<
F<
F<
F<
F<
F<
F<
PERMITTING POLICY:
PERMITS WILL BE DENIED TO EXISTING FACILITIES
BASED ON THE INFORMATION PROVIDED IN THE SITE
ASSESSMENT AND THE PERMITTING POLICY. UP TO 10
PERMITTING RULES MAY BE DEFINED BELOW IN TERMS
OF THE SEVEN RELEASE TYPES (ENTER "T" OR "F")
NUMBER OF PERMITTING RULES (t - 10):
1<
PERMITTING RULES:
RULE
1
RELEASE TYPES
IP DET TOX T/0 DET TOX B-TUB
ON ON ON OFF OFF OFF
F F f F F F , F
NOTE: ALL NEW FACILITIES WILL OPEN WITH PERMITS
-------
=** RUN PARAMETERS **=
ACTION DURATIONS:
FOR EACH ACTION IDENTIFY THE NUMBER OF YEARS FOR:
A. COMPLETION OF CAPITAL INVESTMENT;
B. COMPLETION OF ACTION.
ACTION TYPE:
1. IP MONI TOR ING ON
2. CONST. MONITORING ON
3. PLUME DELINEATION ON
I). CONST. MONITORING OFF
5. PLUME DELINEATION OFF
6. SURFACE SEALING
7. FLUID REMOVAL ON
8. FLUID REMOVAL OFF
9. POST-CLOSURE CARE
5<
5<
B*
999<
30<
30<
30<
30<
K
30<
50<
999<
* 999 ~ ACTION ONGOING IN PERPETUITY
MINIMUM DURATION OF REPONSE ACTIONS IF UNDERTAKEN FOR SECOND TIME:
1. SURFACE SEALING
2. FLUID REMOVAL ON-SITE
3. FLUID REMOVAL OFF-SITE
5<
EFFECT OF RESPONSE ACTIONS ON RELEASES: IDENTIFY FOR EACH RESPONSE
ACTION WHETHER IT INFLUENCES
EACH RELEASE TYPE
0 = NO INFLUENCE; 1 = INFLUENCE
RESPONSE ACTIONS:
SURFACE SEALING
FLUID REMOVAL ON
FLUID REMOVAL OFF
RELEASE TYPES
IP DET TOX T/0 DET TOX B-TUB
ON ON ON OFF OFF OFF
0
1
0
0
1
0
0
1
0
0
1
1
0
1
1
1
0
0
STRENGTH OF INFLUENCE OF
RESPONSE ACTIONS ON RELEASES==>
1 = STRONG
5 = WEAK
3<
-------
PROBABILITY OF DETECTION: PROBABILITY THAT ON-SITE MONITORING
WELLS WILL DETECT RELEASES WITH ROUTINE
MONITORING AND ANALYSIS WHEN THEY OCCUR:
PROBABILITY OF DETECT ION==> 0.75<
PERMITS: PERMITS WILL BE REQUESTED OF EXISTING FACILITIES AT THE
FOLLOWING RATE
1. WITHIN THE NEXT X YEARS, THE PERMIT STATUS WILL BE DETERMINED
AT A GIVEN FRACTION OF FACILITIES
ENTER X ==> 2<
2. BETWEEN YEAR X AND YEAR Y THE REMAINDER OF EXISTING FACILITIES
WILL HAVE THEIR PERMIT RESOLVED
ENTER Y ==> 5<
3. THE FRACTION RESOLVED BY YEAR X IS: .10<
«t. FRACTION OF EXISTING FACILITIES WHO
PASS TEST TO RECEIVE A PERMIT AND
WANT TO OBTAIN A PERMIT
ENTER FRACTION ==> .75<
PROPERTIES OF NEW FACILITIES: NEW FACILITIES DRAWN FROM FACILITY
POPULATION WITHIN THE TOP PERCENT
OF FACILITIES WITH THE FOLLOWING
CHARACTERISTICS:
PERCENTAGE
1. MINIMUM NUMBER OF PEOPLE
DRINKING WATER WITHIN
FACILITY'S ZIP CODE ==> 0.33<
2. MINIMUM COUNTY PROPERTY VALUE ==> 1.00<
3. MAXIMUM TIME TO RELEASE TYPE 2 ==> 0.33<
NOTE: THE RELEASE TYPE IS AN INPUT
-------
==========================** SENSITIVITY FACTORS **====================
COSTS:
MONITORING ACTION TYPE:
1. IP MONITORING ON 1.0<
2. CONST. MONITORING ON 1.0<
3. PLUME DELINEATION ON 1.0<
t. CONST. MONITORING OFF 1.0<
5. PLUME DELINEATION OFF 1.0<
RESPONSE ACTION TYPE:
1. SURFACE SEALING 1.0<
2. FLUID REMOVAL ON 1.0<
3. FLUID REMOVAL OFF 1.0<
POST-CLOSURE CARE 1.0<
CLAIM TYPE:
1. PERSONAL INJURY
MEDICAL COSTS 1.00<
LOST WAGES 1.00<
MEDICAL MONITORING 0.75<
MORTALITY 1.00<
2. REAL PROPERTY 1.00<
3. ECONOMIC LOSS 1.00<
«4. NATURAL RESOURCE DAMAGE 1 .00<
FACILITY POPULATION: (ONLY ONE MAY BE SET TO LESS THAN 1.0)
DECREASE PROBABILITY OF
BEING CHOSEN (0.0 TO 1.0)==> 1.0<
DECREASE PROBABILITY OF NOT
. BEING CHOSEN (0.0 TO 1.0)==> 1.0<
RELEASES:
TIMING OF RELEASES ==> 1.0<
YEARS FOR COUNTING UP RELEASES FOR TABLE 11:
