-oft
HAZARDOUS WASTE TANKS
RISK ANALYSIS
DRAFT REPORT
Prepared for:
The Office of Solid Waste
U.S. Environmental Protection Agency
Prepared by:
ICF Incorporated
Pope-Reid Associates, Inc.
March 1986
-------
PREFACE
This report presents the methodology used and results obtained from an
analysis of the human health risks associated with different regulatory
scenarios for tanks that treat, store, or accumulate hazardous waste. The
Economic Analysis Branch within the Waste Management and Economics Division of
the EPA Office of Solid Waste was responsible for this analysis. EPA retained
Pope-Reid Associates, Inc., for the development of a model that predicts the
timing of various events that lead to tank failure, and, given failure, the
quantity of waste that leaks out to the environment. EPA retained ICF
Incorporated to integrate the results of the tank failure model with
environmental transport, exposure and risk models, and to conduct the analysis
of the risk associated with the various regulatory scenarios using these
models.
The results developed to date are useful in helping determine the
appropriate regulatory strategy to pursue regarding the regulation of
hazardous waste tanks. However, the results are preliminary and subject to
change with refinements currently being considered to the various models
involved in the analysis.
-------
TABLE OF CONTENTS
Page
EXECUTIVE SUMMARY ES-1
1. OVERVIEW OF HAZARDOUS TANK RISK ANALYSIS 1-1
1.1 Introduction 1-1
1.2 Proposed Regulations and Regulatory Alternatives
for Hazardous Waste Tanks 1-1
1.3 Overview of Risk Assessment Methodology 1-8
1.4 Organization of the Report 1-23
2. HAZARDOUS WASTE TANK FAILURE MODEL 2-1
2.1 Overview of Hazardous Waste Tank Failure Model 2-1
2.2 Tank Technologies 2-3
2.3 Protection and Detection Options 2-13
2.4 Characterization of Surrounding Environment 2-20
2.5 Characterization of Waste Stream Physical Properties 2-20
2.6 Events that Lead to the Release of Contaminants 2-22
3. TRANSPORT, EXPOSURE AND RISK MODELS 3-1
3.1 Contaminant Transport in the Unsaturated Zone 3-1
3.2 Contaminant Transport in the Saturated zone 3-5
3.3 Exposure/Risk Model 3-14
4. DEVELOPMENT OF DATA FOR DERIVING DISTRIBUTIONS OF
RISK ESTIMATES 4-1
4.1 Distributions of Risk Estimates 4-1
4.2 Development of Model Tanks 4-16
4.3 Selection of Representative Waste Streams 4-29
4.4 Mapping Tank Population into Generic Hydrogeologic
Scenarios 4-50
4.5 Selecting Representative Release Profiles 4-61
5. RESULTS AND CONCLUSIONS 5-1
5.1 Introduction 5-1
5.2 Hazardous Waste Tank Failure Model Results 5-4
5.3 Risk Results 5-29
5.4 Comparison of Risk Estimates for Hydrogeologic
Settings and Waste Streams 5-46
APPENDIX - FAILURE FREQUENCY DATA AND REPRESENTATIVE RELEASE
PROFILES
-------
LIST OF EXHIBITS
No.
1-1
1-2
1-3
1-4
2-1
2-2
2-3
2-4
2-5
2-6
2-7
2-8
2-9
2-10
2-11
2-12
2-13
2-14
3-1
Title
Proposed Regulations for Hazardous Waste Tanks
Hypothetical Relative Frequency Distribution of
Risk Estimates for a Tank Technology and Regulatory Scenario
Flowchart of the Tank Risk Analysis Model
Methodology to Estimate Frequency Distribution of Risk
Estimates for Each Tank Technology and Regulatory Scenario . .
Potential Release Profile
Characterization of Existing Hazardous Waste Tanks
Operating Parameters
Oxidation/Reduction/Precipitation Treatment Tank
Distillation Treatment Tank
Storage Tank
Inventory Control Cycles
Secondary Containment Assumptions
Soil Parameters for Sand
Generic Waste Stream Characteristics
X
Tank System Failure Matrix: Internal Environmental
Stresses
Tank System Failure Matrix: External Environmental
Stresses
Tank System Failure Matrix: Design, Construction, and
Operation Errors
Catastrophic Events Modeled for Different Tank Types
Physical/Chemical Parameters for Generic Unsaturated
Zone
Page
1-4
1-11
1-13
1-14
2-2
2-5
2-7
2-8
2-10
2-11
2-15
2-18
2-21
2-23
2-25
2-26
2-27
2-29
3-3
-------
- 11 -
LIST OF EXHIBITS (continued)
No. Title Page
3-2 Simulation of Contaminant Transport in the Saturated
Zone by the Prickett-Lonnquist Random Walk Model 3-6
3-3 Standard Breakthrough Curve for Transport Model 3-10
3-4 Graphical Representation of the Methods for Summing
Individual Breakthrough Curves to Obtain the
Concentration Profile at an Exposure Well 3-11
3-5 Hydrogeological Parameters for the Nine Generic Flow
Fields in TRAM 3-12
3-6 Constituent Data Base 3-20
4-1 Percentage of Tank Category Located in Each Ground-
Water Scenario 4-6
4-2 Waste Stream Weights for In-ground, Carbon Steel
Storage Tanks 4-7
4-3 Calculating Representative Release Profile Weights 4-8
4-4 Tank Technology Weights 4-9
4-5 Sample Calculation of Dominant Risk Estimate Weight 4-11
4-6 Hypothetical Dominant Risk Estimates and Their
Weights for Above-ground, Carbon Steel SQG Tanks 4-13
4-7 Hypothetical Frequency Distribution of Risk Estimates
for Above-ground, Carbon Steel SQG Tanks 4-14
4-8 Mean and Median of Risk Distribution 4-15
4-9 Distribution of Storage Tanks by Type and Material
of Construction 4-20
4-10 Distribution of Treatment Tanks by Type and Materials
of Construction 4-22
4-11 SQG Tank Size Distributions 4-26
-------
- Ill -
LIST OF EXHIBITS (continued)
No.
4-12
4-13
4-14
4-15
4-16
4-17
4-18
4-19
4-20
4-21
4-22
4-23
4-24
5-1
5-2
5-3
5-4
Title
Model Tank Characteristics Summary
SQG Wastes Stored in Tanks Ranked According to Total
Waste Quantity Generated and Total Number of Tanks
SQG Wastes Streams Stored in Above-ground and
Underground Tanks
Representative SQG Waste Streams
Representative Waste Streams for RCRA-Permitted
Storage Tanks
Representative Accumulation Tank Waste Streams
Representative Waste Streams for Selected Treatment
Technologies
USGS Regions
Percent of State Within Each USGS Region
Distribution of Ground-water Scenarios Within
USGS Regions
Percent of State Represented by Each Ground-water
Scenario ..
Distribution of Tank Categories Within Each State
Hydrogeologic Setting Weights
Applicable Tank Technologies for Each Regulatory
Scenario
Sample Failure Frequency Data
Sample Representative Release Profiles
Most Representative Release Profiles for Treatment
Tanks ; Baseline Scenario
Page
4-28
4-33
4-34
4-35
4-41
4-45
4-48
4-52
4-54
4-55
4-57
4-58
4-60
5-3
5-5
5-8
5-12
-------
- iv -
LIST OF EXHIBITS (continued)
No. Title Page
5-5 Total Failure Frequency for Treatment Tanks;
Baseline Scenario 5-14
5-6 Most Representative Release Profiles for Storage/
Accumulation Tanks; Baseline Scenario 5-18
5-7 Total Failure Frequencies for Storage/Accumulation
Tanks; Baseline Scenario (Toluene) 5-19
5-8 Most Representative Release Profiles for SQG Tanks;
Baseline Scenario 5-21
5-9 Total Failure Frequencies for SQG Tanks; Baseline
Scenario 5-23
5-10 Most Representative Release Profiles for All Tank
Technologies; Baseline Scenario (Toluene) 5-24
5-11 Comparison of Tank Materials; Most Representative
Release Profiles for Selected Tanks; Baseline
Scenario (Toluene) 5-25
5-12 Comparison of Waste Types; Most Representative
Release Profiles for Selected Tanks; Baseline
Scenario 5-27
5-13 Hypothetical Frequency Distribution of Risk Estimates,
Underground Carbon Steel Storage Tanks (4000 gal.).
All Flow Fields, All Waste Streams 5-31
5-14 Comparison of Risks for Baseline and Secondary
Containment Scenarios; Treatment Tanks 5-34
5-15 Comparison of Risks for Baseline and Secondary
Containment Scenarios; Storage Tanks 5-35
5-16 Comparison of Risks for Baseline and Secondary
Containment Scenarios; Accumulation Tanks 5-36
5-17 Comparison of Risks for Baseline and Secondary
Containment Scenarios; Small Quantity Generator Tanks 5-37
-------
- v -
LIST OF EXHIBITS (continued)
No. Title
5-18 Comparison of Risks for Baseline, Secondary
Containment and Leak Testing with Ground-water
Monitoring Scenarios; Underground Tanks 5-38
5-19 Comparison of Risks for Baseline, Secondary
Containment and Leak Testing With Ground-water
Monitoring Scenario; Treatment Tanks 5-41
5-20 Comparison of Risks for Baseline Secondary
Containment and Leak Testing With Ground-water
Monitoring Scenario; Storage/Accumulation Tanks 5-42
5-21 Comparison of Risks for Baseline and Corrosion
Protection Scenarios; Carbon Steel and Stainless
Steel Tanks 5-44
5-22 Comparison of Risks for Hydrogeologic Settings;
Frequency Distribution of Risk Estimates for
Underground, Carbon Steel Storage Tank (4000 gal.)
All Flow Fields, Waste Stream F003 5-47
5-23 Comparison of Risks for Hydrogeologic Settings;
Frequency Distribution of Risk Estimates for
Underground, Carbon Steel Storage Tank (4000 gal.)
All Flow Fields, Waste Stream F001 5-48
5-24 Comparison of Risks for Waste Streams; Above-ground
On Cradle, Carbon Steel Distillation Tank (2300 gal.),
Flow Field C, All Waste Streams 5-51
5-25 Comparison of Risks for Waste Streams; Above-ground,
On Grade, Carbon Steel ORP Tank (60,000 gal.) Flow
Field C, All Waste Streams 5-52
5-26 Comparison of Risks for Waste Streams; Above-ground,
On Grade, Carbon Steel Storage Tank (210,000 gal.),
Flow Field C, All Waste Streams 5-54
5-27 Comparison of Risks for Waste Streams; Underground,
Carbon Steel SQG Tank (200 gal.), Flow Fields C,
All Waste Streams 5-55
-------
- vi -
LIST OF EXHIBITS (continued)
No. Title
A-l Failure Frequencies for Treatment Tanks . .'. A-2
A-2 Failure Frequencies for Storage/Accumulation Tanks A-9
A-3 Failure Frequencies for SQG Tanks A-24
A-4 Representative Release Profiles for Treatment Tanks A-28
A-5 Representative Release Profiles for Storage/Accumulation
Tanks A-35
A-6 Representative Release Profiles for SQG Tanks A-50
-------
EXECUTIVE SUMMARY
On June 26, 1985, EPA proposed revisions to the Resource Conservation and
Recovery Act (RCRA) regulations for managing hazardous wastes that are stored
or treated in tanks. These revisions address a number of issues for hazardous
waste tanks brought forth in the Agency's January 12, 1981 preamble to the
current regulations, including: (1) the development of permitting standards
under Part 264 for underground tanks that cannot be entered for inspection;
and (2) the requirement under the 1984 RCRA Amendments that new underground
tanks be equipped with leak detection systems. In addition, EPA is proposing
revisions that fulfill aspects of EPA's regulatory strategy for tanks that
were left unaddressed in the January 12, 1981 preamble and for which public
comment was requested when the existing regulations for tanks were promulgated.
Executive Order 12291 and the Regulatory Flexibility Act (P.L. 96-354)
impose requirements that need to be addressed in promulgating revised
regulations for tanks used for hazardous waste storage, treatment, or
accumulation. To fulfill part of these requirements, EPA has developed a
hazardous waste tank risk analysis model to quantitatively assess the risks
associated with releases of contaminants from hazardous waste tanks. This
report presents the first results of EPA's hazardous waste tank risk analysis,
and also discusses the methodologies and assumptions used in the analysis.
PURPOSE
The purpose of this analysis was to assess the human health risks
associated with: (1) the population of hazardous waste tanks under the
current regulations; and (2) the population of hazardous waste tanks under the
various alternative regulatory strategies being considered by EPA. The
population of tanks considered in this analysis included four tank categories:
RCRA-permitted storage tanks; treatment tanks (that are not exempt under EPA's
wastewater treatment exemption); small quantity generator (SQG) tanks; and
accumulation tanks (tanks that store waste for less than 90 days and are
therefore not required to obtain a RCRA permit).
The analysis is focused on five regulatory scenarios, designed to
represent the proposed regulatory alternatives for hazardous waste tanks.1
These scenarios are as follows:
no revised regulatory requirements (baseline);
secondary containment;
1In addition, we are considering alternatives to the proposed
regulations. We plan to evaluate vadose monitoring for appropriate volatile
waste streams, corrosion protection with leak testing, and corrosion
protection with vadose zone monitoring.
-------
ES-2
partial secondary containment and ground-water
monitoring;
corrosion protection for portions of steel tanks in
contact with the soil; and
leak testing and ground-water monitoring.
Analysis of these five regulatory scenarios allows for direct comparisons
between risks presented by hazardous waste tanks as they are currently being
managed and under technical requirements set forth in the proposed
regulations. By analyzing each scenario separately, we are able to evaluate
technical requirements that are being considered as part of alternative
regulatory strategies (e.g., corrosion protection), while also considering
requirements that are central to the proposed regulations (e.g., secondary
containment). Also, further examination of tank regulatory alternatives could
include the evaluation of combinations of technical requirements (e.g.,
partial containment and corrosion protection) and, by using the scenarios
listed above, we have the flexibility to analyze the reduction in risks for
these alternatives.
METHODOLOGY
The main objective of our analysis was to compare the risks associated
with the various proposed regulatory scenarios with the risks associated with
the existing situation. In order to analyze existing tanks, we first needed
to compile information on the existing population of hazardous waste tanks
(i.e., determine what type of tanks exist today, their age, and what type of
hazardous wastes are stored or treated in these tanks). We then needed to
project the type of releases that are occurring or will occur from this
existing population of tanks. These projections on the behavior of tanks
provide an indication of what the future health risks associated with current
operating practices and tank technologies will be.
The regulatory scenarios are based on changes to current operating
practices and technologies. Consequently, we had to project the type of
releases that would occur from tanks regulated under different tank operating
and design specifications. One major part of our methodology, therefore,
focuses on estimating releases from existing tanks regulated under the current
and alternative regulatory scenarios. Another major part of our methodology
focuses on using the projection of releases to estimate human health risk. We
then use these risk estimates to compare the regulatory scenarios.
Hazardous waste treatment and storage tanks may be constructed of a
variety of materials (e.g., carbon steel, stainless steel, concrete, and
fiberglass-reinforced plastic) and may be located completely above-ground (on
cradles or ongrade), partially underground (in-ground), or completely
underground. Also, they may be of various sizes and ages and store a variety
of wastes. Additionally, they may be located over a wide variety of
hydrogeologic conditions, and may vary considerably in their proximity to
human populations.
-------
ES-3
All of these variables may have substantial impact on risk, and thus
needed to be considered in our analysis. Moreover, tank design and operating
practices will have an effect on the risks associated with different hazardous
waste tanks. Thus, we also needed to take these additional factors into
consideration. Therefore, the methodology developed for this analysis differed
from traditional environmental risk assessment methodologies in that it was
necessary to account not only for factors such as waste stream composition and
hydrogeologic conditions, but also for factors that influence the timing and
magnitude of failure events associated with hazardous waste tanks. Due to the
variability in tank designs, operating conditions, and hydrogeologic and waste
stream parameters, it was determined that a simplified approach based on one
or several case histories of failed tanks was not adequate to assess the risk
reductions for the different regulatory scenarios on a nationwide basis.
Rather, we believed it was important to analyze the types of failures that
would most commonly occur for hazardous waste tanks, and to evaluate the
effectiveness of the proposed regulations and alternative regulatory
strategies in (1) preventing these failures; or (2) detecting these failures
before significant damage to public health or the environment has occurred.
Based on this premise, we developed a risk assessment methodology for
hazardous waste tanks. The approach uses broad-based mathematical models to
evaluate the range of environmental conditions, operating standards, and
design scenarios considered necessary for the analysis, and is supported by
data bases containing information on model tank technologies, waste streams,
hazardous chemicals, and hydrogeologic environments. The models we used for
this study are:
The Hazardous Waste Tank Failure (HWTF) Model, a Monte
Carlo simulation model that predicts the timing of failure
events for hazardous waste tanks and estimates release
volumes associated with these failure events;
The Tank Risk Analysis Model (TRAM), an exposure/risk
model that estimates human exposure to hazardous chemicals
via contaminated drinking water and calculates
non-carcinogenic and carcinogenic risk to the exposed
individual. Contaminant transport and degradation in the
unsaturated and saturated zones are estimated using
modified versions of the McWhorter-Nelson Wetting Front and
Prickett-Lonnquist Random Walk models, respectively.
These models reference data bases that we developed to characterize the
existing tank population. We developed the following four data bases:
Model Tank Technology Data Base: contains information on
22 model tank technologies that represent the types of
existing RCRA-permitted storage tanks, treatment tanks,
small quantity generator tanks, and accumulation tanks.
-------
ES-4
Waste Stream Data Base: contains information on 32 waste
streams that represent the range of waste streams handled
in the model tank technologies.
Chemical Data Base: contains information on 120 hazardous
chemicals, of which 36 are representative of chemicals
found in hazardous waste tanks.
Hydrogeologic Data Base: contains information on nine
hydrogeologic settings that represent the range of settings
encountered at tank facilities.
The methodology was developed specifically to allow for comparisons to be
made between regulatory scenarios. It was not designed necessarily to allow
various comparisons to be made between different types of tanks. Such
comparisons could be made if the model tank technologies and waste streams
were revised specifically for this purpose. This report focuses primarily on
comparisons between regulatory scenarios and the reader is cautioned against
attempting to draw final conclusions based on any other comparisons that one
might be tempted to make with data presented in the report.
RESULTS
Before presenting the results, it is important to note that the results
are preliminary, and thus may be subject to significant revisions following
further review and additional modeling work (e.g., inclusion of an algorithm
to account for the development of cracks over time in concrete tanks). The
Tank Risk Analysis Model, and particularly the Hazardous Waste Tank Failure
Model, are large and relatively complex models that rely on major assumptions
resulting from the lack of data, and it is clear that certain model results
are driven by these assumptions. Some of the results seem anomalous and we
are reviewing the assumptions on which these results are based to determine if
revised assumptions should be made. Therefore, we believe that while the
results presented here are useful for comparing regulatory scenarios, they are
less useful for comparing different tank technologies within a scenario. We
further believe our review of the results of this very complex model can be
aided greatly by comments we expect to receive on these preliminary results.
The analysis found that risks were reduced significantly for all tanks,
with a typical reduction of about 50 percent in non-zero risk estimates,
between baseline and secondary containment scenarios. More importantly, high
_3
risk situations (relative risks greater than 10 ) were reduced by 80
percent or more for most tank technologies between baseline and secondary
containment scenarios. Thus, from our analysis, we conclude that secondary
containment systems, equipped with leak detection systems, are a very
effective method of reducing the risks associated with hazardous waste tanks.
Secondary containment systems prevent most releases from escaping to the
environment, in most cases allowing hazardous wastes to be released only as a
result of very low-probability, catastrophic events.
-------
ES-5
Based on the results of the risk analysis, we have ranked the regulatory
scenarios in descending order of risk reduction from the baseline as follows:
Secondary containment for all tanks;
Partial containment and ground-water monitoring for
in-ground and ongrade tanks;
Leak testing and ground-water monitoring for underground
tanks; and
Corrosion protection for steel portions of tank in
contact with the soil.
It is important to note that direct comparisons cannot be made between all
scenarios (for example, partial containment and leak testing scenarios apply
to completely different tank types); however, the above ranking is an overall
indication of the effectiveness of the different scenarios in reducing risks
compared to baseline.
Based on the regulatory scenarios that we examined, our results indicate
that EPA's proposed requirements for full secondary containment- is an
effective strategy for preventing environmental damage caused by leaking
hazardous waste tanks. For in-ground and ongrade tanks, the partial
containment with ground-water monitoring, is in most cases, significantly less
effective than full secondary containment. For underground tanks, leak
testing with ground-water monitoring does not appear to be nearly as effective
as full secondary containment; however, these results must be viewed with
caution due to assumptions in our model that are currently being reconsidered
and which may be significantly affecting our results. Finally, corrosion
protection, without some form of leak detection, does not appear to be very
effective in reducing the risks presented by carbon steel and stainless steel
tanks. Once again, however, we attribute considerable uncertainty to these
results due to model assumptions and limitations.
The risk estimates developed in this analysis are not only a function of
tank failure events and release volumes, but are also controlled by the waste
streams stored in a particular tank and the hydrogeologic settings in which
the tanks are located. In considering regulatory alternatives for tanks, such
as a risk-based variance to secondary containment requirements, it would be
very important to consider these variables on a site-specific basis.
Our analysis was not designed to compare these factors, but rather was
constructed to obtain comparative risk results for different regulatory
scenarios. However, by comparing the generic flow fields and waste streams
included in our analysis, some insight may be gained into what hydrogeologic
and waste parameters most greatly influence risk. Our findings can be
summarized as follows:
-------
Hydrogeologic setting affects risk such that, for a
given technology and waste stream, risk is generally
greater for flow fields with slower ground-water
velocity. This conclusion is somewhat
counter-intuitive, and is dependent on complex
interactions between contaminant mobility and
persistence, ground-water dilution and dispersion, and
the presence of non-aquifer layers. However, it can
be said that, for most waste streams, flow fields with
higher velocities present the lowest risk. In
developing risk-based standards, the ground-water
scenario must be considered in conjunction with
information on potentially exposed populations and
waste stream characteristics.
Many waste streams found in hazardous waste tanks
present relatively high risks, due to their highly
concentrated nature. Among the highest-risk wastes
are spent solvents, certain heavy metal wastes
(particularly those containing thallium and arsenic),
phenol, and reactive wastes (acrylonitrile/cyanide,
2,4-dinitrotoluene, etc.). More dilute wastes such as
spent pickle liquor and petroleum refining wastes
generally present lower risks.
-------
CHAPTER 1
OVERVIEW OF HAZARDOUS WASTE TANK RISK ANALYSIS
1.1 INTRODUCTION
Tanks that are used to store or treat hazardous waste are currently
regulated under the Resource Conservation and Recovery Act (RCRA) [40 CFR Part
264 requirements (January 12, 1981)]. These regulations address minimum
design standards and operating practices. However, evidence suggests that
these tanks, even as currently regulated, may present a potential threat to
public health and the environment. Also, the Hazardous and Solid Waste
Amendments of 1984 (RCRA Amendments) require that more stringent regulations
be promulgated with regard to underground hazardous waste tanks that cannot be
entered for inspection.
In the June 26, 1985, Federal Register, EPA proposed a more stringent and
comprehensive strategy for hazardous waste tanks that addresses both design
and operating standards, by: (1) replacing minimum shell thickness and design
requirements with regulations requiring full secondary containment, or
ground-water monitoring with partial containment, or leak testing with
ground-water monitoring for underground tanks; (2) requiring installation
supervision and corrosion evaluation by certified experts; and (3) requiring
periodic inspections for above-ground and in-ground tanks.
The purpose of this analysis was to assess the human health risks
associated with: (1) the population of hazardous waste tanks under the
current regulations; and (2) the population of hazardous waste tanks under the
various alternative regulatory strategies proposed by EPA. The population of
tanks considered in this analysis included four tank categories;
RCRA-permitted storage tanks; treatment tanks that are not exempt under EPA's
wastewater treatment exemption; small quantity generator (SQG) tanks; and
accumulation tanks (tanks that store waste for less than 90 days and are
therefore not required to obtain a RCRA permit).
In this chapter, we present an overview of our analysis. In Section 1.2,
we outline the proposed regulations for these hazardous waste tanks and
discuss alternative strategies that EPA is considering for these tanks.
Section 1.3 then briefly presents the methodology we used to assess and
compare human health risk associated with the various regulatory scenarios.
The final section describes the organization of the remainder of this report.
1.2 PROPOSED REGULATIONS AND REGULATORY ALTERNATIVES FOR
HAZARDOUS WASTE TANKS
EPA has recently proposed regulations for tanks that are used to store or
treat hazardous wastes (Federal Register, June 26, 1985). These regulations
would supplement and, in some instances, replace current hazardous waste tank
regulations that were promulgated on January 12, 1981. These new regulations
-------
1-2
would complete the Agency's regulatory strategy for managing hazardous waste
tanks, and address requirements of the 1984 RCRA Amendments with respect to
underground tanks. The more stringent controls in the proposed regulations
should substantially improve tank owners' or operators' ability to contain
releases from tank systems and, therefore, should provide needed protection of
human health and the environment. Moreover, EPA has concluded that the
proposed measures, while substantially reducing risks associated with
hazardous waste tanks, are cost-effective when compared to alternative
protection strategies such as shell thickness requirements.
The proposed regulations for tanks that are used to store or treat
hazardous waste are being considered to supplement current requirements
defined in 40 CFR Parts 260, 262, 264, 265, and 270 (SVH-FRL 2791-1);
Hazardous Waste Management Systems; Standards for Hazardous Waste Storage and
Treatment Tank Systems. The proposed regulations address both design and
operating standards and can be summarized as follows:
The installation of all new tanks would be required to be
supervised and observed by a certified expert (registered
professional engineer or equivalent) who would prepare a
written statement attesting to the proper installation of
the tank. Also, the installation site would be required to
be inspected by a corrosion expert, who would also be
required to supervise the installation of a corrosion
protection system (e.g., an impressed current system) if
such a system was considered necessary or was required for
a particular tank.
All new tank systems would be required to be enclosed in
a full secondary containment system that would encompass
not only the body of the tank but all ancillary equipment
(pipes, pumps, valves, etc.)- This secondary containment
system would be required to be equipped with a leak
detection system for detection of leaks within 24 hours of
release.
Facilities with existing tank systems would be required
to conduct integrity assessments of the hazardous waste
tank systems. Tanks that are found to be leaking or unfit
for further use must be repaired or replaced with full
secondary containment.
Facilities with non-leaking tank systems would be exempt
from the full secondary containment requirement if they:
(1) installed monitoring wells and implemented a
ground-water monitoring program; and (2) constructed
partial containment for all above-ground portions of their
tank systems. Because partial containment is not
applicable to underground tanks, these tanks would have the
option of semi-annual leak testing and ground-water
monitoring rather than full secondary containment.
-------
1-3
Inspection requirements would be upgraded to include
inspection of cathodic protection systems and of entire
tank systems for leaks, cracks, corrosion, and erosion that
may lead to releases.
The containment approach outlined in the proposed regulations is presented in
Exhibit 1-1.
These regulations have been proposed partly in response to the 1984 RCRA
Amendments and partly as a result of limitations in the existing regulations.
One limitation relates to the fact that some hazardous waste tanks, as they
are currently being managed, do leak and, therefore, may present a substantial
threat to human health and the environment. Another limitation is that
certain existing tank standards are impractical to implement or may only be
effectively applied at some facilities. An example is the minimum shell
thickness requirement for hazardous waste tanks, which has proven to be
applicable only for above-ground steel tanks, as many or most of the testing
mechanisms for shell thickness do not work adequately for other tank types
(e.g., concrete and underground tanks). It is the intent of the proposed
regulations to address these limitations.
As outlined in the introduction, the purpose of this analysis was to
examine the relative effectiveness of the proposed alternative regulatory
requirements. For this purpose we developed five regulatory scenarios,
designed to represent the range of possible management alternatives for
hazardous waste tanks. These scenarios are as follows:
no revised regulatory requirements (baseline);
secondary containment;
partial secondary containment and ground-water
monitoring for ongrade and in-ground tanks;
corrosion protection for portions of steel tanks in
contact with the soil; and
leak testing and ground-water monitoring for
underground tanks.
Analysis of these five regulatory scenarios allows for direct comparisons
between risks presented by hazardous waste tanks,as they are currently being
managed and under technical requirements set forth in the proposed
regulations. By analyzing each scenario separately, we are able to evaluate
technical requirements that are being considered as part of alternative
regulatory strategies (e.g., corrosion protection), while also considering
requirements that are central to the proposed regulations (e.g., secondary
containment). Also, further examination of tank regulatory alternatives could
include the evaluation of combinations of technical requirements (e.g.,
partial containment and corrosion protection) and, by using the scenarios
-------
EXHIBI I 1-1
PROPOSED REGULATIONS FOR HAZARDOUS WASTE TANKS
Tank Category
RCRA-Permitted
Storage Tanks,
Treatment Tanks a/
New/Existing
New
Existing b/
Accumulation Tanks, c/
SQG Tanks d/
New
Existing b/
Location
above-ground
in-ground
underground
above-ground
in-ground
underground
above-ground,
in-ground,
underground
above-ground,
in-ground,
underground
Containment Requirements
Full Secondary Containment
Full Secondary Containment
Full Secondary Containment,
or Partial Containment with
Ground-Water Monitoring
Full Secondary Containment, or
Semi-Annual Leak Testing with
Ground-Water Monitoring
Full Secondary Containment
Full Secondary Containment
a/ Regulations do not apply to treatment tanks exempt under Part 264 wastewater treatment
exemption.
b/ Existing tanks must comply within one year of effective date of final regulations.
c/ Accumulation tanks are not currently covered under RCRA Part A & B permits and may not store
wastes for more than 90 days.
d/ Although SQG tanks can be either storage or accumulation tanks, we have assumed that all SQG
tanks will be operated as accumulation tanks to avoid the need to obtain a RCRA storage permit.
-------
1-5
listed above, we have the flexibility to analyze the reduction in risks for
these alternatives.1
1.2.1 Baseline Scenario
In the baseline scenario, we assume that all tanks are managed and
operated in compliance with the existing regulations. These regulations
basically involve good operating practices, such as regular visual inspections
and waste compatibility with the tank material. Under this scenario there are
no technical requirements that prevent releases from the primary containment,
nor are there requirements that would assist in the early detection of
releases from underground or in-ground hazardous waste tanks. Under the
baseline scenario, releases from hazardous waste tanks are only detected if:
(1) the leak is above-ground and large enough to be visually detected or (2)
the leak is so large that at the time the contents are normally removed it is
obvious that the tank is not as full as it should be.
1.2.2 Secondary Containment Scenario
Under the secondary containment scenario, we assume that all hazardous
waste tank systems are replaced by new tank systems equipped with full
secondary containment that prevent leaks from the tank and all ancillary
equipment from escaping to the surrounding environment. As prescribed by the
proposed regulations, secondary containment systems in our model scenario are
further fortified by leak monitoring that allows for the rapid detection of
leaks (by remote or visual inspection) that have accumulated within the
secondary containment system. As a result, the full secondary containment
scenario usually prevents releases to the environment from primary tank
failures.
Although full secondary containment is probably the most effective
approach to tank management for minimizing releases to the environment, it is
not 100 percent effective. Tanks with full secondary containment may release
to the environment if there is a catastrophic failure such as an earthquake or
explosion or if there is a failure of the secondary containment system. Thus,
there may still be some risk associated with tanks that have full secondary
containment systems.
Full secondary containment technologies vary depending upon the tank
type. In this analysis, we assume that above-ground, on cradle tanks are
fitted with a lined concrete pad surrounded by diking to prevent leaks from
escaping. We assume that on-grade tanks are replaced with tanks of the same
size and material of construction on top of lined concrete pads surrounded by
diking. Full secondary containment for underground and in-ground tanks
1In addition to the alternative proposed requirements, we plan to
evaluate vadose zone monitoring for appropriate volatile waste streams,
corrosion protection with leak testing, and corrosion protection with vadose
zone monitoring.
-------
1-b
is achieved by replacing existing tanks with double-walled tanks of the same
size and tank material. Finally, we assume that above-ground piping and
ancillary equipment for all tanks is contained by concrete trenches, while
underground piping systems meet secondary containment requirements through the
installation of double-walled pipes.
1.2.3 Partial Secondary Containment and Ground-Water
Monitoring Scenario
In the partial containment and ground-water monitoring scenario of our
analysis, all above-ground portions of a tank system are equipped with
secondary containment, and saturated zone monitoring wells are installed
downgradient to detect releases from portions of the tank not enclosed in the
containment system. In this scenario, we assume that partial secondary
containment is achieved by a lined concrete pad and diking surrounding, but
not underneath, ongrade or in-ground tanks. Above-ground piping is provided
with concrete trenches to complete the partial containment. Partial secondary
containment does not provide protection for the portion of the tank system
that is ongrade or in the ground.
Because releases from portions of tanks not contained within the partial
containment system will be released directly to the environment and may
therefore present risk to human health, the proposed regulations also require
ground-water monitoring for these systems in order to detect releases before
contamination of a drinking water well occurs. In this scenario, we consider
ground-water monitoring to be represented by saturated zone wells that are
located 10 meters downgradient from the tank. We assume that detection occurs
in the year a monitoring well concentration for any of the contaminants in the
tank waste stream exceed a detectable threshold. Once detection has occurred,
we assume that the tank is taken out of operation, thereby removing the source
of further contamination. We then continue to model the transport of wastes
released before tank removal (i.e., no ground-water pumping and treatment is
assumed). Therefore, some released contaminants will not be contained and
will migrate to exposure wells.
1.2.4 Corrosion Protection Scenario
Studies of metal tank systems indicate that the major cause of failure for
systems in contact with the soil is external corrosion. Because the majority
of tanks used to store or treat hazardous waste are made of carbon steel and
are in contact with the soil, it can be seen that corrosion represents a very
significant risk to the integrity of many hazardous waste tanks.
While the installation of full secondary containment effectively isolates
the main tank system from the soil and prevents corrosion, tanks that are
retrofitted with partial containment will still have some portion of their
systems in direct contact with the soil and thus may still be subject to
corrosion. Moreover, installation of full secondary containment or partial
containment and ground-water monitoring is relatively expensive, and, in the
case of the latter, clean-up costs may be substantial for releases that occur
-------
I-/
before detection. Therefore, we have developed a scenario in which cathodic
protection for all metal tank systems in contact with the soil is utilized as
the only protective measure to prevent tank releases. This scenario will
allow us" to evaluate not only the reduction in risk between the baseline and
corrosion protection scenarios, but will also give some indication as to
whether corrosion protection may be a valuable addition to other protective
measures.
In the corrosion protection scenario, all metal tank systems that have
some portion of the tank or ancillary equipment in contact with the soil are
assumed to be fitted with cathodic protection. Thus, the probability of
failures due to corrosion is reduced, while the probability of all other
events remains the same. This allows us to directly assess the reduction in
risk attributable to installing corrosion protection. However, we have not
yet examined the reduction in risk attributable to corrosion protection
combined with other protective measures (e.g., leak testing).
1.2.5 Leak Testing and Ground-Water Monitoring Scenario
The final regulatory scenario considered by our analysis is leak testing
(i.e., integrity assessment) and ground-water monitoring. In this scenario,
we assume that semi-annual leak tests are performed on underground hazardous
waste storage tank systems and that these tank systems are also equipped with
saturated zone monitoring wells.
Leak testing is only applicable for underground tanks because such tests
require stable temperatures to be reliable. The purpose of periodic leak
testing is to detect a leak in an underground tank before extensive
environmental damage or adverse human health effects occur. Currently, there
are no proven leak tests available for tanks containing hazardous waste.
Therefore, for this analysis, we assume that an underground hazardous waste
tank is emptied, cleaned, and filled with water and a hydrostatic leak test is
used.
It is important to note that leak tests are only reliable for leaks
greater than some minimum size; thus, the test will fail to detect very small
leaks. In addition, there is also some probability that the leak test will
fail even if the leak is large enough to be detectable. For this analysis, we
assume that the leak test is 100 percent reliable if the leak is large enough
to be detectable. The more frequently a leak test is required, the more
likely it is that a leak will be detected before excessive environmental
damage or adverse human health effects occur. This analysis assumes a
semi-annual leak testing frequency as proposed in the June 26, 1985, Federal
Register.
In the next section, we present a brief description of the methodology
used to assess the risks associated with each of the five regulatory scenarios
considered in the analysis.
1.3 OVERVIEW OF RISK ASSESSMENT METHODOLOGY
Hazardous waste treatment and storage tanks may be constructed of a
variety of materials (e.g., carbon steel, stainless steel, concrete, and
-------
fiberglass-reinforced plastic (FRP)) and may be located completely
above-ground (on cradles or ongrade), partially underground (in-ground), or
completely underground. Also, they may be of various sizes and ages, and
store a variety of wastes. Additionally, they may be located over a wide
variety of hydrogeologic conditions, and may vary considerably in their
proximity to human populations.
All of these variables may have substantial impact on risk, and thus
needed to be considered in our analysis. Moreover, tank design and operating
practices will have an effect on the risks associated with different hazardous
waste tanks; thus, these factors also needed to be taken into consideration.
Therefore, the methodology developed for this analysis differed from
traditional environmental risk assessment methodologies in that it was
necessary to account not only for factors such as waste stream composition and
hydrogeologic conditions, but also for the range of various failure events
associated with hazardous waste tanks, and for time-dependent processes such
as corrosion. We analyzed the types of failures that would most commonly
occur for hazardous waste tanks, and evaluated the effectiveness of the
proposed regulations and alternative regulatory strategies in terms of (1)
preventing these failures; or (2) detecting these failures before significant
damage to public health or the environment has occurred.
The main objective of our analysis was to compare the risks associated
with the various proposed regulatory requirements to the risks associated with
the existing tank population. In order to analyze the existing tank
population in more detail we divided the tank population into categories of
tanks: RCRA-permitted storage tanks, small quantity generator tanks,
accumulation tanks (less than 90 day storage), small quantity generator tanks
and treatment tanks. For each tank category we then identified types of tank
technologies that are associated with each tank category. For example, we
identified two tank technologies to represent small quantity generator tanks:
(1) underground, carbon steel tanks; and (2) above-ground, carbon steel
tanks. In addition, because the types of waste streams contained in a tank
influence the risk associated with the tank, we identified the most common
types of waste streams "handled by each type of tank technology. Furthermore,
because the hydrogeologic setting surrounding a tank influences contaminant
transport to points of exposure, we identified several generic hydrogeologic
settings to be representative of the settings surrounding tanks.
The four main parts of our risk assessment methodology are summarized
below:
Identify characteristics of the existing hazardous waste
tank population;
Project the type of releases that would occur from tanks
regulated under different operating and design
specifications (i.e., the regulatory alternatives we
examined for this study);
-------
1-9
Estimate the human health risk associated with each tank
technology and regulatory scenario; and
Summarize the range of risk estimates for each tank
technology and regulatory scenario.
Step 1: Identify Tank Population Characteristics
To identify important characteristics for the existing hazardous waste
tank population, we compiled four data bases using available information.
Information on tank technologies; hazardous waste streams most commonly
handled in tanks; physical, chemical, and toxicological properties of toxic
chemicals in these wastes; and hydrogeologic settings surrounding tanks was
compiled into four data bases. These data bases consist of 22 model tank
technologies, 9 generic hydrogeologic settings, 32 waste streams and 36 toxic
chemicals that represent the range of characteristics associated with
hazardous waste tanks.
Step 2: Project Releases
After we characterized the existing tank population, we needed to project
the types of releases that are occurring or will occur from this existing
population of tanks. These projections on the behavior of tanks provide an
indication of what the future health risks associated with current operating
practices and tank technologies will be. Because the regulatory scenarios we
are analyzing are based on changes to current operating practices and
technologies, we had to project the types of releases that would occur from
tanks regulated under different operating and design specifications.
Consequently, the second part of our methodology focuses on estimating
releases form existing tanks regulated under the current and alternative
regulatory scenarios.
In order to project release volumes for these tanks, we developed the
Hazardous Waste Tank Failure (HWTF) Model, a Monte Carlo simulation model to
predict failure events for hazardous waste tanks and to estimate release
volumes associated with these failure events. We used the HWTF model to
predict failure events and release volumes over the 20-year time horizon for
each of the 22 model tank technologies considered in the analysis. In total,
250 "release profiles", or sets of year-by-year releases, are produced for
each modeled tank technology under a particular regulatory scenario. A
statistical technique known as cluster analysis is then used to select five or
six representative release profiles from the 250.2 These representative
2Ideally, we would use all 250 release profiles to obtain 250 risk
estimates for each tank. However, using 250 release profiles to generate risk
estimates for each model tank, regulatory scenario, ground-water setting and
waste stream, would be a computationally intensive task. Therefore, we used
cluster analysis techniques to select the most representative profiles.
as inputs to the Tank Risk Analysis Model.
-------
1-10
profiles are weighted and used as inputs the risk model developed in the next
step.
Step 3: Estimate Human Health Risk
The third part of our methodology focuses on using the projection of
releases to estimate human health risk. We predicted the transport of
contaminants released from tanks to human exposure points (i.e., drinking
water wells). We then generated human health risk estimates associated with
individuals consuming this contaminated drinking water. We used these risk
estimates to compare the regulatory scenarios.
We developed the Tank Risk Analysis Model (TRAM), an exposure/risk model,
to estimate human exposure to hazardous chemicals via contaminated drinking
water and to calculate non-carcinogenic and carcinogenic risk to the exposed
individual.
The Tank Risk Analysis Model is a non-steady-state (i.e., time-varying)
model that consists of three major components: an unsaturated zone transport
model, a saturated zone transport model, and an exposure/risk model.3
Contaminant transport and degradation in the unsaturated and saturated zones
are estimated using modified versions of The McWhorter-Nelson Wetting Front
and Prickett-Lonnquist Random Walk models, respectively. The major components
of TRAM are illustrated in Exhibit 1-2.
Step 4: Summarize Risk Estimates
The technique used to compare regulatory scenarios constitutes the final
part of our methodology. In general, for each tank technology and regulatory
scenario we estimated 324 risk estimates.* These 324 risk estimates
represent the range of risk estimates associated with a tank technology.5
3In order to maintain consistency with other similar analyses (e.g.,
Liner Location Risk and Cost Analysis Model studies), we chose 400 years as
the potential exposure period in our analysis. The 400-year time horizon is
long enough to capture most of the risk associated with releases of
contaminants (i.e., a long enough time horizon such that most contaminants
reach the exposure point within this time frame) and short enough to be
computationally feasible.
'"There are approximately six waste streams per technology, nine
hydrogeologic settings per technology and six sets of annual release volumes
per technology. These factors are described in more detail in the following
chapters of this report.
'Due to the variability in tank designs, operating conditions, and
hydrogeologic and waste stream parameters, it was considered that a simplified
approach based on one or several case histories of failed tanks was not
adequate to assess the risk reductions for the different regulatory scenarios
on a nationwide basis. Consequently, we estimated a range of risk estimates
for each tank technology and regulatory scenario.
-------
EXHIBIT 1-2
FLOWCHART OF THE TANK RISK ANALYSIS MODEL
Tank
Failure/
Release
Component
*
Timinc and
Magnitude or .
Kclcue Volumes*
f '
Unsaturated
Zone
Transport
Component b
Chemical Man
£:lea»ed
luralcd
Zone
Saturated
Zone
Transport
Component b
Chemical
Concenlrallon in
Ground Water
Detection
by Groun
Mooil
Dose
Estimation
Component
(Drinking
Water)
ol Leak
d-waler
>ring
Human Doie Via
Drinking Water
Risk
Estimation
Component
i
Source
Excavation
9 Six representative release profiles, selected using cluster analysis techniques, are input to TRAM.
' Chemical mass can also be degraded or removed from the modeling system in these components.
-------
1-12
There are several ways to summarize these 324 risk estimates. For example, we
could summarize this range using the minimun, mean, median, or maximum risk
estimate associated with the 324 estimates. Because the choice of such a
summary measure aggregates much relevant information, we chose to derive
frequency distributions of risk estimates (e.g., percent of risk estimates
that are 10 ). We derived frequency distributions of risk estimates by
weighting each estimate by its relative frequency of occurrence in terms of
waste stream handled, hydrogeologic setting and representativeness of annual
releases.
Exhibit 1-3 represents a hypothetical relative frequency distribution of
risk estimates for above-ground, carbon steel, small quantity generator
tanks. This distribution represents the range of risk estimates associated
with the population of above-ground, carbon steel small quantity generator
tanks. For example, the distribution indicates that 10 percent of all
above-ground small quantity generator tanks are associated with risk estimates
of 10 . This distribution was based on information on the relative
frequency of factors that influence risk estimates. For example, we
determined that 28 percent of all above-ground small quantity generator tanks
store strong acids or alkaline wastes. A frequency distribution was derived
for each combination of tank technology and regulatory scenario.
Exhibit 1-4 summarizes the methodology we used to derive each frequency
distribution. We first used the HWTF Model to simulate 250 potential release
profiles associated with a tank technology and regulatory alternative. We did
a cluster analysis to select six representative release profiles from the 250
potential release profiles. We also determined, for each representative
release profile, the percent of the 250 potential release profiles that the
representative profile represents.
These six profiles are input to TRAM. Then for each waste stream,
hydrogeologic setting, and representative release profile, TRAM generates a
risk estimate. Approximately 324 risk estimates are generated for each
combination of tank technology and regulatory scenario. Using these risk
estimates and their associated weights, we developed relative frequency
distributions of risk estimates. This process is performed for each
combination of tank technology and regulatory scenario, allowing the risk
estimates for a specific tank under alternative regulatory scenarios (e.g.,
baseline and secondary containment) to be directly compared.
The methodology we used is flexible enough to incorporate a wide variety
of generic situations, but is limited in that it cannot be applied to specific
real-site situations. Thus, it is most useful for making broad comparisons
between regulatory alternatives, tank technologies, waste streams, and
hydrogeologic environments. The methodology was developed specifically to
examine the existing hazardous waste tank population under alternative
regulatory scenarios. It was not designed necessarily to allow various
comparisons to be made between different types of tanks.
-------
1-13
EXHIBIT 1-3
HYPOTHETICAL RELATIVE FREQUENCY DISTRIBUTION OF RISK
ESTIMATES FOR A TANK TECHNOLOGY AND REGULATORY SCENARIO
Relative
Frequency
0.50
0.40
0.30
0.20
0.10
-10
log of Risk Estimate
-------
1-14
EXHIBIT 1-4
METHODOLOGY TO ESTIMATE FREQUENCY DISTRIBUTION OF RISK
ESTIMATES FOR EACH TANK TECHNOLOGY AND REGULATORY SCENARIO
Use HWTF I
Estimate Tin
Magnituc
Release Vc
* rtX-
.,>
vlodel to
ling and
le of
)lumes
Use Cluster Analysis
to Identify Six
Representative
Release Profiles
Ml
(1) Vol.
(2) Vol.
(3) Vol.
Outputs
Time
Use TRAM to Estimate
Risk Estimates and Weights
For Each Combination of
Hydrogeologic Setting. Waste
Stream and Representative
Release Profile
Develop Relative
Frequency Dis-
tributipn of Risk
Estimates
(1) Vol.
(2) Vol.
(3) Vol.
(4) Vol.
Outputs
A
Outputs
Outputs
r1,w1
Relative
Frequency
1
log of Risk Estimate
T324 >W324
(250) Vol.
A
Time
(5) Vol.
(6) Vol.
Time
-------
1-15
The main reason that tank technologies cannot be readily compared is that
they differ in several factors, such as size, age, waste streams handled, and
material of construction. Such comparisons can be made if the model tank
technologies and waste streams were revised specifically for this purpose
(i.e., more of the tank technologies were modeled to have the same size, age,
or waste streams handled). This report focuses primarily on comparisons
between regulatory scenarios, and the reader is cautioned against attempting
to draw final conclusions based on any other comparisons that one might be
tempted to make with data presented in this report.
Below we outline our basic methodology in greater detail, with particular
emphasis on: (1) the HWTF Model; (2) the major components of TRAM; and (3)
the data bases that support the analysis, and how these data bases were
compiled.
1.3.2 Hazardous Waste Tank Failure Model
The Hazardous Waste Tank Failure (HWTF) Model is a Monte Carlo simulation
model that predicts failure events and estimates volumes of released wastes
for hazardous waste tanks.' Its purpose is to identify the various failure
mechanisms associated with hazardous waste tanks, and to predict: (1) the
frequency of occurrence of these failure events; and (2) the magnitude of
release volumes associated with these events. The model simulates the
following failure events:
Overflows;
Leaks and ruptures;
Natural catastrophes;
Secondary containment failures;
Operator errors; and
Detection system malfunctions.
In Monte Carlo simulation, probability distributions are assigned to model
variables for which deterministic values are not available. These probability
distributions reflect the uncertainty regarding a real value for a particular
parameter. The degree to which the model is representative of a particular
situation is dependent upon the availability of data for the construction of
the probability distribution. Once these distributions are constructed, a
value for each variable is randomly chosen from the distribution for that
parameter, and a final model prediction is calculated. By repeating this
procedure a large number of times, a distribution of predicted values for the
final model outcome is constructed. Theoretically, this distribution reflects
the combined uncertainties for the
SA detailed description of the HWTF Model is provided in a separate
report: Pope-Reid Associates, Inc., Hazardous Waste Tank Failure Model:
Description of Methodology, draft report submitted to the Office of Solid
Waste, EPA, January 13, 1986.
-------
1-16
different variables contained within the model. Thus, in situations where
there is significant uncertainty with respect to values for some model
variables, the use of Monte Carlo simulation may be advantageous. However, in
some cases the use of Monte Carlo simulation may ascribe greater reliability
to model results than is actually warranted due to uncertainty over the choice
of probability distributions.
In the HWTF model, distributions for both the occurrence and timing of
failure events are sampled randomly for 250 model iterations. For each
iteration, failure events are noted, and release volumes for failure events
are calculated. The timing and magnitude of release volumes is reported for
each year in which the model has predicted failure to occur. Thus, model
output is a set of 250 iterations with each iteration consisting of annual
release volumes for the tank operating lifetime considered in the analysis.
For this analysis, we used a 20-year time horizon7 for tanks; therefore,
each iteration contains annual release volumes for a total of 20 years. The
250 iterations that are produced by the HWTF model are then sorted using a
statistical technique known as cluster analysis, which allows for the
selection of five or six representative "release profiles".* These profiles
are weighted and serve as inputs to the subsurface transport model.