YEARS SINCE OPENING 1 (>=1): 50<
YEARS SINCE OPENING 2 (>=1): 75<
==========================*» END OF DATA **=
-------
MODEL PARAMETER DATA
THIS FILE CONTAINS THE STANDARD MODEL PARAMETER INPUTS
(NOTE: ALL INPUT MUST BE RIGHT-JUST!FlED.>
COST ALLOCATION DATA
LEGAL LIABILITY REGIMES:
LEGAL REGIME
1 2 3 <4 56 7
PROBABILITY OF LEGAL REGIMES BEING LEGALS(7) 0.0*40 0.0*10 0.300 0.350 0.050 0.150 0.070
ADOPTED BY STATES -
LOWER UPPER
BOUND BOUND
CHANCE OF RECOVERY FOR LEGPER(12,2) 1 0.0 0.0
LEGAL REGIME CATEGORIES 2 0.0 10.0
3 10.0 20.0
t» 20.0 30.0
5 30.0 10.0
6 UO.O 50.0
7 50.0 60.0
8 60.0 70.0
9 70.0 80.0
10 80.0 90.0
11 90.0 100.0
12 100.0 100.0
-------
STATE COVERAGE REGIMES:
PROBABILITY OF CLAIM COVERAGE REGIMES STCOVS{8)
BEING ADOPTED BY STATES
CLAIM COVERAGE REGIME
1 2 3 «t 5678
0.705 0.235 O.OUO 0.000 0,000 0.020 0.000 0.000
STATE PORTION OF COVERAGE UNDER EACH
STATE COVERAGE REGIME FOR EACH CLAIM
TYPE
STAPOR(8,6)
P!
PROP
ECON
ENVI
RESP
O&M
STATE COVERAGE REGIME
3 H 5 6
1
2
3
i|
5
6
0.000
0.000
0.000
0.000
0.000
0.000
0.
0.
0.
0.
1.
1.
000
000
000
000
000
000
0.000
1.000
1.000
1.000
1 .000
1 .000
1.000
1.000
t.ooo
1.000
1 . 000
1.000
0.
1.
1.
1.
0.
0.
000
000
000
000
000
000
1.
1.
1.
1.
0.
0.
000
000
000
000
000
000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
PROBABILITY OF MMC COVERAGE REGIMES BEING
ADOPTED BY STATES
MMC COVERAGE REGIME
1 2 3
STMMCS(3) 0.780 0.110 0.110
PORTION OF MMC COVERAGE PAID FOR BY STATE
UNDER EACH OF THE MMC COVERAGE
REGIMES
MMC COVERAGE PORTION PER REGIME
1 2 3
STAPER(3) 0.000 1.000 1.000
FINANCIAL ASSURANCE:
FINANCIAL ASSURANCE GROUPS:
1 = FINANCIAL RATIOS
2 = TRUST FUND OR INSURANCE
3 = SURETY BOND OR LETTER OF CREDIT
SETTLEMENTS PERCENTAGE FOR FINANCIAL ASSURANCE STTLE1 0.00
GROUP 1
SETTLEMENTS PERCENTAGE FOR FINANCIAL ASSURANCE STTLE2 0.00
GROUP 2
SETTLEMENTS PERCENTAGE FOR FINANCIAL ASSURANCE STTLE3 0.80
GROUP 3
-------
FACILITY CHARACTERISTICS
FACILITY POPULATION DATA:
PARAMETERS OF LOG NORMAL DISTRIBUTION UPON
WHICH FACILITY LIFETIME IS BASED
PARAMETERS FOR EXISTING FACILITIES
PARAMETERS FOR NEW FACILITIES
CLOSE(3)
NCLOSE(3)
MEAN
3.149651
3.*49651
STANDARD
DEV
0.7876
0.7876
MINI MUM
EXPECTED VALUE = EXP(MEAN + 1/2 STD.DEV. ** 2) + MINIMUM
MEDIAN = EXP(MEAN) + MINIMUM
LOWER BOUND = MINIMUM
0.0
15.0
PROBABILITY OF EACH FACILITY
CONFIGURATION IN EACH REGION
REGION
DISFAC< 15, «4) 1
2
3
1
5
6
7
8
9
10
11
12
13
1*4
15
0.
0.
0.
0.
0.
0.
0.
0.
1
6732
1*476
0189
0661
0316
0095
0123
0016
0.0011
0.
0.
0007
0000
0.0020
0.
0.
0.
0036
0002
0016
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
.6732
.1*176
.0*489
.0661
.0316
.0095
.0123
.0016
.0011
.0007
.0000
.0020
.0036
.0002
.0016
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3
.6732
.1*476
.0*489
.0661
.0316
.0095
.0123
.0016
.0011
.0007
.0000
.0020
.0036
.0002
.0016
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
14
.6732
.1*476
.0*489
.0661
.0316
.0095
.0123
.0016
.0011
,0007
.0000
.0020
.0036
.0002
.0016
CURRENT EXCESS CAPACITY FOR
DISPOSAHIN %)
XCAP
50.0
-------
DRINKING WATER SOURCES AND USE
WATER SOURCES;
1 -
2 -
3 -
GROUND WATER ONLY
SURFACE WATER ONLY
BOTH
NO RESPONSE IN DRINKING WATER SURVEY
DISTANCE FROM FACILITY TO EXPOSURE POINT (POTABLE WELL OR SURFACE WATER):
DISTANCE CATEGORY LOWER BOUNDJFT>
1
2
3
H
. U90.
1000.
5280.
26HOO.
UPPER BOUND
-------
THE FOLLOWING TABLES PRESENT THE DISTRIBUTION OF WELL TYPES BY DISTANCE
TO THE FACILITY FOR EACH TYPE OF DRINKING WATER SOURCE
THE DISTRIBUTIONS WERE OBTAINED FROM THE SITE VISIT SURVEY
PTYWELJ 1 ,J, 1
PTYWEL{2,J, 1
PTYWEL(3,J, 1
PTYWEL<5,J,1
1 - GROUND WATER
DISTANCE TO FACILITY
1 2 3 14
0.000 O.U20 O.tl30 0.000
0.000 0.320 0.330 0.000
0.000 0.110 0.100 0.000
0.000 0.150 0.1«40 0.000
1.000 0.000 0.000 1.000
PTYWEL(1,J,2)
PTYWEL(2,J,2)
PTYWEL(3,J,2)
PTYWEL(M,J,2)
PTYWEL(5,J,2)
2 - SURFACE WATER
DISTANCE TO FACILITY
1 2 3 U
0.000 0.500 0.500 0.000
0.000 0.000 0.000 0.000
0.000 0.500 0.500 0.000
0.000 Q.OOO 0.000 0.000
1.000 0.000 0.000 1.000
PTYWEL(1,J,3)
PTYWEL(2,J,3)
PTYWEL(3,J,3)
PTYWEL(t»,J,3)
PTYWEL(5,J,3)
3 - BOTH
DISTANCE TO FACILITY
1 2 3 U
0.000 0.720 0.750 0.000
0.000 0.070 0.070 0.000
0.000 0.140 0.110 0.000
0.000 0.070 0.070 0.000
1.000 0.000 0.000 1.000
PTYWEL(1,J,U)
PTYWEL(2,J,«»)
PTYWEL(3,J,14)
PTYWEL(U,J,4)
PTYWEL(5,J,U)
I* - NO RESPONSE
DISTANCE TO FACILITY
1 2 3 I*
0.000 0.300 0.350 0.000
0.000 0.1HO 0.170 0.000
0.000 0.560 0.180 0.000
0.000 0.000 0.000 0.000
1.000 0.000 0.000 1.000
-------
THE FOLLOWING TABLES PRESENT THE DISTRIBUTION OF SURFACE WATER USE BY DISTANCE
TO THE FACILITY FOR EACH TYPE OF DRINKING WATER SOURCE
THE DISTRIBUTIONS WERE OBTAINED FROM THE SITE VISIT SURVEY
PSWUSE(1,
PSWUSE(2,
PSWUSE(3,
PSWUSE(«»,
PSWUSE(5,
PSWUSE(6,J,
1 - GROUND
DISTANCE TO
1 2
0.000 0.126 0
0.000 0.000 0
0.830 0.500 0
0.170 0.000 0
0.000 0.3714 0
0.000 0.000 0
WATER
FACILITY
3 «l
.168 1.000
.000 0.000
.661 0.000
.168 0.000
.000 0.000
.000 0.000
2 - SURFACE WATER
DISTANCE TO FACILITY
PSWUSE(1,J,2)
PSWUSE(2,J,2)
PSWUSE(3,JF2)
PSWUSEC*.J,2)
PSWUSE(5,J,2)
PSWUSE{6,J,2)
PSWUSE(1,J,3)
PSWUSE(2,J,3)
PSWUSE(3,J,3)
PSWUSE(1,J,3)
PSWUSE(5,J,3)
PSWUSE(6,J,3)
PSWUSE(1,J,
-------
EXISTING FACILITIES:
FACILITY OPEN YEAR:
'
DISTRIBUTION OF GENERAL MAIL SURVEY RESPONSE: POPENY(10)
OPEN YEAR
1850 TO 1899 0.003
1900 TO 1909 0.009
1910 TO 1919 O.O16
1920 TO 1929 0.019
1930 TO 1939 0.028
WO
1950
-1960
1970
1980
TO 1949
TO 1959
TO 1969
TO 1979
TO 1983
RELATIONSHIP BETWEEN PART A RESPONSES AND MAIL
GENERAL
SURVEY
1850
1900
1910
1920
1930
1940
1950
1960
1970
1980
TO
TO
TO
TO
TO
TO
TO
TO
TO
TO
SURVEY
MAIL
1899
1909
1919
1929
1939
1949
1959
1969
1979
1983
0.073
0.130
0.209
0.358
0.155
RESPONSES: PARTA(10,9)
PART A RESPONSE
1900 1910 1920 1930 1940 1950 1960
TO TO TO TO TO TO TO
1909 1919 1929 1939 1919 1959 1969
0.143
0.286
0.000
0.000
0.000
0.000
0.143
0.143
0.000
0.286
0.000
0.000
0.750
0.000
0.000
0.000
0.250
0.000
0.000
0.000
0.000
0.000
0.000
0.400
0.000
0.200
0.000
0.100
0.000
0.300
0.000
0.000
0.000
0.000
0.533
0.067
0.000
0.067
0.133
0.200
0.000
0.000
0.000
0.000
0.000
0.619
0.048
0.048
0.191
0.095
0.000
0.000
0.000
0.000
0.000
0.057
0.509
0.132
0.170
0.132
0.000
0.000
0.033
0.000
0.017
0.017
0.050
0.683
0.183
0.017
1970
TO
1979
0.000
0.000
0.000
0.020
0.000
0.010
0.010
0.092
0.776
0.092
1980
TO
1982
0.000
0.000
0.000
0.000
0.000
0.077
0.000
0.000
0.000
0.923
-------