The HWTF model may be used not only to identify the failure events and
release volumes associated with different hazardous waste tank technologies,
but may also be used to estimate the effectiveness of various protective
measures in preventing releases of hazardous wastes to the environment. This
use may be accomplished within the model by varying the probability
distributions for particular events. However, the model is heavily dependent
upon several key assumptions, and is also limited severely by the lack of data
for the construction of probability distributions used to estimate the
frequency of occurrence and timing of the various events predicted. Important
assumptions include the following:
Tank leaks (with the exception of SQG tanks) are not
detected until 25 percent of the throughput for a given
tank has escaped. For SQG tanks, 100 percent of the tank
throughput must be released before a leak is noticed.
These assumptions, made because of a lack of data on this
subject, have a large impact on the results of the analysis.
7Because we are interested in simulating releases for tanks
representative of the current population of tanks, the initial age
of a tank being analyzed by the HWTF model is the median current age
of such tanks. This age varies by model tank.
*We use the term release profile in this analysis to refer to
the set of 20 annual release volumes generated by an iteration of
the HWTF model.
-------
1-17
Values are assumed for the mean time to failure for many
tank technologies (e.g., concrete tanks) and also for mean
time to occurrence for most failure events (e.g., rupture);
failure times are assumed to be distributed about these
mean values. These values are not strongly supported by
data, and have a major effect on the results of the model.
The model assumes that probability distributions do not
change with time. For example, a tank may have a small
probability of failure due to localized corrosion until
general corrosion thins the tank walls. At that point
there is a high probability of a large number of localized
corrosion failures. This type of event currently is not
accounted for by the model.
Despite these limitations, the HVTF model is currently the only model
available that is capable of analyzing the frequency and severity of the
failure events associated with hazardous waste tanks under the various
regulatory scenarios considered in this analysis. It can be useful for
comparing different regulatory scenarios, but its use is more problematic for
comparing different tank categories and designs. Chapter 2 provides more
detailed documentation of the HWTF model.
1.3.3 Subsurface Transport Model
In the Tank Risk Analysis Model (TRAM), representative release profiles
from the HWTF model are input to a subsurface transport model that predicts
the transport and environmental fate of contaminants released from hazardous
waste tanks in the unsaturated and saturated zones. This transport and fate
model was adapted from EPA's Liner Location Risk and Cost Analysis Model9 (a
risk model used to analyze risks for various hazardous waste land disposal
options). The model uses the McWhorter-Nelson Wetting Front model to simulate
the transport of contaminants in the unsaturated zone, and the Prickett-
Lonnquist Random Walk model to simulate saturated zone transport. These
models are described briefly below.
McWhorter-Nelson Wetting Front Model. In this model, contaminant
transport is represented by the movement of a wetting front generated by the
release of liquid from the tank system. The travel time for the wetting front
through the unsaturated zone is related to the difference in water content in
the layers immediately above and below the wetting front, the distance
travelled in the unsaturated zone, the leakage rate (Darcian flux), and the
retardation of contaminant flow by soil adsorption.
'Liner Location Risk and Cost Analysis Model, draft report to Office
of Solid Waste, EPA, January 1985.
-------
j. - iO
Thus, contaminant transport in the unsaturated zone, as modeled by the
McWhorter-Nelson Wetting Front Model, is dependent not only upon properties of
the unsaturated zone, but also upon source-specific information (i.e., amount
of material released and physical/chemical properties of the released
contaminants). In the Tank Risk Analysis Model, we account for the effects of
soil parameters by assuming that all tanks are located over a generic
unsaturated zone, five meters in thickness, with constant soil moisture,
porosity, and permeability throughout. Furthermore, we assume that the
fractional organic carbon content (Foe) remains constant throughout the
unsaturated zone; therefore, retardation is a function only of the physical
and chemical properties of a particular contaminant.
Contaminant retardation and degradation are predicted based on chemical-
specific information. Retardation factors for all chemicals included in the
TRAM data base (approximately 120 chemicals) are calculated using organic
carbon-water partition coefficients (Koc values). Biological and chemical
degradation is considered only for those chemicals for which data are
available. We used a first-order decay function that accounts for all
relevant degradation processes through the first-order constant, k.
The McWhorter-Nelson model has been used extensively in subsurface
transport studies and is a we11-documented and widely accepted unsaturated
zone model. However, to use this model in TRAM it was necessary to make
several simplifying assumptions with respect to unsaturated zone conditions
and contaminant transport. Although these assumptions tend to oversimplify
transport and attenuation processes occurring in the unsaturated zone, they
are consistent across all situations and thus allow for direct risk
comparisons between different regulatory scenarios to be made. The model,
along with its assumptions and limitations, is documented more thoroughly in
Chapter 3.
The unsaturated zone model provides as its output a mass loading to the
saturated zone for each contaminant released from a particular model tank.
Prickett-Lonnquist Random Walk Model. The Tank Risk Analysis Model
uses the Prickett-Lonnquist ground-water transport model with random walk
particle tracking to simulate the transport of contaminants in the saturated
zone. This model is a numerical (finite difference) model that estimates
transport in two dimensions (longitudinal and vertical) and simulates
dispersion in two dimensions. It is a widely-used and we11-documented
ground-water transport model that has been utilized in many ground-water
contamination studies.
In TRAM, we do not actually use the Prickett-Lonnquist program code.
Instead, TRAM uses a set of "standard breakthrough curves" derived from
Prickett-Lonnquist model output to calculate the concentration of a particular
contaminant at an exposure well, for a given set of hydrogeologic and mass
loading conditions over time. A standard breakthrough curve represents the
annual concentration of contaminants at an exposure point due to a release of
one kilogram of contaminants.
-------
1-19
Hydrogeologic conditions such as horizontal and vertical ground-water
velocities, and aquifer configurations and thicknesses are represented in the
model by a set of nine generic flow fields that were developed for the Liner
Location Risk and Cost Analysis Model. These nine flow fields encompass
horizontal ground-water velocities of 1 to 10,000 m/year, and represent single
and double aquifers with and without intervening nonaquifer layers. All earth
materials in all flow fields are assumed to be homogenous and isotropic.
The transport model has been run for each of the nine generic flow fields.
Results for these runs, in the form of "standard breakthrough curves", are
used to calculate well concentrations based on mass loadings. These factors
are available for three exposure well distances: 60, 600, and 1,500 meters.
For comparative purposes, we used the well concentrations for the 60 meter
distance. For modeling purposes, the 120 contaminants in the TRAM data base
are grouped into four mobility classes based on their retardation coefficients
(R-^ values). The model uses median values for each group rather than
chemical-specific R^ values. Thus, TRAM contains "standard breakthrough
curves" for a total of 108 scenarios (nine flow fields x three exposure well
distances x four mobility classes).
Although the saturated zone model is considered adequate for comparing the
transport of contaminants in hypothetical, generic situations, it should be
stressed that the model, as included in TRAM, is not suitable for site-
specific analyses. Moreover, developing and using the saturated zone
transport model required several simplifying assumptions that affect the
model's results and limit its applicability. These assumptions, however, are
consistent across all situations, and thus the model is considered useful for
comparative purposes. Model assumptions and limitations are discussed in
Chapter 3.
Concentrations of contaminants of concern for a particular tank, waste
stream, and hydrogeological combination are calculated for exposure wells,
located 60 meters from the tank, by the saturated zone model. These
concentrations serve as input to the exposure/risk component model of TRAM,
which is described in the next section.
1.3.4 Exposure/Risk Model
Subsurface transport model results in TRAM are input to an exposure/risk
model, also adapted for TRAM from the Liner Location Risk and Cost Analysis
Model. This model estimates the timing and magnitude of exposure to
individuals using contaminated ground water as a drinking water source This
model also predicts the probability (in terms of a risk estimate) that the
exposed individual will suffer an adverse health effect as a result of using
the contaminated water supply. In this section, we outline this model briefly.
Exposure Component. To estimate drinking water exposure, we assume
that the contaminated water pumped from a well undergoes no treatment, or that
treatment is ineffective in removing a significant portion of the contaminant
mass contained in the water. Also, we assume that individuals continue to
-------
1-20
drink the water despite exceedance of taste and odor thresholds, and
regardless of noticeable health effects.10 To calculate exposure, the
intake rate of drinking water for humans is assumed to be two liters/day, and
the average weight of receptors is considered to be 70 kilograms. One hundred
percent absorption of an ingested contaminant is also assumed to occur.
Thus, the yearly dose to an individual using a particular well is
considered to be equal to well concentration multiplied by the intake rate of
drinking water divided by the receptor weight. In addition to the assumptions
noted previously, we also assume that an individual obtains 100 percent of his
water from the contaminated well, and that the individual consumes this water
over his entire lifetime (population mobility is not considered).
Risk is then calculated for the exposed individual based on the dose and
inherent hazard (H), or potency, of the particular chemical to which the
individual is exposed. For carcinogens, we used a linear dose-response model
(the one-hit model) that assumes no minimum threshold. The model uses EPA
unit risk values, when available, to estimate H for carcinogens. For other
values, we used H values that were based on fitting the one-hit model to
experimental data.
For non-carcinogens, a threshold, or minimum effective dose (MED), is
assumed, and a non-linear dose response model (the Weibull equation) is used
to calculate risk. Although non-linearity of the dose-response curve may vary
somewhat according to the particular chemical of interest, we assume that the
shape factor for the dose-response curve, K, is equal to two. H values for
non-carcinogens were estimated based on experimental MED values and the
frequency response at that dose.
This model is consistent in its approach to calculating carcinogenic risk
with the traditional method used by toxicologists and policymakers. The
methodology for calculating non-carcinogenic risk diverges from the
traditional approach, however, in that it uses derived H values to calculate
an explicit risk, rather than using an experimentally determined threshold to
estimate an allowable daily intake (ADI). This approach was adopted because
the traditional approach is not very useful in explicit risk analyses because
it does not allow actual risk values to be calculated.
The primary limitations and assumptions in the risk model are those
traditionally found in determining inherent hazards and characterizing
dose-response in the low dose regions. Specific sources of uncertainty in the
exposure/risk model are described in Chapter 3. It should be noted that the
assumptions and limitations inherent to this model make it most useful for
comparison purposes, and risks calculated by the model should be considered as
relative and not as absolute risks.
10TRAM currently contains taste and odor threshold values for all 120
chemicals in its data base, and the ability to invoke these thresholds is
incorporated into the model.
-------
1-21
1.3.5 TRAM Data Bases
In order to gain an understanding of the risks associated with hazardous
waste tanks through the use of the Tank Risk Analysis Model, it was necessary
not only to identify failure events and calculate transport times for a model
tank, but to achieve as close an approximation as possible of the real-life
situations in which these tanks are found. Thus, we compiled information to
develop hazardous waste tank technologies, associated hazardous waste streams,
and hydrogeologic settings that are representative of the existing tank
population.
In this section, we present brief descriptions of the data bases that were
developed in support of TRAM, and discuss their uses within the framework of
the analysis.
Model Tank Technology Data Base. To predict tank failure events and
release volumes, and to estimate human health risks for the various regulatory
scenarios considered in the analysis, we developed a data base containing
information on 22 model tank technologies. These model tank technologies were
selected to represent typical tanks reported in the OSW Regulatory Impact
Analysis (RIA) Tank and Small Quantity Generator Surveys, and encompass four
major categories: RCRA-permitted storage tanks, small quantity generator
(SQG) tanks, accumulation (less than 90-day storage) tanks, and treatment
tanks. Model tanks included in each of these four categories were chosen
based on type (i.e., above-ground, in-ground, or underground) and material of
construction (i.e., carbon steel, stainless steel, concrete, and fiberglass
reinforced plastic).
The model tank data base contains information on tank type, material of
construction, and whether a particular model tank is open- or closed-topped.
In addition, the data base contains data on the percentage of each category
represented by each tank type, and includes median age and size for each of
the model tanks. Therefore, although this data base does not represent the
extremes that exist in the population of hazardous waste tanks, it contains a
representative sample of the tank technologies that are most commonly used to
store and treat hazardous waste. The data base is outlined in greater detail
in Chapter 4, Section 2.
Waste Stream Data Base. Waste streams most commonly stored or treated
in tanks were identified for the four major tank categories considered in the
analysis and compiled into a data base containing information on waste stream
characteristics (representative constituents and constituent concentrations)
and on the percentage of each model tank technology storing each waste
stream. In all, 32 waste streams were identified from various data sources as
being among those most commonly reported for hazardous waste tanks, and were
included in the data base.
Of the 32 waste streams, 10 were identified for SQG tanks, six were
selected as being most representative of RCRA-permitted storage tanks, six
were identified for treatment tanks, and 12 were identified for accumulation
-------
1-22
tanks. Thus, there was some overlap between tank categories with regard to
waste streams handled, although many wastes identified for SQGs were unique to
these tanks. The waste streams encompass a broad range of characteristics,
from aqueous wastes containing cyanides and heavy metals to organic solvent
waste streams and concentrated substances such as phenol. Therefore, although
waste constituents and constituent concentrations for many waste streams may
vary markedly from those used in the analysis, the waste stream data base
provides a reasonable representation of the wide variety of wastes that are
commonly handled in hazardous waste tanks.
Chemical Data Base. Transport and fate information, risk data, and
detection limits for 120 hazardous chemicals were compiled in the TRAM
chemical data base. This information was taken from the Liner Location Risk
and Cost Analysis Model chemical data base, and includes data on both organic
and inorganic contaminants. We identified 36 of these chemicals as the most
common chemicals handled in hazardous waste tanks.
Transport and fate data in the data base includes aqueous solubility,
organic carbon-water partition coefficient (Koc) values, and degradation rate
constants for each of the 120 chemicals. Risk information includes unit risk
(potency) values for carcinogens and non-carcinogens, and threshold values for
non-carcinogens. Type of effect (e.g., cancer, liver damage, etc.) is also
identified for each chemical. Chemical detection data includes taste and
odor, and detectable concentrations.
This data base does not contain data on all priority pollutantsr but does
include all of the waste constituents most commonly reported for hazardous
waste tanks. The data base is described in greater detail in the EPA Liner
Location Risk and Cost Analysis Model draft report.11
Hydrogeologic Setting Data Base. As discussed earlier in the chapter,
TRAM does not use site-specific hydrogeologic data, but rather considers a
model tank to be located in one or more of nine generic flow fields that are
believed to represent the range of hydrogeologic conditions that would
typically be encountered at tank facilities. This data base contains
information on the ground-water velocities and aquifer configurations for each
of the nine generic flow fields.
In order to estimate the relative frequency of occurrence of each of these
ground-water flow fields for a particular tank category, we distributed the
population of model tanks over each state, estimated the percentage of
occurrence for each flow field in each state, and from these two distributions
calculated the population of each tank category located in each flow field.
These data allow for the weighting of the risk estimates associated with the
different hydrogeologic settings used in the analysis, and, although it is
11Liner Location Risk and Cost Analysis Model, draft report to Office
of Solid Waste, EPA, January 1985.
-------
1-23
limited somewhat due to the assumptions necessary in constructing the
distribution, it provides some representation of the hydrogeologic conditions
that may be encountered at typical tank facilities.
The following section briefly outlines the organization of the remainder
of the report.
1.4 ORGANIZATION OF THE REPORT
The remainder of this report is organized as follows:
Chapter 2 -- Hazardous Waste Tank Failure Model
This chapter provides an overview of the model used to
simulate the release of hazardous waste over a 20-year time
horizon. The chapter focuses on the type of phenomena that
the model simulates, rather than the parameters and
probability distributions that were used. A more detailed
explanation can be found in a separate report.12
Chapter 3 -- Transport, Exposure and Risk Models
This chapter describes the models we used to estimate the
transport of contaminants in the unsaturated and saturated
zones underlying hazardous waste tanks. The equations and
methodology used to estimate chronic human health risks are
also presented.
Chapter 4 -- Development of Data for Deriving
Distributions of Risk Estimates
This chapter presents the methodologies and data sources
that we used to determine the tank technologies, associated
waste streams, and hydrogeologic settings that we should
model in order to represent the existing population of
hazardous waste tanks. This chapter also presents the
methodology we used to develop distributions of risk
estimates that are representative of existing tank
technologies.
l2Pope-Reid Associates, Inc., Hazardous Waste Tank Failure Model;
Description of Methodology, draft report submitted to the Office of Solid
Waste, EPA, January 13, 1986.
-------
1-24
Chapter 5 -- Results and Conclusions
This chapter reports the results of our current analyses.
Results generated from the HVTF Model for the baseline and
secondary containment regulatory scenarios are presented in
detail. Distributions of risk estimates for every
combination of modeled tank technology and regulatory
scenario are presented and discussed.
Appendix -- Failure Frequency Data and Representative
Release Profiles
The appendix presents all the representative release
profiles that we used to generate risk estimates for each
modeled tank technology and regulatory scenario.
-------
CHAPTER 2
HAZARDOUS WASTE TANK FAILURE MODEL
In this chapter we discuss the Hazardous Waste Tank Failure Model. We
used this model to generate estimates of the timing and magnitude of releases
from hazardous waste tanks over a 20-year time horizon. The purpose of the
model is to simulate the timing and magnitude of various types of tank
failures (e.g., leaks, ruptures, etc.) associated with hazardous waste tank
technologies based on each of the five regulatory scenarios. The model
estimates a potential range of annual release volumes associated with the
hazardous waste tanks. A simulated series of annual release volumes for a
particular tank technology is used as input to the Tank Risk Analysis Model.
In the first section of this chapter we provide a general overview of the
HWTF model. In Section 2.2 we present the representative tank technologies
that we modeled. Section 2.3 identifies various leak detection and prevention
options that were used to model each regulatory scenario. In Section 2.4 we
discuss how we characterized the environment surrounding hazardous waste
tanks, and then in Section 2.5 we describe the generic waste streams that we
used for modeling purposes. Finally, in Section 2.6 we present the types of
events that were simulated and the release volumes associated with each event.
2.1 OVERVIEW OF HAZARDOUS WASTE TANK FAILURE MODEL
The release of hazardous waste from treatment and storage tanks is
contingent on the occurrence of one or more failure events, many of which have
a degree of uncertainty associated with them. Consequently, we used a model
that can account for these uncertainties. The HWTF model is based on the
principles of probability and statistics and can simulate particular aspects
of phenomena that relate to the release of hazardous waste from tanks. This
type of model, generally referred to as a Monte Carlo simulation model,
permits us .to evaluate the potential range of releases from tanks.
Exhibit 2-1 illustrates one representative series of annual release
volumes simulated by the model.2 This series of releases constitutes one
potential release profile associated with a tank. For this study, we used the
model to generate 250 representative series of annual release volumes.1 A
release profile is based on four major factors: (1) timing of events; (2)
Although the choice of the number of estimates (i.e., release profiles)
generated by a Monte Carlo model is traditionally based on the statistical
analysis of the convergence properties of the estimates, this analysis was not
done for this study. Basically, the choice to generate 250 profiles was based
on professional judgment and cost constraints.
2For this release profile, the release was detected in year 12 and the
tank was replaced. No subsequent releases occured following replacement of
the tank.
-------
2-2
EXHIBIT 2-1
POTENTIAL RELEASE PROFILE
5-
Volume
(m3)
34
2-
1-
10 12 14 16
18 20
Time (year)
-------
2-3
magnitude of release volumes; (3) duration of release; and (4-) detection of
release. The model evaluates each of these factors using probability
distributions that represent the phenomena we are modeling. For example, the
occurrence of a tank rupture is based on probability distributions
representative of the ways in which a tank can rupture. The types of events
(i.e., phenomena) that we modeled can be categorized as follows:
Overflows;
Corrosion of the tank and/or pipes;
Rupture of the tank and/or pipes;
Natural catastrophes;
Spills;
Effectiveness of protective measures; and
Effectiveness of detection measures.
The model first simulates the occurrence of events that can lead to
the release of hazardous waste. The ability of protective measures to
prohibit the release from occurring is then simulated. For example, the
effectiveness of an emergency shut-off control system to prevent an overflow
is simulated. If the release is not prevented, a leak rate is simulated.
This leak rate is modeled to continue until the release is detected. The
detection of a release is based on when an available inspection option is
simulated to be effective.
Obviously, whether or not a protective measure prevents a release depends
on what protective measures (e.g., secondary containment) are available and
the effectiveness of the measures. Similarly, the detection of releases,
depends on the frequency and effectiveness of inspections. In a sense, the
protective and detection measures are the focus of this analysis because they
are used to define each regulatory scenario. Consequently, we discuss in this
chapter the assumptions that we used to model each of the regulatory scenarios
in terms of the simulated protective and detection measures. In the next
section we present the hazardous waste tank technologies that we modeled. We
then present the protection and detection measures that we modeled for each
regulatory scenario. In the last section we discuss the failure events that
the model simulates. A more detailed discussion of the probability
distributions and leak rate equations used to model the failure events is
presented in a separate report.1
2.2 TANK TECHNOLOGIES
Clearly, one basic requirement for an assessment of failures of and
releases from hazardous waste tanks is the selection of representative
hazardous waste tank technologies. We divided the universe of hazardous waste
tanks into four categories: RCRA permitted treatment tanks, RCRA permitted
3Pope-Reid Associates, Hazardous Waste Tank Failure Model; Description
of Methodology, draft report to the Office of Solid Waste, EPA, January 13,
1986.
-------
2-4
storage tanks, accumulation tanks (storage for less than 90 days), and small
quantity generator (SQG) tanks. Tanks can be distinguished further by their
location (e.g., above-ground, inground, or underground) and their material of
construction (e.g., concrete, carbon steel, stainless steel, or fiberglass-
reinforced plastic). Using survey data, we selected 22 representative tank
technologies (details of this selection process are presented in Section
4.2). The characteristics of these tank technologies are shown in Exhibit 2-2
and their operating parameters are shown in Exhibit 2-3. For each tank
technology we also identified a median age and size.* These characteristics
are used to simulate the existing tank population. The remainder of this
section provides specific details on the tank technologies that we modeled.
2.2.1 Treatment Tank Systems
We selected two treatment processes, oxidation/reduction/precipitation and
distillation, to represent hazardous waste treatment tanks in this analysis.
In this section we describe each of these processes and the modeled design
specifications.
Oxidation/Reduction/Precipitation
We modeled oxidation/reduction/precipitation as a continuous process using
three tanks. We modeled five different types of oxidation/reduction/
precipitation treatment technologies. These tanks differ in their location
(e.g., above-ground, in-ground) and material (e.g., carbon steel, concrete,
stainless steel) A diagram of the oxidation/reduction/precipitation process
is presented in Exhibit 2-4. Only aqueous waste streams were assumed to be
treated by this process.
The first tank is used to adjust the pH of the influent waste stream to
optimize the oxidation/reduction/precipitation reactions that will occur in
the second tank. The waste is pumped into the top of the tank at a rate that
is equal to the operating capacity of the tank divided by the retention time,
(the length of time the fluid is kept in the tank to complete reactions and
mixing). We assumed that the waste has a thirty minute retention time in the
first tank. We also assumed that the fluid depth in each tank is maintained
at 80 percent of the tank height.
The second tank is used for oxidation, reduction, or precipitation,
depending upon the characteristic of the waste being treated. This tank is
four times as large as the first tank to allow for a two hour retention time.
* We used the median tank age to simulate the conditions of the existing
tank population. For example, for in-ground, concrete, treatment tanks, we
simulated these tanks to have been in operation for the past 7 years (with
respect to model year 1). Some of these tanks have started to corrode and may
be leaking today (i.e., in model year 1) or are likely to develop leaks within
the next few years. Consequently, in model year 1, tanks may be simulated to
be releasing undetected hazardous waste or to be corroding.
-------
EXHIBIT 2-2
CHARACTERIZATION OF EXISTING HAZARDOUS WASTE TANKS
Median
Size
(gal Ions)
TREATMENT:
2,300
2.300
60,000
3,700
3,700
3,700
STORAGE;
5,500
210,000
-------
EXHIBIT 2-2 (continued)
CHARACTERIZATION OF EXISTING HAZARDOUS WASTE TANKS
Median
Size
(gal Ions)
U,000
«4,000
2,100
2,100
Tank Location a/
Underground
Underground
In-ground
In-ground
Material
of Construction b/ Tank Top
FRP
Stainless steel
Concrete Open
Carbon steel Open
Median
Pipe age
Location (years)
underground
unde rg round
above-ground
above-ground
7
7
8
8
Wai 1
Thickness
( inches)
NA c/
.25
6.00
.25
a/ Inground tanks are assumed to be 50 percent below-ground and 50 percent above-ground.
b/ Pipe is constructed of the sane material as the tank for all technologies except concrete tanks, which
have carbon steel pipes.
c/ Tank thickness is not specified for FRP tanks because all failure events are modeled to be independent of
tank thickness.
Source: Median size, tank location, material of construction and median age are based on OSW RIA Tank Survey.
Pipe location and wall thickness are based on engineering judgment.
-------
EXHIBIT 2-3
OPERATING PARAMETERS
Tank Technology
TREATMENT;
On cradles, closed top
On cradles, open top
Ongrade, carbon steel
Ing round, concrete
Ing round, carbon steel
Inground, stainless steel
STORAGE/ACCUMULAT 1 ON ;
On cradles, carbon steel
Ongrade, carbon steel
Underground, carbon steel
Underground, FRP
Underground, stainless steel
1 ng round , cone re te
Inground, carbon steel
SMALL QUANTITY GENERATOR;
Above-ground, carbon steel
Underground, carbon steel
Emergency
Shut-Off
Manual
Automatic
Automatic
Automatic
Automatic
Automatic
Manua 1
Automatic
Manua 1
Manua 1
Manual
Manua 1
Manual
Manua 1
Manua 1
Level Control
Manua I
Automatic
Automatic
Automatic
Automatic
Automatic
Manua 1
Manua 1
Manua 1
Manua 1
Manua 1
Manua 1
Manua 1
Manua 1
Manua 1
Throughput
(ga 1 lons/yr)
2,852,000
18,252,000
476,160,000
29,363,200
29,363,200
29,363.200
1
22,000
6.200,000
16,000
16,000
16,000
8,1400
8,<400
1400
<400
Days Before
Operation Tank Is Emptied
U batches/day
Continuous
Cont i nuous
Continuous
Cont i nuous
Continuous
2 discharges
to tank per day
Continuous
2 discharges
to tank per day
2 discharges
to tank per day
2 discharges
to tank per day
2 discharges
to tank per day
2 discharges
to tank per day
2 discharges
to tank per day
2 discharges
NA
NA
NA
NA
NA
NA
90
NA
90
90
90
90
90
180
180
to tank per day
to
Source: Operating parameters are based on engineering judgment.
-------
FXHIBII ?-M
OXIDAIlON/KfDUCIION/PHI CI PIIAIION IRIAIMINI IANK
Oxidation/Reducing
and Flocculation
Agents
Chemical
Feed System
Waste From
Storage
00
Rapid Mix Tank
Treated Waste
AL high level alarm
FR: flow recorder
LI level indicator
pH pH meter
Clerlfier
Sludge To Treotmpnt
-------
2-9
Following this reaction, another reagent may be added to enhance precipitation
of the waste constituents. We assumed that the pH-adjusted waste from the
first tank is pumped into the second tank at the same rate it is pumped into
the first tank. In this tank, an oxidizing or reducing agent that will render
some component of the waste non-hazardous is added. A flocculating agent is
also added. We assumed that no reactions or phase changes occur in the
reactor except for those pertaining to the destruction of the hazardous
constituents. Also, we assumed that destruction of the hazardous constituent
from both dissolved and suspended portions occurs and that destruction
products are non-hazardous. No appreciable volume change is assumed due to
the addition of reagent chemicals.
The third tank is used for clarifying the waste. This tank was modeled as
being eight times the size of the first tank to allow for a four hour
retention time. The waste containing the flocculating agent is pumped into
the third tank at the same rate that the waste is pumped into the first tank.
In the third tank the clarifier requires a four hour retention time to clarify
the waste. We assumed that all of the dissolved hazardous constituents are
removed to the suspended solids phase upon the addition of the flocculating
agent. We assumed that the effluent from the clarifier contains no suspended
solids and that all suspended solids and all coagulant chemicals are carried
out with the underflow waste volume. We assumed that the underflow waste
volume removes half of the liquid entering the third tank, and consequently
the discharge rates for the effluent and underflow are equal (i.e., each rate
is half the fill rate).
Distillation
We modeled one distillation treatment technology consisting of two carbon
steel tanks of equal volume. The first tank in the process is a 2,300 gallon
distillation tank. The waste (assumed to be 50 percent concentrated) is
pumped into this tank at a rate of 100 gallons per minute. We set the batch
volume equal to the operating capacity of the tank (80 percent of 2,300
gallons) and the fill time equal to the batch volume divided by the pumping
rate. Distillation of a batch was assumed to take four hours, resulting in a
treatment efficiency of 90 percent. Four batches were modeled for each
operating day. The condensed distillate (which is now pure) is subsequently
collected in an accumulation tank, which is assumed to be the same size as the
distillation tank. Residues are pumped out of the bottom of the distillation
tank at the end of the distillation process. The distillation tank system is
illustrated in Exhibit 2-5.
2.2.2 Storage and Accumulation Tank Systems
Storage and accumulation tanks are relatively simple tank systems,
compared to treatment tanks. A schematic diagram is presented in Exhibit
2-6. For this study we have defined storage tanks as tanks that are permitted
under RCRA requirements and are allowed to store wastes for more than 90
days. Accumulation tanks are tanks which are not permitted and store wastes
for less than 90 days. A storage tank may continuously contain some waste,
-------
IXMIBIT 2-5
OISI 111 ANON IRFAIMfNI TANK
Organic Vapor
Waste From
Storage
(containing
organics)
AL : high level alarm
LI : level indicator
SP : sampling point
Tl : temperature indicator
Batch
Distillation
Vessel
Condenser
Receiver Vessel
Steam
>
to
I
O
To Storoge
Sludge Removal From
Vessel To Storage
At End of Process
-------
EXHIBIT 2-6
SIGRACE TANK
.AL
r~^
LI
Waste
X!
Storage Tank
N>
I
AL: high level alarm
LI: liquid level indicator
Closed tank is also equipped with an overflow valve
-------
2-12
but an accumulation tank must be completely emptied at least every 90 days.
However, data indicate that, in actuality, both storage and accumulation tanks
are used to store waste for approximately 90 days. Because the designs of
these two types of tanks are so similar, they have not been modeled separately.
Storage and accumulation tank systems consist of three basic elements: a
fill and discharge system (either using pumps or gravity feed), piping, and
the tank itself. The tank may be above-ground, in-ground, or underground and
may vary in size and material of construction. The pipes may be above-ground
or underground (although we have assumed that underground tanks have
underground pipes and in-ground and above-ground tanks have above-ground
pipes).
Two types of storage and accumulation tank systems were modeled: batch
discharge systems and continuous discharge systems. For batch systems, we
assumed that two batches of waste are discharged to the tank each operating
day and that the tank is full when it is emptied every 90 days. The flow of
waste was assumed to be controlled manually. One accumulation/storage tank, a
210,000 gallon ongrade tank, was modeled as a continuous discharge system.
This tank was assumed to be a settling tank with a 5-day retention time. We
assumed that 20,000 gallons per operating day are pumped into and out of the
tank and that the tank is operated at a reduced capacity of 100,000 gallons.
This tank was also assumed to have manual level controls because minor
fluctuations of the liquid level are not crucial when the tank is operating at
50 percent capacity.
2.2.3 Small Quantity Generator (SQG) Tank Systems
Small quantity generator tanks are basically the same as storage and
accumulation tanks in terms of design. However, storage time has been assumed
to be 180 days, which corresponds to the maximum time allowed in the proposed
SQG regulations for on-site accumulation by SQGs without being required to
obtain interim status or a permit.*
Like storage and accumulation tank systems, SQG tank systems consist of a
fill and discharge system, piping, and a tank. We assumed that SQG tanks are
either located above-ground on cradles or underground, and are constructed of
carbon steel. Above-ground tanks have above-ground pipes and underground
tanks have underground pipes. Because they generate small amounts of waste,
small quantity generators typically utilize small capacity tanks. For
purposes of this study, all SQG tanks were modeled to be 200 gallons (see
Section 4.2 for a detailed discussion).
50 Federal Register 31278, August 1, 1985.
-------
2-13
2.3 PROTECTION AND DETECTION OPTIONS
In this section we describe the modeled protection and detection measures
associated with each tank technology. Basically, tank facility operators can
employ a number of design options and operating procedures to reduce the
possibility of a release from a tank system or to minimize the duration of a
release. We have incorporated a number of these preventive measures into the
HVTF model. These measures include:
Inventory control;
Emergency shut-off control systems;
Visual inspection;
Leak testing;
Secondary containment; and
Cathodic protection.
Once releases are detected, they are minimized by prompt replacement of
failed components. We assumed that if a leak in a particular component is
detected by a visual inspection, then that component alone is immediately
replaced. Similarly, when leaks are detected by leak tests or other
monitoring methods (e.g., inventory control), we assumed that only failed
components are replaced. Tank failures resulting from environmental
catastrophes, however, would entail replacement of all components. We
describe each of these measures in the following sections.
2.3.1 Inventory Control
We assume that, hazardous waste tank operators can detect major shortfalls
(i.e., leaks) in the quantity of stored waste from caused inventory (i.e.,
normal filling and discharging operations). We are uncertain of the
effectiveness of inventory control for detecting small leaks from hazardous
waste tanks. However, we assume that relatively large losses of tank liquid
would be noticed by an operator during normal operations.
We are uncertain of the percent of the tank operating capacity such that a
shortfall will be detectable by casual inventory monitoring. Nevertheless, we
have assumed that a loss of 25 percent or more of the tank contents would be
detected for treatment, storage, and accumulation tanks. The modeled SQG
tanks are considerably smaller than the other modeled technologies, and even
at 25 percent of the capacity of a SQG tank, shortfalls may not be detected.
Consequently, we have assumed that 100 percent of the SQG tank contents must
be released for a leak to be detected.
The algorithm we used to model detection over-simplifies the actual
detection process. In actuality, the detection of releases is likely to be
dependent on an operator's attention to tank operation (i.e., the operator
knows the time required to fill/empty the tank), leakage rate (i.e., if the
leakage rate is greater than the fill rate, the operator would notice that the
tank is not filling up), general inspection procedures (i.e. an open top tank
is likely to be inspected daily in order to prevent overflows), type of tank
-------
2-14
category (i.e. small quantity generators are in more populated areas (parking
lots) than treatment tanks where leaks are more likely to be noticed), and
tank technology (i.e., releases from above-ground tanks are more likely to be
detected than releases from underground tanks. Although our simple approach
does not explicitly incorporate all these factors, other types of modeled
inspections do incorporate some of these factors. For example, casual visual
inspections of above-ground tank components will result in the detection of
releases greater than 250 cmj/sec (see section 2.3.3). For all tanks,
except SQG tanks, we assumed 25 percent of the tank contents must be released
for a release to be detected. Although the detection threshold associated
with SQG tanks is probably less than 100 percent, we could not identify a more
reasonable threshold, and so we chose to use a conservative assumption of 100
percent. In general, for above-ground SQG tanks, releases were detected prior
to the release of 100 percent of the tank contents. However, for underground
SQG tanks, the modeled method for detection required that 100 percent of the
tank contents be released prior to detection.
These thresholds apply to the sum of all losses occurring during the
inventory period. For example, if a tank has two leaks, neither of which is
large enough to trigger detection, their sum may nevertheless exceed the
threshold level, and both leaks would then be detected. However, as casual
inventory monitoring is modeled, it is generally ineffective for treatment
tanks and continuous throughput storage or accumulation tanks, because the
large throughputs of these tanks mean that any leaks large enough to cause
detectable shortfalls would also be large enough to be immediately obvious to
the operator.
Different tank systems are monitored on different cycles. Open-topped
tanks are assumed to be reconciled once a month, while storage and
accumulation tanks are generally reconciled whenever they are pumped out.
Continuous throughput storage tanks (settling tanks) are monitored once a
month, even if they are closed-topped. The monitoring (inventory control)
cycles for each tank technology are shown in Exhibit 2-7.
As an example of how releases are detected by inventory control
monitoring, consider two hypothetical tank systems. One is a closed-topped
storage tank pumped-out semi-annually, the other an open-topped settling tank
pumped-out on a monthly basis. Assume that both tanks have a monthly
throughput of 50,000 gallons. For inventory control to detect a leak in the
first tank, the size of the leak would have to exceed .25 x 50,000
gallons/month x 6 months = 75,000 gallons over a six-month period. In the
second example, the size of the leak would have to exceed .25 x 50,000
gallons/month x 1 month = 12,500 gallons over a one-month period. Note that
the detectable leak rates for these two tanks are the same (in gallons per
month), but that the leak of the second tank would be detected much earlier
than the leak in the first tank.
2.3.2 Emergency Shut-off Systems
Emergency shut-off systems are a means of preventing overflow events by
stopping the inflow of waste into the tank. We modeled two types of emergency
shut-off systems: manual systems and automatic systems with manual backup
-------
2-15
EXHIBIT 2-7
INVENTORY CONTROL CYCLES
Tank Technology
Monitoring Cycle
TREATMENT:
on cradles, carbon steel, closed top
on cradles, carbon steel, open top
ongrade, carbon steel
in-ground, concrete 1 month
in-ground, carbon steel
in-ground, stainless steel
STORAGE/ACCUMULATION:
1 month
1 month
1 month
1 month
1 month
on cradles, carbon steel 90 days
ongrade, carbon steel 1 month
underground, carbon steel 90 days
underground, fiberglass reinforced plastic
90 days
underground, stainless steel 90 days
in-ground, concrete 1 month
in-ground, carbon steel 1 month
SMALL QUANTITY GENERATORS:
on cradles, carbon steel
underground, carbon steel
180 days
180 days
-------
2-16
systems. Manual systems consist of a. high level sensor and an alarm which
signals the operator to manually turn off the inflow pump or shut the inlet
valve if the high-level sensor is activated. Automatic shut-down systems
consist of a high-level sensor connected to a microprocessor which shuts down
the inlet pump and valve. If the inlet system fails to shut down, the
microprocessor adjusts the outlet pump and valve to increase the rate of
outflow.
We applied manual emergency shut-off systems to tanks with batch
processes, and applied automatic emergency shut-off systems to tanks with
continuous processes. These choices seem logical since an operator is
available to respond to an alarm for batch process tanks, while there often is
no operator present for continuous process tanks.
Both of the automatic and manual shut-off systems are subject to failure.
For tank systems with automatic shut-down systems (i.e., continuous treatment
processes), the high level sensor may fail or pumps and valves may not respond
to signals sent by the system to shut down. We assumed a failure probability
of 0.11 per year for automatic shut-off systems. This value is based on
available data for capacitance type level transducers. Manual shut-down
systems fail if an operator does not respond to an overflow alarm condition.
We assumed that the probability for operator error is 0.03, based on available
reliability data. It is also possible for the alarm to malfunction. The
failure probability of the alarm was assumed to be 0.1 per demand, also based
on available reliability data.
2.3.3 Visual Inspections
The Hazardous Waste Tank Failure Model simulates three types of inspec-
tions . One of these is a casual "walk-around" inspection of above-ground
portions of tank systems, which we assume occurs daily at all facilities.
Another is a more thorough weekly inspection of above-ground tanks and piping
which reflects current regulatory requirements.' In addition, we assumed
that during installation of tanks, inspections are conducted to detected any
deficiencies in the tank components. The probability of detecting a leak
depends upon the leak rate, and is different for each type of inspection. The
general assumptions we used to model the effectiveness of each type of visual
inspection are the following:
'For the baseline scenario, we modeled above-ground SQG tanks to be
inspected only monthly, because SQG tanks are not subject to current
regulations. For the other regulatory scenarios, we modeled above-ground SQG
tanks to be inspected weekly.
-------
2-17
Casual inspection. We assumed that casual inspections will
always be sufficient to immediately detect a leak larger than a
running faucet (approximately 250 cm3/sec). For leaks with rates
between 1 drop per second (approximately .05 cm3/second) and 250
cm3/second, we have assumed the probability of detection is 50
percent. Smaller leaks will not be detectable by this method.
Weekly inspection. Weekly inspection is modeled to have a 95
percent chance of detecting any above-ground leak (and a 100 percent
chance of detecting leaks of over 250 cm1/sec).
Installation inspection. Installation inspections are simulated
when tanks or tank components are replaced. Under all regulatory
scenarios, except leak testing combined with groundwater monitoring,
we simulated installation inspections to consist of a visual
inspection and weld testing. We assumed this procedure to be 75
percent effective in detecting a damaged component. Under the
regulatory scenario for leak testing combined with groundwater
monitoring. We assumed that, in addition to visual inspection and
weld testing, leak testing would also be conducted after tank
installation. This inspection procedure was assumed to catch 95
percent of installation defects.
2.3.4 Leak Testing
We simulated leak testing, in combination with ground-water monitoring,
for only one of the regulatory scenarios that we are analyzing. Leak testing
is applicable only to underground tanks and is usually performed by a
contractor experienced in the procedure. Currently, several methods are
commercially available, although their applicability to hazardous waste tanks
is not well documented. As a result, for this analysis we assumed that the
tank is emptied, cleaned, and filled to capacity with water. The water in the
tank is then monitored by a sensitive device which records any changes in the
hydrostatic pressure at a fixed depth within the tank.
We simulated leak testing every six months. This six month frequency of
testing is consistent with the time period prescribed in the Agency's proposed
regulations. Leak testing was modeled to be 100 percent reliable in detecting
leaks that exceed a pre-determined threshold. We used a threshold of Q.I
gallons per hour. When a leak is detected, we assumed that the tank is
replaced.
2.3.5 Secondary Containment
Our assumptions concerning secondary containment for the full and partial
secondary containment regulatory scenarios are shown in Exhibit 2-8.
-------
2-18
EXHIBIT 2-8
SECONDARY CONTAINMENT ASSUMPTIONS
Containment Method
Tank Location
Underground
In-ground
Above-ground,
on cradles
Above-ground,
ongrade
Full Secondary Containment
Scenario
Tank: double-walled
Pipe: double-walled
Tank: double-walled with
concrete pad and berm
surrounding tank
Tank: concrete pad and berm
Pipe: concrete-lined trench
Tank: concrete pad and berm
(pad underneath tank)
Pipe: concrete-lined trench
Partial Secondary
Containment Scenario
NA
Tank:
Pipe: concrete-lined trench Pipe:
concrete pad and
berm surrounding
tank
concrete-lined
trench
Same as secondary
containment.
Tank: concrete pad and
berm (pad not
underneath tank)
Pipe: concrete-lined
trench
-------
2-19
Secondary containment systems are designed to prevent the release of fluids
from the primary containment system (the tank) to the surrounding
environment. Although other containment options are available, we believe
that the ones we selected are representative secondary containment methods
that would reflect future practices. In general, underground components and
in-ground tanks are assumed to be double-walled, and above-ground components
are assumed to have a concrete pad and berm (tanks) or concrete trench (pipes)
beneath them. Ongrade tanks are assumed to be built on top of a concrete pad.
In the case of double-walled tanks and pipes, we simulate continuous
monitoring of the interstitial space by means of a conductivity-type sensor.
We assumed that the liquid sensor has a 90 percent chance of detecting a leak
of any size and can detect breaches in both the inner or outer walls of the
tank and pipe components. Releases contained by above-ground secondary
containment are detected during routine daily inspections.
Although secondary containment systems are designed to prevent all
releases from the primary containment from entering the environment, there is
still a chance that releases may occur. Double-walled tanks and pipes are
still subject to the same corrosive forces as single-walled tanks and pipes.
Howver, because the interstitial monitoring device can detect liquids entering
from a leak in the primary or the secondary containment wall, a release occurs
only if the monitoring device and both walls fail simultaneously.
Above-ground concrete pads and berms are subject to cracking and deterioration
with age. If a concrete pad and berm system is in a failed state at the time
of a primary containment failure, we assumed that the release volume is
equivalent to that from a tank system with no secondary containment. In
addition, external catastrophes may also breach secondary containment.
External catastrophes which may breach secondary containment are the same as
those which may breach the tank system itself. Because these events usually
involve serious structural damage to the entire tank facility, we have assumed
that whenever such catastrophes breach one containment system, they also
breach the other. Therefore, release volumes associated with catastrophic
failures are the same for tanks with and without secondary containment.
2.3.6 Cathodic Protection
Cathodic protection is a method to prevent corrosion of metal tank and
pipe systems in contact with the soil. We modeled the impressed current type
of cathodic protection because it can be retrofitted without excavation of the
tank. For this type of protection, underground portions of tanks and
ancillary pipes are connected to a continuous electrical current to prevent
the electrochemical reactions leading to corrosion from occurring. Properly
installed and maintained, cathodic protection systems should prevent corrosion
indefinitely. However, they require regular monitoring, and prompt repair or
replacement of failed components.
According to the National Association of Corrosion Engineers, the leading
cause of cathodic protection failure is failure to conduct scheduled
maintenance. For this reason, we simulate the quality of maintenance using a
-------
2-20
random multiplicative factor. We simulated the year of failure of cathodic
protection systems using an assumed normal distribution with a mean of 10
years and a standard deviation of five years. The multiplicative factor is a
random number between one and three. A value of one means that no maintenance
is undertaken; a value of three indicates that the operator carefully follows
the prescribed maintenance schedule. This number is multiplied by the sampled
time to failure,- and serves to delay failure for well-maintained systems.
Once cathodic protection fails, the tank begins to corrode in the same
manner as a new, unprotected tank. It should also be noted that cathodic
protection can only prevent exterior corrosion; interior corrosion is not
influenced by cathodic protection systems.
2.4 CHARACTERIZATION OF SURROUNDING SOIL
The soil in contact with a tank influences the rate of corrosion of tank
components in contact with the soil, and the leak rates of contaminants
released from underground tank system components (e.g., piping). We
simplified our approach by modeling only one representative soil type. We
assumed that the soil in contact with underground and on-grade tank systems
consists of a uniform sand. Sand was chosen because it is an abundant
subsurface material, comprising glacial deposits, alluvial materials, and
coastal plain sediments. Consequently, it is typical of many areas of the
country. We also assumed that sand is used as the backfill material for
underground tanks, which is a common practice for tank installations.
Leak rates from underground tanks and pipes are affected by several
different soil parameters, including the void fraction of the particles, soil
particle diameter, the sphericity of the soil particles, and the distance
required for leaking fluid to disperse to a point at which its pressure equals
that of the pore space in the surrounding soils (dispersion length). The
basic soil parameters were readily available from hydrogeologic texts, but
dispersion lengths were not available for underground leaks. Thus, we had to
estimate dispersion lengths based on best engineering judgment.
For purposes of determining corrosion rates, sandy soil can be classified
as benign, or, non-aggressive. There are four basic parameters that determine
the aggressiveness of soil: pH, the presence of suIfides, soil moisture, and
resistivity. We assumed that the soil surrounding underground tanks and pipes
is a benign soil that has a pH of 6 (slightly acidic), has no sulfides, is
unsaturated, and has a resistivity of 15,000 ohm/cm. A summary of the soil
parameters we used is given in Exhibit 2-9.
2.5 CHARACTERIZATION OF WASTE STREAM PHYSICAL PROPERTIES
The physical properties of waste streams handled in tanks influence the
leakage rates for releases from underground, ongrade, and in-ground tanks. In
order to reduce the number of necessary HWTF Model runs, we categorized the 32
model waste streams (i.e., the waste streams we have selected for this
-------
2-21
EXHIBIT 2-9
SOIL PARAMETERS FOR SAND
Parameter Value
Void fraction .53 (dimensionless)
Particle diameter .25 millimeters
Sphericity .65 (dimensionless)
Dispersion length (cm)
circular holes a/ 20 x hole diameter
elongated cracks b/ 40 x crack width
pH 6
Sulfides none
Resistivity 15,000 oho/cm
a/ Dispersion length limited to a maximum of 20 centimeters.
b/ Dispersion length limited to a maximum of 40 centimeters.
-------
2-22
study) according to their physical properties.7 Because viscosity and
density are the physical properties that influence leakage rates, we chose two
representative waste streams that differ in these properties and are
representative of the waste streams we examined in this study. We selected
toluene and aqueous solutions of heavy metals and/or cyanide to represent the
range of physical properties associated with the 32 model waste streams. We
chose toluene and aqueous solutions because a large portion of the wastes
included in the Tank Risk Analysis Model are either aqueous or are comprised
of mixtures of toluene and other organic solvents with similar viscosities and
densities. The viscosities and densities of these two waste streams are shown
in Exhibit 2-10.
For each model underground, in-ground and ongrade tank, the HWTF Model
simulated two sets of 250 representative release profiles, one set
representative of tanks handling aqueous waste streams and the other set
representative of toluene.' Each waste stream modeled in the Tank Risk
Analysis Model was classified as being more similar to either the generic
aqueous or toluene waste stream. Thus, the risk estimates for a particular
waste stream are based on representative release profiles for either the
aqueous or toluene waste stream.
2.6 EVENTS THAT LEAD TO THE RELEASE OF CONTAMINANTS
After identifying the classes of tank systems and the release prevention
and detection technologies to model, we then identified the various possible
failure events. As a preliminary step in this process, we reviewed the
relevant technical literature for information on system and component
reliabilities and failure modes. This review focused on the causes,
consequences, and probabilities of tank system failures.
Our literature search used the following sources and data bases:
Pope-Reid Associates' (PRA) in-house library;
University of Minnesota libraries;
Computerized Engineering Index (a data base containing
engineering abstracts);
7Ideally, we would have run the HWTF Model for each of the 32 waste
streams we have selected for this study.
'Because we modeled all waste streams associated with the distillation
treatment process to be similar to toluene, only one set of HWTF results were
generated for the on cradles distillation treatment tank. Similarly, we
modeled all waste streams associated with the oxidation/reduction/precipitation
process to be similar to the aqueous waste stream. Consequently, only one set
of HWTF Model results were generated for each treatment tank technology.
-------
2-23
EXHIBIT 2-10
GENERIC WASTE STREAM PROPERTIES
USED FOR LEAK RATE CALCULATIONS
Dynamic
Density a/ Viscosity a/
Generic Waste (kg/mj) (centipoise)
Aqueous
998.21
1.00
Toluene
866.90
0.59
a/ At 20°C.
-------
National Technical Information Service (computer
search); and
Pollution Abstracts (computer search).
From this literature search, we identified five general categories of
failure mechanisms:
(1) Stresses due to unusual variations in the internal
environment of the tank;
(2) Stresses due to the external environment of the tank;
(3) Design flaws;
(4) Construction and installation errors; and
(5) Operation or maintenance errors.