FACILITY POPULATION:
PROBABILITY OF SURFACE IMPOUNDMENT
GIVEN PART A SURVEY RESPONSE IS YES
PROBABILITY OF SURFACE IMPOUNDMENT
GIVEN PART A SURVEY RESPONSE IS NO
PROBABILITY OF LANDFILL
GIVEN PART A SURVEY RESPONSE IS YES
PROBABILITY OF LANDFILL
GIVEN PART A SURVEY RESPONSE IS NO
PROBABILITY OF LAND TREATMENT
GIVEN PART A SURVEY RESPONSE IS YES
PROBABILITY OF LAND TREATMENT
GIVEN PART A SURVEY RESPONSE IS NO
PROBABILITY OF INJECTION WELL
GIVEN PART A SURVEY RESPONSE IS YES
PROBABILITY OF INJECTION WELL
GIVEN PART A SURVEY RESPONSE IS NO
PROBABILITY OF A FACILITY EXISTING AS
INDICATED BY THE MAIL SURVEY GIVEN IT
EXISTED AS INDICATED BY THE PHONE
VERIFICATION SURVEY
PROBSI( 1 )
PROBSI(2)
PROBLF( 1 )
PROBLF(2)
PROBLTJ 1 )
PROBLT(2)
PROBIW{1)
PROBIW(2)
0.9168
O.U792
0.7020
0.0604
0.6951
0.0*123
0.8696
0.0196
PROBML
0.41401
-------
PROCESS CHARACTERISTICS DATA:
EXISTING FACILITIES:
PROBABILITY OF FALLING INTO EACH OF
DESIGNS FOR SURFACE IMPOUNDMENTS
PROBABILITY OF FALLING INTO EACH OF
DESIGNS FOR LANDFILLS
PROBABILITY OF FALLING INTO EACH OF
DESIGNS FOR INJECTION WELLS
PROBABILITY OF FALLING INTO EACH OF
DESIGNS FOR LAND TREATMENT
DESIGN
1 2 3 U 5 6 7 89 10
DIMDES(IO) 0.392 0.3«42 0.206 O.OM 0.016 0.000 0.000 0.000 0.000 0.000
LFLDES(10) 0.000 0.000 0.038 0.010 0.003 O.l»7«l 0.475 0.000 0.000 0.000
INJOES(IO) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.000 0.000 0.000
LTMDES(10) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.000 0.000 0.000
NEW FACILITIES:
8 9 10
PROBABILITY OF FALLING INTO EACH OF
DESIGNS FOR SURFACE IMPOUNDMENTS
(NEW FACILITIES ONLY)
PROBABILITY OF FALLING INTO EACH OF
DESIGNS FOR LANDFILLS
(NEW FACILITIES ONLY)
PROBABILITY OF FALLING INTO EACH OF
DESIGNS FOR INJECTION WELLS
(NEW FACILITIES ONLY)
PROBABILITY OF FALLING INTO EACH OF
DESIGNS FOR LAND TREATMENT
(NEW FACILITIES ONLY)
DIMOEN(10) 0.00 0.00 0.3*4 O.U95 0.165 0.0 0.0 0.0 0.0 0.0
LFLDEN(10) 0.00 0.00 0.3tO O.U95 0.165 0.0 0.0 0.0 0.0 0.0
INJDEN(10)
0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0
LTMDEN(IO) 0.00 0.00 0.00 0.00 0.00 1.0 0.0 0.0 0.0 0.0
-------
SIZE (SURFACE AREA):
EXISTING FACILITIES:
PROBABILITY DISTRIBUTION OF
SIZE CATEGORIES FOR EXISTING
DISPOSAL SURFACE IMPOUNDMENT
FACILITIES
DIMSZ (10,3)
PROBABILITY DISTRIBUTION OF
SIZE CATEGORIES FOR EXISTING
LANDFILL FACILITIES
PROBABILITY DISTRIBUTION OF
SIZE CATEGORIES FOR EXISTING
LAND TREATMENT FACILITIES
PROBABILITY DISTRIBUTION OF
SIZE CATEGORIES FOR EXISTING
INJECTION WELL FACILITIES
1
2
3
J|
5
6
7
8
9
10
LDFSZ (10,3) 1
2
3
4
5
6
7
8
9
10
LTMSZ (10,3) 1
2
3
4
5
6
7
8
9
10
INJSZ (10,3) 1
2
3
It
5
6
7
8
9
10
PROB. OF
OCCURRENCE
0.143
0.1451 '
0.181
0.128
0.007
0.030
0.045
0.015
0.0
0.0
0.18
0.24
0.02
0.20
0. 11
0.18
0.07
0.0
0.0
0.0
0.10
0.30
0. 10
0.033
0.067
0.30
0.033
0.034
0.033
0.0
1.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
LOWER BND
OF SIZE
(ACRES)
0.01
0.10
1.0
3.0
7.0
9.0
10.0
26.0
0.0
0.0
0.01
0.50
4.10
6.8
111.7
27.5
142.0
0.0
0.0
0.0
0.10
1.0
7.0
10.0
13.0
15.0
60.0
100.0
188.0
0.0
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