The possible effects of various internal environmental stresses are
tabulated in Exhibit 2-11. This exhibit indicates that internal environmental
factors contributing to failure include extremes of temperature, pressure, and
pH, as well as variations in suspended solids, fluid viscosity, or specific
gravity. Such internal stresses may result in accelerated corrosion,
equipment overload, or failure of control or monitoring components. The
affected components include tanks, pipes, pumps, valves, meters, and other
ancillary equipment. Each type of internal stress, however, does not affect
all components. For example, while pH extremes do affect all of the listed
components, specific gravity variations only affect pumps, mixers, and flow
meters.
External environmental stresses are tabulated in Exhibit 2-12. They
include power fluctuations, exterior temperature extremes, high winds,
humidity extremes, excessive rainfall, adverse soil conditions, earthquakes,
poor air quality, and vandalism. These stresses may result in corrosion,
rupture, spills, or control failure. The specific components affected by each
type of stress are identified in the table.
Types of design, construction, and operation errors are tabulated in
Exhibit 2-13. Such errors include poor material selection, use of improper
components, inadequate structural support, improper process design,
installation damage, use of off-spec materials or components, operator error,
or improper maintenance. These errors may cause structural collapse, rupture,
accelerated corrosion, equipment malfunction, inadequate capacity, improper
process control, or failure to detect on-going releases. The specific
components vulnerable to each of these failures are identified in the table.
Our findings indicate that the releases from tanks are due to a wide range
of events. Because we do not know with certainty the sequence and consequence
(i.e., magnitude of releases) of failure events that will occur at hazardous
waste tanks, we used the HWTF model to simulate the sequence of events at
-------
EXHIBIT 2-11
TANK SYSTEM FAILURE MATRIX: INTERNAL ENVIRONMENTAL STRESSES
AFFECTED COMPONENTS
pH Meters,
Above- In- Under Above- Under ORP Meters, &
Factors Contributing Ground Ground Ground Ground Ground Heat Temperature Conductivity Level Flow
To Failure Type of Failure Tanks Tanks Tanks Pumps Valves Piping Piping Mixers Ex changer a Gauges Metera Indicators Meter
Temperature extremes Corrosion; inac- ******** * * * *
curate system
control
Pressure extremes Rupture or **** * *
collapse; inac-
curate system
control
pH extremes Corrosion; inac- »*** * * *
curate system
control
Excess amounts of Abrasion; corro- ******** * * * * *
solids sion; inaccurate
system control
Viscosity variations Equipment over- * * * * * "
load; inaccurate
system control
Specific gravity Equipment over- * * *
load; inaccurate
system control
* Signifies an effect upon the particular component or upon its operation.
Source: Based on literature search.
-------
EXHIBIT 2-12
TANK SYSTEM FAILURE MATRIX: EXTERNAL ENVIRONMENTAL STRESSES
AFFECTED COMPONENTS
pH Meters,
Above- In- Under Above- Under ORP Meters, 6
Factors Contributing Ground Ground Ground Ground Ground Heat Temperature Conductivity Level Flow
To Failure Type of Failure Tanks Tanks Tanks Pumps Valves Piping Piping Mixers Exchangers Gauges Meters Indicators Meter
Inconsistent power
Temperature
High winds/storms
Humidity extremes
Excess rainfall
Adverse soil
characteristics
Earthquakes
Overflow; vari-
ation or loss
of control
Cracking; corro-
sion; inaccurate
system control
Rupture; spills;
overturning
Corrosion, inac-
curate system
control
Overflow; short
circuits; inac-
curate/loss of
system control;
overturning
Corrosion
Rupture; over-
turning; spills
Adverse air quality Corrosion
Vandalism/
unauthorized entry
Other catastrophic
event
Rupture; overflow *
ignition; inac-
curate/loss of
system control
Complete system *
failure
* Signifies an effect upon the particular component or upon its operation.
Source: Based on literature search.
-------
EXHIBIT 2-1J
TANK SYSTEM FAILURE MATRIX: DESIGN, CONSTRUCTION, AMD OPF.IiAT ION KHHOHS
AFFECTED COMPONENTS ^
pll Meters,
Above- In- Under Above- Under ORP Meters, ».
Factors Contributing Ground Ground Ground Ground Ground l\nat Temperature Conductivity Uw..| F'low
To Failure Type of Failure Tanks Tanks Tanks Pumps Valves Piping Piping Mixers Exchangers Gauges Mnters In.lic.il .>i :-, M<-t ei
DESIGN
Poor Material Collapse or « »
election rupture; corrosion
Poor component Undersized; inap- «
selection proprlato or
Inadequate
Inadequate struc- Non-support; »«. . . .
tural specification structural
weakness
Inadequate process No or Inade- * *
design quate treatment
of waste
CONSTRUCTION/INSTALLATION
Damage during Cracks, scrapes, « «
installation dents, etc.
Off-spec materials Corrosion; rup- . .
Installed ture or collapse
Off-spec component Undersized; ......
Installed. Inadequate or
malfunctioning
Expected service Equipment mal- ........ * « « . .
life function/failure
OPERATION/MAINTENANCE
Operator fails to Overflow; Internal * «
control process environmental
extremes; failure
to detect on-going
release
Operator deliberate- Overflow; Internal
ly ignores safety environmental ex-
or control measures tremes; failure to
detect on-going
release
Improper cleaning. Equipment mal- ........ . . . . .
Inspection, observa- function/failure;
tlon, recordkeelng failure to detect
or replacement on-going release
-------
2-28
tanks. The HVTF model accounts for such random elements by simulating 'the
timing and magnitude of failure events. For each type of failure, we have
assigned probability distributions to those parameters which are subject to
random variation (e.g., initial diameter of a corrosion hole or length and
width of a rupture). These parameters are then used to determine
the timing of the failure events and the magnitude of the releases resulting
from the failure event. In general, the magnitudes of release volumes are
determined by simulating relevant parameters (such as hole size) and using
these parameters as input to a deterministic, mathematical leak rate equation.
In the following sections, we provide a general overview of the
assumptions we used to simulate the timing and magnitude of release volumes
for the major failure events (corrosion, ruptures, catastrophic events,
overflows, and spills). In the last section, we describe the methods we used
to model steady state losses from concrete tanks.1 More detailed
discussions of the distributions we used to simulate the failure events are
provided in a separate report.18
2.6.1 Catastrophic Events
Catastrophic events may result from airplane crashes, high winds (tornadoes
or hurricanes), earthquakes, floods, ignition of the waste, or damage from
nearby fires or explosions. However, not all of these events are possible for
any given system design; non-flammable wastes cannot burn, underground tanks
are less susceptible to damage by explosions, and not all geographic regions
are vulnerable to all of the other hazards. Thus, these events are controlled
by the choice of waste streams and a set of geographic variables.
For this study, we modeled four types of catastrophic events: tornadoes,
vandalism, nearby fires or explosions, and ignition of the waste within the
tank (for toluene only, which is ignitable). Of course, not all of these
events are possible for all system designs. For above-ground and inground
tanks, we modeled all four of these events; however, for underground tanks,
tornadoes and vandalism were disregarded, since the soil cover acts as a
protective shield. In addition, we assumed that catastrophic damage to
underground tanks due to nearby fires or explosions is only one-third as
probable as for above-ground and in-ground tanks. This reduced probability is
because of the protection provided by the soil cover. Information on
catastrophic events is summarized in Exhibit 2-14.
'We assumed that all concrete tanks are porous. Consequently, steady
state losses occur from concrete tanks throughout the tank's operating life.
10Pope-Reid Associates, Inc., Hazardous Waste Tank Failure Model;
Description of Methodology, draft report to Office of Solid Waste, EPA,
January 13, 1986.
-------
2-29
EXHIBIT 2-14
CATASTROPHIC EVENTS MODELED FOR DIFFERENT TANK TYPES
Above-ground In-ground Underground
Event Tanks Tanks Tanks
Tornado X X
Vandalism X X
Nearby fire or explosion X X X a/
Ignition of the waste
(toluene only) X XX
a/ We assumed that the annual probability of nearby fire or explosion for
underground tanks is one-third of the annual probability for above-ground
and in-ground tanks.
-------
2-30
We assumed that these types of events cause the sudden loss of 50 percent
of the volume of storage, accumulation, and SQG tanks, and 80 percent of the
volume of treatment tanks. For storage, accumulation and SQG tanks, we
assumed that on average, over a full year period, these tanks operate at half
capacity. We assumed a higher loss for treatment tanks because they are
usually operated near to their full capacities.
2.6.2 Corrosion of System Components
Corrosion of tanks and pipes involves two basic types of corrosion:
localized corrosion and generalized corrosion. Either of these types of
corrosion can occur on the interior and exterior surfaces of a tank or pipe.
The model accounts for each corrosion mechanism (i.e., interior localized
corrosion, exterior localized corrosion, interior generalized corrosion,
exterior generalized corrosion) by calculating the time to occurrence of a
hole based on a simulated corrosion rate and the assumed tank or pipe wall
thickness. For example, if a localized exterior corrosion pit is simulated to
grow in depth at a rate of 25 mils per year and the wall thickness is .25
inches, after 10 years the pit will perforate the wall. As corrosion
progresses, the hole enlarges and the leak rate increases until the loss is
detected and the component is replaced. In many cases, however, several
corrosion mechanisms act at once, resulting in a hole in the tank or pipe wall
much sooner than from any corrosion mechanism acting alone. For instance,
generalized interior and generalized exterior corrosion may simultaneously
reduce the wall thickness while a localized exterior corrosion pit enlarges.
In this case, corrosion rates could be combined, resulting in a much faster
corrosion rate.
Release volumes from corrosion holes in pipes and tanks are estimated by
using one of two equations, depending on whether the corrosion holes are
located in above-ground or underground components.11 Above-ground leak
rates are dependent on the following factors: the area of the crack or hole,
the acceleration due to gravity, and the equivalent static hydraulic head
(i.e., the equivalent height of the fluid surface above the hole, taking into
account pressure additions due to pumps or losses due to fluid flow through
pipes). The leak rate for underground tanks is different from that for
above-ground tanks because the soil will impede the flow of fluid,
dramatically reducing the leak rate. Thus we modeled the leaks from
underground tanks to be dependent on the following factors: size of the hole,
porosity, density of the released fluid, sphericity of the soil particles,
average diameter of the soil particles; the pressure drop between the inside
of the hole and the surrounding soil; and the distance required for leaking
fluid to disperse. The total release volume for a given leak incident is
"For a more detailed explanation of these equations see, Pope-Reid
Associates, Inc., Hazardous Waste Tank Failure Model: Description of
Methodology, draft report submitted to Office of Solid Waste, EPA, January
13, 1986, pp. 161-165.
-------
2-31
calculated as the product of the leak rate and the duration of the leak. The
latter varies according to the simulated effectiveness of modeled detection
and monitoring options.
2.6.3 Ruptures of Tank Components
Ruptures may result from one or two sources: undetected installation
damage, including latent defects; and normal operating hazards. Normal
operating hazards may produce ruptures at any time during the life of the tank
system; these hazards include strains caused by settling, vibration,
temperature variations, and vehicle collision. Undetected installation
defects will result in immediate failures (i.e., the tank or pipe is installed
in a ruptured condition that is not detected by a visual inspection).
We have assumed that steel and stainless steel are equally likely to
rupture; stainless steel is not significantly stronger than ordinary carbon
steel. Based on discussions with contractors, however, we have concluded that
fiberglass is approximately twice as likely to rupture as is steel. We have
therefore applied this ratio to our rupture and installation damage
probabilities for both tanks and pipes.
Rupture mechanisms are different for above-ground and underground
systems. Above-ground ruptures are most likely to occur due to vehicle
collisions, collisions with fork lifts, freeze/thaw cycles, or latent flaws in
design, fabrication, or installation. Below-ground ruptures are most likely
to result from settling, latent defects, or the driving of heavy equipment
across the site. These rupture mechanisms are different, but we have assumed
that their combined probabilities are approximately equal for above-ground and
underground systems.
In addition, we have assumed that rupture probabilities are independent of
tank capacity and pipe length. We made this assumption because components are
designed to accommodate the normal range of operating stresses. Thus, design
strengths should increase with tank capacity and pipe length. We assumed that
the increased design strength for larger components cancels out the increased
stresses to which they are subject, so that rupture probabilities are
independent of system design.
For double-walled tank system components, we modeled the inner and the
outer walls separately, but we assumed that there is a 50 percent probability
that a breach of the outer wall will also breach the inner wall. Because the
inner wall is subject to fewer stresses than the outer wall, we assumed that a
rupture is less likely to begin with the inner wall than with the outer wall.
In addition, because operating pressures in hazardous waste tanks are
generally low, we assume that only an insignificant fraction of ruptures
initiating with the inner wall also breach the outer wall.
Ruptures include both large cracks and small seam failures. The
corresponding release volumes associated with these types of failures are
computed in the same manner as for corrosion holes, except the hole size does
-------
2-32
not increase with time. Since ruptures are often associated with larger hole
sizes then those due to corrosion, higher leak rates are estimated for these
failures. Consequently, we include a mass balance check to insure that the
loss from a given rupture (or concurrent ruptures) does not exceed the
capacity of the tank system (or throughput).
2.6.4 Overflows
Four conditions must be met before an overflow can occur:
The tank must be almost full to capacity such that an
overflow is possible;
There must be a control error resulting in an attempt
to add too much fluid to the tank;
There must be a failure of the emergency shut-down
system; and
There must be a route by which overflowing fluid may
escape from the tank (e.g. open top, vent).12
Clearly, a tank must first be full enough for an overflow to be possible;
a half-empty 10,000 gallon tank, for example, cannot overflow by adding a
1,000-gallon batch. In general, we have assumed that potential overflow
situations occur at any time for continuous treatment processes, because fluid
is regularly added when the tank is already filled to its normal operating
depth. We assumed that overflows are possible once per batch for batch
treatment (distillation) processes because the operator is adding fluid from a
storage tank which we assume to have larger capacity than the treatment tank
itself. We assumed that overflows have the potential to occur only once per
year for storage tanks, because storage tanks are generally filled in
comparatively small batches and pumped-out before they get dangerously full.
We made the conservative assumption that once per year there is a fluctuation
in waste-generation rates or a failure to follow the pump-out schedule,
thereby allowing the tank to become full.
Once a potential overflow situation develops, there must also be a control
error before a tank can overflow. Such an error would occur if a malfunction
in a level indicator caused too much fluid to be added to the tank (either in
automatic or manual systems), if an automatic level controller failed, if the
operator of a continuous system made an error in the morning start-up routine,
or if the operator of a manual system ignored the liquid level indicator and
added too much fluid.
I2lf no route is available, the overflowing fluid backs up in the piping
system.
-------
2-33
However, even if a control error occurs, an overflow will not occur unless
the emergency shut-down system fails. (See Section 2.3.2 for a description of
the modeling of emergency shut-down systems.) The final requirement for an
overflow to occur is that there be a route by which the excess fluid may
escape. Open-topped tanks, for example, need no other overflow route, and
pump-fed tanks can overflow through their vents.11 Closed, gravity-feed
tanks, however, are less subject to overflow, for the model assumes that their
vent pipes open at a level above the highest possible fluid level in their
fill tanks.lu For such tanks, overflow can only occur through corrosion
holes or ruptures in the fill pipe, the vent pipe, the outlet pipe or valve,
or through faulty flanges or gaskets.
When an overflow occurs, the overflow rate is equal to the fill rate. For
above-ground and inground tanks, the model estimates a detection time randomly
selected from a uniform distribution with a lower bound of zero and an upper
bound equal to the fill time. For underground tanks, if there is not a vent
failure, the fluid simply backs up in the fill pipe and the tank does not
overflow. If the vent does fail, the volume of the overflow is assumed to be
3.3 gallons, which is the model tank system's volume the vent pipe will
contain.
2.6.5 Spills
*
Spills occur during filling and discharging and result when the operator
fails to close a pump or strainer drain after maintenance, or when a portable
flexible hose ruptures or is improperly connected. Which of these spill
mechanisms is possible depends on the details of the system design. Treatment
systems, for example, do not use portable hoses for fluid transfer, nor do
they use strainers. Their pumps, however, are maintained annually, and there
is a chance that the operator will forget to close a pump drain. This type of
maintenance error will produce a spill the next time the pump is used.
Storage, accumulation, SQG tanks, however, are pumped out using a flexible
hose. Spills are possible if the hose ruptures or is improperly attached.
On-site pumps and strainers were not modeled for storage, accumulation, or SQG
tanks, because these pieces of equipment (if they are present) are generally
located at the upstream end of the fill pipe of the tank, and are more closely
associated with the operator's process equipment than they are with the
storage or accumulation tank.
For spills resulting from failure to close the pump drain, the volume is
estimated as the product of the discharge rate and a random detection time
sampled from a uniform distribution with a lower bound of one minute and an
upper bound of three minutes. For spills resulting from hose ruptures or
loose connections, the leak rate is sampled from a uniform distribution with a
11A11 oxidation/reduction/precipitation treatment tanks and in-ground
tanks are assumed to be open-topped tanks.
1(1 All underground tanks are closed, gravity-feed tanks.
-------
2-34
lower bound of zero and an upper bound- of the discharge rate. If the rate is
greater than 0.375 gallons per minute, the detection time is randomly sampled
from a uniform distribution with a lower bound of one minute and an upper
bound equal to 25 percent of the discharge time. If the rate is between 0.002
and 0.375 gallons per minute, the detection time is randomly sampled from a
uniform distribution with a lower bound of one minute and an upper bound equal
to the discharge time. If the rate is less than 0.002 gallons per minute, the
failure is undetected and the rupture leaks for the entire discharge time.
The two rate levels (0.375 and 0.002) were determined using engineering
j udgment.
2.6.6 Steady State Losses from Concrete Tanks
We have assumed that concrete tanks are porous and will, therefore,
gradually release waste to the environment throughout their operating life.
The steady state leakage rate is given by:
(pwaste)
Q = A(l + d/t)K
awaste p^Q
where:
Q = the rate of leakage
A - the surface area of the tank (sides and bottom only)
d = the average fluid depth in the tank
t = the thickness of the concrete (8 inches)
-9
K = the permeability of the concrete to water (2.5 x 10
cm/sec)
ct-j _ = the viscosity of water
H20
o = the viscosity of waste
waste
p = the density of the waste
P-a n - the density of water
H20
Using this equation, we find that a 10,000 gallon unlined concrete tank
with an average aqueous fluid depth of nine feet will have a leak rate of 5.2
gallons/year. This leak rate will continue throughout the operating life of
the tank.
In this chapter, we have presented the type of phenomena that the HWTF
model simulates. Using assumptions about the timing of events, the model
simulates 250 potential release profiles (i.e., the annual release of
contaminants from the tank) associated with a particular tank technology and
-------
2-35
regulatory scenario. We used the model to simulate approximately 73 sets of
potential release profiles.15 From each set of 250 potential release
profiles, a subset was selected to be used to calculate the human health risk
estimates.14 The next chapter presents how we used this subset of release
profiles to estimate human health risks.
11All of the 22 model technologies are not applicable to each of the
regulatory scenarios. For example, concrete tanks are not analyzed to
evaluate the corrosion protection scenario.
"Because using all 250 release profiles to generate risk estimates for
each tank, ground-water setting, and regulatory scenario is a computationally
intensive task, we did not have the Tank Risk Analysis Model generate a risk
estimate for each of the 250 release profiles. See Section 4.5 for a
description of the methodology we used to determine a subset.
-------
CHAPTER 3
TRANSPORT, EXPOSURE, AND RISK MODELS
In this chapter, we discuss the component models of TRAM that are used to
estimate the subsurface transport of contaminants released from hazardous
waste tanks, human exposure to these contaminants via ingestion of
contaminated ground water, and individual human health risks resulting from
these exposures. The exposure pathway in this analysis consists of three
major components:
Transport of contaminants through the unsaturated zone;
Transport of contaminants through the saturated zone; and
Human exposure via consumption of contaminated ground
water.
Simulation of this exposure pathway through the use of component models
within TRAM allows individual exposures to contaminants released from tanks to
be calculated, which in turn allows lifetime chronic non-carcinogenic and
carcinogenic risks to be estimated for the exposed individual. The transport,
exposure, and risk models contained in TRAM have been adapted for the purpose
of this analysis from EPA's Liner Location Risk and Cost Analysis Model1 (a
model that compares risks and costs associated with various land disposal
options for hazardous wastes). TRAM, like the Liner Location Model, is a
non-steady-state model that allows yearly, time-varient exposures and risks to
be estimated.
In this chapter, we describe our adaptation of these models to TRAM, and
discuss their use within the overall framework of the analysis. Section 3.1
describes the unsaturated zone transport model used in TRAM, while Section 3.2
outlines the model used to simulate transport of contaminants in the saturated
zone. Section 3.3 describes the exposure/risk model used in the analysis.
3.1 CONTAMINANT TRANSPORT IN THE UNSATURATED ZONE
In TRAM, contaminant transport in the unsaturated zone is estimated using
the McWhorter-Nelson Wetting Front Model. This model represents contaminant
transport by the movement of a wetting front generated by the release of
liquid from the tank system. A wetting front is defined as the transition
zone between soil layers with different water contents. The travel time for
the wetting front is related to the difference in water content in the layers
above and below the wetting front, the distance travelled in the unsaturated
1Liner Location Risk and Cost Analysis Model, draft report to the
Office of Solid Waste, EPA, January 1985.
-------
3-2
zone, the leakage rate (Darcian flux), and the retardation of contaminant flow
by soil adsorption. The model is analytical (i.e., has a unique solution) and
travel time is estimated using the following equation:
t = (L/q) (n - 9r)(q/k)X/(2 * 3 X) (3-1)
where:
t = travel time from bottom of tank to water table
L = distance traveled by wetting front (i.e., distance between
tank and water table)
q = leakage rate through tank (length/time)
n = soil porosity (dimensionless)
water content below wetting front (dimensionless)
hydraulic conductivity (length/time)
x = pore-size distribution index (dimensionless)
Contaminant transport in the unsaturated zone, as modeled by the
McWhorter-Nelson model, is dependent not only upon properties of the aquifer
but also upon source-specific information (i.e., amount of material released
and physical/chemical properties of the released contaminants). In TRAM, we
account for the effects of soil parameters by assuming that all tanks are
located over a generic unsaturated zone that is assumed to be homogeneous and
isotropic. Soil properties that affect flow (porosity, permeability, and
pore-size distribution) are constant throughout the unsaturated zone. The
generic unsaturated zone is assumed to be five meters in thickness, and soil
moisture below the wetting front is considered to be constant. Also, we
assume that the fractional organic carbon content (Foe) of the unsaturated
zone remains constant throughout. Values for all of the unsaturated zone
parameters used in the model are presented in Exhibit 3-1.
Liquid volume flux in the model is equated to the average annual volume of
waste released from the tank (i.e., the total volume released divided by 20
years). This average liquid flux rate is somewhat of an over-simplification
since in reality liquid release volumes will vary from year to year, and thus
will affect unsaturated zone travel times. However, the model, as used within
TRAM, is unable to account for year-by-year "pulses" of released liquid. In
addition, this averaging has little effect on the overall risk estimate.
Although hazardous waste tanks may contain organic constituents that may
volatilize during transport through the unsaturated zone, we do not consider
this phenomenon in our analysis. For waste constituents with high vapor
pressures, volatilization may be an important process that would result in
-------
3-3
EXHIBIT 3-1
PHYSICAL/CHEMICAL PARAMETERS FOR
GENERIC UNSATURATED ZONE a/
Parameter Model Value
Depth (m) -5
Porosity (dimensionless) 0.25
-3
Hydraulic Conductivity (m/yr) b/ 5 x 10
Pore-size Distribution Index 3.5
(dimensionless)
Fractional Organic Carbon Content 0.02
(dimensionless)
Bulk Mass Density (g/cm1) 2.0
Soil Moisture (dimensionless) c/ 0.1
a/ Unsaturated zone is assumed to be homogeneous, isotropic sand.
b/ Saturated hydraulic conductivity.
c/ Soil moisture below wetting front.
-------
3-5
Thus, all relevant chemical and biological processes (e.g., chemical
hydrolysis, aerobic degradation) are lumped together into the first-order rate
constant, k. In reality, chemical and biological reaction rates are very
site-specific and depend on parameters such as temperature and availability of
oxygen. However, we are unable to account for these site-specific
dependencies in our analysis, and therefore use specified rate constants
(derived from a number of literature estimates) in all situations.
Information on degradation in the subsurface environment, however, is not
available for many hazardous chemicals; we assume that chemicals for which
data are unavailable are infinitely persistent.
In addition to the model assumptions noted above, we also assume that no
interactions between contaminants occur in the unsaturated zone (contaminant
interactions could decrease travel time due to the mobilization of hydrophobic
organics by water-miscible organic solvents or an immiscible solvent phase).
Also, all adsorption reactions are considered to be fast and reversible, and
adsorption is considered to be linear over the range of contaminant
concentrations seen in the analysis. Finally, we assume that no dispersion in
any direction occurs during transport through the unsaturated zone.
Although these assumptions tend to oversimplify transport and attenuation
processes occurring in the unsaturated zone, they are consistent across all
situations and thus allow the model to be used for direct risk comparisons.
Final model output is a mass loading for each contaminant to the saturated
zone transport model, which is described below.
3.2 CONTAMINANT TRANSPORT IN THE SATURATED ZONE
The Prickett-Lonnquist Random Walk model is the component model used to
estimate saturated zone transport of contaminants in TRAM. This model is a
numerical (finite difference) model with variable grid spacings that simulates
advective transport in two dimensions (longitudinal and vertical).
Longitudinal and vertical dispersion are also accounted for in the model.
Exhibit 3-2 presents a schematic representation of the model's simulation of
contaminant transport.
In the Prickett-Lonnquist model, a particular hydrogeologic setting is
divided into a grid network of smaller subregions and ground-water flow
through this grid or network is estimated. Input parameters to the model
include hydraulic gradient, porosity, and hydraulic conductivity for each of
the flow units being considered. Additionally, the model simulates solute
transport through the input of 1,000 particles (which represent a standard
mass input of one kilogram) to the saturated zone. A time step is then
selected for the analysis, and the concentration of particles at each grid
boundary is calculated for each time step. Particle concentrations are used
as a surrogate for contaminant concentrations in the model.
Particle movement (i.e., contaminant movement) is simulated in two steps;
the advective stage, and the dispersive stage. During the advective stage,
-------
3-6
Exhibit 3-2
Simulation of Contaminant Transport in the
Saturated Zone by the
Prickett-Lonnquist Random Walk Model
MODEL
SOURCE AREA>
TOTAL FACILITY
SOURCE AREA
PARTICLE PLUME:
r---- s<
-------
3-7
the distance the particle has traveled due to advective flow is calculated
(this distance is equal to the ground-water velocity multiplied by the time
step). In the dispersive stage, the distance a particle travels is equal to a
random number selected from a distribution multiplied by a longitudinal or
vertical dispersion coefficient (the final value calculated for dispersive
stage movement may be positive or negative). The actual distance traveled by
a particle during a time step is then calculated by summing the distance
calculated for the advective and dispersive stages. This process is repeated
for each particle for each time step for which the model is run.
Particles entering each downgradient grid are counted and converted to a
mass flux, with the average concentration for that grid taken to be the ratio
of the mass flux of particles and the volumetric flux of water (i.e., the
number of particles divided by the volume of liquid contained within the grid
region). Grid volume is multiplied by soil porosity to obtain the percentage
of each grid volume occupied by water.
. Thus, by calculating volumetric and mass fluxes for the grid containing a
drinking water well, concentrations of contaminants at the well can be
estimated. It should be noted that the following assumptions apply in
estimating well concentration:
Dilution and drawdown effects from well pumping are not
evaluated;
Wells are located directly downgradient from the source
of contamination;
Drinking water wells draw only from the contaminated
aquifer(s) of interest, and draw water from the full
thickness of the aquifer(s); and
Transverse dispersion is not modeled within the
Prickett-Lonnquist model.
For the purpose of this analysis, the first three assumptions were
considered reasonable, because well pumping effects are very site-specific and
could not be easily incorporated into such a broad-based analysis, and because
they are consistent and conservative assumptions. However, transverse
dispersion may be a very important process affecting the transport of
contaminants in ground water, particularly for situations where the source of
contamination closely approximates a true point source (e.g., a tank) . Thus,
we modified Prickett-Lonnquist model predicted concentrations for transverse
dispersion according to the following expression:
where:
C .. C .. * (l/ir*n*t*D ) (3-4)
adj. unadj ' y
well concentration adjusted for transverse dispersion
(mg/L)
-------
3-8
C ,. = unadjusted well concentration from Prickett-Lonnquist
model (mg/L)
it =3.14
n = porosity (dimensionless)
t = time of travel from source to well (years)
D = transverse dispersion coefficient (mz/year)
The transverse dispersion coefficient is specific to the aquifer of interest,
and is considered to be a function of distance from the source and aquifer
velocity.
Contaminant degradation (both chemical and biological) within the
saturated zone is accounted for in an identical manner to the unsaturated zone
model. The same first-order rate constants for each contaminant are used to
account for degradation in both the unsaturated and saturated zones, and thus
differences in reaction rates or degradation processes between the two zones
are not taken into account. However, there is sufficient uncertainty with
respect to contaminant degradation to warrant this assumption, even though
physical, chemical, and biological parameters between the two zones may vary
significantly.
The saturated zone model of TRAM also considers contaminant retardation,
but in a somewhat different fashion from the unsaturated zone model. Unique
R_ values are calculated for all contaminants, based on Koc values and
relevant aquifer parameters, and using equation 3-2 as presented in Section
3.1 of this chapter. However, rather than using these contaminant-specific
values in the actual transport model, contaminants are grouped into four
mobility classes based on R_ values, and the model is run using median R_
values for each of the four classes. Contaminant mobility classes and median
R_ values are as follows:
Mobility Class Rp Range Median Rp
1 less than 10 1.3
2 10 to 100 32
3 100 to 10,000 360
4 greater than 10,000 160,000
These mobility classes were used because using chemical-specific R_ values
would require a run of the Prickett-Lonnquist model to generate a standard
breakthrough curve for each contaminant in each flow field. Performing all
the necessary runs woul be a computationally intensive task. Consequently, we
grouped the chemicals into the four mobility classes so that only four runs
were needed for each generic hydrogeologic setting.
-------
3-9
For these four mobility classes, the Prickett-Lonnquist model estimates
well concentrations for a release of one kilogram (1000 particles) of the
source. These results are then adjusted for contaminant degradation and
transverse dispersion. However, the concentration profiles, or breakthrough
curves, (Exhibit 3-3) generated by the model are for only one annual release
of one kilogram. Therefore, for a source releasing for longer than a year,
such as a tank, it is necessary to sum the yearly breakthrough curves to
obtain the actual concentration profile for a particular contaminant at an
exposure well. The methodology for summing individual breakthrough curves is
illustrated graphically in Exhibit 3-4. 2
TRAM does not actually incorporate the Prickett-Lonnquist program code,
but rather references a set of "standard breakthrough curves" of Prickett-
Lonnquist model results. The "standard breakthrough curve" for a particular
mobility class and hydrogeologic setting contains Prickett-Lonnquist model
results for each year in which a contaminant appears at an exposure well. By
multiplying the "standard breakthrough curve" value for a particular year by
the saturated zone mass loading for a contaminant, the well concentration for
that containment in that year is calculated. Thus, these values are used to
calculate yearly, time-varying concentrations of a particular contaminant at
an exposure well for a given set of hydrogeologic and mass loading conditions,
rather than peak or average concentrations. As described in Chapter 1, mass
loading conditions are based on HWTF model release volumes for a particular
tank, and annual disposal rates and concentrations of contaminants in a given
waste stream.
Hydrogeologic conditions such as horizontal and vertical ground-water
velocities and aquifer configurations and thicknesses are represented in TRAM
by a set of nine generic flow fields developed for the Liner Location Risk and
Cost Analysis model. Because of lack of data on actual hydrogeologic
conditions at hazardous waste tank facilities, and because of the need to
simplify the analysis relative to the potential variations in hydrogeology at
real tank facilities, we selected these nine flow fields as being reasonably
representative of conditions that would be most commonly encountered.
The nine flow fields encompass horizontal ground-water velocities of 1 to
10,000 m/year, and represent single and double layer aquifers with and without
intervening nonaquifer layers. These generic flow fields were developed based
on an extensive literature review of ground-water modeling studies and
evaluations in the U.S., and are considered to be representative of the range
of hydrogeologic conditions most commonly encountered at hazardous waste land
disposal facilities (i.e., landfills and surface impoundments) throughout the
country. Physical parameters (i.e., horizontal and vertical velocities,
aquifer configurations, and porosity values) are summarized for the nine
generic flow fields in Exhibit 3-5. Of the note is that all earth materials
in all of the nine flow fields are assumed to be homogenous and isotropic.
2In situations where the summed breakthrough curves exceed the
solubility limit for a contaminant, the well concentration for that
contaminant is limited to solubility (solubility values for TRAM contaminants
are presented in Exhibit 3-6).
-------
3-10
EXHIBIT 3-3
STANDARD BREAKTHROUGH CURVE FOR
TRANSPORT MODEL
(SCENARIO A-l-60)
30 -i
Explanation:
Data generated by
Random-Walk model
2.0 -
Concentration,
(Mg/liter)
1.0 -
50 100
Simulation Time, Years
200
-------
3-11
EXHIBIT 3-4
GRAPHICAL REPRESENTATION OF THE METHODS FOR
SUMMING INDIVIDUAL BREAKTHROUGH CURVES TO
OBTAIN THE CONCENTRATION PROFILE AT AN EXPOSURE WELL
Concentration,
(Mg/liter)
15 n
10 -
03 -
Sum of 10 kg-yr
Breakthrough Curves
for scenario C-1-600
5 10 IS
Simulation Time, Years
20
23
Note: Individual I yr input curves have not been scaled to any particular
facility size/unit mass load
-------
EXHIBIT 3-5
HYOROCEOLOGICAL PARAMETERS FOR THE NINE GENERIC FLOW FIELDS IN TRAM
Flow Field
A
B
C
D
E
F
G
H
1
Layer
upper
upper
upper
upper
upper
upper
lower
upper
lower
upper
lower
upper
iddle
lower
Aqui fer/
Nonaqu i fer .
aquifer
aquifer
aquifer
aquifer
aquifer
nonquifer
aquifer
aquifer
aqui fer
aquifer
aquifer
aquifer
nonaquifer
aquifer
Horizonta 1
Ground -water
Ve 1 oc i ty
(m/yr)
1
10
100
1000
10.0OO
0.05
100
100
10
10
100
10
0.05
100
Vert ica I
Ground-water
Ve 1 oc i ty
(m/yr)
0.1
1
10
100
1000
0.5
2
10
0.1
1
1
0.5
0.5
1
Po ro s i ty
0.30
0.20
0.20
0.1
0.01 a/
O.«i5
0.20
0.20
0.20
0.20
0.20
0.20
O.U5
0.20
Thickness
(ro)
30
30
30
30
30
15
30
30
60
15
15
30
30
30
a/ Assumed to represent homogeneously fractured or karstified rock.
-------
3-13
For the Tank Risk Analysis Model, the Prickett-Lonnquist model was run for
the nine generic flow fields, with assumed exposure well distances set at 60,
600, and 1500 meters, for the four modeled mobility classes.3 Thus, the
table of "standard breakthrough curves" within TRAM contains model results for
108 scenarios (nine flow fields x three exposure well distances x four
mobility classes). Well concentrations are calculated by multiplying the
selected breakthrough curve value for a particular year by the unsaturated
model value for mass loading to the saturated zone, and then adjusting the
predicted concentration for transverse dispersion and degradation.
The saturated zone transport values in TRAM are not suitable for
site-specific analyses, as there is no way to directly incorporate site
hydrogeologic conditions into the model. Moreover, developing .and using the
saturated zone model and generic flow fields require several simplifying
assumptions that affected the model's results and limit its applicability.
Some of these assumptions and limitations are as follows:
Mobilities for a group of contaminants are assumed to be
reasonably represented by a single median R_ value for
the group. This use of a median value leads to significant
under-estimation of transport times for contaminants with
R_s slightly less than the cut-off values for the four
mobility classes.
Degradation occurs only for contaminants whose decay
rates are readily available, and always at the specified
first-order rate. Degradation rates may vary significantly
from these assumed rates, and thus using a single value may
under-estimate or over-estimate the effect of degradation
on contaminant concentrations at exposure wells. Also,
many organic chemicals decay at rates that are so slow that
they are difficult to measure, but which could dramatically
reduce concentrations over long time periods.
The range of hydrogeologic conditions at tank facilities
are assumed to be adequately represented by the nine
generic flow fields selected for the analysis. These flow
fields are representative of a wide range of ground-water
velocities, but do not account for fractured aquifers,
karst aquifers, and other heterogeneous, anisotropic
hydrogeologic situations. These conditions could greatly
affect tranport times and thus have an effect on human
exposure to contaminant releases from tanks.
'In order to compare the regulatory.scenarios, we generated risk
estimates only for exposure at the 60 meter well.
-------
3-14
Despite these limitations, the saturated zone transport approach used in
TRAM is considered useful for comparing the transport of contaminants in
different hypothetical generic situations. In general, the model is
conservative in its assumptions, and in most situations will tend to
over-estimate well concentrations for contaminants. Assumptions are
consistent across all flow fields and contaminants, and, therefore, the model
provides the user with the ability to make direct comparisons between
different hydrogeologic settings and contaminants of concern.
Results from the saturated zone transport model (i.e., contaminant
concentrations at exposure points) are input to the exposure/risk component
model of TRAM. The exposure/risk model is described in the next section.
3.3 EXPOSURE/RISK MODEL
The exposure/risk model converts ambient environmental concentrations of
chemicals to estimates of human health risk. The exposure component calculates
annual and lifetime chemical doses to exposed individuals over the modeling
time period, based on several standard assumptions about intake. The risk
component estimates lifetime individual risk of cancer and other toxic effects
associated with chemical exposures. We estimate lifetime individual risk
estimates for each year of our 400-year model time horizon. To summarize the
lifetime individual risks for a particular chemical over the 400-year time
horizon, we average the 400 lifetime individual risk estimates. The following
sections describe how we calculated the lifetime individual risk for a given
year.
3.3.1 Exposure Component
The exposure component of the model converts chemical concentrations in
drinking water into chemical doses to exposed individuals. TRAM focuses on
the toxic effects of chronic chemical exposures, and therefore is based on
lifetime chemical dose.* As a result, the yearly chemical concentrations
predicted by TRAM must be transformed to average lifetime dose levels, in
mg/kg-day. These lifetime dose values are taken as inputs to the next step in
the risk analysis sequence, risk estimation.
The starting point for dose estimation is a time profile of ground-water
concentration at an exposure well. The value corresponding to each year
represents the estimated drinking water concentration in that year (effects of
water treatment are not included), which is converted to an annual dose by
assuming that average human drinking water consumption is two liters per day,
average human body weight is 70 kilograms, and absorption efficiency is 100
percent.
"Exceptions to this are the few chemicals in the data base classified as
teratogens or subchronic toxicants. Risks for these chemicals are based on
yearly (smallest time increment in the model) doses.
-------
3-15
The yearly dose profile is then transformed to a profile of 70-year-
average doses, with 70 years the assumed human life span for all individuals
over the entire modeling period. This transformation is done by averaging
each yearly dose with the 69 immediately preceding yearly doses (i.e.,
calculating a 70-year moving average over the entire period). The dose is
assumed to be zero for the 69 years preceding the start of the modeling
period. Therefore, each value in the transformed profile represents the
average dose that a hypothetical person reaching 70 years of age in that year
would have been exposed to in their lifetime.5These moving average doses
are appropriate values to use in lifetime risk calculations. Unaveraged
yearly doses are inappropriate as they do not represent lifetime exposures. A
single 400-year (or any other period) average concentration would greatly
limit the type and accuracy of possible risk estimates.
The major advantages to using 70-year moving averages as a measure of
exposure instead of individual yearly doses or single period averages are:
Allows risk calculations for effects having non-linear
dose-response functions (i.e., threshold effects,
carcinogenic effects at high doses);
Allows distribution of risks over time; and
Allows calculation of risk magnitude and variance over
time, and estimation of values such as maximum, median, and
various percentile risks.
The 70-year moving average approach to exposure estimation more realistically
reflects the toxicology of chronic exposures and more accurately represents
the time element in the risk analysis.
3.3.2 Risk Component
Estimating potential human health effects entails predicting
quantitatively the consequences of human exposure to disease-causing agents.
Traditionally, toxicologists use two different approaches to characterize the
potential health effects of chemical exposures. First, they commonly accept a
non-threshold assumption for carcinogenic effects. EPA policymakers have
*Note that the risks estimated based on these lifetime doses are
assigned to the final year of the assumed life span (i.e., the end of the
hypothetical exposed individual's life). Clearly, potential effects from
predicted chronic exposures would actually be distributed within the time
period between initial exposure and the end of the assumed life span. No
attempt is made to more accurately assign chronic risks over an individual's
lifetime because: (1) there is little precedent for this, and considerable
research into latency and timing of effects for specific chemicals would be
necessary to develop an approach, and (2) this level of detail is beyond the
scope of the planned regulatory analysis for tanks.
-------
3-16
followed their lead, using a linear, non-threshold dose-response model to
calculate a unit risk value (risk per unit dose) for a chemical at low doses
which, in turn, enables them to calculate explicit risks to individuals.
Second, toxicologists generally assume that there is a threshold for
noncarcinogenic effects. For these effects, a dose-response model has not
generally been used to estimate human health risks. Rather, an experimentally
determined effects threshold is divided by appropriate safety factors to
estimate an allowable daily intake (ADI). The ADI is compared to projected
exposure levels to provide an indication of the potential for noncarcinogenic
health effects.
The risk estimation component in TRAM diverges from the traditional
approach for chemicals with noncarcinogenic effects because the binary nature
of the traditional approach (i.e., the effect either does or does not occur at
a given dose, as opposed to having a probability of occurrence) is not very
useful for explicit risk, analysis. We have incorporated a dose-response model
that estimates risk from noncarcinogens as a continuous function of dose at
dose levels above the threshold, enabling us to treat these risks in a fashion
more amenable to regulatory analysis. In the following sections we describe
the risk component of this model and summarize the methodology for developing
the necessary dose-response parameters for specific chemicals. A set of
toxicity profiles, available as a supplemental document to the Liner Location
Risk and Cost Analysis Model (LLRCAM) report, documents the derivation of
dose-response parameters for individual chemicals.
Carcinogenic Effects. The model estimates risk for potential
carcinogens using the one-hit equation:
R = 1 - exp (-H * D) (3-5)
where:
R = lifetime individual risk (i.e., probability of an individual
developing cancer over a lifetime)
H = potency, or unit risk (mg/kg/day)
D = lifetime average dose (mg/kg/day)
This equation assumes no threshold (i.e., there is a finite risk for any dose
level) and is linear at low doses, an approach consistent with EPA's proposed
cancer risk assessment guidelines (49 Federal Register 46294-46301, November
17, 1984). The dose, D, used in this equation is developed within the tank
model and is the ultimate output of the release and transport models. The
potency, H, is chemical-specific and represents the response per unit dose of
a chemical. Higher H values are indicative of more potent carcinogens, which
produce higher risks than less potent chemicals at the same dose. For most
potential carcinogens, H values are set equal to the upper-bound unit risk
parameters (q.*) calculated by EPA's Carcinogen Assessment Group (CAG) using
-------
3-17
the linearized multistage model and the most appropriate set of experimental
data. Most of the unit risk parameters are based on animal studies. To use
animal data to estimate human cancer risk, the risk to humans is assumed to be
equal to the risk to animals at the same dose per unit surface area. This
assumption is consistent with current EPA policy for carcinogenic risk
assessment. For a few potential carcinogens that did not have a unit risk
value develoepd by GAG, H values were estimated by applying the one-hit
equations to an appropriate set of experimental data.' The studies these
values are based on are documented in a set of toxicity profiles. These
profiles are available as a supplemental document to the LLRCAM Report.
Noncarcinogenic Effects. The model uses the Weibull equation with a
threshold, as shown below, to estimate risks of noncarcinogenic effects:
R = 1 - exp [-H * (D-t)K] (3-6)
where:
R = lifetime individual risk (i.e., probability of the effect being
modeled occurring in an individual over a lifetime)
_K
H = Weibull dose-response parameter (mg/kg/day)
D = lifetime average dose (mg/kg/day)
t = effect threshold (mg/kg/day)
K * Weibull shape parameter
This equation incorporates an effect threshold, below which risk equals zero.
At doses above the threshold, risk is a funtion of dose and two chemical-
specific parameters, H and K. The lifetime dose, D, for this equation is
defined and calculated in an identical way to the dose for the carcinogenic
risk equation. The threshold, t, is chemical-specific and is derived from the
same experimental data used to develop H. The shape parameter, K, which
theoretically could vary for different chemicals or different effects, is set
to two for all noncarcinogenic effects. Assigning a value for K is an
approximation, but is necessary because of the lack of sufficient dose-
response data for noncarcinogenic effects to derive a unique value of K for
many specific chemicals. The primary basis for assigning K equal to two is a
statistical analysis of a limited number of data sets obtained from the
primary literature,7
'Only one potential carcinogen modeled in this study has an H value that
is not based on a GAG unit risk. Low risk estimates are associated with this
chemical, toluene diisocyanate, because it immediately hydrolyzes in water.
7This study was done by Science Research Systems, Inc. (now K.S. Crump
and Co.). The complete study report is reproduced as an attachment to
Appendix D of the Liner Location Risk and Cost Analysis Model report.
-------
3-18
Thus, the key chemical-specific parameter defining the dose-response curve
for noncarcinogenic effects is H. For each chemical, values for H were
estimated from three toxicologic parameters extracted from an appropriate
study: (1) the chronic minimum effective dose (MED); (2) the effect threshold
(t); and (3) the response fraction at the MED (termed Ro). Both the MED and
threshold were converted from animal values to human equivalent values based
on relative surface area. For chemicals with multiple studies, the study
reporting an adverse effect at the lowest dose was selected as the basis for H.
When chronic studies were unavailable, MED and threshold values from
subchronic studies were divided by five to convert to equivalent chronic
values. For many chemicals the effect threshold was not determined from the
experimental data but was estimated by dividing the human equivalent chronic
MED by a factor of 100 (10 to convert from chronic human equivalent MED to
chronic human equivalent no-effect level, 10 to account for intraspecies
variability). Many studies are reported with insufficient detail to allow
estimation of Ro, and in these cases Ro was set to 10 percent for dichotomous
(quantal) effects and 90 percent for continuous effects.1
After values for MED, t, and Ro are determined from an appropriate study
or estimated as described above, calculation of H is relatively straight-
forward. Rearranging equation 2 and substituting K = 2, R = Ro, and D = MED
gives the following calculation form for H:
H = -In (1 - Ro) (3-7)
(MED - t)2
The toxicity studies used to derive H values are documented in individual
toxicity profiles that are a supplement to the Liner Location Risk and Cost
Analysis report. Values for H, MED, and t used in the tank risk analysis are
listed in Exhibit 3-6.
3.3.3 Limitations of the Exposure/Risk Estimation Model
The primary limitations and assumptions in the exposure/risk estimation
model include those traditionally associated with determining quantitative
toxicity parameters and characterizing dose-response in low dose regions.
Major assumptions are:
'Dichotomous effects either occur or don't occur in each test subject,
and the response frequency is simply the number with the effect as a fraction
of the total. Continuous effects, such as organ weights or enzyme levels, are
often reported as average values for the total test group without data on
individual response. Continuous effects data were "converted" to dichotomous
data by assuming that most (90 percent) of the test invididuals are affected
at the MED.
-------
3-19
Considering human and animal toxic responses to be
similar at similar doses on a surface area basis.
Differences in size, lifespan, metabolism, and physiology
between humans and test animals complicate the
applicability of animal data to estimates of human risk.
However, the scientific community generally accepts animal
responses as a valid indicator of human responses.
Using data from subchronic studies to infer chronic toxic
effects. Because TRAM assesses human health risks due to
chronic exposures, data from long-term studies are
preferable. Often these data are not available, so we have
extrapolated from subchronic studies. This extrapolation
increases the uncertainty in the quantitative assessment of
chronic potencies and also creates the possibility of
ignoring an effect that occurs only after chronic exposures.
Assuming that the dose-response for carcinogens is linear
at low doses and that there is no threshold dose (i.e.,
using the one-hit model). This theoretical approach is
widely accepted and often used by policymakers to provide a
conservative estimate of carcinogenic risk.
Assuming that response from exposure to noncarcinogens at
low doses declines more rapidly with decreasing dose than a
linear function would predict. Similarly, at high doses
the response rises more rapidly than a linear model would
predict.
Assuming that all dose-response curves for
non-carcinogens have the same Weibull shape parameter.
This assumption may under- or over-estimate risk for some
chemicals.
-------
EXHIBIT 3-6
CONSTITUENTS REFERENCED BY TRAM a/
PHYSICAL/CHEMICAL DATA
TOXIC ITY DATA
1 NORG/
CONSTITUENT ORG
2 ACENAPHTHYLENE
3 ACETALDEHYOE
4 ACETONE
5 ACETONITRILE
7 ACRYLON 1 TR 1 LE
8 ALDICARB
12 ARSENIC
lit BENZENE
22 CADMIUM
24 C.TETRACHLORIDE
28 CHLOROFORM
30 CHROMIUM
31 CHRYSENE
32 COPPER
33 CYANIDES
36 12-DCBENZENE
41 DICHLOROMETHANE
51 24-01 NITROTOLUENE
61 HCBENZENE
62 HCBUTADIENE
64 HCCYCLOPENTADIENE
68 LEAD
69 LINDANE
72 METHANOL
75 M.E.KETONE
82 NICKEL
89 PCPHENOL
90 PHENOL
98 TETRACHLOROETHENE
99 THALLIUM
100 TOLUENE
102 TDI
105 111-TCETHANE
107 TCETHENE
112 XYLENE
117 CHLORIDE
1
1
1
1
1
1
0
1
0
1
1
0
1
0
0
1
1
1
1
1
1
0
1
1
1
0
1
1
1
0
1
1
1
1
1
0
SOLUBILITY
(MG/L)
0.3930E+01
0.1000E+07
0.1000E+07
0.1000E+07
0.7900E+05
0.6000E+04
0.1500E+05
0.1780E+04
0.2000E+01
0.7850E+03
0.8200E+04
0.4400E+06
0.1800E-02
0.1100E+06
0.3300E+06
0.1000E+03
0.2000E+05
0.2700E+03
0.6000E-02
0.2000E+01
0.2000E+01
0.1250E+03
0.7800E+01
0.1000E+07
0.3530E+06
0.8200E+02
0. 1400E+02
0.9300E+05
0.2000E+03
0.2400E+06
0.5350E+03
O.OOOOE+00
0.4400E+04
0.1100E+04
0. 1600E+03
0.2200E+06
ATTEN
AK
(YR-1)
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.5
0.0
0.0
0.0
0.1
0.0
0.4
0.0
0.0
0.0
23.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.9
0.0
0.0
130000.0
1.4
0.7
0.0
0.0
UNIT RISK
H
KOC (MG/KG-DAY)-K
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
2500E+04
2190E+01
2190E+01
2190E+01
8500E+00
3650E+02
5000E+01
8300E+02
5000E+03
4390E+03
4400E+02
5000E+01
0.2000E+06
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
5000E+04
5000E+01
1700E+04
8800E+01
4500E+02
3 900 E +04
2900E+05
4800E+04
5000E+04
1100E+04
2190E+01
3880E+01
5000E+03
5300E+06
6000E+01
3640E+03
5000E+03
2500E+03
1550E+01
1520E+03
1260E+03
2680E+03
5000E-01
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.
1150E+02
2500E-01
6600E-02
4900E+00
2400E+00
1040E+05
1500E+02
2900E-01
6600E+04
1300E+00
7000E-01
2400E+01
1150E+02
3600E+01
3100E+00
3500E-01
6300E-03
3100E+00
1670E+01
7750E-01
6100E+09
6200E+05
1330E+01
282CE+03
5500E-05
1820E+02
6300E+01
1030E+01
5100E-01
3050E+03
1800E+00
4300E-02
1600E-02
1100E-01
1400E+00
7700E-05
K
1
2
2
2
1
2
1
1
2
1
1
2
1
2
2
2
1
1
1
1
2
2
1
2
2
2
2
2
1
2
2
1
1
1
2
2
THRESHOLD
(MG/KG-DAY)
O.OOOOE+00
0.9700E-01
0.1900E+00
0.2200E-01
O.OOOOE+00
0.1500E-03
O.OOOOE+00
O.OOOOE+00
0.2900E-03
O.OOOOE+00
O.OOOOE+00
0.1000E-01
O.OOOOE+00
0.4000E-01
0.8700E-02
0.8300E-01
O.OOOOE+00
O.OOOOE+00
O.OOOOE+00
O.OOOOE+00
0.6200E-06
0.1300E-02
O.OOOOE+00
0.9200E-03
0.2300E+01
0.7700E-03
O.6100E-02
0.1500E-01
O.OOOOE+00
0.2800E-03
0.3600E-01
O.OOOOE+00
O.OOOOE+00
O.OOOOE+00
0.4100E-01
0.1700E+02
DETECTABLE
CONCENT
EFFECT (MG/L)
CANCER
TRACHEA
STOMACH
MULT ORG
CANCER
MORTAL IT
CANCER
CANCER
KIDNEY
CANCER
CANCER
K 1 D&L 1 V
CANCER
BLOOD
THYROID
K 1 D&L 1 V
CANCER
CANCER
CANCER
CANCER
BLOOD
BLOOD
CANCER
MULT ORG
EMBRYO
REPROD
EMBRYO
KIDNEY
CANCER
WEIGHT
WEIGHT
CANCER
CANCER
CANCER
LIVER
HYPERTEN
0.1E-01
O.OE+00
0.5E-02
O.OE+00
0. 1E+00
O.OE+00
0.2E-01
0.5E-02
0.2E-01
0.5E-02
0.5E-02
0.2E-01
0.2E-01
0.2E-01
O.OE+00
0.1E-01
0.5E-02
0.2E-01
0. 1E-01
0.1E-01
0.1E-01
0.2E-01
0.5E-05
O.OE+00
O.OE+00
0.2E-01
0. 1E-01
0.1E-01
0.5E-02
O.OE+00
0.5E-02
O.OE+00
0.5E-02
0.5E-02
0.5E-02
O.OE+00
a/ The constituent data base actually contains 120 constituents, but TRAM only references the above 36
constituents.
-------
CHAPTER 4
DEVELOPMENT OF DATA FOR DERIVING DISTRIBUTIONS OF RISK ESTIMATES
In this chapter we describe the methods and data we used to develop the
distributions of risk estimates needed to compare the regulatory scenarios.
These distributions are meant to represent the range of risk estimates
associated with a tank technology. To generate these distributions, we
compiled relevant information on the population of existing hazardous waste
tanks and used this information to develop representative model hazardous
waste tanks, hydrogeologic settings, and waste streams, and to determine the
percentage of the existing tank population that corresponds to each model tank
characterization. These percentages were used to weight the risk estimates
corresponding to each tank characterization. We then used these weighted risk
estimates to derive a distribution of risk estimates.
In the first section of this chapter we describe what we mean by a
distribution of risk estimates and outline the methodology we used to develop
representative distributions. In Section 4.2 we describe the development of
the model tank technologies. Then, in Sections 4.3 through 4.5 we describe
the methodologies used to select the representative waste streams, map the
tank population into hydrogeologic settings, and select representative release
profiles.
4.1 DISTRIBUTIONS OF RISK ESTIMATES
In this section we describe the data and method used to derive
representative distributions of risk estimates. We begin by briefly reviewing
the interpretation of the risk estimates generated by the Tank Risk Analysis
Model (TRAM). We then present and evaluate alternative methods for assessing
the benefits of each regulatory scenario. Finally, we describe in detail the
method chosen for this study.
4.1.1 Interpretation of Risk Estimates
The TRAM generates chronic human health risk estimates. Specifically, the
estimates represent the probable number of individuals per exposed unit of
population that will suffer a specific health effect. For example, a risk
estimate of 0.01 for cancer implies that 1 in 100 exposed individuals is
likely to contract cancer as a result of the exposure. The risk estimates can
also be interpreted as the probability that a given exposed individual will
contract the specified chronic illness. Thus, the risk estimate of 0.01 for
cancer also implies that an exposed individual has a 1 in 100 chance of
contracting cancer as result of his or her exposure.
It bears emphasis that the risk estimates apply only to the exposed
population, i.e., those individuals who come in contact with the contaminants
in question, and not to the entire population of a region. So, for example,
if out of a total population of 100 million, 50 million individuals are
-------
4-2
exposed to the contaminants, and the relevant risk estimate is 0.01, the
number of individuals likely to contract the illness specified as a result of
the exposure is 500,000 (50 million times 0.01). Because we do not have data
on the population of individuals living in the proximity of hazardous waste
tanks, we did not estimate the number of individuals likely to contract an
illness.
4.1.2 Alternatives for Summarizing Risk Estimates
In order to compare the various regulatory scenarios, we focused attention
on developing representative estimates of the human health risk associated
with hazardous waste tanks. Because the risk estimates associated with
hazardous waste tanks vary considerably, depending on the technical
characteristics of the tanks, the wastes they contain, and the hydrogeological
characteristics of the areas in which they are located, it is not clear what
constitutes a representative risk estimate. We identified two basic options
for developing representative risk estimates.
Under the first option, one representative risk estimate would be
derived. This representative estimate would be associated with a single
tank, waste stream, and hydrogeologic setting. For instance, the
representative risk estimate could be based on the most common tank
technology, waste stream, and hydrogeologic setting (e.g., a 10,000 gallon
carbon steel storage tank containing D001 wastes located over an aquifer with
a depth of 30 meters, and a velocity of 10 meters per year). Although such an
estimate would be easy to obtain from a computational standpoint once the
suitable tank technology, waste stream, and hydrogeologic setting had been
chosen, it has two major drawbacks. First, choosing the appropriate tank
technology, waste stream, and hydrogeological setting is likely to be
difficult and controversial. Second, the estimate would not reflect the
considerable variation in risks associated with hazardous waste tanks, and is,
therefore, likely to be misleading.
The second option would be to base representative risk estimates on a
range of risk estimates for several different tank technologies, hazardous
wastes, and hydrogeologic settings. For example, risk estimates for several
representative tank technologies, waste streams (for each technology), and
predominant hydrogeological settings, could be combined in a suitable manner
to obtain a range of risk estimates. Of course, the estimates obtained would
depend on the chosen set of specific tanks, waste streams, and hydrogeologic
settings. In addition, the computational burden of this approach is
undoubtedly higher than that of the first one. However, representative risk
estimates derived using the second approach are likely to better reflect the
variation in risks associated with hazardous waste tanks. For this reason,
the second option was adopted in this study to derive representative risk
estimates. A detailed description of the method used to derive the estimates
is provided below.
-------
4-3
4.1.3 Individual Risk Estimates Underlying the Distribution
We derived risk estimates for numerous combinations of tank technologies,
waste streams, and hydrogeological settings. We considered four major
categories of tank technologies: storage tanks, treatment tanks, accumulation
tanks, and small quantity generator (SQG) tanks. Within each of these four
categories, tanks are characterized by:
the material they are made of -- carbon steel, stainless
steel, concrete, or fiberglass reinforced plastic (FRP);
their location -- above-ground, underground, or in-ground;
their capacity; and
whether they are open or closed topped.
In total, we modeled 22 different tank technologies under each of the
regulatory scenarios. For each of the 22 tank technologies, the Hazardous
Waste Tank Failure Model generated 250 release profiles, with each profile
consisting of a series of 20 numbers representing the annual volume of waste
released from the tank during a 20-year time horizon. We chose five or six
representative profiles from each set of 250 release profiles to depict the
typical performance of each model tank. (The method used is described in
Section 4.5.) Thus, for each of the 22 technologies modeled under the
different regulatory scenarios, five or six representative profiles were
selected. The release profiles are the basic inputs to the Tank Risk Analysis
Model. Within the risk model, a given tank technology could contain any one
of up to 21 different waste streams and be located in any one of 9 possible
hydrogeologic settings.
For a given combination of (1) regulatory scenario, (2) tank technology,
(3) representative release profile, (4) waste stream, and (5) hydrogeologic
setting, the risk model generates annual risk estimates for several different
chronic illnesses (e.g., kidney or liver related illness) over a 400-year
period. An average annual risk estimate is generated for each constituent in
the waste stream. (The precise illnesses depend on the constituents of the
wastes contained in the tank.) Of these average risk estimates for each
constituent, we selected the highest estimate, which we shall refer to as the
dominant risk estimate, to be used for the derivation of the distribution of
risk estimates.lWe use the dominant risk to summarize the risk associated
with a particular combination of tank technology, waste stream, hydrogeologic
setting and release profile. Thus, for each of the approximately 35,640
I0f note is that we did not differentiate between carcinogenic and
non-carcinogenic risk estimates.
-------
4-4
combinations of regulatory scenario, tank technology, release profile, waste
stream, and hydrogeologic setting, we obtained a dominant risk estimate.2
We chose the dominant risk approach to summarize the risk associated with
a particular combination of tank technology, waste stream, and hydrogelogic
setting, for several reasons. First, while there are as many as six toxic
constituents in some of the waste streams selected for the study, risk tends
to be dominated in most cases by one or two chemicals due to differences in
potency and concentration. Furthermore, summing risk estimtes of different
orders of magnitude does not have a significant effect on the outcome,
particularly when one considers that the uncertainty inherent in the models
used to obtain these estimates restricts their use to relative order-of-
magnitude comparisons.
Also, we did not sum risk estimates because, for comparative purposes, we
chose not to differentiate between carcinogenic and non-carcinogenic risks and
between different effects for non-carcinogenic risks. Many toxicologists
consider carcinogenic risks to be additive in the absence of information to
the contrary; therefore, if we had only considered carcinogenic risks, summing
risk estimates would have been considered acceptable.1 However, not only
did we consider both carcinogenic and non-carcinogenic risks, but we also
considered different non-carcinogenic effects. Summary of carcinogenic and
non-carcinogenic risks or of non-carcinogenic risks for different effects
would ignore probable differences in mechanisms and severity.
For these reasons, and because our analysis focused on relative risk
comparisons rather than the estimation of absolute risks, we considered
choosing the dominant risk estimate for a particular waste stream-
representative release profile-hydrogeologic setting combination to be an
adequate representation. This assumption is consistent across all scenarios.
4.1.4 Weighting Dominant Risk Estimates
For a given regulatory scenario and tank technology, we assigned each
dominant risk estimate a weight reflecting its likelihood of occurrence.
These weights sum to one for a given regulatory scenario and technology (e.g.,
the weights assigned to the dominant risk estimates for underground, carbon
steel storage tanks with secondary containment sum to one). We calculated the
2The approximate number of combinations is given by: 5 regulatory
scenarios x 22 tank technologies x 6 representative profiles x 6 waste streams
x 9 hydrogeologic settings = 35,640. The number is approximate because only 5
representative profiles and fewer or more than 6 waste streams were used for
some tank technologies. In addition, not all regulatory scenarios are
applicable to all of the 22 tank technologies.
'The summation of cancer risks would be consistent with EPA's draft risk
assessment guidelines for carcinogens (see Federal Register 49:46294-46301,
November 17, 1984).
-------
4-5
dominant risk estimate weights by multiplying the weights for the relevant
representative release profile, waste stream, and hydrogeologic setting. We
shall refer to the weights used to calculate the dominant risk estimate
weights as the component weights (e.g., the hydrogeologic setting weight).
An example illustrating the computation of the dominant risk estimate weights
is presented below. First, however, we briefly describe the component weights
used in computing, the dominant risk estimate weights. These component weights
are the (1) hydrogeologic setting weights, (2) waste stream weights, (3)
representative release profile weights, and (4) tank technology weights.
Hydrogeologic Setting Weights. The hydrogeologic setting weights
reflect the proportion of a particular category of tanks (i.e., treatment,
storage, SQG, or accumulation tanks) located in each of the nine hydrogeologic
settings. No readily available, detailed information exists on the location
of each modeled tank. Consequently, weights for the hydrogeologic settings
were only compiled for each category of modeled tank. These weights are
presented in Exhibit 4-1. The weights sum to one for a given category of
tanks. (See Section 4.4 for a description of the derivation of the
hydrogeologic setting weights.)
Waste Stream Weights. A waste stream weight indicates the proportion
of a tank technology population that is assumed to handle the waste stream.
For instance, a waste stream weight of 0.24 for D002 handled in in-ground,
carbon steel storage tanks, indicates that 24 percent of these tanks are
assumed to store D002 wastes. As an example, the waste stream weights
corresponding to in-ground, carbon steel storage tanks are presented in
Exhibit 4-2. For a given tank technology, the weights sum to one.
Representative Release Profile Weights. As noted earlier, the
Hazardous Waste Tank Failure Model generates 250 release profiles for each of
the 22 tank technologies (under each regulatory scenario). We divided each
set of 250 profiles into 5 or 6 clusters of similar profiles. We then
selected one representative profile from each of the clusters (the procedure
is described in Section 4.5). These representative profiles were assigned
weights equal to the proportion of profiles in the cluster from which they
were drawn. Exhibit 4-3 provides an example of this weighting scheme.
We assume that each representative release profile represents the releases
associated with a portion (given by the weight) of the modeled tank
population. For the example illustrated in Exhibit 4-3, 7.2 percent of all
tanks, represented by the modeled tank technology, have releases as indicated
by the representative release profile identified for cluster 1.
Tank Technology Weights. The tank technology weights reflect the
assumed distribution of tank technologies within a given tank category. These
weights are presented in Exhibit 4-4. The weights sum to one for each
category of tank. For example, above-ground, carbon steel tanks are assumed
to account for 47.6 percent of all small quantity generator (SQG) tanks.
Moreover, the weights for SQG tanks sum to one, 0.476 + 0.524 =1. We have
not used these weights in any analysis performed to date. However, these
-------
EXHIBIT l»-1
PERCENTAGE OF TANK CATEGORY LOCATED IN EACH GROUND-WATER SCENARIO
Ground-water Scenario
Tank Category
Treatment
Storage
Small Quantity Generators
Accumulation
0.
0.
0.
0.
A
1068
1103
1018
1103
B
0.1U5U
O.WO
0.1500
0.1«»UO
C
0.1555
0.1515
0.1583
0.1515
o
0.072U
0.0720
0.0767
0.0720
£
0.0390
O.OU20
0.0359
0.01)20
F
0. Id U U
0.1/430
0.1182
0.1430
0.
0.
0.
0.
G
1452
1U27
1112
1«»27
H
0.03U6
0.0358
0.0391
0.0358
1
0.156U
0.1590
0. 1U8U
0.1590
-------
4-7
EXHIBIT 4-2
WASTE STREAM WEIGHTS FOR
IN-GROUND, CARBON STEEL STORAGE TANKS
Waste Stream Waste Stream Weight
D002 0.24
D007 0.27
K048 0.04
F006 0.03
D004 0.38
P115 0.01
K062 0.03
-------
4-8
EXHIBIT 4-3
CALCULATING REPRESENTATIVE RELEASE
PROFILE WEIGHTS FOR A TANK TECHNOLOGY
Number of Profiles
Cluster in Cluster
1 18
2 22
3 60
4 50
5 43
6 7
Total 250
Representative
Profile Weight
0.072
0.088
0.240
0.200
0.172
0.228
1.000
-------
4-9
EXHIBIT 4-4
TANK TECHNOLOGY WEIGHTS
Tank
Category
Treatment
Storage
Small Quantity
Generator
Accumulation
Tank
Technology
Above-ground, closed top, carbon steel
Above-ground, open top, carbon steel
Above-ground, ongrade, carbon steel
In-ground, concrete
In-ground, carbon steel
In-ground, stainless steel
Above-ground, on cradles, carbon steel
Above-ground, ongrade, carbon steel
Underground, carbon steel
Underground, FRP
Underground, stainless steel
In-ground, concrete
In-ground, carbon steel
Above-ground, carbon steel
Underground, carbon steel
Above-ground, on cradles, carbon steel
Above-ground, ongrade, carbon steel
Underground, carbon steel
Underground, FRP
Underground, stainless steel
In -ground, concrete
In-ground, carbon steel
Weight
0.238
0.358
0.067
0.198
0.054
0.085
0.582
0.155
0.149
0.018
0.020
0.048
0.029
0.476
0.524
0.582
0.155
0 . 149
0.018
0.020
0.048
0.029
-------
4-10
weights give us the option of summarizing the risk estimates for a category
of tanks. For example, we could use these weights to derive a distribution of
risks associated with all different types of small quantity generator tanks.
Calculating Dominant Risk Estimate Weights. As noted earlier, the
dominant risk estimate weights are obtained by multiplying the relevant
component weights.. The exact procedure is best illustrated by an example.
Suppose the dominant risk estimate for the third representative release
profile for an in-ground, concrete treatment tank when located in
hydrogeologic setting C and storing D001 wastes is 0.00001. The component
weights for calculating the weight to be applied to this dominant risk
estimate are presented in Exhibit 4-5. As shown, the weight associated with
the dominant risk estimate is equal to the product of three component weights.
The weights indicate that: (1) 15.55 percent of all in-ground, concrete
treatment tanks are located in hydrogeologic setting C; (2) 29.4 percent of
all in-ground, concrete treatment tanks handle waste stream D003; and (3)
representative release profile 3 is representative of the types of releases
that are associated with 19.3 percent of all in-ground, concrete treatment
tanks. The product of these three weights (i.e., 0.0088) indicates the
percent (i.e., .88 percent) of all in-ground, concrete treatment tanks that
are associated with the following characterization: (1) are located in
hydrogeologic setting C; (2) handle waste stream D003; and (3) have releases
as indicated by representative release profile 3. In summary, this example
illustrates that .88 percent of all in-ground, concrete treatment tanks are
associated with a risk estimate of 0.00001.
4.1.5 Deriving the Frequency Distribution of Dominant Risk Estimates
Using the procedure described above, we calculated weights for each of the
dominant risk estimates. For a given tank technology and regulatory scenario,
we then used the dominant risk estimates and their weights to develop
frequency distributions of dominant risk. The derivation of the frequency
distribution is best demonstrated via a hypothetical example. Exhibit 4-6
contains the hypothetical dominant risk estimates and their weights for an
above-ground, carbon steel SQG tank. For illustrative purposes, we have
reduced the number of hydrogeological settings, representative release
profiles and waste streams associated with this hypothetical tank.
Consequently, we assumed that there are only two relevant waste streams, two
representative profiles, and two hydrogeologic settings. Therefore, in this
example there are eight dominant risk estimates (two representative profiles x
two waste streams x two hydrogeologic settings). These eight risk estimates
indicate a hypothetical range of dominant risk estimates associated with an
above-ground, carbon steel small quantity generator tank.
The dominant risk estimates in Exhibit 4-6 are arranged in a frequency
distribution in Exhibit 4-7. This distribution can be viewed as a
hypothetical frequency distribution of dominant risk estimates associated with
an aboveground, carbon steel small quantity generator tank. The vertical
scale of the figure gives dominant risk and the horizontal scale gives
-------
4-11
EXHIBIT 4-5
SAMPLE CALCULATION OF DOMINANT RISK
ESTIMATE WEIGHT
Variable
Weizht
Hydrogeologic Setting C
0003 Waste Stream
Representative Profile 3
0.1555
0.2940
0.1930
Dominant Risk
0.0088 - 0.1555 x
0.2940 x 0.1930
-------
4-12
relative frequency. The relative frequency for a given range of dominant risk
-5 -4
estimates (e.g., 5 x 10 to 5 x 10 ) is equal to the sum of the weights
associated with dominant risk estimates in that range. For example, in
Exhibit 4-6, two dominant risk estimates are within the range 5 x 10 to 5
-4 -5 -4
x 10 (specifically, 8 x 10 and 2 x 10 ), and their weights sum to
0.425 (0.125 + 0.300), as shown in Exhibit 4-7. By construction, the relative
frequency of the four "bars" in Exhibit 4-7 sum to one.
Frequency distributions such as the one depicted in Exhibit 4-7 were
prepared for each tank technology under each regulatory scenario, yielding a
total of approximately 110 (twenty-two technologies x five regulatory
scenarios) frequency distributions.* In order to compare the range of
dominant risk estimates for each of these 110 combinations of tank technology
and regulatory scenario, suitable summary measures must be chosen for the
frequency distributions of dominant risk estimates. The two most common
summary measures are the mean and the median. The mean of a dominant risk
estimate distribution is equal to the sum of the products of each risk
estimate and its weight. A simple example is provided in Exhibit 4-8. The
median dominant risk estimate is the risk estimate such that (roughly) 50
percent of the weight of the distribution is accounted for by risk estimates
greater than or equal to it and 50 percent by risk estimates less than or
equal to it. The derivation of the median is also illustrated in Exhibit
4-8. The median dominant risk estimate can be interpreted as the dominant
risk estimate such that 50 percent of the time the actual dominant risk
estimate exceeds it and 50 percent of the time it is lower than it. For
example, if the median dominant risk estimate is 0.001, the actual dominant
risk estimate will exceed 0.001 50 percent of the time, and be lower than
0.001 50 percent of the time.
The example presented in Exhibit 4-8 demonstrates that the mean and median
of a distribution can differ considerably (mean = 0.0001043, median =
0.000001). The mean is approximately two orders of magnitude higher than the
median. In general, the value of the mean dominant risk estimate will be more
sensitive than the median to extreme dominant .risk estimate values with small
weights. In the example presented, the mean dominant risk estimate primarily
reflects the dominant risk estimate of 0.001. But note that 90 percent of the
dominant risk estimates are lower than the mean dominant risk estimate. The
choice between the mean and median as a summary measure for dominant risk
estimates will depend largely on the importance that is given to extreme
dominant risk estimates with low weights.
An alternative to either the mean or the median that may be a suitable
summary measure, given the characteristics of the dominant risk estimate
distributions obtained, is the proportion of non-zero risk dominant
estimates. For example, if the dominant risk estimates obtained are (0,
'"The number of frequency distributions is approximate because all the
regulatory scenarios are not applicable to each tank technology. For example,
partial containment is not applicable to above-ground, on cradles tank
technologies.
-------
4-13
EXHIBIT 4-6
HYPOTHETICAL DOMINANT RISK ESTIMATES AND THEIR
WEIGHTS FOR ABOVE-GROUND, CARBON STEEL SQG TANK
Dominant Risk Estimate
Hydrogeo logic
Setting
A
B
Representative
Profile
1
2
1
2
F001
Waste Stream
3 x 10"3
(0.025)
4 x 10"5
(0.050)
2 x 10"4
(0.300)
2 x 10"6
(0.150)
X900
Waste Stream
8 x 10"5
(0.125)
6 x 10"6
(0.250)
3 x 10~5
(0.066)
6 x 10"7
(0.034)
Note: Weights are presented in parentheses.
-------
4-14
EXHIBIT 4-7
HYPOTHETICAL FREQUENCY DISTRIBUTION OF
RISK ESTIMATES FOR ABOVE-GROUND, CARBON STEEL SQG TANK
5 X ICT7 to
-6
Risk
Estimate
5 x 10
5 X 1(T6 to
5 x icr5
5 X 10'5 to
5 X 10'4
5 X 10"* to
5 X 10*3
0.050 0.100 0.150 0.200 0.250 0.300 0.350 0.400 0.450 0.500
Percent
-------
4-15
EXHIBIT 4-8
MEAN AND MEDIAN OF DOMINANT RISK DISTRIBUTION
Risk Estimate Weight
ID'8
lO'7
icf6
ID'5
ID'5
ID'5
ID'3
0.10
0.15
0.25
0.10
0.15
0.15
0.10
Cumulative Weight Risk Estimate x Weight
0.10
0.25
0.50
0.60
0.75
0.90
1.00
1.0 x 10~9
1.5 x 10"8
2.5 x 10"?
1.0 x 10"6
1.5 x 10~6
1.5 x 10"6
1.0 x 10"4
Median = 10
-6
Mean ~ 1.043 x 10
-4
-------
4-16
10 ,10 ,10 ), the proportion of non-zero dominant risk estimates is
0.75. If limited importance is given to the precise magnitude of non-zero
dominant risk estimates, the proportion of non-zero dominant risk estimates is
a useful summary measure for the dominant risk estimate distribution.
Although single summary measures such as the mean, median, or proportion
of non-zero dominant risk estimates are convenient, they generally fail to
adequately reflect all the relevant features of a frequency distribution, as
the example presented above demonstrates. An alternative to using a single
summary measure is to use several, such as the 25th percentile, the 50th
percentile, and the 75th percentile of the distribution. For the example in
Exhibit 4-8 we would have:
Dominant
Percentile
25th
50th
75th
Risk Estimate
-7
10
-6
10
io"5
Thus, 25 percent of the time, dominant risk estimates are less than or equal
to 10 , 50 percent of the time they are less than or equal to 10 , and
75 percent of the time they are less than or equal to 10 . (Note that, by
definition, the 50th percentile is equal to the median.) This approach to
summarizing frequency distributions of dominant risk estimates, although
slightly more cumbersome, will reflect the variation in dominant risk
estimates.
Having discussed the construction and interpretation of dominant risk
estimate distributions, the remainder of this chapter is devoted to examining
the data needed to derive the frequency distributions.
4.2 DEVELOPMENT OF MODEL TANKS
4.2.1 Introduction
To estimate the human health risks associated with alternative regulatory
requirements for hazardous waste tanks, we developed 22 model tanks that
represent typical tanks in the Office of Solid Waste Regulatory Impact
Analysis Tank Survey (OSV RIA) and the Small Quantity Generator Survey (SQG
Survey). These models represent typical tanks for each of four categories of
hazardous waste tanks: (1) storage; (2) treatment (excluding wastewater
treatment tanks); (3) accumulation (storage fox less than 90 days at
facilities that generate more than 1000 kg of waste per month); and (4) small
quantity generators (that generate between 100 and 1000 kg of hazardous waste
per month).
-------
4-17
We identified model tank technologies for each of these four categories
based on differences among tank characteristics, especially type (i.e.,
above-, in-, or underground) and material of construction (e.g., carbon steel,
stainless steel, concrete, fiberglass reinforced plastic), that affect risks,
and on the data available to describe each category. We selected the 22 model
tank technologies to represent the predominant tank types and materials of
construction for each of the four tank categories listed above in order to
minimize the number of model tanks developed. It was important to minimize
the number of model tanks included in the analyis due to the complexity of the
release, transport, and exposure calculations that are required.
The basic steps we used in developing the model tank technologies are as
follows:
Define the tank categories of interest (i.e., SQG,
storage, accumulation and treatment);
Determine the tank characteristics most likely to
affect releases to the environment and compliance costs;
Identify the data available to describe these tank
characteristics; and
Analyze the available data to describe the population
and select appropriate model tanks.
In the remainder of section 4.2 we discuss the selection of the model
tanks in more detail, present the characteristics of the model tanks selected,
and highlight the limitations and assumptions that are incorporated in the
development of the model tanks.
4.2.2 Methodology
The purpose of the selection methodology was to minimize the number of
model tanks that would adequately represent the national population of tanks.
To ensure that the development of model tanks reflected the objectives of the
risk analysis effort, we first considered the regulatory requirements to be
examined by the model. Consideration of these requirements (e.g., secondary
containment, ground-water monitoring, corrosion protection, leak testing,
initial integrity assessment) provided general information on the type of tank
characteristics that needed representation.
Next, we identified the types of tanks that would be subject to the
requirements of the hazardous waste tank rulemaking. First, we identified
three broad categories of facilities with hazardous waste tanks: (1) RCRA
interim status or permitted hazardous waste storage or treatment tank
facilities; (2) facilities that generate more than 1000 kg of hazardous waste
per month and accumulate hazardous waste in tanks for periods of less than 90
days; and (3) small quantity generator facilities that generate between 100
and 1000 kg of hazardous waste per month and use tanks to store or treat the
waste.
-------
4-18
We then identified the types of tank characteristics that influence the
frequency and magnitude of releases from tanks to further characterize the
model tanks.5J8J7J*J The characteristics identified include tank type
(i.e., above-ground, underground, in-ground), use (e.g., accumulation, storage
or treatment), design capacity, quantity of waste contained, type of waste
contained, material of construction, operating procedures (e.g., inspection
schedules, inventory control practices), age, safety features (e.g., overfill
protection devices), installation practices, design features (e.g., secondary
containment), and piping length.
The next step in the development of model tanks was to identify readily
available sources of information that would describe as many of the tank
characteristics of interest as possible for each of the three broad categories
of facilities potentially affected by the proposed rulemaking. We used the
QSW RIA Generator Survey, Tank Survey and SQG Survey as the most direct
sources of the appropriate information. Although these were the best
available data sources, the information on hazardous waste tanks contained in
these sources is inconsistent with respect to the level of detail provided.
In particular, the Generator Survey provides information on the number of
accumulation tanks used nationwide, the number of tanks per facility and the
average size of the tanks, while the SQG survey provides information on the
number of facilities with tanks, the types of waste managed in the tanks, the
average storage period, and the tank throughput. The Tank Survey, on the
other hand, provides information on many of the tank characteristics of
interest.
Based on the available data sources and consideration of the analytical
complexity (and the associated need to minimize the number of model tank
facilities selected for analysis, we decided to describe each model tank in
terms of tank type (i.e., above-, in-, or underground), design capacity,
material of construction, age, the median number of tanks at facilities with
each type of model tank, and whether the tank is open- or closed-topped. The
use of the available data to describe the model tanks in terms of these
selected characteristics is described below for each of the three hazardous
5F. G. Bercha and Associates, Limited, Bulk Plant Risk Optimization,
report for the Environmental Protection Service, Hull, Quebec, December 1982.
*JRB Associates, Failure Incident Analysis; Evaluation of Storage
Failure Points, a report to the Office of Solid Waste, EPA, 1982.
7SCS Engineers, Assessment of the Technical, Environmental, and Safety
Aspects of Storage of Hazardous Waste in Underground Tanks, a report to the
Office of Solid Waste, EPA, 1984.
*SCS Engineers, Assessment of the Technical. Environmental, and
Management Aspects of Storage and Treatment of Hazardous Waste in Aboveground
and Underground Tanks, a report to the Office of Solid Waste, EPA, August
1984.
-------
4-19
waste tank facility categories identified above. We present RCRA storage and
treatment tanks separately because experience and the available data indicate
that these are distinct tank populations. In addition, we present our
estimation of the type and length of piping associated with the model
facilities.
Storage Tanks
We first selected model storage tanks by categorizing the storage tanks
(from the OSW RIA data base) by tank type and material of construction as
shown in Exhibit 4-9. We used two categories of above-ground tanks because we
assume that above-ground tanks less than or equal to 25,000 gallons are
installed off the ground on cradles or legs and that tanks larger than 25,000
gallons are installed on-grade.* We then identified the combinations of
tank type and material that best represent the storage tank population.
As shown in Exhibit 4-9, nearly 80 percent of above-ground tanks with
capacities less than 25,000 gallons are made of carbon steel. Thus, we
selected carbon steel as the material of construction for the entire
population of above-ground tanks with capacities less than 25,000 gallons. We
did not model tanks with capacities less than 25,000 gallons that are made
from other materials because they each represent a small proportion of this
type of tank.
For above-ground tanks larger than 25,000 gallons, we also selected carbon
steel as the construction material for the model tank because 94 percent of
these types of tanks are constructed with carbon steel. For in-ground tanks,
Exhibit 4-9 shows that both concrete and carbon steel are commonly used. Based
on these data and judgement concerning the impacts of tank material on risks
for in-ground tanks, we selected concrete and carbon steel in-ground model
tanks to represent 62.3 and 47.7 percent of the in-ground storage tank
population, respectively.
For underground tanks, Exhibit 4-9 shows that carbon steel is the most
common construction material, followed by stainless steel, fiberglass,
concrete and "other". Based on these data and judgement, we selected carbon
steel, stainless steel, and fiberglass model tanks to represent 79.8, 10.6 and
9.6 percent of the underground storage tank population, respectively.
Next, we obtained data on the distribution of tank size, age and top
(closed or open), and the median number of tanks at facilities with storage
tanks for each of the following four types of tanks:
Above-ground storage tanks less than 25,000 gallons;
SICF Incorporated, Estimated Costs of Compliance with Proposed RCRA
Regulations for Hazardous Waste Storage, Treatment, and Accumulation Tank
Facilities, a draft report to the Office of Solid Waste, EPA, March 1985.
-------
EXHIBIT 4-9
DISTRIBUTION OF STORAGE TANKS BY TYPE AND MATERIAL OF CONSTRUCTION
Frequency of Occurrence/Percent of AlI Storage Tanks/Percent of Type of
Storage Tanks/Percent of Storage Tank Material of Construction
Material of Construction
Type
Above-ground
(< 25,000 gal.)
Above-ground
(> 25,000 gal. )
In-ground
Underground
TOTAL
Ca rbon
Stee 1
2,311
15.5
78.3
61.0
740
14.6
91.0
19.5
104
2.0
26.7
2.7
632
12.4
66.6
16.7
3,790
74.6
Sta inless
Steel
292
5.7
9.9
65.8
28
0.6
3.6
6.3
23
0.5
5.9
5.2
101
2.0
10.6
22.7
111
8.7
Concrete
21
0.1
0.7
6.3
1
0.2
1.3
3.0
243
1.8
62.3
72.5
61
1.2
6.1
18.2
335
6.6
Fiberglass
251
1.9
8.5
70.1
0
0
0
0
16
0.3
1.1
4.5
91
1.8
9.6
25.4
358
7.0
Aluminum
9
0.2
0.3
10.0
0
0
0
0
0
0
0
0
0
0
0
0
9
0.2
Other a./
68
1.3
2.3
46.9
9
0.2
1.1
6.2
4
0.1
1.0
2.8
64
1.3
6.7
44.1
145
2.9
Tota 1 b/
2,955
58.2
787
15.5
390
7.7
949
18.7
5,081 c/
100
a/ Tanks indicated by the survey respondents to be made of wood or "other" materials.
b/ Totals may not agree with source data due to rounding.
c/ Does not include 210 "other" types and 164 "Not Accounted For"; total = 5,455.
Source: OSW RIA Tank Survey Data Base.
I
to
O
-------
4-21
Above-ground storage tanks larger than 25,000
gallons;
In-ground storage tanks; and
Underground storage tanks.
We considered obtaining these data separately for each of the seven selected
model storage tanks. Our review of the survey methodology used to develop the
OSW RIA Data Base, however, indicated that these data would not support
distinctions among tanks of the same type that are constructed of different
materials. Based on the data obtained, we characterized the seven model
storage tanks as follows:
Location
Above-ground
<25,000
Above-ground
£25,000
Underground
Underground
Underground
In-ground
In-ground
Material
Carbon steel
Carbon steel
Carbon steel
Fiberglass
Stainless steel
Concrete
Concrete
Open or
Closed Top
Closed
Closed
Median
Age
(Years)
Open
Carbon steel
21
7
7
7
8
8
Median
Size
(Gallons)
5,500
210,000
4,000
4,000
4,000
2,100
2,100
Tanks
Per
Facility a/
4
4
4
4
4
a/ Median number of tanks of all types per facility with storage tanks.
Treatment Tanks
The procedure we used to select model treatment tanks was essentially
identical to that for storage tanks because the source of readily available
data is the same. Data from the OSW RIA Data Base on tank type and material
is summarized in Exhibit 4-10. As shown, the majority of above-ground tanks
less than 25,000 gallons are made of carbon steel. Thus, we selected a carbon
steel model tank to represent the entire population of above-ground treatment
tanks with capacities less than 25,000 gallons.
-------
EXHIBIT U-10
DISTRIBUTION OF TREATMENT TANKS BY TYPE AND MATERIAL OF CONSTRUCTION
Frequency of Occurrence/Percent of All Treatment Tanks/Percent of Type of
Treatment Tanks/Percent of Treatment Tank Material of Construction
Material of Construction
Type
Above-ground
(< 25.000 gal.)
Above-g round
(>25,000 gal.)
In-ground
Unde rg round
TOTAL
Ca rbon
Steel
1.317
39.1
66.1
80,2
181
5.1
81.2
11.0
132
1.0
12.5
8.0
12
0.1
16.1
0.7
1,612
19.1
Stainless
Steel
160
1.8
8.0
38.1
12
0.1
5.1
2.9
218
7.1
23.5
59.0
0
0
0
0
120
12.6
Concrete
71
2.2
3.7
9.7
26
0.8
11.7
3.1
606
18.1
57.1
79.5
56
1.7
76.7
7.3
762
22.8
Fiberglass
320
9.6
16.1
89.9
__
--
--
--
36
1.1
3.1
10.1
0
0
0
0
356
10.6
Aluminum
13
0.1
0.6
10.6
_-
--
-
--
19
0.6
1.8
59.1
0
0
0
0
32
1.0
Other
107
3.2
5.1
81.7
1
0.1
1.8
3.1
15
0.1
1.1
11.5
5
0.1
6.8
3.8
131
3.9
Total a/
1,991
59.6
223
6.7
1,056
31.6
73
2.2
3,313 b/
100
a/ Totals nay not agree with source data due to rounding.
b/ Does not include 11 "other" types and 185 "Not Accounted For"; total = 3,569.
Source: OSW RIA Tank Survey Data Base.
-------
4-23
For above-ground treatment tanks larger than 25,000 gallons, we also
selected carbon steel as the construction material for the model tank because
the vast majority (81.2 percent) of this type of tank are constructed with
carbon steel. For in-ground tanks, Exhibit 4-2 shows that concrete, stainless
steel and carbon steel are all relatively common. In addition, Exhibit 4-2
shows that underground tanks comprise only a very small portion of the
treatment tank population. Thus, we decided to let underground treatment
tanks be represented by in-ground model tanks. Based on these data, we
selected concrete, stainless steel and carbon steel in-ground model tanks to
represent 58.6, 25.2, and 16.2 percent of the combined in-ground and
underground treatment tank population, respectively.
Next, we obtained information from the OSW RIA Data Base on the distribu-
tion of tank size, age, and top (closed or open), and the median number of
tanks at facilities with treatment tanks for each of the following three types
of treatment tanks:
Above-ground treatment tanks smaller than 25,000 gallons;
Above-ground treatment tanks larger than 25,000 gallons; and
In-ground treatment tanks.
We used these data to further characterize each of the selected model treat-
ment tanks in a manner identical to that described above for storage tanks.
Additional information from the data base on commonly used treatment
technologies was also used.
The data on treatment technologies indicated that two treatment processes
most commonly associated with above-ground tanks less than 25,000 gallons are
distillation and oxidation/reduction/precipitation. Because distillation
requires a closed-top tank while oxidation/reduction/precipitation is normally
conducted in open tanks, we used two model treatment tanks to represent
above-ground treatment tanks less than 25,000 gallons. Because 60 percent of
the above-ground treatment tanks less than 25,000 gallons are open-topped, the
open-top model tank represents 35.8 percent (0.6 x 59.6) of the treatment tank
population while the closed-top model tank represents 23.8 percent (0.4 x
59.6) of the population.
-------
4-;
The resulting model treatment tank, characteristics are as follows:
Location
Above-ground
<25,000 gal.
Above-ground
<25,000 gal.
Above-ground
>25,000 gal.
In-ground
In-ground
In-ground
Material
Carbon steel
Carbon steel
Carbon steel
Concrete
Carbon steel
Stainless steel
Open or
Closed Top
Open
Closed
Open
Open
Open
Open
Median
Age
(Years)
10
10
10
Median
Size
(Gallons)
2,300
2,300
60,000
3,700
3,700
3,700
Tanks
Per
Facility a/
5
5
5
a/ Median number of tanks of all types per facility with treatment tanks.
Accumulation Tanks
Data collected as part of the OSW RIA Generator Survey and maintained in
the OSW RIA Data Base indicate that the accumulation tank population consists
of approximately 6,400 tanks. The data base, however, does not provide
information on tank capacity, material of construction, age or other
characteristics of interest for estimating human health risks. Thus, we chose
to represent accumulation tanks in this analysis with the same seven model
tanks presented above for storage tanks. In doing so, we also assumed that
these seven model tanks are distributed the same for both accumulation and
storage tanks (e.g., the concrete in-ground model tank represents 4.8 percent
of all accumulation tanks).
Small Quantity G«n«rator (SQG) Tanks
As noted above, the source of readily available information on the
population of tanks used to store hazardous waste at facilities that generate
between 100 and 1000 kg per month of hazardous waste is the OSW SQG Survey.10
The survey indicates that there are about 9,240 SQG facilities with tanks in
18Abt Associates, Inc., National Small Quantity Hazardous Waste
Generator Survey. EPA, 1985.
-------
the primary Standard Industrial Classifications (SICs) included in the SQG
Survey. Inclusion of information from' the survey on secondary SICs increases
the total number of facilities with tanks to 11,388. The distribution of
these 11,388 facilities with above-ground and underground tanks is as follows:
Number of
Type of Tank Faciliites
Above-ground only 4,745
Underground only 4,935
Both above-ground and
underground
TOTAL
However, the data do not indicate the number of tanks at each facility. For
this analysis, we assumed that each facility has two tanks, so the total SQG
tank population is estimated to be 22,776, or about 23,000. We believe that
this is a conservative (i.e., high-side) assumption because our experience
indicates that there are relatively few instances in which quantities of waste
generated by a small quantity generator facility will be large enough to
require two tanks.
The survey also does not provide information on tank construction
materials. Because the data presented in Exhibits 4-9 and 4-10, and
engineering judgement, indicate that the vast majority of hazardous waste
tanks are carbon steel, we selected carbon steel as the construction material
for both the underground and above-ground SQG model tanks.
The survey does provide some information on tank size in that it indicates
the average amount of waste generated per month and the average storage
duration. Using these data for facilities that only store wastes in tanks, we
calculated the estimated tank size distributions shown in Exhibit 4-11.ll
We assumed that all SQG tanks will be operated as accumulation tanks (to avoid
the need to obtain a RCRA storage permit) if the proposed tank standards are
applied to SQG facilities. Therefore, we used the distribution for
above-ground tanks that store for less than 180 days as the basis for the SQG
llWe estimated tank capacity by assuming that it was equal to the
average quantity of waste accumulated before it is removed for disposal. We
calculated the average quantity of waste accumulated by multiplying the
storage time in months by the storage rate in kg/mo. We converted the
resulting waste quantity (in kg) to gallons using a conversion factor of
0.2688519 (which assumes a waste specific gravity equal to that of water).
-------
4-26
EXHIBIT 4-11
SQG TANK SIZE DISTRIBUTIONS
Above-ground Tanks (greater than 180-day storage)
Size Cumulative
(Gallons) Percent
500 15
750 82
1,500 88
1,750 100
Above-ground Tanks (less than 180-day storage)
Size
(Gallons)
S50
100
250
500
1,000
1,500
Cumulative
Percent
18
31
82
90
98
100
Underground Tanks
Size Cumulative
(Gallons) Percent
50 25
200 75
500 100
-------
4-27
above-ground model tank size. The majority of above-ground SQG tanks (82
percent) appear to be 250 gallons or less and the majority of underground SQG
tanks (75 percent) appear to be 200 gallons or less. Thus, we selected 200
gallons as the tank size for both the above-ground and underground SQG model
tanks. To characterize tank age, we used the median ages for underground and
above-ground storage tanks from the OSW RIA tank survey.
Based on the foregoing data and assumptions, we selected two model tanks
to represent the SQG tank population with the following characteristics:
Location
Material
Above-ground Carbon steel
Underground Carbon steel
Open or
Closed Top
Closed
N/A
Median
Age
(Years)
6
7
Median
Size
(Gallons)
200
200
Tanks
Per
Facility a/
2
2
a/ Number of tanks of all types per facility.
Piping
Because some of the regulatory requirements under consideration affect
ancillary equipment as well as tanks, development of the model tanks required
inclusion of information on associated piping characteristics. Because the
data sources discussed above do not contain any information on piping, we
obtained additional information and made assumptions as necessary to develop
piping characteristics. Based on public comments on the proposed regulations
for hazardous waste tanks (50 CFR 26444, June 26, 1985) and conversations with
trade associations, tank manufacturers, and other industry representatives, we
estimated that a typical storage, treatment, or accumulation tank has 50 feet
of piping for each tank if the facility has fewer than 5 tanks and 200 feet of
piping per tank if the facility has five or more tanks.
Because somewhat less than half of the facilities have five or more tanks,
we used 100 feet as the length of piping associated with each storage,
treatment, SQG, and accumulation model tank. In addition, we assumed that all
piping associated with underground tanks is underground. For above-ground and
in-ground tanks, we assumed that all of the piping is above-ground and located
within the tank secondary containment area (if there is one).
For all of the model tanks, we assumed that the piping is made of the same
material as the tank, except in the case of concrete tanks where we assumed
that the associated piping is carbon steel.
Summary
The cumulative result of the process described above was the selection of
22 model tanks for use in estimating human health risks using the Hazardous
Waste Tank Risk Analysis Model. These 22 model tanks are summarized in
Exhibit 4-12.
-------
EXHIBIT 4-12
MODEL TANK CHARACTERISTICS SUMMARY
TECHNOLOGY
TREATMENT :
STORAGE:
SMALL QUANTITY
ACCUMULATION:
LOCATION
above-ground; on cradles
above-ground; on cradles
above-ground; ongrade
in-ground
in-ground
in-ground
above-ground; on cradles
above-ground; ongrade
be 1 ow-g round
below-ground
below-ground
in-ground
in-ground
GENERATORS:
above-ground
below-ground
above-ground; on cradles
above-ground; ongrade
be 1 ow-g round
below-ground
below-ground
in- ground
in-ground
OPERATING
CAPAC 1 TY
(gal Ions)
<25,000
<25,000
>25,000
--
--
--
<25,000
>25,000
--
--
--
--
200
200
<25.000
>25,000
--
--
--
--
MATERIAL
carbon steel
carbon steel
carbon steel
concrete
carbon steel
stainless steel
carbon steel
carbon steel
carbon steel
fiberglass
stainless steel
concrete
carbon steel
carbon steel
carbon steel
carbon steel
carbon steel
carbon steel
fiberglass
stainless steel
concrete
carbon steel
OPEN OR
CLOSED TOP
closed
open
open
open
open
open
closed
closed
--
--
--
open
open
closed
closed
closed
--
--
open
open
MEDIAN
AGE
(Years)
5
5
9
10
10
10
6
21
7
7
7
8
8
6
7
6
21
7
7
7
8
8
MEDIAN
SIZE
(ga 1 Ions)
2,300
2,300
60,000
3,700
3.700
3,700
5,500
210,000
4,000
4,000
4,000
2,100
2,100
200
200
5,500
210,000
1,000
4,000
4,000
2,100
2,100
PERCENIAGL OF
TECHNOLOGY
.238 a/
.358
.067
. 198
.054
.085
.582 b/
.155
. 1U9
.018
.020
.OU8 ft
.029 N
o
.1476 c/
.524
.582 d/
. 155
. 149
.018
.020
.048
.029
a/ Total number of treatment tanks is 3,369.
b/ Total number of storage tanks is 5,455.
c/ Total number of small quantity generator tanks is 22,776.
d/ Total number of accumulation tanks is 6,400.
-------
4-29
4.2.3 Assumptions and Limitations
As indicated throughout the preceeding discussion, data limitations
required that we make several assumptions in order to select the model tank
technologies. The more important of these assumptions are:
Above-ground tanks with a capacity less than or equal
to 25,000 gallons are installed above the ground on
legs or cradles;
Above-ground tanks with a capacity greater than
25,000 gallons are installed ongrade;
The model storage tanks derived from the OSW RIA Data
Base are also representative of accumulation tanks;
All piping for underground tanks is underground while
all piping for above-ground and in-ground tanks is
above-ground;
SQG tanks are all 200 gallon, carbon steel tanks, and
either on cradles or completely underground; and
SQG facilities with tanks have two tanks.
In addition to these assumptions, the approach used to develop the model
tanks results in several limitations that have implications for the overall
analysis. The most significant limitation is that the approach used does not
represent the extremes that exist in the tank population. For example,
representing all above-ground storage tanks smaller than 25,000 gallons as
carbon steel fails to capture any risk differences that may exist for such
tanks made from other materials. Similarly, use of median characteristics
such as age and size does not capture the extremes in the population.