UPPER BND
OF SIZE
(ACRES)
0.1
1.0
3.0
7.0
9.0
10.0
26.0
100.0
0.0
0.0
0.50
4.1
6.8
14.7
27.5
142.0
468.0
0.0
0.0
0.0
1.0
7.0
10.0
13.0
15.0
60.0
100.0
188.0
680.0
0.0
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-------
NEW FACILITIES:
PROBABILITY DISTRIBUTION OF
SIZE CATEGORIES FOR NEW
DISPOSAL SURFACE IMPOUNDMENT
FACI LI T I ES
DIMSZN(10,3)
1
2
3
l*
5
6
7
8
9
10
PROBABILITY DISTRIBUTION OF
SIZE CATEGORIES FOR NEW
LANDFILL FACILITIES
LDFSZN(10.3)
PROBABILITY DISTRIBUTION OF
SIZE CATEGORIES FOR NEW
LAND TREATMENT FACILITIES
PROBABILITY DISTRIBUTION OF
SIZE CATEGORIES FOR NEW
INJECTION WELL FACILITIES
1
2
3
l*
5
6
7
8
9
10
LTMSZN(10,3) 1
2
3
5
6
7
8
9
10
INJSZN(10,3) 1
2
3
H
5
6
7
8
9
10
PROB. OF
OCCURRENCE
0.0
.526
.211
,1<49
.008
.035
.053
.018
0.0
0.0
0.0
0.29
0.03
0.2H
0.13
0.22
0.09
0.0
0.0
0.0
0.00
0.333
0.111
0.037
0.074
0.333
0.037
0.038
0.037
0.0
1.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
LOWER BND
OF SIZE
(ACRES)
0.01
0.10
1.0
3.0
7.0
9.0
10.0
26.0
0.0
0.0
0.01
0.50
1».10
6.8
11.7
27.5
1*42.0
0.0
0.0
0.0
0.10
1.0
7.0
10.0
13.0
15.0
60.0
100.0
188.0
0.0
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
UPPER BND
OF SIZE
(ACRES)
0. 10
1.0
3.0
7.0
9.0
10.0
26.0
100.0
0.0
0.0
0.50
4.1
6.8
1*4.7
27.5
1U2.0
1468.0
0.0
0.0
0.0
1.0
7.0
10.0
13.0
15.0
60.0
100.0
188.0
680.0
0.0
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-------
DECONTAMINATION OF FACILITIES AT CLOSURE:
AT CLOSURE THE 0/0 MAY CHOOSE TO REMOVE ALL WASTES TO AN OFF-SITE
LOCATION OR TO A DISPOSAL LOCATION ON-SITE. THIS REMOVAL ELIMINATES
THE NEED FOR POST-CLOSURE CARE FOR THAT PROCESS. ONLY SMALL
PROCESSES WILL BE DECONTAMINATED DUE TO THE HIGH COSTS OF EXCAVATION
AND TRANSPORTATION.
SURFACE IMPOUNDMENTS:
ENTER X ==>
SURFACE IMPOUNDMENTS WITH AREA LESS
THAN X ACRES WILL DECONTAMINATE.
SIZMAX(I) 0,20
SURFACE IMPOUNDMENTS WITH AREA EXCEEDING X ACRES BUT
LESS THAN Y ACRES WILL DECONTAMINATE IFF A LANDFILL OF
SIZE EXCEEDING Z EXISTS ON-SITE.
ENTER Y ==> SIZMAX<2) 0.50
ENTER Z ==> LFZMIN 5.00
LANDFILLS: LANDFILLS WITH AREA LESS THAN X ACRES
- WILL DECONTAMINATE.
ENTER X ==> LFZMAX(1)
0.10
LANDFILLS WITH AREA EXCEEDING X ACRES BUT LESS THAN
Y ACRES WILL DECONTAMINATE IFF A SURFACE IMPOUNDMENT OF
SIZE EXCEEDING Z EXISTS ON-SITE.
ENTER Y ==> LFZMAX<2) 0.25
ENTER Z ==> SIZMIN 5.00
LAND TREATMENT: LAND TREATMENT WITH AREA LESS THAN X ACRES
WILL DECONTAMINATE.
ENTER X ==> LTZMAX(I) 0.00
LAND TREATMENT WITH AREA EXCEEDING X ACRES BUT LESS THAN
Y ACRES WILL DECONTAMINATE IFF A SUITABLE DISPOSAL
LOCATION IS AVAILABLE ON-SITE.
ENTER Y ==> LTZMAX(2) 0.00
-------
COST DATA
POST-CLOSURE CARE COSTS:
123456789
UPPER BOUND, MEDIAN AND LOWER CRCOEF(9,3) 1 60. 60. 2.4 355. 591. 1030. 210. 0.6 200.
BOUND FOR EACH OF THE 9 COEFFICIENTS 2 89. 118. 5.9 473. 1182. 1415. 830. 1.8 1017.
OF THE CARE COST EQUATION 3 118. 176. 11.8 591. 1773. 1800. 1450. 3.0 3074.
CRCOEF{9,1)=LOWER BOUND - -
CRCOEF(9,2)=MEDIAN
CRCOEF(9,3)=UPPER BOUND
ALL COEFFICIENTS ARE FOR COSTS IN FIRST QUARTER 1982 DOLLARS
THE COEFFICIENTS ARE AS FOLLOWS: .