4.3 SELECTION OF REPRESENTATIVE WASTE STREAMS
Hazardous waste tanks handle an extremely wide variety of substances, and
it is probably safe to assume that any hazardous waste, with the exception of
pure gases and solids, may at one time be stored in a tank. However, for the
purpose of this analysis, we considered it most important to identify a
relatively small number of waste streams that would represent the wastes most
commonly stored and treated in hazardous waste tanks. Our objective was to
analyze the differences in risk associated with different hazardous wastes
handled in tanks, while minimizing the complexity of the analysis. Because
our analysis addressed chronic human health risks due to contaminant releases
from hazardous waste tanks, we focused only on wastes containing chronically
toxic chemicals. Thus, our selection criteria for waste streams included not
only total quantity handled in tanks, but also toxicity of the constituents in
each waste stream. By selecting waste streams with the most toxic
-------
4-30
constituents, we were able to obtain conservative risk estimates. We did not
consider concentrations of toxic constituents in a particular waste stream in
our selection criteria because, although concentration does influence risk,
there is a great deal of uncertainty as to the concentration of a constituent
in any particular waste, and we were able only to identify generic average
values for our analysis.
We did not begin our selection process with any set number of waste
streams in mind for any tank category, but rather with the objective of
including as representative a sample as possible of wastes handled in tanks
without allowing the analysis to become overly computationally intensive.
Thus, our general selection methodology ususally involved the compilation of
one or more ranked lists of waste streams for a particular category. The
number of waste streams for a category was then selected based on an obvious
cut-off point (according to total volume handled) in these lists.
For this analysis we identified waste streams for each technology, even
though we only had data for each tank category. The most detailed available
data characterizes waste streams by tank location (i.e., above-ground or
underground). Consequently, we identified waste streams for each tank
category using professional judgment and available information on the type of
wastes stored in above-ground and underground tanks. We assumed that some
wastes streams were incompatible with tank materials.
In this section, we describe the methodologies used to select
representative waste streams for each of the four modeled tank categories, and
to characterize the selected waste streams. In the following four sections we
present the selection methodology and representative waste streams for SQG
tanks, RCRA-permitted storage tanks, accumulation tanks, and treatment tanks,
respectively.
4.3.1 Small Quantity Generator Tank Waste Streams
Although only a small number of data sources containing information on SQG
wastes were available, these sources generally were quite comprehensive in
terms of information on waste streams and waste storage and treatment
technologies. We identified two major sources:
National Small Quantity Hazardous Waste Generator
Survey. (February, 1985). This survey provides estimates
of the number and type of small quantity generators, types
of waste generated, and their waste management practices.
The report contains data on 22 industry groups and the
number of generators in each group, the amount of waste
generated, and the percentages of wastes stored and treated
on- and off-site. This survey also gives information on
the percentage of each waste type stored in above-ground
and underground tanks. It does not provide detailed waste
characterization information, only general descriptions of
the included waste streams.
-------
.-31
Economic Analysis of Resource Conservation and Recovery
Act Regulations for Small Quantity Generators, (June,
1985). This report contains information on SQG industries
(e.g., dry cleaning), and also includes data on specific
SQG waste streams. These data include total quantity
generated (kg/month) and total number of generators for
each of 39 waste streams; also included for each waste
stream are characteristics such as constituents of concern
and concentrations.
By combining these data sources, we compiled information on the quantity
and number of generators for the most common SQG waste streams, waste stream
characteristics (i.e., most toxic constituents, or constituents of concern and
concentrations of these constituents), waste stream storage in above-ground or
underground tanks, and the relative frequency of occurrence for each waste
stream on a nationwide basis, i.e., the percentage of the total SQG waste
quantity handled in tanks that each waste stream represents.
Due to the availability of information on SQG waste streams that are
stored and/or treated in tanks, the selection of waste streams for inclusion
in the Tank Risk Analysis Model was relatively straightforward. We selected
waste streams using the following methodology:
Ranked all SQG waste streams in terms of total waste
quantity generated (kg/month) and number of generators.
Identified which SQG waste streams are stored in tanks.
Ranked all SQG waste streams that are stored in tanks in
terms of total waste quantity generated and total number of
generators, and number of tanks storing each waste stream.
Divided the waste streams into those usually stored in
above-ground and underground tanks.
Characterized each waste stream by its constituents of
concern.
Assigned each waste stream a frequency of occurrence
(i.e., percentage of the total SQG waste quantity handled
in tanks that each waste stream represents).
We first identified all wastes generated by small quantity generators. We
identified 39 waste streams. We then compared the list of generated wastes
with wastes that are stored in above-ground or underground tanks. We
eliminated from the ranked list all waste streams that are not stored in tanks
according to SQG survey data. The SQG waste streams that we eliminated
included the highest-ranked waste stream (in terms of quantity and number of
generators), lead-acid batteries, and the seventh-ranked waste stream, waste
formaldehyde.
-------
4-32
To reduce the number of waste streams to be analyzed, we ranked the waste
streams according to: 1) the total quantity of waste generated and total in
number of generators; and 2) the total number of tanks storing each waste
stream. We then assigned a composite ranking to each waste stream based on
its position in each of these two lists. The final list of ranked waste
streams that we used for this study is illustrated in Exhibit 4-13.
These 16 waste streams were then categorized according to whether they
were commonly stored in above-ground or underground tanks. It was found that
only one waste stream, waste ink with solvents or heavy metals, was unique to
underground tanks, while cyanide wastes, spent plating wastes, solutions or
sludges with photo silver, and heavy metal wastewater sludges are stored only
in above-ground tanks. The waste streams included in each category are listed
in Exhibit 4-14.
Using available information, we identified the constituents of concern and
constituent mass fractions for each waste stream. As expected, some of the
waste streams had very similar, or, in some cases identical, constituents.
For example, ignitable paint wastes and ignitable waste both contained toluene
and methyl ethyl ketone. Including each of these waste streams in the risk
assessment increases the complexity of the analysis but provides very little
additional information on risk. In order to reduce the complexity of the
analysis, we combined waste streams rather than including individual waste
streams in cases where several of the waste streams have very similar
characteristics. Similar waste streams were combined, using a weighted
average based on total waste quantity generated, to determine characteristics
for the composite, or aggregate, waste streams.
We identified a total of five combined waste streams. The waste streams
and associated characteristics are presented in Exhibit 4-15. The combined
waste streams are: 1) spent solvents/ignitable wastes/ignitable paint wastes;
2) waste pesticides/pesticide washing and rinsing solutions; 3) heavy metal
wastewater sludges/cyanide wastes/spent plating wastes/other reactive wastes;
4) photographic wastes/solutions and sludges with photo silver, and 5) waste
ink with solvents or heavy metals/ink sludges with chromium or lead. These
waste streams were found to contain similar constituents at similar concentra-
tions, and therefore combining them will not result in an over- or under-
estimation of risk, which can occur when dissimilar waste streams are combined.
For example, ignitable wastes, ignitable paint wastes, and spent
non-halogenated solvents all contain waste solvents at similar concentrations,
and, while the solvents are not exactly the same in all cases, their
toxicities, mobilities, and degradation rates do not vary significantly, and
thus they present similar risks. Also, while these wastes may contain
different constituents in some cases (e.g. ignitable paint wastes may contain
heavy metals in addition to solvents) the constituents that dominate risk are
very similar for these wastes. If, however, these waste streams were aggre-
gated with a pesticide waste stream containing a high-risk compound (e.g.,
lindane), the risk associated with this composite waste stream would probably
be greatly over-estimated when compared to the real situation. Therefore, we
avoided combining any but the most similar waste streams in this analysis.
-------
4-33
EXHIBIT 4-13
SQG WASTES STORED IN TANKS RANKED ACCORDING TO TOTAL WAST
QUANTITY GENERATED AND TOTAL NUMBER OF TANKS
Composite
Ranking
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
a/ Source:
and Recovery
Total Waste Total Number
Quantity of Tanks
Waste Stream Ranking a/ Ranking b/
Spent Solvents
Strong Acid or Alkaline Wastes
Photographic Wastes
Sludges or Solutions w/Photo Silver
Ignitable Paint Wastes
Filtration Residue from Dry Cleaning
Heavy Metal Wastewater Sludges
Waste Pesticides
Spent Plating Wastes
Ignitable Wastes
Other Reactive Wastes
Cyanide Wastes
Pesticide Washing & Rinsing Solution
Ink Sludge w/ Chromium or Lead
Waste Ink with Solvents and Heavy Metal
Wastewater Wood Preservatives
ICF Inc., et al., Economic Analysis of
1
2
3
4
5
9
6
7
10
14
8
11
15
13
13
15
1
2
3
5
7
4
10
12
9
6
13
11
8
15
15
16
Resource Conservation
Act Regulations for Small Quantity Generators,
report to the
Office of Solid Waste, EPA, June 1985.
b/ Source: SQG Survey.
-------
EXHIBIT 4-14
SQG WASTE STREAMS STORED IN
ABOVE-GROUND AND UNDERGROUND TANKS a/
Spent Solvents
Strong Acid or Alkaline Wastes
Photographic Wastes
Sludges or Solutions w/Photo Silver
Ignitable Paint Wastes
Filtration Residue from Dry Cleaning
Heavy Metal Wastewater Sludges
Waste Pesticides
Spent Plating Wastes
Ignitable Wastes
Other Reactive Wastes
Cyanide Wastes
Pesticide Washing & Rinsing Solutions
Ink Sludge w/Chromium or Lead
Waste Ink with Solvents and Heavy Metals
Wood Preservative Wastewater
Above-ground Tanks
X
X
X
X
X
X
X
X
X
X
X
X
X
X
None
X
Underground Tanks
X
X
X
None
X
X
None
X
None
X
X
None
X
X
X
X
a/ Wastes are not ranked in this Exhibit.
-------
EXHIBIT U-15
REPRESENTATIVE SQG WASTE STREAMS
Waste Stream Description a/
Number of
Tanks b/
Constituent Name c/d/
Const!tuent
Mass
Fraction c/
Relat ive
Frequency of
Occurrence b/e/
Spent Solvents/
Ignitable Paint Wastes/
Ignitable Wastes
1967 (ag)
1286 (ug)
Strong Acid or Alkaline
Wastes
Photographic Wastes/
Solutions or Sludges with
Photo SiIver
Filtration Residues from
Dry Cleaning
Heavy Metal Wastewater
Studges
Spent Plating Wastes/
Cyanide Wastes/
Other Reactive Wastes
1358 (ag)
295 (ug)
820 (ag)
56 (ug)
63 (ag)
31 (ug)
305 (ag)
Halogenated stream
(category 1)
Chloroform 0.10
Carbontetrachloride 0.10
1,2-Dichlorobenzene 0.10
1,1,1-Trichloroethane 0.10
Trichloroethylene 0.10
Non-halogenated stream
(category 2)
Toluene 0.10
Toluene diisocyanate 0.10
Methyl ethyl ketone 0.10
Benzene 0.10
Xylene 0.10
Acetaldehyde 0.10
Lead 0.0005
Hydrochloric acid 0.05
(or) Sodium hydroxide 0.05
Methanol 0.14
Acetone 0.1U
PhenoI 0.09
Cyanide 0.00000057
Tetrachloroethylene 0.53
Chromium (VI) 0.0110
Copper 0.0051
Cadmium 0.0031
Lead 0.00018
Nickel 0.0105
Cyanide 0.0093
.20U (ag)
.32) (ug)
.204 (ag)
.320 (ug)
.282 (ag)
.147 (ug)
.170 (ag)
.028 (ug)
.013 (ag)
.015 (ug)
.06*4 (ag)
£-
Ul
-------
EXHIBIT «4-15 (continued)
REPRESENTATIVE SQC WASTE STREAMS
Waste Stream Description a/
Waste Pesticides/
Pesticide Washing and
Rinsing Solutions
Wood Preservative
Wastewaters
Waste Ink with Solvents
or Heavy Metals/
Ink Sludge with Chromium
or Lead
Number of
Tanks b/
160 (ag)
180 (ug)
16 (ag)
< (ug)
158 (ug)
Constituent Name c/d/
Lindane
(category 1)
A 1 d i ca rb
(category 2)
Acenaphthene
Chrysene
Pentachlorophenol
Chromium VI)
Lead
To 1 uene
Const i tuent
Mass
Fraction c/
0.01
0.01
0.00030
0.00010
0.00078
0.00061
0.00032
0.00010
Relative
Frequency of
Occurrence b/e/
.017
.015
.017
.0«»5
.028
.002
.079
(ag)
(ug)
(ag)
(ug)
(ag)
(ug)
(ug)
§/ In order to simplify the analysis, similar waste streams were combined in some cases.
b/ Obtained from data supplied to OSW by OPRA (ag = above-ground, ug = underground) .
c/ Constituents and mass fractions were identified from SQG Economic Analysis Report (OSW-ICF/DPRA/PRA); For waste streams
whose characteristics were not identified in this report, professional judgement was used to identify constituents and
mass fractions.
d/ Categories were identified using professional Judgement.
§/ Assumes that selected wastes represent 100 percent of all wastes handled in SQG tanks.
-------
4-37
Finally, after identifying the important SQG waste streams that are stored
in above-ground and underground tanks, we assigned characteristics to these
waste streams from data contained in the economic analysis report. We then
combined similar waste streams and calculated the relative frequency of
occurrence for each waste stream. We obtained the total number of
above-ground and underground SQG tanks, and divided the number of tanks for
each waste stream by the total number of SQG storage tanks for that tank type
(above-ground or underground). For example, DPRA, Inc. identified 4,809
above-ground SQG storage tanks, of which 1,358 are used to store strong acid
or alkaline wastes. Thus, the relative frequency of occurrence for that waste
stream in above-ground tanks is 1,358/4,809, or 28.2 percent. These
frequencies represent the waste stream weights assigned to the two model SQG
tanks.
Exhibit 4-15 also presents waste stream characteristics and frequency of
occurrence for each of the identified SQG waste streams. We identified a
total of eight types of waste streams, five of which are combined waste
streams. Of note is that in some cases it was necessary to supplement the
available data with our judgement as to what was contained in a particular
waste stream. For example, information on waste pesticides and pesticide
washing and rinsing solutions did not specify which pesticides are disposed of
more frequently than others. Since constituents must be specified in order to
estimate risk, we included two pesticide waste streams; one containing a
highly toxic, highly persistent, but relatively non-mobile pesticide, such as
lindane; and one containing a more mobile but less toxic pesticide, such as
aldicarb. These two waste streams represent a large variety of pesticides
currently in use. We assumed that each of these waste stream categories have
equal frequencies of occurrence.
A similar problem arose in the case of spent solvent wastes, where a total
of 11 solvents may commonly be found. The toxicities, mobilities, and
degradation rates of these substances vary considerably, and it is also
unlikely that all would be found in any single waste stream. Thus, if all of
these constituents are included in all situations, the risk associated with
this waste stream may be significantly over-estimated as a result of the
inclusion of high-risk compounds such as carbon tetrachloride that would
probably not always be present. Therefore, we have divided spent solvents
into two waste stream categories, halogenated and non-halogenated solvents,
with equal frequencies of occurrence, as is illustrated in Exhibit 4-15.
These assumptions made with respect to waste stream constituents and
constituents' concentrations have a significant impact on the results of the
risk analysis. However, we believe that the waste streams selected are
reasonably representative of the SQG waste streams most commonly stored in
tanks, and thus are useful in deriving representative risk distributions for
these tanks.
4.3.2 RCRA-Permitted Storage Tank Waste Stream*
We used data from several sources to identify waste streams that are
generated in the greatest quantity and are stored in RCRA-permitted tanks.
-------
Our selection process was based on ranking the waste streams according to
available information on total waste stream quantity generated, total quantity
handled in tanks, number of facilities handling a waste stream, and the
chronic toxicity of the waste streams' constituents of concern. The data
sources that we used included the following:
The RCRA Risk-Cost Analysis Model Waste Stream Data
Base, (July 1984) (referred to in this analysis as
the W-E-T Model data base). This data base contains
information on 267 hazardous waste streams collected
from 106 references. It contains data on waste
quantity generated, number of generators, and waste
stream characteristics, but does not contain specific
information on the processes used to treat, store,
and/or dispose of the included waste streams.
The National Survey of Hazardous Waste Generators
and Treatment, Storage, and Disposal Facilities
Regulated Under RCRA in 1981. (April 1984) (referred
to in this analysis as the OSW RIA Mail Survey). This
survey, while lacking specific information on waste
stream characteristics is a more comprehensive data
source in that it identifies waste management
practices, and the waste streams managed in these
practices, at RCRA Part A-permitted facilities as of
1981. Waste streams in the survey are identified by
EPA waste code, and are characterized by treatment,
storage, and/or disposal method.
Appendices for Supporting Documentation for the
RCRA Incinerator Regulations. 40 CFR 264 Subpart D -
Incinerators, (July 1984) (referred to in this
analysis as the Incinerator Data Base). This data
base contains information on characteristics for
unspecified RCRA waste streams (e.g., D001).
Documentation for the Development of Toxicity and
Volume Scores for the Purpose of Scheduling Hazardous
Wastes. (March 1985) (referred to in this analysis as
the ENVIRON Toxicity-Volume Report). This data source
contains information on chronic toxicity for more than
100 hazardous chemicals.
The W-E-T model data base served two major purposes in this analysis: (1)
it provided a great deal of information on specific waste stream characteris-
tics; and (2) it represented a "master list" of the largest quantity hazardous
waste streams generated in the U.S., which was useful for comparison purposes
to the more tank-specific information included in other data sources.
-------
In our selection process, we first compiled five lists of relevant waste
streams by identifying the largest quantity of waste streams handled in tanks
or disposal processes. Using information from the OSW Mail Survey, we
identified and listed the top 50 waste streams (in terms of total quantity
handled) for each of the following waste categories:
stored in tanks;
treated in tanks;
treated in other processes;
disposed of by injection; and
disposed of by incineration.l2
We eliminated wastes that are most commonly landfilled, treated or disposed of
in surface impoundments, and handled in land treatment systems and waste
piles. A further reduction in the total number of wastes considered was made
by eliminating solid or gaseous waste streams (e.g., heavy metal dust,
industrial gases, etc.).
These five lists, each containing 50 ranked waste streams, provided
tank-specific information on storage tank waste streams, and, along with the
W-E-T Model "master list", allowed for extensive cross-referencing in
identifying the waste streams most commonly handled in tanks. However,
characteristics for EPA D waste codes, which were ranked highly in nearly all
the above categories, were not included in either the OSW or W-E-T Model data
bases. The Incinerator Data Base was therefore used to characterize D wastes
(i.e., to identify the constituents of concern for D wastes).11
Finally, we evaluated the waste streams in terms of the toxicity of the
associated constituents. We assigned toxicity scores on a relative scale
based on available information and used the most toxic constituent in each
waste stream as an overall indicator of waste toxicity.1*. These toxicity
scores were used to supplement waste stream quantity and tank-specific
information in the final selection process.
12For the purposes of the analysis it was assumed that wastes disposed
of by injection, incineration, and other processes would be handled in tanks
at some point in the treatment/storage/disposal process. Consequently, we
included these waste streams in our analysis.
1'Wastes that are to be incinerated must have relatively high contents
of flammable organic liquids. Consequently, we identified constituents of
concern such as toluene, acetone, and methanol for D wastes. However, D
wastes may contain a wide variety of substances, some of which may pose very
little chronic risk. Because we assigned relatively high-risk constituents to
D wastes, our analysis likely over-estimates the risks attributable to waste
streams D001, D002, and D003.
lfcToxicity values were obtained from Documentaion for the Development of
Toxicity and Volume Scores for the Purpose of Scheduling Hazardous Waste,
report to the Office of Solid Waste, EPA, by Environ Corp., March 1985.
-------
4-40
Exhibit 4-16 presents information on 21 waste streams that we identified
as the most representative of the wastes commonly handled in RCRA-permitted
storage tanks. We used total quantity handled in tanks as our primary
selection criterion. In cases where several wastes were handled in similar
quantities, we chose the waste containing the most toxic constituents. The
list includes aqueous waste streams containing heavy metals and cyanides,
spent solvents, and concentrated U wastes such as phenol and dichloromethane.
Once representative waste streams were selected, it was necessary to
distribute them over the population of model storage tanks considered in the
analysis. In order to accomplish this, we used the dominant RCRA
characteristic (ignitability, corrosivity, reactivity, EP toxicity, or
toxicity) assigned by EPA for each waste code. We then used data from the OSW
Hazardous Waste Tank RIA Survey (part of OSW Mail Survey) on the percentages
of above-ground, in-ground, and underground tanks storing each type of RCRA
characteristic waste to determine the percentage of each tank type storing a
particular waste stream. Of note is that approximately 50 percent of the
facilities surveyed in the RIA reported that their tanks were used to store
waste mixtures, and, therefore, we developed five waste streams (X500 through
X504 in Exhibit 4-16) to represent these mixtures. Based on information in
the RIA data base, we also determined the percentage of above-ground,
in-ground, and underground tanks storing each waste stream.
4.3.4 Accumulation Tank Waste Streams
Sources of data for hazardous waste accumulation tanks were neither as
comprehensive nor as plentiful as they were for hazardous waste treatment and
storage tanks (RCRA-permitted tanks) or for small quantity generator (SQG)
tanks. However, by cross-referencing between the available data sources, we
selected a reasonably representative sample of waste streams that are handled
in accumulation tanks. We used the following data sources:
RCRA Risk-Cost Analysis Model data base (W-E-T Model
data base). This data base contains detailed information
on 264 hazardous waste streams at the generator stage. It
is very comprehensive but does not contain information on
storage and disposal for the included waste streams; thus,
no direct evidence linking a particular waste stream with
hazardous waste tanks is available in these data.
EPA Hazardous Waste Generator Survey. This survey
contains information on waste streams (waste stream ID code
and frequency reported (number of facilities reporting a
particular waste stream)) and also contains information on
whether accumulation tanks are used at each of the surveyed
facilities. Thus, while there is no direct link between
-------
EXHIBIT *4-16
REPRESENTATIVE WASTE STREAMS FOR RCRA-PERMITTED STORAGE TANKS a/
Waste Stream EPA
Category Waste Code
Corrosive, NOS 0002
Reactive, NOS 0003
Acrylonitri le Wastes K011
EP Chromium 0007
Spent Halogenated F001
Solvents
Spent Non-ha logenated F003
Solvents
Ignitable, NOS 0001
Petroleum Refining Wastes K048
Constituent Dominant
b/ Constituent Name c/ Mass Fraction c/ RCRA Characteristics d/
Chloride
Me t ha no 1
Copper
Toluene
Acetone
To 1 uene
Cyanide
2,
-------
EXHIBIT 4-16 (Continued)
REPRESENTATIVE WASTE STREAMS FOR RCRA-PERMITTED STORAGE TANKS a/
Percentage of
Waste Stream
Category
Spent Pickle Liquor
Electroplating Wastes
Non-Cyanide Electroplating
Wastes
EP Arsenic
Dichlo rone thane
Pheno 1
Tha 1 1 i un Su 1 fa te
Chlorinated Pesticide
Production Wastes
Igni table Waste Mixture
Corrosive Waste Mixture
EPA
Waste Code
K062
F007
F006
D004
U080
U188
P115
X907
X500
X501
Constituent Dominant
b/ Constituent Name c/ Mass Fraction c/ RCRA Characteristics
Chromium
Lead
Cadmium
Copper
Cyanide
Chromium
Nickel
Copper
Ch roro i urn
Cadmium
Lead
Nickle
Arsenic
Dichlorome thane
Pheno 1
Tha 1 1 i urn Su 1 fa te
Hexach 1 enobu tad i ene
Hexach 1 o rocyc 1 opentad i ene
Hexach 1 o robenzene
Chloroform
Carbon Tetrachlorlde
Toluene
Toluene Diisocyanate
Methyl Ethyl Ketone
Benzene
Acetaldehyde
Acetone
Chloride
Methanol
To 1 uene
Ch row i urn
Lead
0.00001 Corrosive
0.0001
0.001 Reactive
0.00014
0.000075
0.00017
0.0015
0.0018 Toxic
0.00068
0.00035
O.OOOM
0.003365
0.0005 EP Toxic
1 . 000 Tox i c
1.000 Toxic
0.01 Toxic
0.10 Toxic
0.10
0.10
0.10
0.10
0.0894 Igni table
0.10
0.123
0.110
0.10
0.155
0.05 Corrosive
0.043
0.038
0.00001
0.0001
lanks Storii
d/ Waste Streai
0.010
0.0001
0.01 1
0.0001
O.OOI'I
0.0?5
0.0005
0.0056
o. i n
0. 1 18
-------
EXHIBIT U-16 (Continued)
REPRESENTATIVE WASTE STREAMS FOR RCRA-PERMITTED STORAGE TANKS a/
Waste Stream EPA
Category Waste Code
Reactive Waste Mixture X502
EP Toxic Waste Mixture X503
Toxic Waste Mixture X50U
Constituent Dominant -
b/ Constituent Name c/ Mass Fraction c/ RCRA Characteristics
Cyanide
Toluene
2,l4-Dini tro toluene
Acetoni tri le
Ac ry 1 on i t r i 1 e
Chromium
Arsenic
Chloroform
Carbon Tetrachloride
Trichloroethylene
1,1, 1-Trichlo roe thane
Hexachlorobenzene
Tetrachloroethylene
0.00733 Reactive
0.0838
0.1610
0.006
O.OOOU
0.0005 EP Toxic
0.0005
0.10 Toxic
0.10
0.10
0.10
0.10
0.10
Percentage of
Tanks Storing
d/ Waste Stream e/
0.0<48
0.076
0. 12*4
a/ For both the model above-ground storage tanks, we assumed they handle all the listed waste streams. The carbon
steel, underground tanks store all listed wastes, except for D003, K011, D007, F007, and D004. The fiberglass
reinforced plastic tanks store K062, D007, KOU8, F006, P115 and X504. The stainless steel underground tanks store all
wastes except for 0003, K011, 0007, F007 and 000*4. The concrete in-ground tanks store KOM8, F006, P115, and X50U.
The carbon steel in-ground tanks store K062, 0007, K048, F006, DOOU, P115 and 0002.
b/ Waste codes are combined in some cases for similar waste streams (e.g., F003 is used to represent F003-F005).
c/ Constituents and mass fractions taken from W-E-T Model data base.
d/ Dominant RCRA characteristics identified based on EPA waste code information and professional judgement.
e/ Calculated based on dominant RCRA characteristics; assumes that selected wastes represent 100 percent of all wastes
stored in RCRA-permitted storage tanks.
-------
4-44
waste streams and accumulation tanks in this data base, we
assume that for facilities having accumulation tanks, the
waste streams generated by these facilities most frequently
are most likely to be handled in accumulation tanks.
Using the above mentioned data bases we compiled three lists: a list of
the 50 waste streams generated in the greatest quantity (from the W-E-T Model
data base); a list of the most frequently reported waste streams for
facilities handling waste under the 90-day exclusion rule (from the Hazardous
Waste Generator Survey); and a list of the most frequently reported waste
streams for facilities handling waste in accumulation tanks.
These lists were cross-referenced, with most weight given to wastes
reported at facilities having accumulation tanks. For the most part, there
was generally good agreement between the three lists, although some wastes
that ranked high on the W-E-T Model list were not reported by hazardous waste
generators included in the Generator Survey. This probably indicates that
some blending of wastes occurs at generator facilities before on-site or
off-site disposal.
Based on a cross-reference of the three lists, a total of 12 waste streams
considered to be representative of those handled in accumulation tanks were
selected according to their rankings on each of the three lists. We excluded
wastes not represented on all three lists, and established a cut-off for
wastes based on frequency of reporting. These selected wastes were then
distributed over the population of model accumulation tanks according to the
frequencies reported for the waste streams from the Hazardous Waste Generator
Survey.
The 12 waste streams that we selected to represent wastes stored in
accumulation tanks are presented in Exhibit 4-17. For the most part, these
waste streams are very similar to the representative waste streams identified
in our analysis for RCRA-permitted storage tanks. However, as expected, more
EPA U code waste streams were identified for accumulation tanks than for
storage tanks. In general, accumulation tanks are more closely associated
with hazardous waste generators than with treatment, storage, and disposal
(TSD) facilities and thus one would expect to find more "generator" wastes
reported for accumulation tanks.
Accumulation tank waste stream information is not as reliable as the waste
stream data compiled for SQG and RCRA-permitted storage tanks in this
analysis, due to lack of data that would allow for the direct identification
of wastes most commonly handled in accumulation tanks. Therefore, the results
of our risk analysis are significantly more uncertain for these tanks. Also,
we had no data that would allow us to determine the percentages of above-
ground, in-ground, and underground tanks storing these waste streams and used
professional judgement to compile these percentages.
-------
EXHIBIT 1-17
REPRESENTATIVE ACCUMULATION TANK WASTE STREAMS a/
Percentage of
Waste Stream
Description Waste Code
Corrosive, NOS §/ 0002
Igni table, NOS D001
EP Chromium £/ D007
EP Lead D008
Spent Pickle Liquor K.062
Spent Non-Ha logenated F003
Solvents
Spent Ha logenated Solvents F001
t
1,1, 1-Trichloroethane U226
Electroplating Wastes FOOT
Constituent
b/ Constituent Name c/ Mass Fraction c/
Chloride
Me t ha no 1
Coppe r
To 1 uene
Methanol
To 1 uene
Acetone
Xy 1 ene
Benzene
Me thy 1 e thy I ke tone
Ch rora i urn
Lead
Chromium
Lead
Benzene
To 1 uene
Toluene diisocyanate
Methyl ethyl ketone
Xy 1 ene
Aceta Idehyde
Chloroform
Carbon tetrachloride
1 , 2-0 i ch I o robenzene
1,1, 1-Trichloroethane
Trichloroethylene
1, 1 , 1-Trichloroethane
Cadmium
Copper
Cyanide
Nickel
0.05
0.0132
0 . 0005
0.0381
0.0911
0.0695
0.1552
0.1912
0.1176
0.1101
0.0005
0 . 0005
0.000010
0.000100
0. 10
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.10
1.00
0.00010
0.00011
0.000075
0.0015
Tanks Storing
Waste Stream i
0.225
0.190
0.075
0.051
0.036
0.138
0.112
0.031
0.077
I
F-
t_n
-------
EXHIBIT U-17 (continued)
REPRESENTATIVE ACCUMULATION TANK WASTE STREAMS a/
Waste Stream
Description
Ac ry 1 on 1 1 r 1 1 e
D i ch 1 o rone thane
Pheno 1
Waste Code b/
U009
U080
U188
Const i tuent Name c/
Ac ry 1 on i t r i 1 e
Dichlorome thane
Pheno 1
Const i tuent
Mass Fraction c/
1.00
1.00
1.00
Percentage of
Tanks Storing
Waste Stream d/
0.002
0.022
0.010
a/ We assumed that both model above-ground accumulation tanks store all the listed waste
streams. The carbon steel, underground tanks store all listed wastes except for D002,
0007, 0008, and F007. The fiberglass reinforced plastic tanks store 0002 and K062. The
stainless steel underground tanks store all wastes except for 0007, 0008, and F007. The
concrete in-ground tanks store 0007 and DOOB. The in-ground carbon steel tanks store D002,
0007, 0006 and K062.
b/ Waste codes combined in some cases for similar waste streams (e.g., F003 represents F003
through F005.
c/ Constituents and mass fractions obtained from W-E-T Model data base.
d/ Calculated from total quantities reported in Small Quantity Generator Survey; assumes that
selected waste streams represent 100 percent of all wastes stored in accumulation tanks.
e/ NOS = not otherwise specified.
£/ EP = extraction procedure (waste has been determined to exceed the maximum contaminant
level allowed under EPA's Extraction Procedure Toxicity Test).
-------
4.3.5 Treatment Tank Waste Streams
Treatment tanks differ from the other tank categories included in the
analysis in that a wide variety of treatment processes are carried out in
tanks, and the type of waste treated in a particular tank is dependent upon
the treatment process for which that tank is used. For example, an oxidation/
reduction/precipitation system may be used to treat wastes containing heavy
metals and cyanides, but will not treat waste streams containing high
concentrations of organic solvents. Therefore, in order to simplify our
analysis, we restricted our selection to two treatment processes that we
believed were most representative of hazardous waste treatment tanks, and
selected waste streams according to their compatibility with the two treatment
processes. Data sources used in this selection process included:
The Hazardous Waste Tank RIA Data Base, compiled by OSW
in 1984; and
The National Survey of Hazardous Waste Generators and
Treatment, Storage, and Disposal Facilities Regulated Under
RCRA in 1981, (April, 1984) and referred to in this
analysis as the OSW Mail Survey.
The hazardous waste tank RIA data base was used to compile data on the
most common hazardous waste treatment processes employing tanks, while the OSW
Mail Survey provided a list of waste streams reported as most commonly treated
in tanks.
Based on data from the RIA data base and the selection criteria described
in Section 4.2, we selected oxidation/reduction/precipitation (ORP) and
distillation as the two representative treatment technologies to be included
in the risk analysis. Oxidation/reduction/precipitation is a relatively
complex process that is used mainly to treat aqueous wastes containing heavy
metals and cyanides, where as distillation is considerably less complex and is
used to treat highly concentrated mixtures of organic solvents. Oxidation/
reduction/precipitation was among the most highly ranked waste treatment
technologies reported in the hazardous waste tank RIA data base, and was
compatible with a large number of the highest-volume waste streams from the
OSW mail survey. Distillation, while not as highly ranked in the RIA, was
also compatible with a large number of waste streams from the OSW Mail Survey,
and, because it is used to treat highly concentrated wastes, is not generally
exempt under the RCRA wastewater treatment exemption. Both treatment
processes are outlined in greater detail in Chapter 2 and in Section 4.2.
We selected 10 waste streams from the OSW Mail Survey list as compatible
with ORP, and six waste streams as compatible with distillation. These waste
streams are summarized in Exhibit 4-18. All 16 waste streams had previously
been included in the analysis as representative RCRA-permitted storage tank
waste streams, and, therefore, we used the constituents of concern and average
concentrations for these wastes previously identified from the W-E-T model
data base. The wastes were distributed over the model tank population based
on the total quantity reported for each of the "waste treated in tanks" list
from the OSW Mail Survey. The distribution was carried out according to the
-------
EXHIBIT 1-18
REPRESENTATIVE WASTE STREAMS FOR SELECTED TREATMENT TECHNOLOGIES
Percentage of
Treatment Waste Stream Constituent Tanks Storing
Process Description Waste Code a/ Constituent Name b/ Mass Fraction b/ Waste Stream c/
Distillation Spent Halogenated Solvents F001 Chloroform
Carbon tetrachlor ide
1 , 2-D ichlo robenzene
1,1, 1-Trichloroethane
T r i ch 1 o roe thy 1 ene
Spent Non-Ha logenated F003 Benzene
Solvents Toluene
To 1 uene d i i socyana te
Methyl ethyl ketone
Xy 1 ene
Aceta Idehyde
Chlorinated Pesticide X907 Hexach lorobenzene
Production Wastes Trichloroethy tene
Chloroform
Hexachlorobutad iene
Carbon Tetrachlor ide
Hexachlorocyclopentad iene
Igni table, NOS D001 Methanol
Toluene
Acetone
Xylene
Benzene
Methyl ethyl ketone
0.10 0.016
0. 10
0. 10
0.10
0. 10
0.10 0.107
0. 10
0.10
0.10
0.10
0. 10
0.10 0.556
0. 10
0.10
0. 10
0. 10
0.10 ^
0.0911 0.283 °°
0.0695
0.1552
0. 1912
0. 1176
0.1101
Ox i da t i on/Reducti on
Precipitation
Dichloronethane
1,1,1-Trichloroethane
Reactive, NOS
Acrylonitrile Wastes
U080
U226
D003
K011
Dichloromethane 1.00
1,1,1-Trichloroethane 1.00
Acetone 0.0891
Toluene 0.838
Cyanide 0.0100
2,4-Dinitrotoluene 0.1601
Acetonitrite 0.006
Cyanide 0.003
Ac ryI on i t r iIe 0.0001
0.001
0.001
0.291
0.031
-------
EXHIBIT U-18 (continued)
REPRESENTATIVE WASTE STREAMS FOR SELECTED TREATMENT TECHNOLOGIES
Percentage of
Treatment
Technology
Ox i da t i on/Reduc t i on
Precipitation (continued)
Waste Stream
Description
EP Chromium
Petroleum Refining
Sludges/Emul sions
Spent Pickle Liquor
Non-cyanide Electroplating
Wastewaters
Electroplating Wastewaters
EP Arsenic
Thai 1 ium Sulfate
EP Lead
Waste Code
D007
K048
K062
F006
F007
D00>4
P115
D008
a/ Constituent Name b/
Ch rom i urn
Chromium
Lead
Chromium
Lead
Copper
Chromium
Cadmium
Lead
Nickel
Cadmium
Copper
Cyanide
Chromium
Nickel
Arsenic
Tha 1 1 i urn
Lead
Const i tuent
Mass Fraction b/
0.0005
0.000004
0.000026
0.000010
0.000100
0.0018
0.0068
0.00035
O.OOOU
0.00365
0.00010
0.0001*4
0.000075
0.00017
0.0015
0.0005
0.10
0.0005
lanks Storing
Waste Stream c
0. 152
0.317
0.02U
0.161
0.018
j
4
v
0.001
0.0002
0.0025
a/ Waste codes combined in some cases for similar waste streams (e.g., F003 represents F003 through F005).
b/ Constituents and mass fractions from W-E-T Model data base.
c/ Assumes that selected wastes streams represent 100 percent of all wastes treated by that particular process.
-------
4-50
treatment process assigned to a particular tank, rather than by tank type
(e.g., above-ground, in-ground, or material). Consequently, all the model ORP
treatment tanks have the same distribution of waste streams.
4.3.5 Assumptions and Limitations
Although a large number of data sources were identified and reviewed as
part of the waste stream selection process, a lack of information about
certain waste characteristics and tank management practices required that
several assumptions be made. Major limitations resulting from these
assumptions are as follows:
Average waste stream characteristics (constituents of
concern and concentrations for these constituents)
identified from available data sources were used to
represent waste streams in the analysis. In reality, waste
characteristics may vary markedly from these average
characteristics, and thus we may be over-estimating or
under-estimating risks for the waste streams included in
our analysis. In particular, the assumption regarding
waste mixture characteristics adds significant uncertainty
to the analysis.
The waste streams selected for our analysis were assumed
to represent 100 percent of the hazardous waste stored or
treated in tanks. The selected waste streams represent the
highest-volume chronically toxic wastes handled in tanks,
but do not account for acutely toxic wastes or low-volume
high toxicity wastes.
In order to distribute waste streams over the population
of model tanks, it was necessary to make assumptions
regarding tank management practices. In general, we
assumed that tanks were managed according to good
engineering practices, and that wastes were compatible with
tank materials. In reality, this would not always be the
case, and tank-waste incompatibilities would lead to tank
failures in some instances.
4.4 MAPPING TANK POPULATIONS INTO GENERIC HYDROGEOLOGIC SCENARIOS
The hydrogeologic setting underlying a tank influences the transport rate
of contaminants through the ground water to potable wells downgradient of the
tank. The setting is defined in terms of the unsaturated and saturated zones
underlying the tank. Within the framework of the Tank Risk Analysis Model one
representative unsaturated zone and nine different saturated zones are
modeled.11 In order to weight the risk estimates associated with each
lsSee Chapter 3, Transport, Exposure and Risk Models, for a description
of the modeled unsaturated and saturated zones.
-------
4-51
hydrogeologic scenario, we estimated the percentage of each tank category
(i.e. SQG, storage, accumulation and treatment tank) population that is
located within each scenario.16 For example, we estimated that 15 percent
of all storage tanks are located in a hydrogeologic setting with a vertical
ground-water velocity of 10 meters per year. The derivation of these
percentages is a three-step procedure as follows:
Identify percentage of each ground-water scenario in
each state;
Identify percentage of tank category population in
each state; and
Calculate percentage of tank category population
located in each ground-water scenario.
The methodologies and necessary data sources used to perform these three steps
are described below. Section 4.4.4 then summarizes the assumptions we used to
determine the percentages of the tank populations that we assigned to each
ground-water scenario.
4.4.1 Ground-Water Scenarios Located Within Each State
The assignment of ground-water scenarios to each state was based on
available information relating to the ground-water scenarios assigned to the
widely used United States Geological Survey (USGS) regions. These regions are
defined by similar features that influence the presence and availability of
ground water (e.g., geologic conditions and the nature of recharge and
discharge areas).17 Assuming that the hydrogeologic scenarios identified to
exist within a USGS region exist uniformly within the USGS region, and
estimating the percentage of each state that is comprised of each USGS region,
we estimated the percentage of ground-water settings within each state.
Exhibit 4-19 presents the USGS regions. Most USGS regions are contiguous
areas. By visual inspection, we estimated the percent of each state covered
by each USGS region. For example, approximately 20 percent and 80 percent of
Massachusetts is comprised of USGS regions 7 and 9, respectively. Exhibit
4-20 presents the USGS regions within each state and the percent of the state
that the USGS region comprises.
lsFor this study, we did not determine the distribution of hydrogelogic
settings associated with each tank technology. Consequently, we assumed that
that tank technologies associated with a particular category of tanks are
distributed the same with respect to hydrogeologic settings. For example, the
distribution for hydrogeologic settings that surround small quantity generator
tanks, is the same for both above-ground and underground small quantity
generator tanks.
1TFor a detailed description of the USGS regions see Heath, R.C.,
Groundwater Regions of the United States. USGS Water Supply Paper 2242, 1984.
-------
4-52
EXHIBIT 4-19
USGS Regions a/
2. Alluvial Basin
^.smi**** --
refliou 1 1 western
; 1 /_, \Mountam
ange*
Non9laciafed
Central- region
** >v' ,0 dentra(Vegton
14 ALASKA
s~ ''"!~* ^
' s~^*&'
? 13.
, c;^ HAWAII
"^- ' "^^.
' % i » *
''' ^ ^ *=> ,.«,
'= ^ '- ^ J !
15. PUERTO mCO
AND
VIRGIN ISLANDS
X^ - - ^
0
0
x^i'
Jt\
r^
%.
500 MILES
800 KILOMETERS
7. Glaciatec
Central
region
6. Nonglaciat
Central
region
a/ Source: Heath, R.C., 1984.
-------
We obtained the generic ground-water scenarios representative of the USGS
regions using information compiled from another study.11 In order to make
reasonable assignments of ground-water scenarios to each USGS region, this
study divided each USGS region into a non-alluvial area and alluvial valley.
All alluvial valleys were assigned ground-water scenarios C, D, G, and I,
except for the southeast coastal plain. The alluvial valley within the
southeast coastal plain has several swampy areas and was assigned ground-water
scenarios A, B, C, D, E, F, G, and H. The assignment of ground-water
scenarios to the non-alluvial areas within each USGS region was based on the
characteristics of the region. Exhibit 4-21 presents the percent of
non-alluvial areas and alluvial valleys assigned to each USGS region and the
representative ground-water scenarios within each USGS region. We assumed
that the ground-water scenarios are distributed uniformily within a USGS
region.
We determined the percentage of each ground-water scenario within a state
by multiplying the percentage of the state within each USGS region (Exhibit
4-20) by the percentage of the USGS region that is represented by the
ground-water scenario (Exhibit 4-21). For example, four percent of
Massachusetts is comprised of ground-water scenario C: .20 * (.68 * .167 +
.32 * .25) + .80 * (.80 * . 25 + .20 * .25). The percents used to derive the
estimate that four percent of Massachusetts is comprised of ground-water
scenario C are the following:
20 percent and 80 percent of the state are comprised of
USGS regions 7 and 9, respectively (see Exhibit 4-20);
68 percent and 32 percent of USGS region 7 are
non-alluvial and alluvial valley areas, respectively (see
Exhibit 4-21);
16.7 percent and 25 percent of the non-alluvial and
alluvial areas, respectively, in USGS region 7 are
comprised of ground-water scenario C (see Exhibit 4-21);
80 percent and 20 percent of USGS region 9 are
non-alluvial and alluvial valley areas, respectively, ( see
Exhibit 4-21).
25 percent and 25 percent of the non-alluvial and
alluvial areas, respectively, in USGS region 9 are
comprised of ground-water scenario C (see Exhibit 4-21).
IIJRB Associates, Comparative Risk Analysis for Ground-Water Sources,
draft report, under EPA contract No. 658-01-6558, October 16, 1984.
-------
4-54
EXHIBIT 4-20
PERCENT OF STATE WITHIN EACH USGS REGION
USGS Region
State
Alabama
Arkansas
Arizona
California
Colorado
Connecticut
Delaware
Florida
Georgia
Iowa
Idaho
Illinois
Indiana
Kansas
Kentucky
Louisiana
Massachusetts
Maryland
Maine
Michigan
Minnesota
Missouri
Mississippi
Montana
North Carolina
North Dakota
Nebraska
New Hampshire
New Jersey
New Mexico
Nevada
New York
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Virginia
Vermont
Washington
Wisconsin
West Virginia
Wyoming
1
0
0
0
25
25
0
0
0
0
0
75
0
0
0
0
0
0
0
0
0
0
0
0
25
0
0
0
0
0
10
0
0
0
0
40
0
0
0
10
0
0
10
0
0
50
0
0
35
2
0
0
60
75
5
0
0
0
0
0
10
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
90
0
0
0
0
0
0
0
0
0
10
40
0
0
15
0
0
0
3
0
0
0
0
0
0
0
0
0
0
15
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
10
0
0
0
60
0
0
0
0
0
0
0
0
0
35
0
0
0
4
0
0
40
0
30
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
15
0
0
0
0
0
0
0
0
0
0
0
50
0
0
0
0
0
30
5
0
0
0
0
15
0
0
0
0
0
0
0
0
30
0
0
0
0
0
0
0
0
0
0
0
0
65
0
0
10
0
0
0
10
0
0
0
0
5
0
10
0
0
0
0
0
0
5
6
40
55
0
0
25
0
0
0
15
5
0
10
15
55
90
0
0
10
0
0
0
50
0
60
10
20
15
0
20
65
0
0
40
90
0
60
0
0
40
70
50
0
55
0
0
10
100
30
7
0
0
0
0
0
30
0
0
0
95
0
90
0
15
0
0
20
0
0
80
50
50
0
15
0
80
20
0
15
0
0
50
60
0
0
30
0
0
45
0
0
0
0
0
0
60
0
0
8
5
0
0
0
0
0
0
0
30
0
0
0
85
0
0
0
0
50
0
0
0
0
0
0
55
0
0
0
0
0
0
0
0
0
0
10
0
50
0
0
0
0
25
0
0
0
0
0
9
0
0
0
0
0
70
0
0
0
0
0
0
0
. 0
0
0
80
0
100
20
50
0
0
0
0
0
0
100
15
0
0
50
0
0
0
0
100
0
0
0
0
0
0
100
0
30
0
0
10
50
45
0
0
0
0
100
5
25
0
0
0
0
0
10
100
0
40
0
0
0
0
100
0
35
0
0
0
50
0
0
0
0
0
0
0
0
50
0
30
30
0
20
0
0
0
0
0
11
5
0
0
0
0
0
0
95
30
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
a/ Source: ICF estimates.
-------
EXHIBIT 1-21
DISTRIBUTION OF GROUND-WATER SCENARIOS WITHIN USGS REGIONS a/
Percent of Representative Ground-Water Scenarios b/
USGS Regions
1. Western Mountain Ranges
2. Alluvial Basins
3. Columbia Lava Plateau
4. Colorado Plateau and
Wyoming Basin
5. High Plains
6. Non-Glaciated Central
7. Glaciated Central
8. Piedmont and Blue Ridge
9. Northeast and Superior
Uplands
10. Atlantic and Gulf Coasta
Plain
11. Southeast Coastal Plain
a/ Source: JRB Associates,
Percent Land Area
Non-AI luvial Area
Alluvial Val ley
Non-AI luvia 1 Area
Alluvial Val ley
Non-Alluvial Area
Alluvial Valley
Non-AI luvial Area
Alluvial Valley
Non-Alluvial Area
A 1 1 uv i a 1 Va 1 1 ey
Non-Alluvial Area
Alluvial Valley
Non-Alluvial Area
Alluvial Valley
Non-Alluvial Area
A 1 1 uv i a 1 Va 1 1 ey
Non-AI luvial Area
Alluvial Valley
1 Non-AI luvial Area
Al luvial Valley
Non-Alluvial Area
Al luvial Valley
Comparative Risk Analysis
84
16
92
8
82
18
85
15
84
16
81
19
68
32
85
15
80
20
76
24
A
0.0
0.0
12.5
0.0
0.0
0.0
11.1
0.0
11.1
0.0
20.0
0.0
16.7
0.0
33.3
0.0
0.0
0.0
14.3
0.0
80 0.0
20 0.0
for Ground-water
i
25.0
0.0
12.5
0.0
20.0
0.0
11.1
0.0
11.1
0.0
20.0
0.0
16.7
0.0
33.3
0.0
25.0
0.0
14.3
0.0
C
25.0
25.0
12.5
25.0
20.0
25.0
11.1
25.0
11.1
25.0
0.0
25.0
16.7
25.0
0.0
25.0
25.0
25.0
14.3
25.0
B
0.0
25.0
12.5
25.0
20.0
25.0
11.1
25.0
11.1
25.0
0.0
25.0
0.0
25.0
0.0
25.0
0.0
25.0
0.0
25.0
0.0
0.0
0.0
0.0
0.0
0.0
11.1
0.0
11.1
0.0
20.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
20.0 20.0 20.0 0.0
0.0 25.0 25.0 0.0
Sources, draft report. EPA
£
25.0
0.0
12.5
0.0
0.0
0.0
11.1
0.0
11.1
0.0
20.0
0.0
16.7
0.0
33.3
0.0
25.0
0.0
14.3
0.0
G
0.0
25.0
12.5
25.0
20.0
25.0
11.1
25.0
11.1
25.0
0.0
25.0
16.7
25.0
0.0
25.0
25.0
25.0
14.3
25.0
20 . 0 0.0
0.0 25.0
contract No.
H
25.0
0.0
12.5
0.0
20.0
0.0
11.1
0.0
11.1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
11.3
0.0
1
0.0
25.0
12.5
25.0
0.0
25.0
11.1
25.0
11.1
25.0
20.0
25.0
16.7
25.0
0.0
25.0
0.0
25.0
11.3
25.0
0.0 20.0
0.0 25.0
658-01-6558,
October 16, 198U, p. 1-9.
b/ Generic hydrogeoIogic settings assigned to USCS regions are assumed to be uniform!ly distributed within the USGS region.
-------
4-56
Exhibit 4-22 presents the percent of the state that is comprised of each of
the ground-water scenarios.