1 MAINTAIN BENCHMARKS: $
2 MAINTAIN SOIL INTEGRITY (CAP): $/ACRE
3 -- SECURITY: S/ACRE
4 -- FENCE MAINTENANCE: S/ACRE 0.5
5 GAS MONITORING: $
6 LEACHATE MONITORING: $
7 LEACHATE COLLECTION AND REMOVAL: $/ACRE 0.477
8 LEACHATE TREATMENT: C/GAL
9 "IMPROVED" EROSION CARE: S/ACRE 0.310
PROBABILITY THAT SECURITY SECUR 1.000
IS REQUIRED IN THE POST-
CLOSURE PERIOD
PROBABILITY THAT FENCE FENCE 1.000
MAINTENANCE IS REQUIRED IN
THE POST-CLOSURE PERIOD
PROBABILITY THAT GAS MONITORING GAS 0.050
IS REQUIRED IN THE POST-
CLOSURE PERIOD.
PROBABILITY THAT LEACHATE . LEACH1 1.000
MONITORING IS REQUIRED AT
FACILITIES WITH LEACHATE
COLLECTION SYSTEMS THROUGHOUT
THE POST-CLOSURE PERIOD
PROBABILITY THAT LEACHATE LEACH2 1.000
COLLECTION AND REMOVAL IS
REQUIRED AT FACILITIES THAT
MONITOR LEACHATE COLLECTION SYSTEMS.
PROBABILITY THAT LEACHATE TREATMENT LEACH3 1.000
IS REQUIRED AT FACILITIES THAT
COLLECT AND REMOVE LEACHATE DESIGN
123456789 10
SPECIFICATION OF THOSE DESIGNS LEACHD(10) 1111101000
WITH LEACHATE COLLECTION -
SYSTEMS (1=YES, 0=NO)
-------
MONITORING COSTS:
RANGE OF WELL SAMPLING COSTS IN DOLLARS
FOR MONITORING INDICATING PARAMETERS
WELL LIFE IN YEARS
ISAMP(2) 1000. 2000.
WELL IF 30
RANGE OF CAPITAL COSTS FOR WELL CONSTRUCTION WELCAP(2) 25000. 50000.
CONSAM(2) 1750. 3000.
RANGE OF WELL SAMPLING COSTS IN DOLLARS
FOR MONITORING CONSTITUENTS
PLUME DELINEATION AND TRACKING COSTS:
THE FOLLOWING TABLE MAPS THE CAPITAL COST(K) OF PLUME DELINEATION
AND TRACKING BASED ON AFFECTED AREA AND DEPTH.
DEPTH
(FT)
25.
50.
75.
100.
200.
20000,
15.0
22.5
26.0
34.0
60.0
PD/T K COSTS (IN 1,000 OF 1982$)
PLUME SIZE(FT**2)
50000. 500000. 1250000. 2000000. 5000000. 8000000- 20000000.
15.0
22.5
26.0
34.0
60.0
32.5
44.0
55.0
67.5
120.0
37.5
50.0
65.0
82.5
138.0
40.0
56.0
73.0
91.0
153.0
46.0
65.0
85.0
104.0
187.5
59.0
80.0
104.0
126.0
225.0
80.0
108.0
138.0
170.0
300.0
-------
RESPONSE COSTS:
RANGE OF THE COEFFICIENT FOR
SURFACE SEALING COSTS ($/ACRE)
SEAL(3)
LOWER MEDIAN UPPER
6500. 20000. 40000.
THE FOLLOWING TABLE CONTAINS COST ESTIMATES (CAPITAL EXPENDITURES, K,
OPERATION AND MAINTENANCE, O&M) FOR FLUID REMOVAL AND TREATMENT,
AND PLUME DELINEATION.
PROBABILITIES OF ON-SITE AND OFF-SITE FLUID REMOVAL AND TREATMENT
PLUME SIZE {FT**2)
31495.
-K-
0.0000
0.1875
0.5000
0.8125
1.0000
-O&M-
0.0000
0.1875
0.5000
0.8125
1.0000
PLUME CHARACTERISTICS:
OFF-SITE PLUME DISPERSION COEFFICIENTS USED TO SIMULATE PLUME SIZE
(1000S OF 1982 $)
20198. 89102, 279992. 538838. 987773. 2428574.
PROBABILITY
0.0
.5
1.0
DISPERSION
COEFFICIENT
.05
.1
.2
5102554.
173.6
237.9
291.1
387.7
434. 1
22.7
30.5
39.2
53.3
59.7
202.7
262.5
326.2
418. 7
490.2
31.5
38.1
48.8
59.6
70.3
210.7
276.6
351.4
440.2
603.5
34.9
43.5
55.6
74.2
111.4
242.5
314.5
422.8
683.5
1293.8
36.0
47.8
75.4
118.8
241.9
262.3
448.0
686.4
1167.6
3108.4
48.0
73.0
121.9
226.3
464.1
314.8
438.9
675.2
2452.8
3990.4
53.1
76.0
121.9
338.5
649.3
417.0
697.3
1111.2
2452.8
4465.9
76.4
113.2
210.7
338.5
767.0
417.0
697.3
1197.9
6548.0
9697.7
76.4
118.0
243.3
1299. 1
2565.6
-------
CLAIMS COSTS:
PERSONAL INJURY:
THE FOLLOWING TABLES DESCRIBE CHARACTERISTICS OF THE U.S. POPULATION.
1980 POPULATION DISTRIBUTION FOR EACH AGE CATEGORY
AGE GROUP(YRS.) LABEL POPULATION
0-9 POPIYR(1,1) 0.1459
10 - 19 POPIYR(1,2) O.mo
20 - 29 POPIYR(1,3) 0.1803
30 - 39 POPIYR{1,1») . 0.1392
UO - «*9 POPfYR|1,5) 0.1005
50 - 59 POPtYR(1,6) 0.1030
60 - 69 POPIYR{1,7) 0.0833
70 - 79 POPIYR(1,8) 0.0512
80 - 89 POPIYR(1,9) 0.0197
90 + POPIYR(1,10) 0.0032
NUMBER OF BIRTHS PER 1000 PEOPLE OVER 10 YEAR PERIOD
AGE GROUP(YRS.) LABEL NUMBER OF BIRTHS
10 - 19 BIRTH(I) 147.0
20 - 29 BtRTH(2J 576.7
30 - 39 B(RTH(3) 212.3
l»0 - 149 BIRTH(H) 10.8
PROBABILITY OF DEATH WITHIN NEXT 10 YEARS BY AGE CATEGORY
AGE GROUP(YRS.) LABEL PROBABILITY OF DEATH
0-9 PDEATH(1\ 0.0198
10 - 19 PDEATH(2) 0.0070
20 - 29 PDEATH(3) 0.0129
30 - 39 PDEATH(M) 0.0159
1*0 - 49 PDEATH(5) 0.03?t»
50 - 59 PDEATHJ6) 0.0892
60 - 69 PDEATH(7} 0.1901*
70-79 PDEATH(8) 0.3827
80 - 89 PDEATH(9) 0.6850
90 + PDEATH(IO) 1.0000
-------
DISABILITY CHARACTERISTICS BY AGE CATEGORY
AGE GROUP(YRS. )
0-9
10 - 19
20 - 29
30 - 39
MO - l»9
50 - 59
60 - 69
70 - 79
80 - 89
90 +
AGE GROUP(YRS.)
0-9
10 - 19
20 - 29
30 - 39
HO - 49
50 - 59
60 - 69
70 - 79
80 - 89
90 +
VALUE OF
DISABILITY DAY
(1982 $)
0.00
13.36
5^.51
67.77
70.90
68.61
64.52
0.00
0.00
0.00
MEDICAL COSTS
LABEL
MEDICL(I)
MEDICL<2)
MEDICL(3)
MEDICL(il)
MEDICL(5)
MEDICL(6)
MEDICL(7)
MEDICL(8)
MEDICL(9)
MEDICL(10|
NO. OF DISABILITY
DAYS PER YEAR
0.00
0.55
1.22
0.95
1.20
1.76
2.17
0.00
0.00
0.00
MEDICAL COSTS PER YEAR
(S/PERSON/YEAR)-
1U5.16
81.96
1U2.H3
225.52
239.91
378.32
686.03
1139.20
11U0.13
767.11*
AGE GROUP(YRS.)
0-9
10 - 19
20 - 29
30 - 39
UO - t»9
50 - 59
60 - 69
70 - 79
80 - 89
90 +
MEDICAL MONITORING COSTS
LABEL MEDICAL MONITORING COSTS
($/PERSON/YEAR)
MEDMON(I) 160.
MEDMON(2) 160.
MEDMON|3) 160.
MEDMONCO 160.
MEDMON(5) 160.
MEOMON(6) 160.
MEDMON{7) 160.
MEOMON<8} 160.
MEDMON<9) 160.
MEDMON(IO) 160.
-------
REAL INCREASES OVER TIME IN PRICES AND WAGES (CUMULATIVE CHANGES)
BY YEAR
1
11
21
31
11
51
61
71
81
91
101
111
121
131
111
151
161
171
181
IN PRICES FOR
PERSONAL HEALTH
CARE SERVICES
RLMED(19)
1.0
1.219
1.161
1.666
1.859
2.071 ,
2.313
2.581
2.791
3.212
3.583
3.583
3.583
3,583
3.583
3.583
3.583
3.583
3.583
AGE GROUP(YRS.
0-9
10 - 19
20 - 29
30 - 39
10 - 19
50 - 59
60 - 69
70 - 79
80 - 89
90 +
IN INTENSITY IN WAGES
OF HEALTH FOR
CARE USE DISABILITY
MEDINT(19) RLDISA(19)
1.0 1.0
1.105 1.122
.220 1.302
.318 1.511
.189 1.751
.615 2.035
.817 2.362
2.007 2.711
2.217 3.181
2.11)9 3.692
2.705 1.285
2.7O5 1.285
2.705 1.285
2.705 1.285
2.705 1.285
2.705 1.285
2.705 1.285
2.705 1.285
2.705 1.285
MEDIAN ANNUAL WAGES BY AGE
ANNUAL
| WAGE(1982 $)
WAGE(10)
0.0
8687.0
11181.0
17621.0
18133.0
17838.0
16776.0
0.0
0.0
0.0
IN WAGES
FOR
MORTALITY
RLMORT(19)
1.0
1.122
1.302
1.511
1.751
2.035
2.362
2.711
3.181
3.692
1.285
1.285
1.285
1.285
1.285
1.285
1.285
1.285
1.285
REMAINING LIFE
(YEARS)
REMLIF{10)
70
61
51
12
33
21
17
11
3
0
-------
FRACTION OF DEATHS ATTRIBUTABLE TO DISEASES OF INTEREST BY AGE
AGE GROUP(YRS. )
0-9
10 - 19
20 - 29
30 - 39
10 - <49
50 - 59
60 - 69
70 - 79
80 - 89
90 +
FRACTION
DIFRAC(10)
0.7796
0.1867
0.1705
0.3089
OJ4228
O.*4805
0,<4685
0.4121
0.3569
0.276U
FRACTION OF ATTRIBUTABLE DEATHS THAT REMAIN ATTRIBUTABLE IN FUTURE
BY YEAR
1
11
21
31
U1
51
61
71
81
91
101
111
121
131
1l»1
151
161
171
181
FRACTION
OIFAOJ(19)
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
.0000
,0000
.0000
.0000
.0000
.0000
.0000
MINIMUM AGE FOR COMPUTING MORTALITY CLAIMS MINAGE
-------
REAL PROPERTY:
INITIAL HOUSING VALUES NEAR THE SITE
HOSVL1
0.50
LArntootu HO « riwoi ii/n vr PIC.L/IWI i u- i
INITIAL FARMLAND VALUES NEAR THE SITE
EXPRESSED AS A FRACTION OF MEDIAN
COUNTY VALUES ( 0- 1 )
FRACTION OF PROPERTY DEVALUED DUE TO
DETECTABLE RELEASE OFF SITE
FRACTION OF PROPERTY DEVALUED DUE TO
DETECTABLE AND TOXIC RELEASE OFF SITE
FREQUENCY OF REAL PROPERTY CLAIMS
DUE TO DETECTABLE RELEASE OFF SITE
FREQUENCY OF REAL PROPERTY CLAIMS
DUE TO DETECTABLE AND TOXIC RELEASE
OFF SITE
ECONOMIC LOSS:
1
FRMVL1
RPDEV(1)
RPOEV(2)
RPFREQ(1 )
RPFREQ(2)
THE FOLLOWING TABLE CONTAINS THE COSTS OF AN ALTERNATE
1.