4.4.2 Tank Category Population Within Each State
Based on a variety of data sources, we estimated the percentages of each
tank category (i.e., treatment, storage, small quantity generator and accumu-
lation tank) population within each state. We obtained the distribution of
treatment and storage tanks from Part A Permit Application data. Because we
had insufficient information on the location of accumulation tanks we assumed
that the distribution for accumulation tanks was equivalent to the distribution
for storage tanks. Exhibit 4-23 presents the distribution of the tank
categories throughout the contiguous United States.
We based the distribution of small quantity generator tanks on the
distribution of waste associated with small quantity generators. We used
information compiled for a previous study on small quantity genertors.1'
For this previous study, the Small Quantity Generator Survey and County
Business Patterns data from 1982 were used to determine the geographical
distribution of SQG wastes. Because some SQG wastes are not stored in tanks,
according to the SQG survey, we eliminated several waste streams from the list
of generated wastes.20 We assumed that the percent of SQG tanks located
within each state was distributed in the same way that the quantity of wastes
are distributed. Exhibit 4-24 presents the distribution of SQG tanks among
the states.
4.4.3 Distribution of Tank Categories Within Each Ground-water Scenario
The calculation of the percentage of each tank category population
assigned to each ground-water scenario is a straightforward procedure. For
each tank category and ground-water scenario, we multiplied the percentage of
the tank population located in each state by the percentage of the state that
is comprised of the ground-water scenario. For each scenario, the products
were summed across all states to obtain the percentage of the tank category
population that is represented by the ground-water scenario. Exhibit 4-24
presents the distribution of the tank categories among the ground-water
scenarios. We used these percentages to weight the risk estimates in order to
obtain distributions of risk estimates.21
lsSee IGF Incorporated, et al., Economic Analysis of Resource
Conservation and Recovery Act Regulations for Small Quantity Generators.
prepared for the Office of Solid Waste, EPA, June 1985.
2"The following wastes were eliminated: arsenic wastes, heavy metal
dust, heavy metal waste materials, paint waste with heavy metals, solvent
still bottoms, waste formaldehyde, and wastes with ammonia.
2lSee Section 4.1, Distributions of Risk Estimates, for a description of
the use of the percentages in calculating national risk estimates.
-------
EXHIBIT 4-22
PERCENT OF STATE REPRESENTED BY EACH GROUND-WATER SCENARIO
Ground-water Scenario
STATE
Alabama
Arkansas
Arizona
California
Colorado
Connecticut
Delaware
Florida
Georgia
Iowa
Idaho
Illinois
Indiana
Kansas
Kentucky
Lous i ana
Massachusetts
Maryland
Maine
Michigan
Minnesota
Missouri
Mississippi
Montana
North Carolina
North Dakota
Nebraska
New Hampshire
New Jersey
New Mexico
Nevada
New York
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennesse
Texas
Utah
Virginia
Vermont
Washington
Wisconsin
West Virginia
Wyoming
A
13
13
10
8
8
3
10
2
14
11
1
11
12
13
15
10
2
20
0
9
5
13
10
11
21
12
10
0
10
13
10
5
13
15
0
16
0
19
12
14
13
9
18
0
1
8
16
8
.5
.8
.7
.6
.9
.4
.9
.9
.4
.2
.2
.8
.1
.4
.7
.9
.3
.1
.0
.1
.7
.8
.9
.4
.0
.3
.8
.0
.4
.6
.4
.7
.3
.5
.0
.0
.0
.6
.0
.6
.4
.3
.2
.0
.7
.4
.2
.2
8
14
13
10
13
14
17
10
18
19
11
19
11
12
13
15
10
18
20
20
13
15
13
10
16
21
12
10
20
13
15
12
15
13
15
18
16
20
19
14
14
13
11
18
20
18
14
16
15
.3
.8
.7
.9
.1
.4
.9
.1
.2
.2
.4
.8
.1
.4
.7
.9
.3
.1
.0
.1
.7
.8
.9
.7
.0
.3
.8
.0
.4
.7
.0
.7
.3
.5
.2
.0
.0
.6
.1
.6
.4
.4
.2
.0
.0
.4
.2
.5
C
11.4
10.2
13.4
16.4
14.1
23.3
16.9
18.4
11.6
19.0
23.2
17.9
17.1
9.5
6.0
16.9
23.9
9.1
25.0
20.5
22.2
12.0
16.9
12.0
8.4
16.4
13.2
25.0
16.0
8.1
14.2
22.2
13.5
5.6
22.5
9.0
25.0
10.3
13.8
8.4
10.1
14.5
6.9
25.0
21.8
19.6
4.7
14.8
D
6.0
5.3
13.4
11.1
8.8
5.9
6.0
17.9
8.9
8.3
7.5
7.7
7.5
7.8
4.9
6.0
5.6
4.8
5.0
7.4
6.5
6.4
6.0
5.0
4.6
7.4
11.0
5.0
5.9
6.0
14.2
6.5
6.7
5.6
14.1
5.6
5.0
4.9
6.6
5.1
6.9
12.4
4.8
5.0
11.3
6.8
4.7
7.5
E
6.6
8.9
3.8
0.0
8.3
0.0
0.0
2.4
3.2
0.5
0.0
1.6
2.4
11.7
14.6
0.0
0.0
1.6
0.0
0.0
0.0
8.1
0.0
9.7
1.6
3.2
8.5
0.0
3.2
13.6
0.0
0.0
6.5
15.5
0.0
9.7
0.0
0.0
6.9
11.3
9.0
4.7
8.9
0.0
0.0
1.6
16.2
8.2
F
14.3
13.8
10.7
13.9
14.1
17.4
10.9
18.1
19.2
11.2
16.9
11.8
12.1
13.4
15.7
10.9
18.3
20.1
20.0
13.1
15.7
13.8
10.9
16.7
21.0
12.3
10.8
20.0
13.4
15.7
10.4
15.7
13.3
15.5
8.4
16.0
20.0
19.6
14.1
14.6
13.4
11.4
18.2
20.0
12.2
14.4
16.2
15.5
G
10.6
10.2
13.4
11.1
8.8
23.3
16.9
3.2
6.8
19.0
7.5
17.9
17.1
9.5
6.0
16.9
23.9
9.1
25.0
20.5
22.2
12.0
16.9
6.8
8.4
16.4
13.2
25.0
16.0
6.0
14.2
22.2
13.5
5.6
14.1
9.0
25.0
10.3
11.7
8.4
10.1
12.4
6.9
25.0
11.3
19.6
4.7
7.5
H
5.6
4.9
10.7
13.9
10.1
0.0
10.9
2.9
3.5
0.5
19.4
0.0
0.0
2.8
1.1
10.9
0.0
4.3
0.0
0.0
0.0
0.0
10.9
5.3
3.8
0.0
6.1
0.0
5.4
3.5
12.0
0.0
0.0
0.9
18.2
0.0
0.0
5.4
2.6
3.3
5.3
11.4
2.2
0.0
18.0
0.0
0.0
10.7
I
17.8
19.1
13.4
11.1
12.9
9.3
16.9
16.0
13.3
19.0
5.0
19.5
19.6
18.4
20.5
16.9
7.9
10.7
5.0
16.5
12.2
20.1
16.9
16.5
10.1
19.7
15.7
5.0
16.3
18.1
12.6
12.2
20.0
20.2
4.3
18.7
5.0
10.3
18.1
19.7
18.2
12.4
15.8
5.0
5.6
15.2
21.0
12.3
-------
4-58
EXHIBIT 4-23
DISTRIBUTION OF TANK CATEGORIES WITHIN EACH STATE
State
Alabama
Arkansas
Arizona
California
Colorado
Connecticut
Delaware
Florida
Georgia
Iowa
Idaho
Illinois
Indiana
Kansas
Kentucky
Louisiana
Massachusetts
Maryland
Maine
Michigan
Minnesota
Missouri
Mississippi
Montana
North Carolina
North Dakota
Nebraska
New Hampshire
New Jersey
New Mexico
Nevada
New York
Ohio
Treatment a/
Tanks
0.0153
0.0084
0.0091
0.0745
0.0076
0.0668
0.0035
0.0223
0.0153
0.0111
0.0000
0.0661
0.0494
0.0084
0.0069
0.0202
0.0236
0.0160
0.0007
0.0439
0.0042
0.0174
0.0011
0.0014
0.0091
0.0000
0.0007
0.0014
0.0759
0.0028
0.0000
0.0717
0.0703
Storage a/
Tanks
0.1350
0.0056
0.0088
0.0776
0.0088
0.0423
0.0038
0.0139
0.0154
0.0070
0.0027
0.0749
0.0516
0.0083
0.0130
0.0204
0.0223
0.0163
0.0009
0.0525
0.0065
0.0158
0.0093
0.0014
0.0098
0.0014
0.0027
0.0023
0.0711
0.0023
0.0005
0.0581
0.0702
SQG b/
Tanks
0.0131
0.0078
0.0085
0.1113
0.0119
0.0234
0.0020
0.0419
0.0197
0.0132
0.0029
0.0538
0.0219
0.0121
0.0110
0.0129
0.0367
0.0142
0.0053
0.0377
0.0205
0.0216
0.0074
0.0026
0.0219
0.0031
0.0075
0.0055
0.0516
0.0035
0.0026
0.0915
0 . 0485
Accumulation c/
Tanks
0.0135
0.0056
0.0088
0.0776
0.0088
0.0423
0.0038
0.0139
0.0154
0.0070
0.0027
0.0749
0.0516
0.0083
0.0130
0.0204
0.0223
0.0163
0.0009
0.0525
0.0065
0.0158
0.0093
0.0014
0.0098
0.0014
0.0027
0.0023
0.0711
0.0023
0.0005
0.0581
0.0702
a/ Source: Part A Permit Application Data (1979-1980).
b/ Source: ICF Incorporated, et a_l., Economic Analysis of Resource
Conservation and Recovery Act Regulations for Small Quantity Generators.
prepared for EPA, OSW, June 1985, pp. C-l to C-16.
c/ We assumed that the distribution for accumulation tanks is equal to the
distribution for storage tanks.
-------
EXHIBIT 4-23
(Continued)
DISTRIBUTION OF TANK CATEGORIES WITHIN EACH STATE
State
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Virginia
Vermont
Washington
Wisconsin
West Virginia
Wyoming
Treatment a/
Tanks
0.0105
0.0035
0.0508
0.0028
0.0125
0.0
0.0056
0.1149
0.0049
0.0118
0.0021
0.0105
0.0195
0.0140
0.0014
Storage a/
Tanks
0.0158
0.0060
0.0423
0.0033
0.0107
0.0
0.0079
0.1339
0.0051
0.0174
0.0019
0.0058
0.0233
0.0135
0.0019
SQG b/
Tanks
0.0121
0.0094
0.0618
0.0085
0.0100
0.0029
0.0160
0.0615
0.0046
0.0162
0.0026
0.0139
0.0253
0 . 0045
0.0015
Accumulation c/
Tanks
0.0158
0.0060
0.0423
0.0033
0.0106
0.0
0.0079
0.1339
0.0051
0.0174
0.0019
0.0058
0.0233
0.0135
0.0019
a/ Source: Part A Permit Application Data (1979-1980).
b/ Source: ICF Incorporated, et al., Economic Analysis of Resource
Conservation and Recovery Act Regulations for Small Quantity Generators,
prepared for the Office of Solid Waste, EPA, June, 1985, pp. C-l to C-16.
c/ We assumed that the distribution for accumulation tanks is equal to the
distribution for storage tanks.
-------
Tank Category
Treatment
Storage
Sma11 quant Ity Gene ra to rs
Accumulation
EXHIBIT 4-24
HYDROGEOLOGIC SETTING WEIGHTS
Ground-water Scenario a/
E F G
H
I
0.1068 0.1154 0.1555 0.0721 0.0390 0.1444 0.1452 0.0346 0.1564
0.1103 0.1440 0.1515 0.0720 0.0420 0.1430 0.1427 0.0358 0.1590
0.1018 0.1500 0.1583 0.0767 0.0359 0.1482 0.1412 0.0391 0.1484
0.1103 0.1440 0.1515 0.0720 0.0420 0.1430 0.1427 0.0358 0.1590
a/ See Chapter 3 for a description of these ground-water scenarios.
ON
O
-------
4-61
4.4.4 Assumptions and Limitations
The approach for determining the percentage of a tank category located
within each ground-water scenario is based on the assumption that the tank
population is distributed across the whole state. For example, suppose four
ground-water scenarios are located within a state. We have assumed that the
tank population within the state is distributed evenly among the settings. In
actuality, although four scenarios exist in the state, all the tanks may be
located in only two of the settings. Because we do not have information on
the ground-water settings underlying the existing tank population, we do not
know what type of effect this assumption will have on the national risk
estimate (i.e., underestimate or overestimate the risk). Of note is that the
modeling framework that we have developed is flexible such that we can use
better information if it becomes available. We need only to modify the
percentages in Exhibits 4-20 through 4-23.
4.5 SELECTING REPRESENTATIVE RELEASE PROFILES
As noted in Section 4.1, the Hazardous Waste Tank Failure model generates
up to 250 release profiles for each of the 22 tank technologies under each
regulatory scenario.22 We initially considered running each of these 250
profiles through the Tank Risk Analysis Model (TRAM), however, this would be
computationally intensive. We subsequently decided to use a weighted
representative sample of the 250 profiles generated by the HWTF model in the
risk analysis.
Several alternative techniques for choosing representative release
profiles were considered. These included simply characterizing each profile
by its total release volume, then dividing the profiles into groups with
similar total releases, and, finally, picking the profile with the median
total release for each group. Other more involved techniques that were also
based on first computing summary measures for each profile, such as the
standard deviation and number of zero releases, and then grouping profiles on
the basis of these summary measures were examined and subsequently discarded
in favor of the technique used, namely, cluster analysis. The latter
technique, which is described below, was considered superior because it takes
into account all the characteristics of a profile and not just a handful, such
as the total release or the number of zero releases. In particular, it takes
into account the timing of releases, which is an important component (i.e.,
purpose) of the HWTF model.
The cluster analysis procedure we used to select representative release
profiles consisted of the following three steps:
22 For tanks whose age was less than the median tank age, fewer than 250
profiles were generated. However, for ease of exposition, the above
discussion assumes 250 profiles were generated in all cases.
-------
4-62
1. Determine the appropriate number of profile groups.
2. Divide each set of 250 profiles into groups of similar
profiles, where the number of groups is that determined in
step 1.
3. Select a representative profile from each group.
Thus, we divided each set of 250 profiles into five or six clusters (i.e.,
groups) of similar profiles, and chose a single representative profile from
each cluster. The three basic steps are discussed in turn below.
4.5.1 Determining the Number of Profile Groups
The number of profile groups was determined by the presence or absence of
profiles with zero release volumes for the entire 20-year time horizon. If
such "zero profiles" were absent in a set of 250 profiles, we divided the
profiles into five groups. On the other hand, if zero profiles were present,
we divided the 250 profiles into six groups, with the sixth group consisting
entirely of zero profiles.
Test runs indicated that, in general, little was gained in terms of better
grouping of profiles by increasing the number of profile groups to seven or
more. However, if fewer than five groups were used, dissimilar profiles would
frequently be placed in the same group.
4.5.2 Grouping Similar Profiles
Profiles were divided into groups (or "clusters") using a standard
statistical technique known as cluster analysis. Specifically, the SAS
(Statistical Analysis System) .statistical package program FASTCLUS was used.
Before inputting the profiles to FASTCLUS, all zero profiles (if any) were
removed and placed in a separate group. The remaining profiles were then
divided into five groups of similar profiles using the FASTCLUS procedure.
FASTCLUS performs a variant of the k-means method for non-hierarchical
cluster analysis.21 In this particular application, FASTCLUS performed the
following four steps:
1. To initiate the clustering process, FASTCLUS selects five
profiles to be used as "cluster seeds." These profiles are
chosen so that they reflect the variation in the total set
of 250 profiles.
2*A detailed description of the FASTCLUS procedure is presented in
Chapter 18 of the SAS User's Guide: Statistics. Version 5 Edition (Gary,
North Carolina: SAS Institute, 1985). The description presented here is
adapted from Romesburg, C.H., Cluster Analysis for Researchers, Belmont,
California: Lifetime Learning Publications, 1984, pp. 295-296.
-------
4-63
2. FASTCLUS then forms temporary clusters (i.e., groups of
profiles) by sequentially assigning each of the remaining
profiles to the cluster seed closest to it.2" As profiles
are assigned to clusters, the cluster seed is recomputed and
set equal to the mean of the profiles assigned to the
cluster. Consequently, the cluster seeds generally change
as profiles are assigned to clusters.
3. After all the profiles have been assigned to clusters, the
first iteration is complete. The final cluster seeds from
the first iteration are then taken as the initial seeds for
the second iteration. The process described in steps 1 and
2 is repeated: profiles are sequentially assigned to the
nearest cluster seed, and the seeds are updated as the
process continues. This interative procedure ensures that
the final clusters obtained are the best possible ones.
4. A maximum of 10 iterations are carried out. If the change
in cluster seeds from one iteration to the next satisfies
the standard (default) convergence criterion in FASTCLUS,
the program terminates. More specifically, iterations
terminate when the maximum distance by which any seed has
changed is less than or equal to the minimum distance
between initial seeds times 0.02. In virtually all cases,
the program terminated before reaching the tenth iteration.
4.5.3 Selecting Representative Profiles from Clusters
The output of the FASTCLUS procedure is a set of five clusters of similar
(non-zero) release profiles. The final step of the selection procedure is to
pick a representative profile from each cluster. The profile closest to the
2*The criterion used to determine whether or not profiles were similar
(i.e., close) was the "distance" between profiles. The distance between two
profiles is calculated by first subtracting one profile from the other
profile, then squaring each of the differences, adding up the squared
differences, and finally taking the square root of the resulting sum. More
precisely, the Euclidean distance measure, (i.e., the distance between two
profiles) is given by the square root of the sum of squared deviations:
20
i 12*
fZ (vt1 - vtj)2] *
t « 1
where v is the release volume in year t for profile i, and v ^ is
similarly interpreted.
-------
4-64
cluster seed, which by construction is the mean of all the profiles in the
cluster, was picked as the representative profile. In cases where zero
profiles were present, a zero profile was selected as the sixth representative
profile.
In some cases, the presence of outlier profiles, that is, profiles with
very high releases relative to the majority of the profiles in the set, would
force the clustering program to place dissimilar profiles into a single
cluster so that the outlier profiles could be placed in separate clusters. We
found in these cases that the profile closest to the cluster mean was
sometimes not representative of the profiles in the cluster. Consequently, in
these cases, we chose the profile that was at the median distance from the
cluster mean. We found this profile was consistently representative of the
profiles in a cluster.
4.5.4 Assumptions and Limitations
In general, the profiles chosen by the technique described above were very
representative of the larger set of profiles from which they were drawn.
Moreover, the technique ensured a systematic and consistent approach to
selecting representative profiles. The primary limitation of the technique is
the reliance on a fixed number of non-zero representative release profiles
(five). As noted earlier, in the majority of cases, five clusters were found
to be sufficient in that similar profiles were placed in the same cluster and
dissimilar profiles were placed in different clusters. However, in some
cases, the presence of outlier profiles induced dissimilar profiles to be
placed in the same group. As described above, this problem was at least
partly remedied by altering the criterion used to choose representative
release profiles in these cases.
In the situation just described, five non-zero profile clusters proved to
be too few. In other cases encountered, less than five clusters were needed.
This was especially true for some of the sets of release profiles developed
for the secondary containment regulatory scenario. For these sets, a very
large proportion of the release profiles generated were zero profiles. When
dividing the small number of non-zero profiles for these cases into clusters,
it was therefore common for outlier profiles to be placed in separate
clusters, rather then being placed together in a single cluster. This implied
that more than one of the representative profiles chosen was an outlier
profile. However, this does not constitute a significant problem as long as
the representative profiles are considered along with their weights, i.e., the
proportion of the total number of profiles that they represent (see Section
4.1.4 for a detailed description of these weights).
-------
CHAPTER 5
RESULTS AND CONCLUSIONS
5.1 INTRODUCTION
In this chapter, we present the results of our risk analysis for hazardous
waste tanks under the five regulatory scenarios outlined in Chapter 1. As
discussed in Chapter 1, our analysis consisted of two major components:
An analysis of some of the more common failure events
associated with hazardous waste tanks, and an estimation of
the volumes of liquid released due to these events over a
20-year time horizon; and
A quantitative assessment of the relative risks over a
400-year period to individuals living in close proximity to
hazardous waste tanks, via the exposure pathway of
contaminated ground-water, resulting from the release of
hazardous waste from these tanks.
In this chapter, we present results for the HWTF model analysis and the risk
analysis separately, so that some insight may be gained into both the failure
events and release volumes, and the risks associated with the population of
model tank technologies under the different regulatory scenarios.
This chapter is divided into three sections after this Introduction
section. HWTF model results are presented in Section 5.2, and results of the
risk analysis are presented in Section 5.3. Finally, Section 5.4 presents a
brief discussion of the effects of hydrogeologic settings and waste stream
characteristics on the risks associated with hazardous waste tanks. Before
presenting these results, however, it is important to note that the results
presented in this chapter are preliminary, and thus may be subject to
significant revisions following further review and additional modeling work
(e.g., inclusion of an algorithm to account for the development of cracks over
time in concrete tanks and an algorithm to model vadose monitoring for
appropriate volatile constituents).
TRAM, and particularly the HWTF component model, is a large and relatively
complex model that relies on major assumptions resulting from the lack of data
in some cases, and it is clear that certain model results are driven by these
assumptions (e.g., assumptions on leak detection through casual inventory
control). Some of the results seem anomalous and we are reviewing the
assumptions on which these results are based to determine if revised
assumptions should be made. However, we believe our review of the results of
this very complex model can be greatly aided by comments we expect to receive
on these preliminary results.
The five regulatory scenarios considered in our analysis and presented in
detail in Chapter 1 are as follows:
-------
5-2
Baseline. Under this scenario, all 22 model tank
technologies are included, and are considered to be
operated according to good management practices under
current regulations.
Secondary Containment. In this scenario, all 22 model
tank technologies are included; all tanks are considered to
be new, and are equipped with full secondary containment
systems and interstitial monitoring (leak detection)
devices.
Partial Secondary Containment with Ground-Water
Monitoring. This scenario applies to above-ground,
ongrade tanks and in-ground tanks, and includes 10 of the
22 model tank technologies included in the analysis. These
tanks are retrofitted with concrete pads, trenches, and
diking that contain all above-ground portions of the tank,
and are equipped with saturated zone monitoring wells
located 10 meter downgradient from the tank. Of note is
that in-ground and ongrade portions are not contained
within the containment system.
Corrosion Protection. This scenario includes 12 steel
tank systems that have some portion of their systems in
contact with soil (12 of 22 model tanks). These tanks are
assumed to be equipped with cathodic protection (i.e.,
impressed current) that significantly reduces external
corrosion of the tank and ancillary equipment.
Leak-Testing with Ground-Water Monitoring. This
scenario includes all underground tanks (7 of the 22 model
tanks), and assumes that all tanks are leak tested
semi-annually. In addition, all tanks are equipped with
saturated zone monitoring wells located 10 meters
downgradient from the tank.
Exhibit 5-1 presents the model tank technologies included in each of the five
regulatory scenarios.
In the next section, we present HWTF results for these five regulatory
scenarios. Risk results for the scenarios are compared in Section 5.3. Once
again, however, we caution that these results are preliminary, and are
considered useful only for comparing differences in release and risk estimates
for the different scenarios described above. Differences in waste streams,
tank sizes and tank ages currently preclude the use of these results for
rigorous tank-to-tank comparisons within a scenario. However, the models used
in the analysis are flexible enough to be used for tank-to-tank comparisons if
slight adjustments to the model tank data bases are made. Thus, we have the
capability to perform other comparisons using our analytical approach.
-------
EXHIBIT 5-1
APPLICABLE TANK TECHNOLOGIES FOR EACH REGULATORY SCENARIO
Partial Containment
Tank Technology
Treatment:
On cradles, closed top, carbon steel
On cradles, open top, carbon steel
Above-ground, ongrade, carbon steel
In- ground, concrete
In-ground, carbon steel
In-ground, stainless steel
Storage:
On cradle, carbon steel
Ongrade, carbon steel
Underground, carbon steel
Underground, FRP
Underground, stainless steel
In-ground, concrete
In-ground, carbon steel
Small Quantity Generator:
Above-ground, carbon steel
Under-ground, carbon steel
Accumulat ion:
On cradles, carbon steel
Ongrade, carbon steel
Underground, carbon steel
Underground, FRP
Underground, stainless steel
In-ground, concrete
In-ground, carbon steel
Basel ine
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Secondary
Conta inment
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
with Ground-water
moni to ring
X
X
X
X
X
X
X
X
X
X
Corros ion
Protect ion
X
X
X
X
X
X
X
X
X
X
X
X
with Ground-water
Mon i tor i ng
X
X
X
X
X
X
X
-------
5-4
5.2 HAZARDOUS WASTE TANK FAILURE MODEL RESULTS
In this section we discuss the Hazardous Waste Tank Failure (HWTF) model
results for the five proposed regulatory scenarios. The primary purpose of
this discussion is to provide an overview of the types of results generated by
the HWTF model. Consequently, we do not discuss the results for all five
proposed regulatory scenarios in detail. Instead, we focus on the results for
the baseline and secondary containment scenarios. A detailed discussion and
comparison of the HWTF model results for all five scenarios would, in large
part, be redundant since those results are mirrored in the risk estimate
results presented and discussed in Section 5.3 for all five scenarios.
Differences in risk estimates across the five scenarios primarily reflect
differences in the HWTF model results.
We summarize the HWTF model results using two basic types of summary
statistics: aggregate failure frequency data and representative release
profiles (RRPs). The frequency data provide an indication of the percentage
of model tanks for which a specific failure event is likely to occur. The
representative release profiles for the tank technologies indicate the
releases likely to result from the various failure events. These are the
release profiles that we used to generate the risk estimates.1 Below we
first describe how we summarized the HWTF model results and then discuss the
baseline and secondary containment results. (Summaries of the HWTF model
results for all five regulatory scenarios are presented in the Appendix to
this report.)
5.2.1 Types of Summary Statistics
Aggregate failure frequency data. Selected failure frequency data for
a hypothetical tank technology are presented in Exhibit 5-2. 2 A brief
description of each of the failure events is provided below.
Hose ruptures. The pump-out hose may rupture when the
tank contents are periodically removed. Hose ruptures are
modeled only for storage/accumulation and SQG tanks.
1 See Section 4.5 for a detailed description of the methodology we used
to select the RRPs.
2 It should be emphasized that the list of failure events in Exhibit 5-2
is not exhaustive: several events included in the HWTF model that were
considered to be of lesser importance are not presented in the table. The
omitted events are first, second, and third pipe and tank replacements; and
overflows, pipe/tank ruptures, pipe/tank corrosion, and weld/gasket failures
beyond the first occurrence (e.g., second and third overflows).
-------
EXHIBIT 5-2
SAMPLE FAILURE FREQUENCY DATA a/
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
I
FAILURE CATEGORY
HOSE
RUPTR
0.0
O.ll
0.4
1.2
O.ll
1 (WELD | PIPE
LOOSE | OVER IGASKETI INSTAL
HOSEI FLOW | INSTALI DEFIC
0.9
8.0
7.5
4.8
8.5
I TOTAL | 2.4 | 29.7
0.0
1.3
0.9
2.7
0.4
0.4
0.0
0.0
0.0
0.4
5.3 I 0.8
0.4
0.0
0.0
0.0
0.0
0.4
TANK | WELD/
INSTALI GASKT
OEFIC) FAIL
0.0
0.4
0.0
0.0
0.4
15.2
13.4
13.8
8.0
9.3
PIPE
RUPTR
4.5
1.7
1.2
3.0
2.5
TANK
RUPTR
0.9
2.6
2.1
3.5
3.0
PIPE
CORR
4.9
1.6
1.7
3.1
3.1
0.8 I 59.7 I 12.9 1 12.1 | 14.4
TANK
CORR
5.8
21.5
23.3
18.3
10.1
79.0
CATAS-
STROPH
RLS
0.0
0.4
1.3
0.4
0.0
2.1
a/ Frequencies are expressed as percent of iterations in which specified failure
event occurred.
-------
5-6
Loose hose connection. The flexible hose connection may
become loose during pump-out. As for hose ruptures, this
failure event is modeled only for storage/accumulation tanks
and SQG tanks. (We do not model the transport of
contaminants to and from treatment tanks, consequently hose
ruptures and loose hose connections are not modeled for
treatment tanks.)
Overflow. This event corresponds to the overflow of
waste from the tank. It can result from a level indicator
or level controller failure, an operator error, or a
mechanical failure of the inlet valve. However, an overflow
can occur only if the tank is also nearly full, the
shut-down system fails, and there is an escape route for the
waste (e.g., an open top or vent).
Improper weld or gasket installation. An improper weld
installation might be caused by use of an insufficient weld
thickness, while an improper gasket installation may be due
to use of a defective gasket or damage to the gasket during
installation.
Pipe installation deficiency. A pipe installation
deficiency may result from the use of pipe that was damaged
prior to installation or by damage to the pipe during
installation.
Tank installation deficiency. A tank installation
deficiency may occur due to the same reasons cited for the
pipe installation deficiency.
Weld or gasket failure. This event may be caused by the
rupture of the weld, or erosion or corrosion of the gasket.
Pipe rupture. A pipe may rupture due to poor material
selection, inadequate structural support, settling action
after installation, freeze/thaw cycles, or vehicle
collisions.
Tank rupture. A tank rupture may be caused by inadequate
structural support, insufficient material strength, or
vehicle collisions.
Pipe corrosion. Pipe corrosion leaks may occur due to
localized exterior corrosion, localized interior corrosion,
or generalized corrosion.
Tank corrosion. As for piping, tank corrosion leaks may
be caused by localized exterior corrosion, localized
interior corrosion, or generalized corrosion.
-------
Catastrophic events. Several events may cause a
catastrophic failure of a tank system. These include acts
of vandalism, high winds, earthquake, floods, or an
explosion.
The entries in Exhibit 5-2 represent the percentage of the approximately
250 profiles generated by the HWTF model in which specific failure events
occurred during a given interval of time. For example, the second entry in
the second column of the table indicates that in 0.4 percent of the iterations
a hose rupture occurred during years two to five of the model period.
Proceeding down the second column of the table, it can be seen that hose
ruptures were just as frequent during years six to ten and years sixteen to
twenty, but were three times as frequent during years eleven to fifteen (1.2
percent of all profiles). In total, hose ruptures occurred in 2.4 percent of
the profiles at some time during the modeled 20-year time horizon. Exhibit
5-2 also reveals that the most frequent failure event is tank corrosion (79
percent of all profiles), followed by weld/gasket failures (59.7 percent) and
loose hose connections (29.7 percent).
It should be noted that in order to model the existing population of
tanks, the HWTF simulates the occurrence of failure events (e.g., corrosion)
prior to the first year of the model period. In Exhibit 5-2, events occuring
prior to the first year of the model period are included with those occuring
in the first year itself. Thus, the first row of data in the exhibit
represents failures occurring before and during year one of the model period.
For some events, the data in the first row only represent failures
occuring in the first year. These events include hose ruptures, loose hose
connections, overflows and catastrophic events. These types of events either
do not result in releases for a long period of time (i.e., a year) or result
in replacement of the tank. Consequently, these events are not representative
of the physical state of the existing tank population and their frequency of
occurrence in the past is not reported.
Representative Release Profiles (RRPs). Representative release
profiles for a hypothetical tank technology are presented in Exhibit 5-3. For
purposes of exposition, release volumes are not presented for each year;
instead they are summed over 5-year intervals. For instance, for the first
representative release profile (RRP) listed in Exhibit 5-3, releases to the
environment occurred only during the first ten and last five years of the
model period. Over this period, a total of 5.666 m1 of contaminants were
released. In contrast, for the second RRP, releases occurred during each of
the four time intervals, and the total release over the model period is 10.608
m3.
It should be emphasized that even if the entries for two different RRPs
are identical for a given time interval, this does not imply that the time
pattern of releases is the same for the two RRPs over that time interval. For
example, suppose the release volumes for the first RRP during years one to
five are (5, 5, 5, 5, 5 mj), while those for the second RRP are (0, 0, 0, 0,
-------
EXHIBIT 5-3
SAMPLE REPRESENTATIVE RELEASE PROFILES
pDAr i i r
1
2
3
1
5
6
WEIGHTED
AVERAGE
Iur i PUT
.759
.127
.051
.012
.017
.OOU
1.000
VOLUME
YRS. 1- 5 1 YRS. 6-10
1.730 | 1.316
0.926 I 1.378
3.620 8.677
0.658 I 1.080
0 . 000 0 . 000
13.«431 25.100
1
1.697 1 1.763
RELEASED (cubic
YRS. 11-15
0.000
2.551
3.500
0.000
0.000
6.211
0.528
: meters)
YRS. 16-20
2.620
5.750
2.199
6.923
0.000
0.511
3.139
TOTAL
5.666
10.608
18.296
8.661
0.000
15.589
7. 127
-------
5-9
25 m3). For both profiles, the total volume released over the 5-year
interval is 25 mj, however the time pattern of releases is very different
for the two profiles. This difference in the time pattern can influence the
risks associated with the releases.
In addition to the RRPs, Exhibit 5-3 presents the weights associated with
each RRP. As discussed in Section 4.1.4, these weights reflect the proportion
of the approximately 250 profiles that each RRP represents. Thus, the first
RRP in Exhibit 5-3 represents 75.9 percent of the roughly 250 profiles; or,
stated somewhat differently, 75.9 percent of the 250 profiles generated are
similar to the first RRP. The weights of the other RRPs can be interpreted in
the same way. Note that the RRPs in Exhibit 5-3 are listed in order of
decreasing weight. This convention is adopted in all the RRP exhibits
contained in this report. The first RRP listed in each of these exhibits is
therefore the RRP with the largest weight and will be referred to as the most
representative release profile (MRRP). The MRRP is not necessarily the
representative release profile with the largest release volume or the profile
that results in the highest risk.
By multiplying each RRP by its weight, and summing over all RRPs, a
weighted-average representative release profile can be derived. This provides
a convenient summary of all 250 release profiles. The weighted-average
release profile for the RRPs in Exhibit 5-3 is presented at the bottom of the
table. Although the weighted-average profile is a convenient summary measure,
it should be interpreted with caution since it is not an actual, simulated
release profile. As such, it may not reflect a plausible combination of
failure events. We do not use the weighted-average release profile at any
point in our analysis, and it is presented in the exhibits only to facilitate
comparison of the RRPs.
In the next section, we discuss the failure frequency data and RRPs for
each of the 22 technologies modeled under the baseline regulatory scenario.
Since accumulation tanks and storage tanks have the same release profiles and
failure rates, only 15 sets of RRPs and failure rate data are discussed. In
the final section, we summarize the results for the secondary containment
regulatory scenario. Throughout our discussion we present summary data and
general trends in the results. In order to reduce the length of this chapter,
we present the detailed failure frequency data and actual representative
release profiles for each regulatory scenario in the Appendix.
5.2.2 Results for the Baseline Regulatory Scenario
In this section, we summarize the baseline results for each tank category
modeled (i.e., treatment, accumulation/storage, and small quantity
generators). For each tank technology, we discuss the failure frequency data
and representative release profiles.
Treatment Tanks
The design specifications for each of the six model treatment tanks are
summarized in the following table. Except for the distillation tank, the
aqueous generic waste stream was used when modeling all treatment tanks.
-------
5-10
Location Process
on cradles distillation
on cradles oxidation/reduction/precipitation
ongrade oxidation/reduction/precipitation
in-ground oxidation/reduction/precipitation
in-ground oxidation/reduction/precipitation
in-ground oxidation/reduction/precipitation
Size
Material (gallons)
carbon steel 2,300
carbon steel 2,300
carbon steel 60,000
concrete 3,700
carbon steel 3,700
stainless steel 3,700
Size
(m3)
8,
8.
227.
14.0
14.0
14.0
Exhibits A-l (pp. A-2 to A-8) and A-4 (pp. A-28 to A-34) of the Appendix
present the failure frequencies and RRPs, respectively, for each of the above
treatment technologies. We briefly discuss the results for each technology
below.
On cradles, carbon steel, 2,300 gal., distillation tank. The most
common failure events for this technology are overflows, pipe corrosion, and
weld/gasket failures. Except for overflows and weld/gasket failures, failures
tend to occur more frequently after the first five years of the 20-year time
horizon modeled. This increase in the occurrence of events is due to the
aging of the tank's components. Overflows are simulated to occur frequently
for distillation tanks because the distillation tank is modeled as a batch
distillation process with no automatic shut-off system to prevent overflows.
It should be noted that the frequency data for overflows corresponds to the
first overflow. The frequencies of subsequent overflow events were not
summarized, but they were simulated to occur throughout the modeled 20-year
time horizon.
On the whole, the RRPs for this technology are quite similar, with total
releases only ranging from 50 m3 to 70 mj. Moreover, the releases are
quite uniformly distributed over the four time intervals for all five RRPs.
On cradles, carbon steel, 2,300 gal., oxidation/reduction/precipitation
(ORP) tank. The frequency data for this technology are very similar to those
for the distillation tank. The primary differences are the much lower
occurrence of pipe corrosion (38.7 percent vs. 91.8 percent), and the more
uniform occurrence of overflows in the four time intervals for the
oxidation/reduction/precipitation (ORP) tank. We cannot explain this
difference in the frequency of pipe corrosion at present, and we are in the
process of investigating it. The difference in the occurrence of overflows is
due to the differences in the process control systems that we modeled for each
tank. The ORP tank is modeled as a continuous process with an automatic
shut-off system to prevent overflows.
Although the failure data for the distillation and ORP tanks are similar,
the RRPs for these technologies differ markedly. The first RRP for the ORP
tank (see Exhibit A-4, p. A-28), which represents nearly 84 percent of all
generated profiles, has a total release of only 5.67 mj. The smaller
releases for the ORP tank, relative to the distillation tank, are due
primarily to the more frequent occurrence of overflows for the distillation
tank.
-------
5-11
Ongrade, carbon steel, 60,000 gal., ORP tank. The frequency data for
this tank technology are very similar to those for the 2,300 gallon, ORP
tank. The only notable difference is the lower occurrence of tank corrosion
for the 60,000 gallon, ongrade tank (48.5 percent vs. 67.4 percent). This
difference is due to the fact that the ongrade tank has a thicker wall than
the on cradles tank (one-half inch vs. one-quarter inch). However, examining
the relevant RRPs in Exhibit A-4 (p. A-29), it can be seen that the releases
for the 60,000 gallon, ongrade tank are far larger than those for the 2,300
gallon tank on cradles. For instance, the total release volume for the most
representative release profile (MRRP) for the 2,300 gallon, tank on cradles is
5.67 mj, whereas the corresponding volume for the 60,000 gallon, ongrade
tank is 67.53 m1. This difference is primarily a function of the tank sizes
(2,300 gallons vs. 60,000 gallons).
In-ground, 3,700 gal., concrete, carbon steel, and stainless steel ORP
tanks. These three technologies differ only in terms of the material from
which the tanks are made. With a few notable exceptions, the failure data for
the three technologies are very similar to each other and to the failure data
for the treatment technologies discussed above.3 The primary differences
are in the occurrence of pipe and tank corrosion. The lowest occurrence of
pipe corrosion is for the stainless steel tank (28.1 percent), followed by
the carbon steel tank (35.7 percent), and the concrete tank (44.1 percent).
The frequency of tank corrosion is lowest for the concrete tank (0 percent)
since concrete tanks do not corrode. Carbon steel tanks have the least
corrosion resistance and consequently experience the most tank corrosion (97.9
percent).
Exhibit A-4 (pp. A-30 to A-34) reveals that for all three technologies a
single profile is representative of approximately 94 percent of all the
profiles generated. The total releases for the MRRPs range from 25 mj for
the stainless steel tank to 43 m1 for the carbon steel tank. The larger
releases from the carbon steel tank are due, largely, to the higher occurrence
of tank corrosion failures for this technology. For all three technologies,
there are a small proportion of profiles with release volumes that far exceed
those of the MRRPs.
Comparison of treatment tank technologies. In order to highlight the
similarities and differences between the treatment tank technologies, the most
representative release profile (MRRP) for each technology is plotted in
Exhibit 5-4. The exhibit reveals that the MRRP for the open top, on cradles,
1 It should be noted that we consider the HWTF model results for
concrete tanks to be considerably less reliable than the results for carbon
steel, stainless steel, and FRP tanks. The HWTF model currently does not
include an algorithm to account for cracking of concrete tanks as a result of
normal aging and weathering processes. This failure mechanism is very
important and the failure of the model to account for it may cause concrete
tanks to appear more reliable than they really are. We are currently adapting
the concrete tank algorithms of the HWTF model to account for normal cracking.
-------
EXHIBIT 5-4
MOST REPRESENTATIVE RELEASES PROFILES FOR
TREATMENT TANKS
BASELINE SCENARIO
Above-ground, on cradles,
carbon steel, closed
2,300 gal.
Above-ground, on cradles,
carbon steel, open
2,300 gal.
Above-ground, ongrade,
carbon steel,
60,000 gal.
In-ground.
concrete,
3,700 gal.
In-ground,
carbon steel,
3,700 gal.
In-ground,
stainless steel,
3.700 gal.
83.7%
96.7%
94.2%
40.3%
84.0%
Years 1-5
Years 6-10
\ 1 Years 11-IS
| ] Years 16-20
10
20
30
40
50
60
70
80
Volume Released
Note: Numbers next to bars indicate the percent of the 250 release profiles that
the MRRP represents.
-------
carbon steel, 2,300 gallon, ORP tank has the smallest total release volume.
Along with tbe distillation tank, the ORP tank is the smallest treatment tank
modeled. The MRRP with the largest release is that for the ongrade, 60,000
gallon, carbon steel tank.
Total failure frequency data for the treatment technologies are tabulated
in Exhibit 5-5." The failure data demonstrate that there is little if any
variation in the frequencies of pipe ruptures across the six treatment
technologies. With the exception of the distillation tank, this is also true
for overflows and weld/gasket failures. The greatest absolute variation in
failure frequencies is in the tank rupture, and tank/pipe corrosion
categories. The frequency of catastrophic releases is roughly similar for all
the technologies. Examining the data in Exhibit 5-5, it is clear that there
is no technology for which failure frequencies are uniformly lowest. For
example, although the distillation tank has relatively low frequencies of
occurrence for a number of failure events, it has by far the highest frequency
of pipe corrosion and overflows.
Storage/Accumulation Tanks
The hazardous waste tank Regulatory Impact Analysis (RIA) conducted by the
Office of Solid Waste in 1984 indicated that although RCRA-permitted storage
tanks and accumulation tanks have been regulated in a significantly different
manner, the actual management practices for these tanks, such as storage time
and annual throughput, are not significantly different. While the RIA
indicated slight differences in storage time, these differences were on the
order of a few days, and thus were too small to have a significant impact on
HWTF model output. Therefore, the failure and release profile results for
storage tanks were also assumed to hold for otherwise identical accumulation
tanks. In total, seven storage/accumulation tank technologies were modeled.
These are summarized below.
Location
on cradles
ongrade
underground
underground
underground
in-ground
in-ground
Size
Material (gallons)
carbon steel 5,500
carbon steel 210,000
carbon steel 4,000
fiberglass 4,000
reinforced plastic
stainless steel 4,000
concrete 2,100
carbon steel 2,100
Size
20.8
794.8
15.1
15.1
15.1
7.9
7.9
* Hose ruptures and loose hose connections are not simulated to occur
for treatment tanks because we do not model the transport of wastes to and
from treatment tanks.
-------
EXHIBIT 5-5
TOTAL FAILURE FREQUENCIES FOR TREATMENT TANKS ''
BASELINE SCENARIO
Hose Loose Weld/Gasket
Technology Rupture Hose Overflow Installation
On cradles, carlxm
steel, distillation 0.0 0.0 100.0 2.5
On cradles, carbon
steel, ORP 0.0 0.0 76.6 4.3
Ongrade, carbon steel ,
ORP 0.0 0.0 73.2 4.0
In-ground, concrete,
ORP 0.0 0.0 76.9 2.8
In-ground, carbon
steel, ORP 0.0 0.0 76.7 4.4
In-ground, stainless
steel, ORP 0.0 0.0 76.9 2.0
Pipe
Installation
Defect
0.4
I .0
0.8
0.0
0.0
0.0
Tank
Installat ion
Defect
0.8
I .5
3.3
2.L
0.8
0.8
Wei
-------
5-15
For all storage/accumulation tanks, except the tank on cradles, separate
release profiles were generated for the generic aqueous and toluene waste
streams.5 (The generic aqueous waste stream was used in modeling releases
from the tank on cradles.) Thus, a total of 13 sets of failure data and
representative release profiles (RRPs) were generated for storage/accumulation
tanks. For all six of the technologies for which separate results were
generated for aqueous and toluene wastes, failure frequencies were essentially
independent of waste type, therefore failure data for only the toluene waste
stream are discussed below. On the other hand, representative release
profiles did vary significantly with waste type, given the importance of waste
density and viscosity in determining leak rates. As a result, RRPs for all 13
runs are discussed. Failure frequency data for the storage/accumulation tanks
are presented in Exhibit A-2 of the Appendix (pp. A-9 to A-23), while the RRPs
are presented in Exhibit A-5 (pp. A-35 to A-49).
On cradles, carbon steel, 5,500 gal. tank. The most common failure
events for this technology are weld/gasket failures, tank corrosion, loose
hose connections, and overflows. Failures appear to occur primarily during
the first ten years of the model period because Exhibit A-2 only presents the
first occurrence of these events, and not the second, third or fourth
occurrences.
Examining the RRPs for this technology, it can be seen that the total
release for the MRRP is quite small (5.67 m1). The total releases for the
other RRPs range from 0 m3 to approximately 46 m3.
Ongrade, carbon steel, 210,000 gal. tank. The failure frequencies for
this technology differ markedly from those for the 5,500 gallon tank on
cradles. For both technologies, weld/gasket failures and tank corrosion are
very frequent; however, the frequency of hose ruptures and loose hose
connections are zero for the ongrade tank since hoses are not used to empty
the tank. On the other hand, pipe corrosion occurs much more frequently for
the ongrade tank given the high throughput for this tank, which results in
faster pipe erosion and corrosion. Also, the pipes for the ongrade tank are
21 years old at the beginning of the model period and age contributes to high
failure rates.
The RRPs for the ongrade tank, which are displayed in Exhibit A-5 (pp.
A-35, A-36), reveal that relatively large volumes of waste are released from
this tank. Total releases range from 342 m3 to 82,000 m3 for aqueous
wastes, and 977 m1 to 104,000 m1 for toluene wastes. This may be
attributed to the large size of the tank and its high throughput. Note,
however, that for both waste types, the MRRPs have the smallest releases. For
* For these technologies, tanks are in direct contact with soil. The
leak rate from a tank to soil depends on the density and viscosity of the
waste it contains, hence the use of two generic waste streams. Only one waste
stream was necessary to estimate leak rates from tanks not in contact with
soil.
-------
5-16
nearly all the RRPs, releases increase over time. This is especially true of
the MRRPs for the two waste types. This time pattern of releases is a result
of undetected leaks from the bottom of the ongrade tank that increase in
volume over time. The model assumes that detection does not occur until 25
percent of the monthly tank throughput leaks out over a one month period.
Underground, 4,000 gal., carbon steel, FRP, and stainless steel tanks.
These three technologies differ only in terms of the material from which the
tanks are made. The major effect of tank material appears to be on the
frequency of pipe/tank ruptures and corrosion. Frequencies of pipe/tank
ruptures are similar for the carbon steel tank and the stainless steel tank,
but are substantially higher for the fiberglass reinforced plastic (FRP)
tank. This tank experiences more pipe/tank ruptures because the material
strength of FRP is not as high as that of steel. However, since FRP tanks
donot corrode, the frequency of pipe/tank corrosion is zero for these tanks.
The carbon steel tank has more failures due to corrosion than the stainless
steel tank. This difference in the occurrence of corrosion and the relative
magnitudes of the release volumes discussed below indicate that corrosion is
the over-riding factor influencing the magnitude of release volumes. Other
events, such as overflows and pipe ruptures, do not lead to the high releases
associated with corrosion.
Comparing the profiles for the three technologies in Exhibit A-5 (pp. A-38
to A-45), it can be seen that the smallest releases are associated with the
FRP tank, with the stainless steel tank a close second. The total releases
for the MRRPs of these two technologies are on the order of 2 m1. On the
other hand, the total releases for the carbon steel tank MRRPs are roughly two
to three times this magnitude, primarily due to the higher occurrence of
corrosion.
In-ground, 2,100 gat., concrete and carbon steel tanks. These two
technologies also differ only in terms of the material of which the tanks are
made. As the failure data for these tanks reveal, there is a somewhat higher
occurrence of both pipe and tank installation deficiencies for the carbon
steel tank, and a much higher occurrence of tank corrosion (79.9 percent vs. 0
percent) since concrete does not corrode. The higher occurrence of
installation deficiencies is due to more frequent replacements of the carbon
steel tank.
The RRPs for these tank technologies, which are presented in Exhibit A-5
(pp. A-45 to A-49), indicate that releases are larger for the carbon steel
tank. This is due primarily to the higher occurrence of corrosion for the
carbon steel tank, and is also a result of the HWTF model's incomplete
modeling of the dominant failure mechanisms associated with concrete tanks.'
'See footnote three for a discussion of the deficiency of the model in
this regard.
-------
5-17
Comparison of storage/accumulation tank technologies. In order to
facilitate comparison of the various storage/accumulation tank technologies,
the MRRP for each technology is plotted in Exhibit 5-6. For simplicity, only
the toluene waste stream MRRPs are shown since they convey virtually the same
information as the aqueous waste stream MRRPs. Total failure data for the
storage/accumulation technologies are presented in Exhibit 5-7. Failure data
are also presented only for the toluene waste stream since failure frequencies
did not vary significantly with waste type.
Exhibit 5-6 shows that the MRRP with the smallest total release is for the
underground, stainless steel, 2,100 gallon tank. Note that this is the
smallest storage/accumulation tank modeled. The MRRP with the largest release
is for the ongrade, 210,000 gallon tank made of carbon steel, which is the
largest tank modeled.
The total failure data in Exhibit 5-7 reveal that there is considerable
variation in the frequency of failure events across the various storage/
accumulation technologies. The highest variation is in the pipe and tank
corrosion categories and the overflow category. The lowest occurrence of
corrosion is in the underground FRP and stainless steel tanks, and the
in-ground concrete tank; however the in-ground concrete tank has a high
frequency of overflows. As in the case of the treatment tanks, no technology
has uniformly lowest failure frequencies. But, the underground FRP and
stainless steel tanks have relatively low failure frequencies in nearly all
the categories.
The frequency of catastrophic releases and overflows differ considerably
for the various storage/accumulation tank technologies. This is due to
differences in the assumed decision and operating parameters for the
technologies. For the above-ground and in-ground tank technologies, waste is
pumped into the tanks and has a direct overflow path through either the vent
pipe or the open top. In contrast, for the underground tank technologies,
waste is gravity fed and the vent pipe is normally placed above the height of
the fluid in the source tank. As a result, waste cannot overflow from the top
of the vent pipe. The only overflow paths are through corroded vent pipes or
failed weld/gasket connections.
As for the differences in the occurrence of catastrophic releases, these
are less frequent for underground tanks because they are assumed not to be
subject to vandalism. Moreover, a nearby fire or explosion is assumed to be
one-third as likely to damage an underground tank as it is to damage an
above-ground tank.
Small Quantity Generator (SQG) Tanks
Two small quantity generator (SQG) tank technologies were modeled: a 200
gallon, above-ground, on cradles, carbon steel tank, and a 200 gallon (8 m3),
underground carbon steel tank. For the underground tank, two sets of results
were generated: one for the aqueous waste stream and the other for the toluene
waste stream. For the above-ground tank only one waste stream is used since
-------
EXHIBIT 5-6
MOST REPRESENTATIVE RELEASE PROFILES FOR
STORAGE/ACCUMULATION TANKS
BASELINE SCENARIO
Above-ground, on cradles,
carbon steel
5,500 gal.
Above-ground, ongrade,
carbon steel,
210.000 gal.
Underground,
carbon steel,
4,000 gal.
Underground,
fiberglass reinforced plastic,
4,000 gal.
Underground,
stainless steel,
4,000 gal.
In-ground,
concrete,
2,100 gal.
In-ground,
carbon steel,
2,100 gal.
Years I-S
Years 6-10
1 1 Years 11-15
| ] Years 16-20
Volume Released
Note: Numbers next to bars indicate the percent of the 250 release prot;le:
ihe \1RRP represents.
-------
EXHIBIT 5-7
TOTAL FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
BASELINE SCENARIO (TOLUENE)
Pipe Tank Weld/ ci ta -
Hose Loose Held/Gasket Installation Installation Gasket Pipe Tank pipe Tink stroplm:
Technology Rupture Hose Overflow Installation Defect Defect Failure Rupture Rupture Corrosion Corrosion Release
On cradles, carbon
steel
On grade, carbon
steel
33.8 53.2
0.0
Underground, carbon
steel 24.9
Underground FRP
24.5
Underground, stainless
steel 25.4
In-ground, concrete 12.0
In-ground, carbon
steel
0.0
54.6
55.6
52.5
52.9
12.4 49.3
48.5
2.0
4.9
5.6
5.7
42.8
41.0
0.8
1.2
0.4
0.4
0.4
0.8
1.0
0.0
1.2
0.8
1.3
0.8
0.0
0.5
0.0
0.8
0.4
0.0
0.0
0.4
2.0
74.4
81 .2
61.0
59.1
59.0
80.7
78.9
10.4
13.1
13.0
24.3
13.0
10.0
10.5
7.7
11.7
12.1
17.3
12.4
11.0
9.3
9.0
98.5
15.0
0.0
4.5
8.0
12.2
56.7
51 .0
Hd .'»
II. O
4.0
2.1 v£>
2.1
4.H
7'J.'»
* Frequencies are expressed as percent of 250 iterations in which the failure event specified occurred.
-------
5-20
the tank is on cradles and not in contact with soil As a result, the leak
rate from the tank does not depend on the physical properties of the waste it
contains. The failure data for the SQG technologies are presented in Exhibit
A-3 (pp. A-24 to A-27) of the Appendix, and the RRPs are presented in Exhibit
A-6 (pp. A-50 to A-53).
Underground, 200 gal., carbon steel tank. Examining the failure data
for the underground tank first, it can be seen that the failure frequencies
are virtually identical for the two waste streams. The most common failure
events are tank corrosion (approx. 80 percent), weld/gasket failures (approx.
60 percent), and loose hose connections (approx. 30 percent). The remaining
failure events occur quite infrequently.
The RRPs for the underground tank technology suggest that smaller releases
are associated with the toluene waste stream. As discussed further below,
this difference may be attributed to the earlier detection of toluene releases.
Above-ground, 200 gal., carbon steel tank. The failure frequency data
for the above-ground SQG tank technology indicate that weld/gasket failures,
tank corrosion, and loose hose connections are also among the most common
failure events for this technology. However, in contrast to the underground
SQG tank, overflows occur very frequently for the above-ground tank (48
percent vs. approx. 5 percent). As discussed previously in the context of
storage/accumulation tanks, above-ground tanks have direct overflow paths
through either the vent pipe or open top. In contrasts, underground tanks do
not tend to overflow since we assume they are gravity fed and overfilling just
results in the wastes backing up in the vent pipe. Thus, the only overflow
paths we model for underground tanks are through corroded vent pipes or failed
weId/gasket connnections.
In addition, catastrophic releases are twice as frequent for the above-
ground tank since it is assumed to be more susceptible to vandalism than an
underground tank. However, tank corrosion is far less frequent for the
above-ground tank (44 percent vs. 80 percent) since it is only in contact with
the atmosphere and is not in contact with the corrosive environment of the
soil, unlike the underground tank.
The RRPs for the above-ground SQG tank indicate that lower volumes of
wastes are released relative to the underground tank.
Comparison of SQG tank technologies. A comparison of the MRRPS for the
three SQG runs is presented in Exhibit 5-8. The MRRP for the above-ground SQG
tank clearly has the smallest release. The MRRPs for the underground SQG tank
differ somewhat in that these are positive releases during years one to five
for the aqueous wastes stream but no the toluene waste stream. It should be
noted, however, that the aqueous wste stream MRRP is only representative of
58.5 percent of all the release profiles. For all but one of the other RRPs
for the aqeous waste stream, there are no releases during years one to five
(see Exhibit A-6, p. A-51.)
-------
5-21
Above-ground,
:arbon steel, (Aqueous)
200 gai.
Underground,
carbon steel (Toluene)
200 gal.
Underground,
carbon steel, (Aqueous)
200 gal.
EXHIBIT 5-8
MOST REPRESENTATIVE RELEASE PROFILES FOR
SQG TANKS
BASELINE SCENARIO
67.2%
73.5%
Years 1-5
Year; 6-10
Years 11-15
Years 16-20
58.5
Volume Released (m
Note: Numbers next to bars indicate the percent of the 250 release profiles that
the MRRP represents.
-------
5-22
The total failure frequencies for the SQG tank technologies are shown in
Exhibit 5-9. This tabulation simply highlights the observations made above
regarding the relative frequency of failure events for the two technologies.
Comparison of Tank Technologies
Exhibit 5-10 shows the total releases for the MRRPs of the fifteen tank
technologies modeled. In the case of the storage/accumulation technologies
and the underground SQG technology, the data plotted are for the toluene
generic waste stream. As the exhibit clearly demonstrates, the largest total
release is associated with the 210,000 gallon, ongrade, carbon steel storage/
accumulation tank, which is also the largest tank modeled. However, with the
exception of this tank, the largest releases are for treatment tanks, while
SQG tanks have the smallest releases. It should be noted that these
conclusions also hold with the RRPs for the aqueous waste streams, which are
not plotted.
Comparison of Tank Materials
Several of the tank technologies modeled differ only in terms of tank
material. The MRRPs for these technologies are plotted in Exhibit 5-11. The
data plotted for the storage/accumulation tanks are for the toluene generic
waste stream profiles, however the conclusions drawn below also hold for the
aqueous waste stream profiles.
The first group of bars in the figure are for the 2,700 gallon, in-ground
treatment tanks. For these three technologies, the largest release is from
the carbon steel tank, and the smallest release is from the stainless steel
tank. The release from the concrete tank is slightly larger than that from
the stainless steel tank.7
The second group of bars in the figure are for the 4,000 gallon,
underground storage/accumulation tanks. The largest release for this group is
also from the carbon steel tank. The releases from the stainless steel and
fiberglass reinforced plastic (FRP) tanks are similar in magnitude, however
the FRP tank has a slightly larger release volume.
The final group of bars plotted in Exhibit 5-11 are for the 2,100 gallon,
in-ground storage/accumulation tanks. For this group, the larger release is
also from the carbon steel tank.
The above material comparisons suggest that the largest releases are
consistently associated with carbon steel tanks. Conclusions regarding the
material with the smallest releases are difficult to make given the
combinations of tank technologies modeled and the"assumptions used in the HWTF
model (e.g., the cracking of concrete tanks was not incorporated in the model).
7 See footnote three regarding the accuracy of the HWTF model results
for concrete tanks.
-------
EXHIBIT 5-9
TOTAL FAILURE FREQUENCIES FOR SQG TANKS *
BASELINE SCENARIO
Technology
Above -ground
Underground (toluene)
Underground (aqueous)
Hose
Rupture
0.8
2.4
2.4
Loose
Hose
34.1
29.7
29.0
Overflow
48.3
5.3
5.7
Weld/Gasket
I nstalla tion
0.8
o.e
0.8
Pipe
Installation
Defect
0.0
0.4
0.8
Tank
Insta I lation
Defect
0.8
0.8
I .6
Meld/
Gasket
Fa 1 1 are
74.3
59.7
60.4
Pipe
Rupture
10.4
12.9
13.2
Tank
Rupture
6.9
12.1
11 .2
Pipe
Corrosion
9.0
14.4
14.3
T.I nk
Cor t os ion
44. L
7S» .0
BO.O
Ci la -
str nph ic
Kel e<»se
4.0
2.1
2.1
* Frequencies are expressed as percent of 250 iterations in which the failure event specified occurred.
-------
5-24
EXHIBIT 5-10
MOST REPRESENTATIVE RELEASE PROFILES FOR
ALL TANK TECHNOLOGIES
BASELINE SCENARIO (TOLUENE)
Treatment, <
Above-ground
Treatment, 4
In-ground
Storage/Accumulation,
Above-ground
Storage/Accumulation,
Underground
on cradles,
distillation
on cradles,
ORP
ongrade
Storage/Accumulation,
In-ground
SQG.
Above-ground
SQG.
Underground
concrete
S>.XX'V ' ' \?t'/N.^\SSN^^V\Ny\!''''^ carbon steel
"'; """:, '" ",", '",' """';<-'" "": stainless steel
on cradles
carbon steel
^fiberglass reinforced plastic
~]stainless steel
concrete
I carbon steel
142m
ongrade
\. | Treatment
{ I Storage!Accumulation
10
20
30 40
50 60 70 80 90
Volume Released (m
-------
5-25
EXHIBIT 5-11
COMPARISON OF TANK MATERIALS
MOST REPRESENTATIVE RELEASE PROFILES FOR SELECTED TANKS
BASELINE SCENARIO (TOLUENE)
Treatment,
In-ground
Storage/Accumulation,
Underground
Storage/Accumulation,
In-ground
I | Concrete
II Carbon Sttel
[ ] Stainless Steel
Fiberglass
Reinforced Plastic
10 20 30
Volume Released (m ^ )
40
50
-------
5-26
Comparison of Waste Types (Aqueous vs. Toluene)
As noted previously, for six of the seven storage/accumulation tank
technologies and for the underground SQG tank technology, two sets of failure
frequencies and release profiles were generated: one for aqueous wastes and
another for toluene wastes.1 The failure frequencies obtained for these
technologies do not vary significantly with waste type, however there is some
variation in the RRPs for the two sets of results. The total releases of the
toluene and aqueous waste stream MRRPs for each technology are compared in
Exhibit 5-12. As the figure reveals, there is no consistent relationship
between total release volume and waste type. For the first technology (the
5,500 gallon, ongrade tank) and the last two (the in-ground carbon steel tank
and the underground SQG tank), total releases are larger for the aqueous waste
stream MRRPs, substantially so in the case of the tank on cradles. However
for the remaining technologies, total releases are somewhat larger for toluene.
The absence of a consistent relationship between waste type and release
volume is the result of two opposing effects. Although the release rate for
the toluene waste stream is higher than that for the aqueous waste stream, an
aqueous release may remain below the assumed detection limit for a longer
period of time. As a result, the total volume released may be larger or
smaller for the toluene waste stream relative to the aqueous waste stream.
5.2.3 Results for the Secondary Containment Regulatory Scenario
In this section, we present and discuss the representative release
profiles (RRPs) for the secondary containment simulations.
The failure frequency data for the secondary containment scenario are not
discussed here (although they are presented in the Appendix) since the primary
purpose of secondary containment is not to alter the frequencies of failure
events, but to prevent releases from reaching the environment. Our
presentation is, therefore, limited to discussing the representative release
profile results.
Although secondary containment is not (necessarily) designed to affect
failures, the failure results in the Appendix indicate differences between
baseline and secondary containment failure frequencies (see Exhibits A-l, A-2,
and A-3). These difference may be attributed to the following three factors.
First, the implementation of secondary containment assumes the installation of
a new tank. Therefore, the frequency of tank corrosion and weld/gasket
failures is lower than under the baseline scenario since corrosion is a
time-dependent process and all the baseline tanks are at least five years old
at the start of the model period. Second, since the secondary containment
tank technologies have double walls, a single wall will not be subjected to
simultaneous interior and exterior corrosion, as would be true for the
* Only the aqueous waste stream was used in modeling the storage/
accumulation tank on cradles since it is not in contact with soil.
-------
5-27
EXHIBIT 5-12
COMPARISON OF WASTE TYPES
MOST REPRESENTATIVE RELEASE PROFILES FOR SELECTED TANKS
BASELINE SCENARIO
Above-ground, ongrade,
carbon steel,
5,500 gal.
Underground,
carbon steel,
4,000 gal.
Underground,
fiberglass reinforced plastic,
4,000 gal.
Underground,
stainless steel,
4,000 gal.
In-ground,
concrete,
2,100 gal.
In-ground,
carbon steel,
2,100 gal.
Underground,
carbon steel*,
200 gal.
10
Total Volume Released (m )
* SQG tank, all others are storage/accumulation tanks
-------
5-28
single-walled baseline tank technologies. Finally, the frequency of
installation deficiencies is higher for the secondary containment scenario
because new tanks are being installed.
Below we provide a general discussion of the representative release
profiles for the secondary containment scenario.
Treatment Tanks
The RRPs for treatment tanks with secondary containment are presented in
Exhibit A-4 of the Appendix (pp. A-28 to A-24). A comparison of these RRPs
with those for the baseline treatment tanks reveals a pronounced reduction in
release volumes due to secondary containment. This is especially true for the
MRRPs. The MRRPs for all secondary containment treatment tanks have zero
total releases, whereas the baseline MRRPs have total releases ranging from
approximately 6 mj (the 2,300 gallon, open top, ORP tank on cradles) to 70
m3 (the 2,300 gallon, closed top, distillation tank on cradles).
Storage/Accumulation Tanks
Exhibit A-5 of the Appendix (pp. A-35 to A-49) contains the RRPs for
storage/accumulation tanks with secondary containment. Once again, a
comparison of these profiles with the corresponding baseline profiles reveals
a marked reduction in total releases due to secondary containment, especially
for the MRRPs, The total releases for the secondary containment MRRPs range
from 0 m1 to approximately 0.023 m3, whereas the total releases for the
baseline MRRPs range from 1.39 mj to 977 mj.
SQG Tanks
The RRPs for the three secondary containment SQG runs are shown in Exhibit
A-6 of the Appendix (pp. A-50 to A-53). Comparing these with the baseline SQG
RRPs, it can be seen that secondary containment results in a definite
reduction in releases, but a less dramatic one than that observed in the case
of treatment and storage/accumulation tanks.
Total releases over the 20-year time horizon for the secondary containment
MRRPs are zero in all cases, while total releases for the baseline MRRPs range
from 0.38 m3 (above-ground tank) to 1.96 mj (underground tank).
5.2.4 Other Regulatory Scenarios
In this section we provide a very brief discussion of the HWTF model
results for the partial secondary containment with ground-water monitoring,
corrosion protection, and leak-testing with ground-water monitoring
scenarios. The summary results for these three scenarios are presented in
their entirety in the Appendix to the report.
The frequencies of failure events under the partial secondary containment
scenario are nearly identical to those under the baseline scenarios. This is
due to the fact that partial secondary containment does not entail any changes
-------
5-29
in the tank itself but involves installing structures to contain releases and
prevent them from reaching the environment.
In nearly all cases, partial secondary containment resulted in a
substantial reduction in release volumes relative to the baseline scanario.
However, the reduction was not as marked as with secondary containment. This
may be a result of the inability of partial containment to prevent releases
from the on-grade or in-ground portion of a tank. . Thus, such releases may not
be detected until they are detected by ground-water monitoring at the
monitoring well.
The frequencies of failure events under the corrosion protection scenarios
are virtually identical to those for the baseline scenario, with the exception
of pipe and tank corrosion. The frequencies of these two events were far
lower with corrosion protection. However, the release volumes for the RRPs
under the corrosion protection scenarios were generally not smaller than those
for the baseline scenario. This is due, in part, to the fact that corrosion
protection alone does not reduce the frequency of many of the most important
failure events, such as ruptures and overflows. These events can account for
a very large portion of total releases. It is also important to note that we
modeled corrosion protection without any form of leak detection.
The failure frequencies for the leak testing scenario also do not differ
significantly from those for the baseline scenario. However, the total
releases for the RRPs for this scenario are substantially smaller than those
for the baseline scenario. But the releases are not as small as those for
secondary containment. The implementation of leak testing does not prevent
failures from occurring, nor does it prevent releases from contaminating the
environment. However, if releases are eventually detected through either leak
testing or ground-water monitoring, the leaking tank is removed, resulting in
lower release volumes compared to the baseline scenario.
Although we do not discuss the HWTF model results for these regulatory
approaches in detail, we present the risks associated with each alternative in
the following section. Since the HWTF model results for each model tank is
the only variable between regulatory scenarios, the detailed risk comparisons
will directly reflect the HWTF model results under the alternative regulatory
scenarios. The next section presents the risk estimates that were generated
based on the representative release profiles discussed above and presented in
the Appendix to the report.
5.3 RISK RESULTS
Based on the representative release profiles from the HWTF model, we
constructed frequency distributions of dominant risk estimates for the 22
model tank technologies under the five regulatory scenarios considered in the
analysis. To construct these distributions, we performed TRAM runs for each
of the waste streams and generic hydrogeologic settings associated with a
-------
5-30
model tank technology. Each of the release profile-waste stream-
hydrogeologic setting dominant risk estimates were weighted, and these
weighted estimates were then used to construct a frequency distribution of
dominant risk estimates for a particular tank technology.' An example
distribution is illustrated in Exhibit 5-13. This hypothetical example
illustrates that >0 percent of all underground, carbon steel, storage tanks
-3 -1
are associated with a dominant risk estimate between 10 and 10
We derived the frequency distribution of dominant risk estimates for a
particular combination of tank technology and regulatory scenario as follows;
Generated risk estimates for each combination of chemicals
in identified waste streams, hydrogeologic settings, and
representative release profiles. This step resulted in
approximately 1,294 risk estimates.10
Determined dominant risk estimates for each combination of
waste stream, hydrogeologic setting, and representative
release profile. This step resulted in approximately 324
dominant risk estimates.
Assigned weights to each dominant risk estimate. This
weight represents the percentage of the tank population that
handles the associated waste stream, is located in the
associated hydrogeologic setting, and is represented by the
representative release profile. The weight assigned to a
dominant risk estimate for a partcular combination of waste
stream, X, and hydrologic setting, Y, and representative
release profile, Z, is the product of: (1) the percentage of
the tank technology that handles waste stream X; (2) the
percentage of the tank technology that is located in
hydrogeologic setting Y; and (3) the percentage of the tank
technology that has releases as indicated by representative
release profile Z.
9To represent the risks for a particular waste stream, we chose the risk
for the dominant chemical, rather than summing the risks for all chemicals in
that waste stream. In addition, we did not differentiate between carcinogenic
and non-carcinogenic risk estimates. Because we did not differentiate
carcinogenic and non-carcinogenic risk estimates, we are evaluating both these
health effects on an equal basis. For example, a situation with a dominant
.3
risk estimate of 10 with a carcinogenic effect is evaluated on the same
.3
scale as a dominant risk estimate of 10 with a non-carcinogenic effect.
It is important to note that we did estimtae risk estimates for each chemical
in a particular waste stream.
10Per technology there are approximately six waste streams, four
constituents per waste stream, nine hydrogeologic settings and six
representative release profiles.
-------
HYPO I Ml I ICAI HUQUfNCY 01 SIM I HUI ION Of
RISK I SI IMAU S UNOfRCROUNO CAKBON SI ML SIOHACf
IANK (ii.Oim «jal.) All HOW MHOS. All WASIf SIHfAMS
R
E
L
A
T
1
V
E
F
R
E
Q
U
E
N
C
y
1.0 »
I
I
0.9
1
1
0.1
1
1
1
Q.T »
1
1
1
0*6 »
1
1
0.5 »
1
1
1
0.4 »
1
1
1
0.3 »
1
|
1
0.2 »
. 1
1
|
0.1
1
| *****
| *****
*****
*****
*****
***««
!**«*«
*****
*****
**?** **«*«
«l44l** *****
***** #*4**
****> ***««
***** *****
****c *****
****« ***** ***** *****
***** ***** ***** *****
*****
*****
*****
*****
*****
*****
«*««3
*****
*****
3A*«*
*****
««***
Ui
I
-16
-12
-10 -8
LOG OF RISK FSriHATF
-6
-------
Calculated frequency distribution of dominant risk
estimates using the dominant risk .estimates and their
associated weights.
It is important to note that the frequency distribution of dominant risk
estimates for a particular model tank technology (under any regulatory
scenario) is a function of the representative release profiles, the waste
streams handled, and the generic hydrogeologic settings for that tank. All of
these factors are dependent on a tank's size, age, material of construction,
location, and category, and thus are unique to each of the 22 model tank
technologies. Therefore, the results presented in this section are mainly
useful for comparing relative risks among regulatory scenarios, rather than
for direct comparisons between tank technologies.
Below, we present several risk comparisons for the scenarios considered in
this analysis. These comparisons include:
A comparison of risks for baseline and secondary
containment scenarios for storage, treatment, accumulation,
and SQG tanks;
A comparison of risks for underground tanks for baseline,
secondary containment, and leak testing with ground-water
monitoring scenarios;
A comparison of risks for baseline, secondary
containment, and partial containment with ground-water
monitoring for treatment, storage, accumulation and SQG
tanks; and
* A comparison of risks for baseline and corrosion
protection scenarios for carbon steel and stainless steel
tanks.
Rather than presenting frequency distributions, such as presented in
Exhibit 5-13, for each model tank technology under each scenario, we have
summarized results for the different scenarios and have presented them in bar
graphs so that direct comparisons between risk estimates for different
scenarios can be made more readily. The bar graphs present the percent of
-9
non-zero risk estimates (in this analysis we used 10 as the zero/non-zero
risk cut-off) associated with a particular tank and regulatory scenario.
Also, in some graphs, we have divided the non-zero risk estimates into low
-9 -5 -5 -3
risk (10 to 10 ), medium-risk (10 to 10 ) and high-risk (greater
_o
than 10 ).11 Thus, these graphs also allow for the reduction in high
risk situations to be compared for different scenarios.
llThese divisions are arbitrary and were selected to aid in presenting
the results. Also, it should be noted that these risks are relative (i.e.,
for comparative purposes only), and have no significance in an absolute sense.
-------
5-33
5.3.1 Comparison of Risks for Baseline and Secondary
Containment Scenarios
To compare risks for existing hazardous waste tanks managed according to
the baseline scenario to new tanks with full secondary containment and
interstitial monitoring systems, we evaluated risks under each scenario for
all of the 22 model tank technologies included in the analysis, and used these
risk estimates to construct bar graphs comparing non-zero risk estimates for
the two scenarios. These bar graphs are presented in Exhibits 5-14 through
5-17, for treatment, storage, accumulation, and SQG tanks, respectively.11
As illustrated, risk estimates were reduced significantly for all tanks,
with a typical reduction of about 50 percent in non-zero risk estimates
between baseline and secondary containment scenarios. More importantly, high
.3
risk situations (relative risks greater than 10 ) were reduced by 80
percent or more for most tank technologies. Thus, from our analysis, we
conclude that secondary containment systems, equipped with leak detection
systems, are a very effective method of reducing the risks associated with
hazardous waste tanks. Secondary containment systems prevent most releases
from escaping to the environment, allowing hazardous wastes to be released -
only as a result of catastrophic events, hose ruptures and loose hose
connections (for underground tanks), and cracking of concrete pads and berms.
Our risk analysis for these scenarios produced one unusual set of
results. One tank technology, an in-ground, concrete, accumulation tank, was
found to present zero risk under both the baseline and secondary containment
scenarios. Although this tank stores relatively low-risk, dilute waste
streams, we believe that this result is driven mainly by a lack of resolution
within the HWTF model with regard to failure events for concrete tanks.
Therefore, we consider this result, and the results of our analysis for all
concrete tanks, to be very preliminary and subject to revision as we obtain
better information. (This holds true not only for the baseline and secondary
containment scenarios, but for all other scenarios as well.)
5.3.2 Comparison of Risks for Underground Tanks for Baseline,
Secondary Containment, and Leak-Testing with Ground-Water
Monitoring Scenarios
In Exhibit 5-18, we compare relative risks for underground carbon steel,
stainless steel, and FRP tanks for the baseline, secondary containment, and
leak-testing with ground-water monitoring scenarios. As illustrated, leak
testing with ground-water monitoring, while reducing risks for underground
tanks compared to the baseline scenario, is not nearly as effective as
secondary containment in reducing risks associated with these tanks.
12These exhibits present the percent of risk estimates categorized as
low, medium, and high-risk situations. For example, as illustrated in Exhibit
5-14, for an above-ground, ongrade, carbon steel tank approximately 10, 10 and
35 percent of the risk estimates are low, medium and high-risk, respectively.
-------
5-34
EXHIBIT 5-14
COMPARISON OF RISKS FOR
BASELINE AND SECONDARY CONTAINMENT SCENARIOS*
TREATMENT TANKS
Above-ground, on cradles,
carbon steel, open,
2,300 gal.
Above-ground, on cradles
carbon steel, closed,
2,300 gal.
Above-ground, ongrade,
carbon steel,
60,000 gal.
In-ground,
concrete,
3,700 gal.
In-ground,
carbon steel,
3,700 gal.
In-ground,
stainless steel,
3,700 gal.
| | Bastlint
E^ll Sicondary Containment
10
20
30
40
50
60
70
80
90
100
Percent Non-Zero Risk Estimates
* Lines on bars represent cut-off between arbitrarily selected low, medium, and
high risk categories.
b For this technology, 13 and 71 percent of the risk estimates are medium and
high risk, respectively.
-------
5-35
EXHIBIT 5-15
COMPARISON OF RISKS FOR
BASELINE AND SECONDARY CONTAINMENT SCENARIOS'
STORAGE TANKS
Above-ground, on cradles,
carbon steel,
5,500 gal.
Above-ground, ongrade,
carbon steel,
210,000 gal.
Underground,
carbon steel,
4,000 gal.
Underground,
fiberglass reinforced plastic,
4,000 gal.
Underground,
stainless steel,
4.000 gal.
In-ground,
concrete,
2,100 gal.
In-ground,
carbon steel,
2,100 gal.
| 1 Bastlint
mil Secondary Containment
10 20 30 40 50 60 70
Percent Non-Zero Risk Estimates
80
90
10
a Lines on bars represent cut-off between arbitrarily selected low, medium, and
high risk categories.
-------
5-36
EXHIBIT 5-16
COMPARISON OF RISKS FOR
BASELINE AND SECONDARY CONTAINMENT SCENARIOS
ACCUMULATION TANKS
Above-ground, on cradles,
carbon steel,
5,500 gal.
Above-ground, ongrade,
carbon steel,
210,000 gal.
Underground,
carbon steel,
4,000 gal
Underground,
fiberglass reinforced plastic,
4,000 gal.
Underground,
stainless steel,
4,000 gal.
In-ground,
concrete,
2,100 gal.
In-ground,
carbon steel,
2,100 gal.
I 1 Bastlint
WJJi Stcondary Containment
10 20 30 40 50 60 70
Percent Non-Zero Risk Estimates
80
90
100
a Lines on bars represent cut-off between arbitrarily selected low, medium, and
high risk categories.
-------
5-37
EXHIBIT 5-17
COMPARISON OF RISKS FOR .
BASELINE AND SECONDARY CONTAINMENT SCENARIOS
SMALL QUANTITY GENERATOR TANKS
Above-ground,
carbon steel,
200 gal.
Underground,
carbon steel,
200 gal.
| | Baseline
Secondary
Containment
10 20 30 40 50 60 70
Percent Non-Zero Risk Estimates
80
90
100
* Lines on bars represent cut-off between arbitrarily selected low, medium, and
high risk categories-
-------
5-3d
EXHIBIT 5-18
COMPARISON OF RISKS FOR BASELINE, SECONDARY CONTAINMENT AND
LEAK TESTING WITH GROUND-WATER MONITORING SCENARIOS
UNDERGROUND TANKS
Accumulation,
carbon steel,
4,000 gal.
Accumulation,
fiberglass reinforced plastic,
4,000 gal.
Accumulation,
stainless steel,
4,000 gal.
Small Quantity Generator,
carbon steel,
200 gal.
Storage,
carbon steel,
4,000 gal.
Storage,
fiberglass reinforced plastic,
4,000 gal.
Storage,
stainless steel,
4,000 gal.
| | Baseline
[ ] Leak Testing
with Ground-water
Monitoring
Fg-3 Secondary
'Iffihl Containment
10 20 30 40 50 60 70 80 90
Percent Non-Zero Risk Estimates
100
-------
5-39
Our leak testing and ground-water monitoring scenario assumes that leak
tests are performed semi-annually, using a hydrostatic pressure test that is
capable of detecting leaks of 3.8 x 10 m3 (0.1 gallons) per hour.
Ground-water monitoring is provided by wells located 10 meters downgradient,
and contamination is assumed to be detected as soon as the concentration of
any chemical stored in the tank exceeds a detectable concentration at the
monitoring well. The tank is considered to be removed upon detection of a
leak and all subsequent releases are set to zero. IJ
Thus, our results for leak testing and ground-water monitoring are
somewhat driven by our assumptions. However, we believe that both of the
assumptions mentioned above are reasonable, and that our results illustrate
several real problems that limit the effectiveness of leak testing with
ground-water monitoring. The two major problems are:
Leak testing is not effective in detecting small
leaks. A leak of 0.0001 m3/hour seems trivial, but
considering the highly concentrated nature of some wastes
stored in underground tanks, such an amount may present
significant risks to human health. For example, a tank
leaking at 0.0001 mj/hour will release approximately 1
m3 (264 gallons) of hazardous waste over a year. If this
waste contains 10 percent benzene,1* the relative
lifetime risk to an individual using ground water
-4
contaminated by this leak may be as high as 10
Ground-water monitoring is limited in its effective-
ness in detecting releases of contaminants from hazardous
waste tanks. For highly mobile contaminants in
hydrogeologic settings where horizontal ground-water
velocities are rapid and unsaturated zone depths are small,
detection of releases by ground-water monitoring is
1'Because we assume that the tank is removed (instead of replaced) and
subsequent release volumes are set to zero, the distribution of risk estimates
represents a lower bound of nonzero risk estimates. In actuality, the tank
is likely to be replaced and future releases may occur. For this study, we
did not model tank replacement (following detection by ground-water
monitoring) because we used two different models to model detection and
releases. TRAM is used to model subsurface transport and detection, and HWTF
is used to model releases. We could not integrate the results from both
models within a reasonable framework.
14Benzene is a common constituent of many hazardous waste streams. It
is a carcinogen and is moderately mobile in the subsurface environment
relative to other hazardous organic chemicals.
-------
5-40
relatively rapid, and thus quick action may be taken to prevent
further contamination. However, for less mobile contaminants and/or
hydrogeologic settings with slow horizontal ground-water velocities
or deep unsaturated zones, detection times may be on the order of
years or tens of years. This allows for the release of relatively
large volumes of hazardous waste before detection, thus presenting
very significant human health risks or requiring extensive corrective
action.1S
The combined protective measures of leak testing and ground-water
monitoring somewhat counteract these effects; for example, a leak too small to
be detected by leak testing may be detected by ground-water monitoring, and a
release of a low-mobility contaminant may be detected quickly if the leak is
large enough to be detected by a semi-annual leak test. However, small leaks
of low-mobility, highly toxic, persistent chemicals (e.g., lindane) in low
velocity ground-water settings may go undetected for long periods of time, and
may eventually present substantial health risks to individuals using ground
water as a source of drinking water.
Finally, although we do not generally consider the results of our analysis
to be appropriate for comparisons between tanks, it is interesting to note jthe
difference in the effectiveness of leak testing with ground-water monitoring
between steel and FRP tanks. As illustrated, leak testing with ground-water
monitoring is much more effective in reducing risks for FRP tanks than for
steel tanks. We believe that this may be due to differences in failure
mechanisms for the two tank materials; steel tanks may develop small leaks due
to corrosion that are undetectable by leak testing, whereas FRP tank failures
are generally large seam cracks or ruptures that would generally be detected
by leak testing. Thus, these results indicate that for FRP tanks, leak
testing with ground-water monitoring is nearly as effective as secondary
containment. However, due to the fact that leak testing with ground-wter
monitoring represents lower-bound estimates (see footnote 13), these results
should be viewed with caution.
5.3.3 Comparison of Risks for Baseline, Secondary Containment and
Partial Containment With Ground-Water Monitoring
Risks for baseline, secondary containment, and partial containment with
ground-water monitoring are compared for on-grade and in-ground treatment
tanks and for storage and accumulation tanks in Exhibits 5-19 and 5-20,
respectively. Our results indicate that partial containment with ground-water
monitoring is effective in reducing risks for hazardous waste tanks, but in
general does not reduce risks as greatly as does the installation of full
secondary containment systems.
isln our analysis, we did not consider corrective action (i.e.,
contaminated soil excavation and/or ground-water treatment) for tanks. Upon
detection, we assumed only that the tank was removed, preventing further
releases.
-------
5-41
EXHIBIT 5-19
COMPARISON OF RISKS FOR BASELINE, SECONDARY CONTAINMENT AND
PARTIAL SECONDARY CONTAINMENT WITH GROUND-WATER MONITORING SCENARIOS
TREATMENT TANKS
Above-ground, ongrade,
carbon steel,
210,000 gal.
In-ground,
concrete,
2,100 gal.
In-ground,
carbon steel,
2,100 gal.
In-ground,
stainless steel,
2,100 gal.
| | Baseline
r"~"l Partial
Containment
{1|J Secondary
&"£ Containment
10 20 30 40 50 60 70 80
Percent Non-Zero Risk Estimates
90 100
-------
5-42
EXHIBIT 5-20
COMPARISON OF RISKS FOR BASELINE, SECONDARY CONTAINMENT AND
PARTIAL SECONDARY CONTAINMENT WITH GROUND-WATER MONITORING SCENARIOS
STORAGE/ACCUMULATION TANKS
Storage, ongrade,
carbon steel,
210,000 gal.
Storage, in-ground,
concrete,
2,100 gal.
Storage, in-ground,
carbon steel,
2,100 gal.
Accumulation, ongrade,
carbon steel,
210,000 gal.
Accumulation, in-ground,
concrete,
2,100 gal.
Accumulation, in-ground,
carbon steel,
2.100 gal.
n
Bast line
{"""I Partial
' ' Containment
FT£1 Secondary
&££il Containment
10 20 30 40 50 60 70
Percent Non-Zero Risk Estimates
80
90
100
-------
5-43
As noted earlier, partial containment prevents releases only from the
above-ground portions of hazardous waste tanks and ancillary equipment.
Ground-water monitoring is therefore necessary to detect leaks from in-ground
or ongrade portions of these tanks. Thus, the assumptions and limitations
with respect to ground-water monitoring that were mentioned in the previous
section also apply to this scenario. Leaks are assumed to be detected when
the detectable concentration of any contaminant is exceeded at the monitoring
well. Time to detection is a function of the liquid volume flux from the
tank, the mobility of the contaminant(s) of concern, and the horizontal
ground-water velocity and depth of the unsaturated zone16 for the
hydrogeologic setting in which the tank is located. Once again, we assume no
corrective action, only that the tank is removed upon detection of a leak and
all release volumes for the remaining years of the 20-year time horizon are
set to zero.17
For some tanks presented in Exhibits 5-19 and 5-20, partial containment
with ground-water monitoring appears to be more effective than secondary
containment in reducing relative risks. These results are counter-intuitive
in that it is difficult to understand how a system in which releases to the
environment must occur in order to facilitate leak detection could be more
effective than a system that prevents releases under all but the most extreme
circumstances. The differences in risk reduction are minor, however, and may
be due in part to our assumptions regarding detection and tank removal, in
which all subsequent releases are set to zero. (The underlying assumption is
that a tank is no longer operated at that site, thus biasing the results
toward partial containment and ground-water monitoring.) Conversely, tanks
with full secondary containment are modeled to operate over the entire 20-year
time horizon and may release contaminants over the entire modeling period.
This assumption may result in greater release volumes for secondary
containment than for partial containment over the 20-year modeling period,
and, therefore, may affect the results of the risk analyses. Also, these
results may be due to a lack of resolution within the HWTF model with regard
to the modeling of certain secondary containment systems. We are conducting
further analyses in this area.
5.3.4 Comparison of Risks for Baseline and Corrosion Protection
Scenarios
In Exhibit 5-21, we compare frequency distributions of risk estimates for
the baseline and corrosion protection scenarios. Corrosion protection was
considered to be applicable to any metal (i.e., carbon steel and stainless
steel) tank system in which the tank or ancillary equipment is in contact with
the soil (e.g., above-ground, ongrade tanks; in-ground tanks; and underground
14We model a single generic unsaturated zone depth of 5 meters; however,
in reality this parameter may vary greatly, and will have an impact on
detection time.
17See footnote 13.
-------
5-44
EXHIBIT 5-21
COMPARISON OF RISKS FOR BASELINE AND
CORROSION PROTECTION SCENARIOS
CARBON STEEL AND STAINLESS STEEL TANKS
Treatment, above-ground,
carbon steel, 60,000 gal.
Treatment, in-ground,
carbon steel, 3,700 gal.
Treatment, in-ground,
stainless steel, 3,700 gal.
Storage, above-ground,
carbon steel, 210,000 gal.
Storage, underground,
carbon steel, 4,000 gal.
Storage, underground,
stainless steel, 4,000 gal.
Storage, in-ground,
carbon steel, 2,100 gal.
Accumulation, above-ground,
carbon steel, 210,000 gal.
Accumulation, underground,
carbon steel, 4,000 gal.
Accumulation, underground.
stainless steel. 4.000 gal.
Accumulation, in-ground,
carbon steel, 2,100 gal.
Small Quantity Generator,
underground, carbon steel,
2,100 gal.
10
20
30 40 50 60 70
Percent Non-Zero Risk Estimates
80
-------
5-45
tanks). Corrosion protection was assumed to be provided by cathodic
protection (i.e., impressed current).
As illustrated in Exhibit 5-21, our results indicate that corrosion
protection is not very effective in reducing the risks associated with
hazardous waste tanks, and overall risk reduction may be considered negligible
for this regulatory scenario. This neglible reduction is, in part, due to the
fact that corrosion protection does not aid in preventing many of the failure
events associated with hazardous waste tanks, such as seam cracks, and
overflows. Also, for this regulatory scenario we have assumed no method for
the detection of releases (e.g., no leak testing or ground-water monitoring)
once they occur. Finally, it should be noted that the effectiveness of
corrosion protection may depend on site-specific soil parameters, and it is
difficult to include such site-specific data in an analysis of this sort.
Thus, assumptions within our model (e.g., our use of a benign rather than an
aggressive soil to represent site soil conditions) may cause us to
under-estimate the effectiveness of corrosion protection.
5.3.5 Summary and Conclusions
Based on the results of the risk analysis, we have ranked the regulatory
scenarios in descending order of risk reduction from the baseline as follows:
Secondary Containment for all tanks;
Partial Containment and Ground-Water Monitoring for
ongrade and in-ground tanks;
Leak Testing and Ground-Water Monitoring for underground
tanks; and
Corrosion Protection for carbon steel and stainless
steels tanks in contact with soil.
It is important to note that direct comparisons cannot be made between all
scenarios (for example, partial containment and leak testing scenarios apply
to completely different tank groups); however, the above ranking is an overall
indication of the effectiveness of the different scenarios in reducing risks
compared to baseline.
Thus, our results indicate that EPA's proposed requirements for full
secondary containment is an effective strategy for preventing environmental
damage caused by leaking hazardous waste tanks. For in-ground and ongrade
tanks, the partial containment with ground-water monitoring is, in most cases,
significantly less effective than full secondary containment. For underground
tanks, leak testing with ground-water monitoring does not appear to be nearly
as effective as full secondary containment. Finally, corrosion protection,
without any detection options, does not appear to be very effective in
reducing the risks presented by carbon steel and stainless steel tanks.
-------
5-46
5.4 COMPARISON OF RISK ESTIMATES FOR HYDROGEOLOGIC SETTINGS
AND WASTE STREAMS
The risk estimates developed in this analysis are not only a function of
tank failure events and release volumes, but are also controlled by the waste
streams stored in a particular tank and the hydrogeologic settings in which
the tank is located. In considering regulatory alternatives for tanks, such
as a risk-based variance to secondary containment requirements, it would be
very important to consider these variables on a site-specific basis.
Our analysis was not designed to compare thes.e factors, but rather was
constructed to obtain comparative risk results for different regulatory
scenarios. However, by comparing the generic flow fields and waste streams
included in our analysis, some insight may be gained into what hydrogeologic
and waste parameters most greatly influence risk. Below, we briefly discuss
the effects of these parameters on our analysis.
5.4.1 Comparison of Ground-Water Flow Fields
TRAM uses nine generic ground-water flow fields to approximate the range
of hydrogeologic settings encountered across the U.S. In order to compare the
relative risks for tanks located in these different environments, we evaluated
risks under the baseline scenario for all nine flow fields for a single tank
technology, an underground carbon steel storage tank, and two waste streams,
F003 (spent non-halogenated solvents, containing benzene, methyl ketone, and
four other solvents) and F001 (spent halogenated solvents, containing carbon
tetrachloride, trichloroethylene, and several other chlorinated compounds).
Differences in risks among ground-water scenarios depend on complex
interactions between contaminant mobility and persistence, ground-water
dilution and dispersion, and the presence of non-aquifer layers.
From similar Liner Location Risk and Cost Analysis Model analyses we have
determined that comparisons using waste streams with highly mobile, highly
persistent constituents (such as F003) show a different distribution of risks
for the nine generic flow fields than waste streams with less mobile, less
persistent constituents (such as F001). Therefore, two waste streams were
chosen to account for differences in contaminant mobility and persistence that
may affect comparisons between different ground-water scenarios. For
comparative purposes, we used an exposure well distance of 60 meters.
Risks for each ground-water scenario were determined for the selected
model tank technology and waste streams F003 and F001, and are summarized in
Exhibits 5-22 and 5-23. As illustrated, risks are distributed from 1 to
10 for ground-water flow fields A through I for waste stream F003, and
-2 -in
from 10 to 10 for waste stream FOOl. The highly mobile, highly
persistent chemicals in waste stream F003 result in the highest risk being
exhibited by flow field A, which has the slowest ground-water velocity of any
of the nine settings (1 m/year). This result is due to two factors: (1) the
slow ground-water velocity in flow field A results in minimal dilution, thus
causing well concentrations of contaminants to be very high; and (2) the slow
-------
ixiiinii 5-??
COMI'AKISON Or RISKS M)R IIYDRUCroi OC 1C SI I I INGS
IRIQIILNCY 1)1 SI XI III! I ION Of RISK ISIIMAIFS (OR UNOLKCKOIINU CARtlON
SUM SIORACI IANK (l|, ()()(» CAI ) All HOW MMDS, WAS 11 SIRIAM I (HH
R
E
I
A
r
i
V
E
F
R
£
Q
U
E
N
C
y
1 .0
1
1
1
0.9
1
1
O.fl
1
1
1
0.7 »
1
1
1
O.b »
1
1
1
0.5 »
1
1
1
0.4 »
1
1
1
0.} »
1
1
1
o.z
. 1
1
1
O.I »
1
| ****,!
1 *****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
******
*****
***** *****
****>.: *****
***#:< ****£
***** **«:**
:»***<:
* * * ^ ;=
****:»
*****
*****
****:>
* * * * <«
*«**£
****tt
*****
**»** «*->>.
*£**£ *>*->^t
***** *****
-16 -I* -12 -10 -rt -6
LUG OF RISK CSIIMATr
-------
fximm 5-?i
COMI'AKISON 01 RISKS I OH HYDROCIOl OC1C SI I I INC.S
JRIQIIINCY OISIRIHIM ION Ol RISK fSIIMAICS FOR IINDI RCROUNI). CARI1ON SUM
SIOKACI IANK ('1.00(1 CAI ) All IIOW MMDS. WASH SIKIAM I Oil I
R
E
L
A
f
1
V
E
F
R
E
Q
U
t
H
C
V
1.0 >
1
1
1
0.9 »
1
1
O.H
1
1
1
0.7 »
1
1
1
0.6
1
1
1
o.s *
1
. 1
1
0.4 »
1
1
1
0.)
1
1
1
0.? «
1
1
1
0.1 »
1
1
1
*****
*****
***««
*****
*****
****«
*****
***A*
******
*****
* « 41 * *
***** *****
***** ****3
***** ***** ****«:
**«*<: ***** ****>.<
****< ***<--> ****:.t
****W ***** ****»
***«« ***** *****
***** **(.'** «4*«r*
***** ****** ****<>
***»« ***** ***** *****
I
*-
do
-^^ -10 -a -6 -ft -2
I
I OC OF RISK rSTIMATF:
-------
5-49
ground-water velocity also results in the longest exposure period (e.g.,
persons served by wells in flow field A have an average exposure period of 200
years, as opposed to flow field B, with an average exposure period of 140
years, and flow field E with an average exposure period of 16 years.)
Therefore two waste streams were chosen to account for differences in
contaminant mobility and persistence that may affect comparisons between
different ground-water scenarios.
Other high-risk flow fields are B, H, and I, while lowest-risk flow fields
are E and F. It should be noted, however, that these results are a function
of both the distance to the exposure point (distance from source to well) and
the choice of the time horizon over which risk is considered. For example, if
a well distance of 1500 meters and a 200-year time horizon were chosen, a flow
field with a moderately ground-water velocity (e.g., flow field C; 100
meters/year) would appear to present greater risks than flow field A, because
the containments being transported in flow field A may not reach the 1500
meter well within the 200-year time horizon.
Results for waste stream F001 are somewhat different, in that flow field A
exhibits the lowest risk. This result is due to the lower mobility of F001
constituents, which results in only a small portion of the contaminant mass
that is released by the tank reaching the 60 meter well by year 400, the final
year we are considering in our analysis. Due to slightly greater ground-water
velocity, contaminants do reach the 60 meter well in flow fields B, G and I,
which are the highest-risk flow fields for waste stream FOOl. Once again,
flow fields E and F present the lowest risks; this result is due to the rapid
ground-water velocity in flow field E (resulting in more dilution and
dispersion and a shorter period of exposure) and the presence of a non-aquifer
overlying the affected aquifer in flow field F. (The presence of the
non-aquifer layer greatly inhibits the movement of contaminants in this flow
field.)
Thus, this analysis indicates that, for the 60 meter well distance,
hydrogeologic settings with slower ground-water velocities present higher
risks1* in most cases than hydrogeologic settings with more rapid
velocities. However, it is important to note that; (1) well distance greatly
influences the results of this analysis; and (2) the analysis does not account
for the amount of ground-water contaminated in each flow field, or the
proximity of the hazardous waste tanks to ground-water users. For example,
even highly mobile contaminants do not reach wells at 1,500 meters in flow
field A, and therefore one would expect risks for this well distance to be
dominated by flow fields, where velocities are sufficient to allow for some
exposure. Also, while high ground-water velocity flow fields present the
lowest risk, in our analysis, contamination in these aquifers will also result
11 Our analysis uses as the measure of risk, the 400-year average
individual lifetime risk. A different risk measure (e.g., 200-year average
individual lifetime risk) might produce somewhat different results.
-------
5-50
in the largest volume of contaminated ground water, which translates into
greater probability of eventual exposure, more resource loss, and much greater
clean-up costs.
Therefore, we conclude from our analysis that the effect of hydrogeologic
setting on the risks presented by hazardous waste tanks is dependent not only
on properties of .the underlying aquifer but also on waste stream
characteristics and local ground-water use considerations. For a risk-based
variance approach, the interactions between these parameters would need to be
considered on a site-specific basis.
5.4.2 Comparison of Waste Streams
Ideally, in comparing risks for different waste streams, a single
comparison would be made across all waste streams for a single tank technology
and a hydrogeologic setting. However, because different tank categories
handle different groups of waste streams (e.g., SQG wastes differ
significantly from treatment tank wastes), it is not possible to use our
results to make a broad comparison of all the different waste streams included
in the analysis. Also, comparisons among different waste streams within a
particular tank category were complicated by differences in release volumes.
for the tank technologies within that category. Therefore, in order to make a
direct comparison of waste streams within a particular category (i.e.,
identify high-risk and low-risk waste streams), we selected one tank'
technology from that category and compared risks for all wastes handled by
that technology. By choosing tank technologies that handle a large number of
waste streams, we were able to make direct comparisons among essentially all
waste streams within each of the four major tank categories. Hydrogeologic
setting was held constant by using risk results for flow field C in all cases.
Below, we present these comparisons for each tank category.
Treatment Tank Waste Streams. For a comparison among the different
organic wastes treated in tanks, we selected an above-ground, on cradle,
carbon steel, distillation tank system. Based on our initial waste stream
analysis, it was determined that this tank technology handled six waste
streams: F001, spent halogenated solvents; D001, ignitable wastes, N.O.S.;
X907, chlorinated pesticide still bottoms; F003, spent non-halogenated
solvents; U080, dichloromethane; and U226, 1,1,1-trichloroethane. Risks were
calculated for each of these waste streams, and are summarized in Exhibit 5-24.