2.
3.
5!
6.
7.
8.
9.
NATURAL RESOURCE LOSS:
DREDGING ($/SITE)
FISH LOSS (S/SITE)
RECREATION LOSS (S/YEAR)
POPULATION
SERVED
<=499.
<=999.
<=4999!
<=99999.
<=999999.
>=1000000.
DREDGE
FISH
RECR
0.50
0.15
0.30
1.00
1.00
WATER SUPPLY:
CAPITAL
COSTS
300.0
360.0
540.0
910.0
1585.0
2120.0
9135.0
45265.0
292115.0
588530.
1500.
16640.
WATSUP(9,2)
O&M
COSTS
9.3
11.2
17.9
34.1
52.8
87.1
269.0
1760.0
15900.0
(1000S OF 1983 S)
-------
KNOWLEDGE/ACTION DATA
THE FOLLOWING TABLE IDENTIFIES THE ABILITY OF EACH OF THE
7 RELEASE TYPES TO BE DETECTED BY THE 10 ACTIONS, WHERE
TRUE = CAPABLE OF DETECTION AND FALSE = NOT CAPABLE
THE ACTIONS ARE:
1.
2.
3.
U.
5.
IP-ON
CONST-ON
PD/PT-ON
CONST-OFF
PD/PT-OFF
6. SURFACE SEALING
7. FLR/T-ON
8. FLR/T-OFF
9. CARE
10. EXPOSURE
LABEL: MONABL(7,10J
RELEASE TYPE
IP-ON
DET-ON
TOX-ON
T/0 OFF
DET-OFF
TOX-OFF
BATHTUB
ACTIONS
1* 5 6 7
10
T
F
F
F
F
F
T
T
T
T
F
F
F
T
T
T
T
F
F
F
T
F
F
F
F
T
T
T
F
F
F
F
T
T
T
F
F
F
F
F
F
T
T
T
T
F
F
F
T
F
F
F
F
T
T
T
F
F
F
F
F
F
T
F
F
F
T
F
F
F
THE FOLLOWING TABLE INDICATES THE HIERARCHY OF ACTIONS, WHERE
0 = NEW ACTION HAS NO EFFECT ON EXISTING ACTION AND
EXISTING ACTION HAS NO EFFECT ON NEW ACTION
1 = NEW ACTION TURNS OFF EXISTING ACTION
-1 = .EXISTING ACTION PREVENTS NEW ACTION FROM STARTING
THE ACTIONS ARE: 1. IP-ON 6. SURFACE SEALING
2. CONST-ON 7. FLR/T-ON
3. PD/PT-ON 8. FLR/T-OFF
H. CONST-OFF 9. CARE
5. PO/PT-OFF 10. EXPOSURE
LABEL: HIERAC(10,10)
NEW ACTION
01
02
03
Oil
05
06
07
08
09
10
EXISTING ACTION
3 t 5 6 7 8
10
0
1
1
0
0
0
1
0
0
0
-1
0
1
0
0
0
1
0
0
0
-1
-1
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
1
0
0
0
0
0
-1
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
-1
-1
-1
0
0
0
0
0
0
0
0
0
0
-1
-1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
-------
35
32
36
39
140
41
>42
72
*|I4
'15
146
l»7
H8
1»9
51
78
50
53
55
54
56
NEW MEXICO
NEVADA
NEW YORK
OHIO
OKLAHOMA
OREGON
PENNSYLVANIA
PUERTO RICO
RHODE ISLAND
SOUTH CAROLINA
SOU7H DAKOTA
TENNESSEE
TEXAS
UTAH
VIRGINIA
VIRGIN ISLAND
VERMONT
WASH 1 NGTON
WISCONSIN
WEST VIRGINA
WYOM 1 NG
NM
NV
NY
OH
OK
OR
PA
PR
Rl
SC
SD
TN
TX
UT
VA
VI
VT
WA
Wl
WV
WY
H .
H
1
3
2
4
1
1
1
2
ti
2
2
«4
1
1
1
u
3
1
U
------- |