As illustrated, risk estimates for the various waste streams treated by
-4
this tank technology range from 1 to 10 . The highest risk estimates are
exhibited by waste stream D001, ignitable wastes, N.O.S. Low-risk waste
streams included U080, dichloromethane, and U226, 1,1,1-trichloroethane. Both
of these contaminants are assumed to degrade fairly rapidly in
ground-water-soil environments.
Another treatment tank technology, an above-ground, on-grade, carbon steel
oxidation/reduction/precipitation (ORP) tank was selected in order to compare
aqueous waste streams treated in tanks (heavy metals, reactive wastes, etc.).
Ten waste streams were included in this comparison. The frequency distribution
of risk estimates for the ten waste streams handled in this ORP tank is
summarized in Exhibit 5-25. Due to the very large release volumes associated
-------
rxniBi i 5-?'i
COMPAHISON Ol RISKS I OK WASH SlfUAMS ABOVE -CIUMINI),
ON CKAIJI I . CAKBON SIML UISIIILAIION IANK (?,»(!(! C.AI )
HOWIIfin C. AIL WAS 1C SiniAMS
1.0
0.9
O.B
R
E
I
A 0.7
T
1
V
E 0.6
F
R
E 0.5
Q
U
E
N 0.4
C
V
0.3
0.2
0.1
****=:
**>*;
*****
*****
*****
***«>
*****
****
*** ><
**<-*«
*****
*****
******
*****
**«*«
*<***»
«****
*****
**«* >
*****
****v *****
***** *****
***** *****
****<
«*<>«::
*****
*****
*****
*****
*?***
*****
*****
*'<***
to
I
-16
-12
-10 - -6
LOG OF RISK ESTIMATE
-6
-*»
-------
fxinnti
COMI'AKISON Of RISKS r OR WASH S IMAMS
ON I.KAW . CAKBON sun OHP IANK u.u.n.m CAI >
MOW Mill) C. All WASIE SIIUAMS
R
E
t
A
T
1
V
E
F
R
E
Q
U
E
N
C
t
1.0 »
1
1
1
0.9 »
1
1
1
O.fl *
1
1
1
0.7 »
1
1
1
0.6 *
1
1
1
0 i *
I
1
1
0.<« *
1
1
1
0.3 »
1
1
1
0.2 *
1
!
0.1 »
1
I
1
*****
*****
*****
*****
*****
*****
*****
***** >>«<»
***** *****
***** «44*4
***** *»»*>
***** ft****
***** *****
***** ***** «*0«.)i
***** ***** *****
***** ***** *=>»**
***** ****> *«**=>
***** ***** ***** **«*«
***** ***** **«»< ***** **#««
I
Ul
-16
-12
-10 -8
LOG or RISK rsTimrc
-6
-------
5-53
with the selected technology, dose-response thresholds for essentially all
waste stream constituents were exceeded by a wide margin, and risks are very
high for all wastes. However, particularly high risks are presented by waste
streams P115, thallium sulfate (a low-mobility but very toxic metal), D004
(arsenic, a highly mobile and potent carcinogen), and D003 (reactive wastes,
N.O.S.). Low-risk waste streams included K062, spent pickle liquor, and F007,
electroplating wastes. Both of these low-risk waste streams are relatively
dilute, with concentrations of lead and chromium in the low parts per million
range.
Storage and Accumulation Tank Waste Streams. For a comparison of the
waste streams handled in storage and accumulation tanks, we analyzed
above-ground, on grade carbon steel storage tanks. Because release volumes
are identical for storage and accumulation tanks, and the types of waste
streams handled are very similar, we used this comparison to represent waste
streams for both tank categories.
Twenty-one waste streams were selected for this technology. Once again,
due to high release volumes, risks are relatively high for all waste streams
(Exhibit 5-26). Waste streams with especially high risks included U188,
phenol (a highly mobile and very common organic chemical); P115, thallium
sulfate; D001, ignitable wastes, N.O.S.; and D003, reactive wastes, N.O.S.
Once again, low risk waste streams included readily degradable chlorinated
solvents such as U080, dichloromethane, and dilute aqueous wastes such as
K048, petroleum refining wastes, and K062, spent pickle liquor.
SQG Waste Streams. Our final waste stream comparison was made among
the nine wastes handled in SQG underground, carbon steel tanks. Waste streams
for this technology included pesticide wastes, spent solvents, wood preserving
wastes, strong acid wastes, and printing and photographic wastes.
The frequency distribution of risk estimates for these waste streams is
-2
summarized in Exhibit 5-27. As illustrated, risks range from 10 to
-8
10 , with the highest risks exhibited by pesticide wastes (particularly
aldicarb, which is much more mobile than the other selected pesticide,
lindane) and wood preserving wastewaters, which contain relatively immobile
but very toxic constituents. Low-risk waste streams include K086, waste ink
sludges and solvents, and strong acid/alkali wastes.
One caution that should be observed in evaluating the results of the waste
stream comparison is our lack of data on waste stream characteristics for
waste streams D001, D002, and D003 (unspecified ignitable, corrosive, and
reactive wastes, respectively). As outlined in Chapter 4, we determined the
characteristics for these wastes using only one data source, which was based
on information on incinerated wastes. Because incinerated wastes must
generally have a high content of flammable (i.e., organic) constituents, the
wastes used in the analysis may have artificially high concentrations of some
of the constituents that are driving risk (e.g., methanol, xylene, acetone,
etc.). The use of these characteristics for D wastes in our analysis probably
over-estimates risks for these waste streams.
-------
t.o »
0.9
IXIMOir 5-26
COMPARISON 0» RISKS fOK WASfC SIRfAMS AOOVl-CKOUNI)
ON CKAUt. CAKIION SI I [I SIORACL IANK
(Pin.nun OAI ) now nrio c. ALI WASH SIRIAMS
R
E
L
A
T
I
V
E
F
R
E
0
U
E
N
C
V
0.1
o.r
0.6
0.3
0.2
O.I
t_n
l
t_n
»«,**<<
*«..i«;>*,)
*****
tttftf**:
**«*>
* *c * * if
**«*»
<»***
**»>*
«***«
-16
-12
-10 -8
LOG OF RISK ESTIMATE
-6
-2
-------
^-
«,-
9-
31VW11S3 XSIU JO 001
9- Ot-
i\-
VI-
91-
«**«*
*****
*«»«*
:.-.«««*
*****
i»**...*
*****
*****
*****
***<>*
***«*
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
4****
1
1
1
1
1
1
1
1
1
«
1
1
I
1
1
1
1
1
«
1
1
1
1
1
1
1
1
1
1
1
1
«
fO
«*0
0*1
N
3
n
o
3
3
A
I
1
1
3
SHVIHIS IISVM I IV 'j (Illll HO I I
(|
-------
5-56
5.4.3 Summary and Conclusions
Based on this analysis we have drawn the following conclusions:
Hydrogeologic setting affects risk such that, for a given
technology and waste stream, risk is generally greater for
flow fields with slower ground-water velocity. This
conclusion is somewhat counter-intuitive, and is dependent
on complex interactions between contaminant mobility and
persistence, ground-water dilution and dispersion, and the
presence of non-aquifer layers. However., it can be said
that, for most waste streams, flow fields with higher
velocities (e.g., scenarios B, C, I) present the lower
risks, while flow fields such as F (non-aquifer overlying
the aquifer) and E (very rapid ground-water velocity),
present the lowest risk. However, in developing risk-based
standards, the ground-water scenario must be considered in
conjunction with information on potentially exposed
populations and waste stream characteristics.
Many waste streams found in hazardous waste tanks present
relatively high risks, due to their highly concentrated
nature. Among the highest-risk wastes are spent solvents,
certain heavy metal wastes (particularly those containing
thallium and arsenic), phenol, and reactive wastes
(acrylonitile/cyanide, 2,4-dinitrotoluene, etc.). More
dilute wastes such as spent pickle liquor and petroleum
refining wastes generally present lower risks.
-------
APPENDIX
FAILURE FREQUENCY DATA AND
REPRESENTATIVE RELEASE PROFILES
This appendix contains the summary output from the Hazardous Waste Tank
Failure (HWTF) Model. The appendix consists of six exhibits. Exhibits A-l
through A-3 present the failure frequency data for treatment tanks, storage/
accumulation tanks, and SQG tanks, respectively. Exhibits A-4 through A-6
present the representative release profiles for the three tank categories. An
explanation of the data presented in the exhibits is provided in Section 5.2.1
of the report.
With the exception of the risk estimates for the partial secondary
containment with groundwater monitoring regulatory scenario and the leak
testing with groundwater monitoring regulatory scenario, the risk estimates
presented in Section 5.3 of the report are based on the representative release
profiles in Exhibits A-4, A-5, and A-6. In the case of the two regulatory
scenarios with groundwater monitoring, the representative release profiles
used to generate the risk estimates are modified versions of those presented
in Exhibits A-4, A-5, and A-6. The modification entailed setting release
volumes to zero once releases were detected via groundwater monitoring and
abated. This modification was necessary because groundwater monitoring was
not incorporated in the HWTF model.
-------
EXHIBIT A-l
FAILURE FREQUENCIES FOR TREATMENT TANKS
Treatment Tank Technology 1
Above-ground, on cradles, carbon steel, closed, 2,300 gallons
Baseline, Toluene
1
1
1 YEAR
I or
I FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 1 IWELP
HOSE | LOOSE | OVER (CASKET
RUPTRI HOSEJ FLOW | INSTAL
0.0
O.O
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
70.6
29.1
0.0
0.0
0.0
TOTAL I 0.0 | 0.0 (100.0
0.1
0.1
1.3
0.1
0.0
2.5
PIPE | TANK
INSIALIINSTAL
DEFIC) OEFIC
0.0
O.O
0.0
0.1
0.0
0.1
0.0
O.O
0.0
0.1
0.1
WELD/I
CASKTJ PIPE
FAIL | RUP1R
8.2
52.1
13.9
5.6
2.6
0.8 | 82.7
0.1
1.3
6.1
5.2
1.7
1
1
TANK | PIPE
RUPTRI CORR
1.3
2.2
6.9
5.3
3.9
20.7 | 19.6
1.3
11.8
28.1
26.1
21.2
TANK
CORR
0.1
3.5
16.5
16.9
20.8
,
CATAS-I
STROP!)
RLS
0.1
0.0
1.3
2.2
0.1 |
91.8 I 58.1 | 1.3 !
Treatment Tank Technology 1
Above-ground, on cradles, carbon steel, closed, 2,300 gallons
Secondary Containment, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
1 IWELO
LOOSE) OVER | CASKET
HOSEI FLOW | INSTAL
0.0
0.0
0.0
0.0
0.0
75.6
21.1
0.0
0.0
0.0
1.2
0.0
0.8
0.1
0.0
PIPE
INSIAL
DEFIC
0.8
0.0
O.O
0.0
0.0
TANK
i INSTAL
OEFIC
1.2
O.O
0.0
0.1
0.1
WELO/I
CASMI PIPE
FAIL | RUPTR
0.0
10.1
29.2
11.2
3.2
1.2
3.6
1.1
1.1
1.8
TANK
RUPTR
0.8
1.1
1.1
5.6
2.0
PIPE
CORR
0.8
5.2
18.0
27.6
29.6
ICATAS-
TANK JSTUOPII
CORR j RLS
0.0
1.2
9.6
15.6
21.6
0.0
0.1
2.0 1
1.6 I
1.2 1
I
TOTAL | 0.0 | 0.0 1100.0 | 2.1 I 0.8 I 2.0 I 81.0 I 18.1 I 17.2 I 81.2 I 18.0 I 5.2 I
Treatment Tank Technology 2
Above-ground, on cradles, carbon steel, open, 2,300 gallons
Baseline, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
LOOSE
HOSE
0.0
0.0
0.0
0.0
0.0
OVER
FLOW
9.5
19.1
25.3
15.9
6.5
WELD
CASKET
INSTAL
0.0
1.9
1.0
0.5
0.9
PIPE
INSTAL
Of FIC
0.0
0.0
0.0
1 .0
0.0
TANK
INSTAL
OEFIC
0.0
0.0
O.O
1.0
0.5
WELD/
CASKT
FAIL
10.9
61.7
7.8
5.2
1.9
1
l PIPE
RUPTR
2.3
3.3
8.7
7.3
5.0
1
TANK | PIPE
RUPTRI CORR
1.1 1 1.1
1.6 1 10.0
5.0 I 15.8
6.6 1 ,10.1
6.0 | 1.1
ICAIAS-
TANK ISTROPH
CORR | RLS
0.5
10.8
17.1
15.0
21.0
0.0
0.0
1.5
1.1
2.1
TOTAL | 0.0 I 0.0 I /6.6 I 1.3 I 1.0 I 1.5 I 90.5 I 26.6 I 23.8 I 38.7 I 67.1 I 5.3
-------
EXHIBIT A-1 (continued)
FAILURE FREQUENCIES FOR TREATMENT TANKS
Treatment Tank Technology 2
Above-ground, on cradles, carbon steel, open, 2,300 gallons
Secondary Containment, Aqueous
YEAR
or
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
I IWELO | PIPE
LOOSE) OVER ICASKETI INSIAL
HOSEl FLOW | INSTALI OEFIC
0.0 I 5.6
0.0 1 25.6
0.0 I 21.2
0.0 I 15.2
0.0 1 8.8
0.1
0.0
0.1
2.0
2.0
TOTAL | 0.0 I 0.0 I 76.1 I 1.8
1.2
0.0
0.0
0.1
0.0
TANK
INSTAL
DEFIC
2.0
0.0
0.0
0.0
0.0
1.6 1 2.0
WELD/
GASKT
FAIL
0.0
15.2
27.2
6.8
6.1
85.6
1
PIPE I TANK
RUPTRl RUPTR
2.0
1.1
8.1
1.1
6.0
0.8
1.0
6.8
9.6
3.2
PIPE
CORR
2.0
5.2
11.0
16.1
7.2
25.2 I 21.1 | 11.8
TANK
CORR
0.1
1.2
10.1
16.8
17.6
16.1
ICATAS-
[STROPH
RLS
0.0
0.1
2.0
1.6
1.2
5.2
Treatment Tank Technology 1
Above-ground, ongrade, carbon steel, 60,000 gallons
Ba seIi ne. Aqueou s
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1
HOSE I LOOSE
RUPTRl HOSE
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
TOTAL | 0.0 1 0.0
IWELD | PIPE
OVER ICASKETI INSTAL
FLOW | INSTALI OEFIC
7.8
20.1
22.6
11.8
7.6
0.0
2.0
0.0
0.1
1.6
0.0
0.0
0.8
0.0
0.0
TANK | WELD/
INSTALI CASKT
DEFIC! FAIL
3.3
0.0
0.0
0.0
0.0
35.2
10.2
11.3
1.0
3.2
1
PIPE I TANK
RUPTRl RUPIR
1.2
5.7
6.5
5.2
5.6
11.1
1.1
7.3
1.8
3.6
PIPE
CORR
1.2
13.2
15.7
1.6
2.8
73.2 | 1.0 | 0.8 I 3.3 I 93.9 1 21.2 I 30.9 1 31.5
TANK
CORR
9.8
10.2
8.5
10.2
9.8
18.5
CA1AS-
SIROPH
RLS
0.0
1.6
0.1
1.6
2.8
6.1
Treatment Tank Technology 1
Above-ground, ongrade, carbon steel, 60,000 gallons
Corrosion Protection, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
LOOSE
HOSE
0.0
0.0
0.0
0.0
0.0
OVER
FLOW
7.8
20.1
22.6
11.8
7.6
WELD
CASKET
INSTAL
0.8
0.8
0.1
1.6
0.1
PIPE
INSTAL
OEFIC
0.0
0.0
0.0
0.0
0.0
IANK
INSTAL
DLFIC
3.3
0.0
0.0
0.0
0.0
WELD/
CASKT
FAIL
31.0
11.8
9.7
1.8
2.1
1
PIPE I TANK
RUPTRl RUPTR
1.2 1 16.0
5.7 1 5.7
6.5 1 3.2
5.2 1 5.2
5.6 1 3.2
PIPE
CORR
0.0
0.0
2.8
1.6
3.2
ICAIAS-
TANK ISTROPH
CORR | RLS
8.2
0.1
0.0
0.0
0.8
0.0
1.6
0.1
1.6
2.8
TOTAL I 0.0 I 0.0 I 73.2 | 1.0 I 0.0 | 3.3 I 92.7 I 21.2 I 33.3 I 7.6 I 9.I I 6.1
-------
EXHIBIT A-1 (continued)
FAILURE FREQUENCIES FOR TREATMENT TANKS
Treatment Tank Technology H
Above-ground, ongrade. carbon steel, 60,000 gallons
Partial Containment with Ground-water Monitoring, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE
1 1 IWELO
HOSE | LOOSE | OVER (CASKET
RUPTR) HOSE) FLOW | INSTAL
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
TOTAL | 0.0 | 0.0
7.8
20.4
22.6
14.8
7.6
73.2
0.0
0.8
1.2
0.8
1.2
CATEGORY
PIPE | TANK | WELD/
INSTALI INSTALI GASKT
DEFIC) DEFICI FAIL
0.0
0.4
0.0
0.0
0.0
3.3
0.0
0.0
0.0
0.0
4.0 1 0.4 1 3.3
26.6
45.2
12.7
4.8
2.4
91.7
PIPE
RUPTR
1.2
5.7
6.5
5.2
5.6
TANK
RUPTR
15.6
6.9
4.4
4.0
4.0
24.2 I 34.9
PIPE
CORR
4.5
13.9
15.4
3.6
3.6
TANK
CORR
10.2
6.2
8.6
4.8
13.5
CATAS-
STROPH
RLS
0.0
1.6
0.4
1.6
2.8
41.0 I 43.3 I 6.4
Treatment Tank Technology 4
Above-ground, ongrade, carbon steel, 60,000 gallons
Secondary Containment, Aqueous
YEAR
OF
FAIIURE
1
2- 5
6-10
11-15
16-20
FAILURE CAIEGORY
1 I IWELD | PIPE | TANK
HOSE | LOOSE) OVER (CASKET) INSTALI INSTAL
RUPTRI HOSEI FLOW |INSfAL| OEFICl OEFIC
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
5.6
25.6
21.2
15.2
8.8
0.4
0.0
0.4
2.0
0.8
TOTAL | 0.0 | 0.0 I 76.4 1 3.6
1.2
0.0
0.0
0.4
0.0
1.6
2.0
0.0
0.0
0.4
0.0
2.4
WELD/
CASKT
1 FAIL
0.0
45.2
26.8
7.6
6.0
1 1
PIPE I TANK
RUP1RI RUPTR
2.0
4.4
8.4
4.4
6.0
85.6 I 25.2
0.8
4.0
6.8
8.8
4.0
PIPE
CORR
2.0
5.2
14.0
16.4
7.2
1 CA 1 AS*
TANK ISTROPH
CORR 1 RLS
0.4
1.2
7.6
10.0
5.2
0.0
0.4
2.0
1.6
1.2
24.4 | 44.8 | 24.4 1 5.2
Treatment Tank Technology 5
In-ground, concrete, 3,700 gallons
Baseline, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
LOOSE
HOSE
0.0
0.0
0.0
0.0
0.0
OVER
FLOW
7.0
25.9
22.6
15.3
6.1
WELD I PIPE | TANK
CASKET) INS1AL) INSfAL
INSTALI OEFICl OEFIC
0.0 1 0.0
0.4 I 0.0
0.0 I 0.0
1.6 1 0.0
0.8 1 0.0
2. 1
0.0
0.0
0.0
0.0
WELD/
GASKT
FAIL
28.8
46.9
10.3
4.0
4.4
PIPE
RUPTR
2.9
6.9
3.2
4.0
4.4
TANK
RUPTR
0.0
0.0
0.8
12.7
25.6
PIPE
CORR
3.7
18.5
13.9
2.8
'5.2
ICATAS-
TANK ISTROPH
CORR | RLS
0.0
0.0
0.0
0.0
0.0
0.4
1.2
1.2
1.6
1.2
TOTAL I 0.0 I 0.0 I 76.9 I 2.8 I 0.0 | 2.1 | 94.4 | 21.4 I 39.1 I 44.1 I 0.0 | 5.6
-------
EXHIBIT A-1 (continued)
FAILURE FREQUENCIES FOR TREATMENT TANKS
Treatment Tank Technology 5
In-ground, concrete, 3,700 gallons
Partial Containment with Ground-water Monitoring, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1
HOSE | LOOSE
RUPTRI HOSE
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
OVER
FLOW
7.0
25.9
22.6
15.3
6.1
IWELO | PIPE I TANK
GASKET | INSTALI INSTAL
INSTALI OEFICI DEFIC
0.1
0.8
0.1
0.1
1.2
0.0
0.0
0.0
0.0
0.1
2. 1
0.0
0.0
0.0
0.0
WELD/I
GASKTI PIPE
FAIL | RUPTR
31.2
12.0
11.8
1.0
0.8
2.9
6.9
3.2
1.0
1.1
TANK
RUP1R
0.0
0.0
2.1
10.6
29.6
PIPE
CORR
1.9
17.3
11.0
2.8
2.8
ICAIAS-
TANK ISTROPM
CORR | RLS
0.0
0.0
0.0
0.0
0.0
0.1
1.2
1.2
1.6
1.2
TOTAL I 0.0 I 0.0 I 76.9 I 3.2 I 0.1 I 2.1 I 92.8 I 21.1 I 12.6 I 38.8 I 0.0 I 5.6
Ireatment Tank Technology 5
In-ground, concrete, 3,TOO gal Ions
Secondary Containment, Aqueous
I
I
FAILURE CATEGORY
YEAR | | I
OF | HOSE | LOOSE | OVER
FAILURE) RUPTRI HOSE I FLOW
1
2- 5
6-10
11-15
16-20
0.0 1 0.0
0.0 I 0.0
0.0 1 0.0
0.0 I 0.0
0.0 I 0.0
5.6
25.6
21.2
15.2
8.8
WELD
CASKET
INSTAL
1.2
1.6
0.0
1.6
0.8
TOTAL I 0.0 1 0.0 I 76.1 I 5.2
PIPE
INSTAL
DEFIC
0.8
0.0
0.0
0.0
0.0
0.8
TANK
INSTAL
DEFIC
0.1
0.0
0.0
0.0
0.0
0.1
WELO/I
CASKTI PIPE
FAIL I RUPTR
0.0
11.1
31.2
7.6
1.8
2.0
1.1
8.1
1.1
6.0
1
TANK | PIPE
RUPTRI CORR
0.0 1 2.8
0.0 I 7.6
0.0 I 12.8
0.8 1 17.6
3.2 I 1.0
TANK
CORR
0.0
0.0
0.0
0.0
0.0
88.0 I 25.2 | 1.0 | 11.8 I 0.0
CAIAS-
STROPM
RLS
0.0
0.1
2.0
1.6
1.2
5.2
Treatment Tank Technology 6
In-ground, carbon steel, 3,700 gallons
Baseline, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
LOOSE
HOSE
0.0
0.0
0.0
0.0
0.0
0.0 1 0.0
IWELO | PIPE
OVER (GASKET | INSTAL
FLOW I INSTALI DEFIC
7.1
26.1
21.9
15.3
6.3
76.7
0.0 I 0.0
1.2 I 0.0
1.2 | 0.0
1.6 1 0.0
O.H 1 0.0
TANK | WELD/
INSTALI CASKT
DEFIC! FAIL
0.8
0.0
0.0
0.0
0.0
1.14 | 0.0 | 0.8
32.1
11.9
10.3
5.8
2.8
93.2
PIPE
RUPTR
2.9
7.1
3.2
1.1
1. 1
TANK
RUPTR
17.8
3.7
6.6
1.0
1. 1
21.1 I 36.2
PIPE
CORR
0.1
18.1
13.7
1.2
2.0
35.7
TANK
CORR
53.1
17.0
15.3
9.2
3.3
97.9
CATAS-
STROPH
RLS
0.1
1.2
1.2
1.6
1.2
5.6
-------
._.. .. > (ouni imieo)
FAILURE FREQUENCIES FOR TREATMENT TANKS
Treatment Tank Technology 6
In-ground, carbon steel, 3,700 gallons
Corrosion Protection, Aqueous
I
FAILURE CAFECORY
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
1 1 IWELO
HOSE | LOOSE) OVER (CASKET
RUPTR) HOSEI FLOW | INSTAL
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
7.1
26.1
21.9
15.3
6.3
0.0
1.6
0.8
0.8
O.O
PIPE I TANK
INSTALI INSTAL
DEFICl DEFIC
0.0
0.0
0.0
0.0
0.0
0.8
0.0
0.0
0.0
0.0
WELD/I
CASKTJ PIPE
FAIL | RUPTR
33.6
11.1
8.7
2.8
2.8
2.9
7.1
3.2
1.1
1.1
1
TANK | PIPE
RUPTR I CORR
13.3
5.0
5.8
6.2
3.6
0.1
0.0
O.O
2.1
2.0
TANK
CORR
19.8
2.0
1.6
2.8
2.8
CATAS-
STROPH
RLS
0.1
1.2
1.2
1.6
1.2
TOTAL | 0.0 I 0.0 I 76.7 I 3.2 I 0.0 I 0.8 I 92.3 I 21.1 I 33.9 I 1.8 I 59.0 I 5.6
Treatment Tank Technology 6
In-ground, carbon steel, 3.7OO gallons
Partial Containment with Ground-water Monitoring. Aqueous
YEAR
or
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CAIFCORY
1 1 IWELO
HOSE | LOOSE I OVER (CASKET
RUPTR | HOSEI FLOW | INSTAL
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
6.7
25.9
22.2
15.5
6.3
0.0
0.8
0.1
0.8
0.1
PIPE I TANK | WELO/
INSTALJ INSIALI CASKT
DEFICl OEFICI FAIL
O.O
0.1
0.0
0.0
0.0
0.8
0.0
0.0
0.0
0.0
36.1
38.5
9.6
5.1
2.9
1
PIPE | TANK
RUPTRJ RUPTR
2.9
7.2
3.3
1.1
1.2
13.8
2.9
7.6
1.7
7.1
PIPE
CORR
2.5
18.0
12.2
2.1
3.7
TANK
CORR
17.3
21.7
18.1
6.3
2.8
CAtAS-
STKOPII
RLS
0.1
1.2
1.2
1.6
1.2
TOTAL I 0.0 I 0.0 I 76.6 I 2.1 I 0.1 I O.8 I 92.8 I 21.7 I 36.1 I 38.8 I 99.5 I 5.6
Treatment Tank Technology 6
In-ground, carbon steel. 3,700 gallons
Secondary Containment, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
I 0.0
LOOSE
HOSE
0.0
0.0
0.0
0.0
0.0
I 0.0
OVER
FLOW
5.6
25.6
21.2
15.2
8.8
76.1
WELD
CASKET
INSTAL
0.1
0.0
0.1
2.1
0.8
1.0
1
PIPE
INSTAL
DEFIC
1.2
0.0
0.0
0.1
0.0
| 1.6
AILURL
TANK
INSTAL
DEFIC
11.2
0.0
0.0
0.0
0.0
11.2
CA 1 tCW
WELD/
CASKT
FAIL
0.0
15.2
27.2
7.2
7.6
87.2
IY
PIPE
RUPTR
2.0
1.1
8.1
1.1
6.O
25.2
1
TANK | PIPE
RUPTR j CORR
2.1 I 2.0
7.2 | 8.1
11.6 I 12.1
5.6 I 19.2
9.2 I , 7.2
36.0 I 19.2
TANK
CORR
0.0
8.1
29.2
11.8
11.8
67.2
CATAS-
SFROPH
RLS
0.0
0.1
2.0
1.6
1.2
5.2
-------
EXHIBIT A-1 (continued)
FAILURE FREQUENCIES FOR TREATMENT TANKS
Treatment Tank Technology 7
In-ground, stainless steel, 3,700 gallons
Baseline, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
| IWELU | PIPE | TANK | WELD/
LOOSE I OVER IGASKETI INSTALI INSTALl GASKT
IIOSEI FLOW | INSTALI OEFIC) DEFIC) FAIL
0.0
0.0
0.0
0.0
0.0
7.0
26.1
22.2
15.3
6.3
0.0 | 0.0
0.1 I 0.0
0.0 I 0.0
1.2 | 0.0
0.1 I 0.0
0.8
0.0
0.0
0.0
0.0
33.9
12. 1
10.1
5.3
2.9
1
PIPE | TANK
RUPTRI RUPTR
2.9 1 16.1
7.1 1 1.6
3.2 | 5.8
1.1 1 3.2
1.1 1 3.6
PIPE
CORR
2.5
6.1
7.5
1.9
7.1
TANK
CORR
2.5
1.6
7.0
10.3
13.6
CA1AS-
STROPH
RLS
0.1
1.2
1 .2
1.6
1.2
TOTAL | 0.0 I 0.0 I 76.9 I 2.0 I 0.0 I 0.8 I 91.6 I 21.1 | 33.3 I 28.1 I 38.0 I 5.6
Treatment rank Technology 7
In-ground, stainless steel. 3,700 gallons
Corrosion Protection, Aqueous
1
YEAR
or
FAILURE
1
2- 5
6-10
11-15
16-20
HOSE
RUPTR
ooooo
ooooo
LOOSE
HOSE
0.0
0.0
0.0
0.0
0.0
OVER
FLOW
7.0
26.1
22.2
15.3
6.3
WELD
GASKET
INSTAL
0.0
1.2
0.1
0.1
0.1
1
PIPE
INSTAL
OEFIC
0.0
0.0
0.0
0.0
0.0
"AILURE
TANK
INSIAL
OFF 1C
0.8
0.0
0.0
0.0
0.0
CAIEGOF
WELD/
CASKT
FAIL
31.0
15.9
12.1
2.8
1.2
IY
PIPE
RUPTR
2.9
7.1
3.2
1.1
1.1
1
TANK
RUPTRI
11.5
1.5
8.7
5.1
5.3
PIPE
CORR
0.0
0.8
1.6
1.2
1.2
TANK
CORR
1.2
0.0
0.0
0.1
0.1
CA1AS-
S1ROPM
RLS
0.1
1.2
1.2
1.6
1.2
1
TOTAL I 0.0 I 0.0 I 76.9 I 2.1 I 0.0 I 0.8 I 93.3 I 21.1 I 38.1 I 1.8 I 2.0 I 5.6 I
Treatment Tank Technology 7
In-ground, stainless steel, 3,700 gallons
Partial Containment with Ground-water Monitoring, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL
FAILUKL CAIIUUKY
1
HOSE | LOOSE
RUPTRI HOSE
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0 I 0.0
OVER
FLOW
7.0
26.1
22.2
15.3
6.3
76.9
WELD | PIPE
GASKET) INSIAL
INSTALI OEFIC
0.1 1 0.0
1.2 I 0.0
0.1 1 0.0
0.8 1 0.0
0.0 I 0.0
2.8 | 0.0
TANK
INSTAL
OEFIC
0.8
0.0
0.0
0.0
0.0
WELD/
GASKT
FAIL
38.0
37.1
11.2
2.1
3.7
0.8 I 92.1
PIPE
RUPTR
2.9
7.1
3.2
1.1
1.1
1
TANK 1 PIPE
RUPTRI CORR
13.2
5.1
6.2
5.2
7.1
21.1 1 37.1
0.8
1.1
1.0
6.2
' 3.6
TANK
CORR
2.9
2.0
9.5
11.6
9.1
CATAS-
STROPII
RLS
0.1
1.2
1.2
1.6
1.2
18.7 35.1 | 5.6
-------
EXHIBIT A-1 (continued)
t
FAILURE FREQUENCIES FOR TREATMENT TANKS
Treatment Tank Technology 7
In-ground, stainless steel, 3,700 gallons
Secondary Containment, Aqueous
1
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 I (WELD | PIPE
HOSE | LOOSE! OVER JCASKET | INSTAL
RUPTR) HOSEI FLOW |INS1AL| DEFIC
0.0
O.O
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
5.6
25.6
21.2
15.2
8.8
O.ll
1.2
0.8
1.6
1.6
1.2
0.0
0.0
0.0
O.It
TOTAL I 0.0 I 0.0 1 76.1 I 5.6 I 1.6
TANK
INSTAL
DEFIC
11.2
O.O
0.
-------
EXHIBIT A-2
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 8 and Accumulation Tank Technology 17
Above-ground, on cradles, carbon steel, 5,500 gallons
Baseline, Aqueous and loluene
1
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
2.1
7.2
6.8
10.1
7.6
1 IWELD
LOOSE) OVER | CASKET
HOSEJ FLOW IINSTAL
1.6
13.9
12.7
12.3
9.7
2.1
12.6
15.2
9.7
8.9
0.0
0.0
0.0
0.1
0.1
PIPE
INSTAL
OEFie
0.0
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
WELD/I
GASKTJ PIPE
FAIL | RUPTR
16.9
27.0
11.1
8.1
6.0
0.0
2.1
3.2
2.1
2.1
TANK
RUPTR
0.1
1.2
3.3
1.6
1.2
PIPE
CORR
0.8
2.1
2.1
2.9
0.8
TANK
CORR
0.0
9.3
19.1
13.1
11.9
CATAS-
STROPH
RLS
0.1
1.2
1.6
0.0
0.8
I TOTAL | 33.8 I 53.2 I 18.5 I 0.8 I 0.0 I 0.0 I 71.1 I 10.1 I 7.7 I 9.0 I 56.7 I 1.0
Storage Tank Technology 8 and Accumulation Tank Technology 17
Above-ground, on cradles, carbon steel, 5,500 gallons
Secondary Containment. Aqueous and Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CAIEGORY
1 1
HOSE | LOOSE | OVER
RUPTR | HOSEI FLOW
1.6
7.6
9.2
9.2
8.0
2.8
17.2
15.2
11.6
6.8
2.0
8.1
8.8
8.1
5.6
WELD | PIPE | 1ANK I WELD/
CASKET | INSTALI INSTALI GASKT
INSTALI OEFICI DEFIC| FAIL
0.8
0.0
0.0
1.2
0.1
0.8
0.0
0.0
0.0
0.0
0.8
0.0
0.1
0.0
0.0
0.0
20.8
26.8
17.2
8.0
PIPE
RUPTR
0.8
2.0
2.1
1.6
2.8
TANK
RUPfR
0.8
1.2
1.6
2.1
2.8
PIPE
CORR
0.8
1.2
1.1
6.0
1.6
TANK
CORR
0.0
0.1
6.1
11.1
16.0
CAIAS-
SIROI'H
Rl S
0.0
0.1
2.0
1.6
1.2
TOTAL I 35.6 I 53.6 I 33.2 I 2.1 I 0.8 t 1.2 I 72.8 I 9.6 I 8.8 I 11.0 I 37.2 I 5.2 I
Storage Tank Technology 9 and Accumulation Tank Technology
Above-ground, ongrade, carbon steel, 210,000 gallons
Baseline, Toluene
18
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CAJLGORY 1
__ 1
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
TOTAL I 0.0
1
LOOSE! OVER
HOSEI FLOW
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.0
0.8
0.8
0.0
2.0
WELD
GASKET
INSTAL
0.0
0.8
0.0
0.1
0.0
1.2
PIPE
INSTAL
DEFIC
0.0
0.1
0.0
0.8
0.0
1.2
TANK
INSTAL
DEFIC
0.1
0.1
0.0
0.0
0.0
0.8
WELD/I
GASKT | PIPE
FAIL | RUPTR
17.8
33.3
17.3
7.9
1.9
81.2
1.3
3.9
0.1
1.1
3.1
1
TANK | PIPE
RUPTR | CORR
1.9
1.2
0.8
2.6
2.2
13.1 1 11.7
9.8
35.0
28.8
11.7
10.2
TANK
CORR
16.9
7.6
10.6
9.3
6.6
98.5 1 51.0
CATAS-I
STROPHI
RLS I
0.1
0.8
2.6
1.8 1
0.1 1
6.0 1
-------
EXHIBIT A-2 (continued)
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210,000 gallons
Baseline, Aqueous
1 1
YEAR
OF I
FA 1 LURE 1
1
2- 5
6-10
1 11-15
1 16-20
FAILURE CAIEGORY
1 | (WELD | PIPE I TANK
HOSE | LOOSE) OVER JCASKET j INSTALI INSTAL
RUPTRI HOSEI FLOW | INSTALI OEFIC) DEFIC
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.0
0.8
0.8
0.0
0.0 | 0.0
0.8 I 0.1
0.0 | 0.0
0.1 | 0.8
0.0 | 0.0
0.1
0.1
0.0
0.0
0.0
1 TOTAL | 0.0 I 0.0 I 2.0 1 1.2 1 1.2 1 0.8
WELD/
GASKT
FAIL
17.5
33.8
17.1
7.8
5.1
81.6
PIPE
RUPTR
1.3
3.9
0.8
1.1
3.5
TANK
RUPTR
6.1
1.2
0.8
2.6
2.2
PIPE
CORK
9.6
35.0
28.9
11.1
10.1
ICAIAS-
TANK IS1ROPII
CORR 1 RLS
17.1
7.5
10.5
8.7
6.9
13.9 1 12.9 J 98.3 1 50.7
0.1
0.8
2.6
1.8
0.8
6.1
Storage lank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210,000 gallons
Corrosion Protection, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 | IWELD | PIPE
HOSE | LOOSE) OVER (GASKET | INSTAL
RUPTRI HOSEI FLOW | INSTALI OEFIC
0.0 I
0.0
0.0
0.0
0.0
TOTAL | 0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.0
0.8
0.8
0.0
0.0 I 0.0
1.2 I 0.0
O.O | O.O
0.1 | 0.0
0.0 | 0.0
TANK
INSTAL
DEFIC
0.1
0.1
0.0
0.0
0.0
WELD/
GASKT
FAIL
18.2
37.1
15.6
8.0
3.1
PIPE
RUPTR
1.3
3.9
0.1
1.1
3.1
TANK
RUPTR
5.3.
1.7
0.8
1.2
2.2
PIPE
CORR
0.0
2.6
12.1
11.6
11.2
ICATAS-
TANK (STROP!)
COKR | RLS
11.7
1.2
1.7
0.1
1.3
0.0 I 2.0 I 1.6 | 0.0 1 0.8 1 82.3 1 13.1 1 11.2 | 10.8 1 19.3
0.1
0.8
2.6
1.8
0.1
6.0
Storage Tank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210.000 gallons
Corrosion Protection, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-1O
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
0.0
0.0
0.0
LOOSE
HOSE
0.0
0.0
0.0
0.0
0.0
OVER
FLOW
0.1
0.0
0.8
0.8
0.0
TOTAL | 0.0 I 0.0 I 2.0
WELD
GASKET
INSTAL
0.0
1.2
O.O
0.1
0.0
1.6
PIPE 1 TANK
INSTALI INSTAL
DEFICl DEFIC
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.1
0.0
0.0
0.0
WELD/
GASKT
FAIL
18.0
37.8
15.3
7.9
3.5
0.8 I 82.5
1
PIPE | TANK
RUPTRI RUPTR
1.3 I 6.6
3.9 1 1.7
0.8 I 0.8
1.1 } 1.2
3.5 1 2.2
PIPE
CORR
0.0
2.6
12.2
11.1
'11.0
TANK
CORR
11.9
1.2
1.7
0.1
1.3
CATAS-
STROPH
RLS
0.1
0.8
2.6
1.8
0.8
13.9 1 12.5 1 10.2 | 19.5 1 6.1
-------
EXHIBIT A-2 (continued)
FAILURE FREQUENCIES FOR SIORAGE/ACCUMULATION TANKS
Storage lank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade. carbon steel, 210,000 gallons
Partial Containment with Ground-water Monitoring, Toluene
1
1
1 YEAR
1 or
FAILURE
1
2- 5
6-10
11-15
16-20
1
FAILURE CATEGORY
1
HOSE | LOOSE
RUP1RI HOSE
0.0
0.0
0.0
O.O
0.0
0.0
0.0
0.0
0.0
0.0
IWELD | PIPE
OVER IGASKEH INSTAL
FLOW IINSIALI DEFIC
0.0
0.0
0.0
0.0
0.0
0.
-------
EXHIBIT A-2 (continued)
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel, 1.000 gallons
Baseline, Toluene
YEAR
OF
FAILURE
1
1 2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 | IWELO
HOSE | LOOSE) OVER j GASKET
RUPTR) HOSEI FLOW (INSTAL
0.0
6.1
5.9
6.3
6.3
1.2
11.3
16.1
11.3
8.1
TOTAL I 21.9 1 51.6
0.0
1.2
1.3
1.2
1.2
0.1
0.0
0.0
0.0
0.0
1.9 1 0.1
PIPE | TANK
INSTALI INSIAL
DEFIC) DEFIC
0.8
0.0
0.0
0.0
0.0
0.8
0.0
0.0
0.0
0.0
0.1
WELD/
GASKT
FAIL
22.3
13.8
9.2
9.7
6.0
0.1 1 61.0
PIPE
RUPTR
1.6
2.0
1.2
2.8
2.1
TANK
RUPTR
0.0
2.5
2.5
2.9
1.2
PIPE
CORR
5.5
2.5
0.0
3.3
3.7
13.0 I 12.1 | 15.0
TANK
CORR
16.8
26.9
21.8
13.1
5.0
86.9
CA FAS-
SI ROPH
RLS
0.0
0.1
1.3
0.1
0.0
2. \
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel, 1,000 gallons
Baseline, Aqueous
1
1
1 YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL I
FAILURE CATEGORY
I | IWELO | PIPE | TANK
HOSE I LOOSE! OVER (GASKET 1 INSTAL | INSIAL
RUPTRI HOSEI FLOW | INSTAL 1 DEFIC) DEFIC
0.0 I 2.9
6.3 I 12.1
6.2 | 18.3
6.2 | 7.9
6.2 | 12.1
0.0
1.2
1.2
2.1
O.8
0.1 I 0.8 I 0.0
0.0 I 0.0 | 0.0
0.0 1 O.O I 0.0
0.0 | 0.0 I 0.1
0.0 1 0.0 1 0.0
WELD/
GASKT
FAIL
16.2
13.3
13.7
8.7
9.6
I
PIPE j TANK
RUPTRI RUP1R
1.6
1.6
2.0
2.8
2.1
0.1
3.2
1.2
2.8
3.7
PIPE
CORR
1.6
2.0
1.6
2.8
3.3
21.9 1 53.3 I 5.6 I 0.1 I 0.8 1 0.1 1 61.5 1 13.1 1 11.3 1 11.3
1
TANK
COHR
16.7
27.5
21.6
11.3
5.8
85.9
CATAS-
STROI'H
RLS
0.0
0.1
1.2
0.1
1.0
2.0
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel. 1,000 gallons
Corrosion Protection, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1
HOSE | LOOSE
RUPTRI HOSE
0.0 I 3.0
6.1 I 11.5
6.0 1 15.5
6.3 I 9.7
6.3 I 10.5
TOTAL I 25.0 1 53.2
OVER
FLOW
0.0
1.2
1.3
2.0
1.2
5.7
WELD
GASKET
INSTAL
0.1
0.0
0.0
0.0
O.O
PIPE
INSTAL
DEFIC
0.8
0.0
0.0
0.0
0.0
0.1 1 0.8
TANK
INSTAL
DEFIC
0.0
0.1
0.0
0.0
0.0
0.1
WELD/
GASKT
FAIL
16.0
13.6
11.0
8.5
9.7
61.8
PIPE
RUPTR
1.6
1.6
1.6
2.8
2.1
TANK
RUPTR
0.1
2.9
0.8
2.9
3.7
13.0 | 10.7
PIPE
CORR
3.8
0.0
0.8
0.0
0.1
5.0
TANK
CORR
11.8
1.6
1.6
2.0
1. 1
21. 1
CATAS-
STROPH
RLS
0.0
0.1
1.3
0.1
0.0
2. 1
-------
f XIIIIII I A-;- (com i lined)
IAIIURL IRtQUlNCItS »OR SIORACI/ACCUMUlAlION IANKS
SLOI.KJI; lank technology 10 and Accuiuiilnt ion lank lucliiioloyy 19
Ondurgfound, carbon steel, 1.OUO gallons
Corrosion Protection, Aqueous
VI AH
01
IAIIUIU
1
2- 5
6-10
ll-lr>
i6-;'»
IAIIUIU CAILGORY
IIOSl
Hll I'll*
O.O
6.3
6.2
6.?
6.2
loose
HOSE
2.9
13.7
15.1
10.3
10.8
OVfR
FLOW
0.0
1.2
1.2
2.0
1.2
Hill)
CASK! 1
INSIAI
U.'l
0.0
U.O
0.0
0.0
PI PI I TANK
INSIAI | INSIAI.
Oil IC| DEI 1C
0.8
0.0
0.0
O.O
0.0
0.0
0.0
0.0
0.0
0.0
WE IO/
CASK!
IAIL
16.2
13.3
13.7
9.1
9.6
PIPE
RUPIR
1.6
1.6
2.0
2.8
2.1
1ANK
RUPIR
0.1
3.3
0.8
2.8
3.7
PIPf
CORR
3.7
0.0
0.8
0.0
0.1
ICAIAS-
IANK ISIROPII
CORK j RLS
12.9
1.6
. 1.6
?.o
1.0
0.0
0.1
1 .2
0.1
0.0
IOIAI | 21.9 | 53.1 I 5.6 I 0.1 I 0.8 I 0.0 I 61.9 I 13.1 I U.O I 1.9 I 22.1 I 2.O
:>loi,i<|r I.ink liM.-hnoloijy in :ind AcctiHinlni ion lank lochuoloyy 19
Undo i <|i omul, cat bon steel. 1,000 gallons
Si.Toiula ry Containment, loltiene
-.1 Alt
01
lAIIUlif
1
?- 5
6- 10
11 -15
16-20
lAUURt C AUGURY
nosr
RUPIR
1.6
5.2
6.8
6.1
6.8
LOOSf
HOSE
5.6
15.2
11.6
10.0
11.2
OVIR
now
0.0
0.8
1.2
1.2
0.0
WELD
CASKl 1
INSIAL
U.I
U.O
0.0
0.0
0.0
PIPE
INSIAL
DEE 1C
2.0
0.0
0.0
O.O
0.1
IANK | WILD/
INSIALI CASK)
OEFICI EAIL
2.8
0.0
0.0
0.0
0.0
0.0
11.1
16.1
11.1
12.1
PIPE
RUPIR
0.0
1.6
3.6
7.2
7.6
IANK
RUPIR
0.1
2.0
6.1
2.8
1.6
PIPE
CORR
0.1
2.8
1.6
1.6
0.0
IANK
COItR
0.1
1.8
18.1
11.0
11.1
CAIAS-
SIKOI'll
RLS
0.0
0.1
0.1
0.8
0.1
IOIAI | 26.8 I 53.6 I 3.2 I 0.1 I 2.1 I 2.8 I 5r.6 I 2O.O I 13.2 I 6.1 I 52.0 I 2.O
lank lechnology 10 arid Accumulation lank technology 19
Undeiground, carbon steel, 1,000 gaI Ions
Secondary Containment, Aqueous
VI AH
01
1 A 1 1 UUl
1
?- 5
6-10
11-15
16-20
FAILURE CA 11 GORY
HOSE
RUPIR
1.6
5.2
6.8
6.1
6.8
1
LOOSE! OVIR
HOSEI ELOW
5.6
15.2
11.6
10.0
11.2
0.0
0.8
1.2
1.2
0.0
WELD
CASKE 1
INSIAL
U.I
U.O
U.O
0.0
U.O
PIPE
INSIAL
DEE 1C
2.0
0.0
0.0
0.0
0.1
IANK | WELD/
INSIALI GASKI
DEEICI EAIL
2.8
0.0
0.0
0.0
0.0
0.0
11.1
16.1
11.1
12.1
PIPE
RUPIR
0.0
1.6
3.6
7.2
7.6
1ANK
RUPIR
0.1
2.0
6.1
2.8
1.6
PIPE
CORR
0.1
2.8
1.6
1.6
0.0
IANK
CORR
0.1
1.8
18.1
11.0
11.1
CAIAS-
SIKOI'tl
RLS
0.0
0.1
0.1
0.8
0.1
IOIAI | 26.8 I 53.6 I 3.2 | 0.1 I 2.1 I 2.8 I 5/.6 I 20.0 I 13.2 I 6.1 I 52.0 I 2.O
-------
EXHIBIT A-2 (continued)
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel, 1,000 gallons
Leak Testing with Ground-water Monitoring, Toluene
1
1
I YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
1
HOSE
RUP1R
0.0
6.1
5.9
6.3
6.3
FAILURE CATEGORY
1 IWELD
LOOSE 1 OVER | GASKET
HOSEI FLOW | INSTAL
1.2
11.5
15.1
12.6
10.5
0.0
1.2
1.3
1.2
1.2
TOTAL | 21.9 1 56.9 1 1.9
0.0
0.0
0.0
0.0
0.0
PIPE
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
0.0 I 0.0
TANK | WELD/
INSTALI GASKT
OEFICl FAIL
0.0
0.0
0.1
0.0
0.0
22.6
13.0
8.8
9.6
6.1
0.1 | 60.1
I 1
PIPE I TANK I PIPE
RUPTRJ RUPTRJ CORR
1.6 I 0.0 I 5.1
2.0 | 2.0 I 1.6
1.2 | 2.0 I 0.0
2.8 I 2.9 1 3.3
2.1 | 3.3 1 3.3
TANK
CORR
16.7
27.2
25.2
12.8
5.0
CATAS-
STROPH
RLS
0.0
0.1
1.3
0.1
0.0
13.0 | 10.2 | 13.6 I 86.9 1 2. 1
Storage Tank Technology to and Accumulation Tank Technology 19
Underground, carbon steel. 1,000 gal Ions
Leak Testing with Ground-water Monitoring, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
10IAL
FAILURE CATEGORY
1 1
HOSE 1 LOOSE) OVER
RUPTRl HOSEI FLOW
0.0
6.3
6.2
6.2
6.2
1.1
11.8
15.7
12.9
9.9
21.9 1 57.1
0.0
1.2
1.2
1.2
1.2
WELD
CASKE1
INSIAL
0.0
0.0
0.0
0.0
0.0
PIPE
INSIAL
OEFIC
0.0
0.0
0.0
0.0
0.0
1.8 | 0.0 I 0.0
TANK
INSTAL
OEFIC
0.0
0.0
0.1
0.0
0.0
WELD/I
GASKT | PIPE
FAIL | RUPTR
22.7
13.3
8.6
9.1
6.5
1.5
2.0
1.2
2.8
2.1
TANK
RUP1R
0.0
2.1
2.0
2.8
3.3
PIPE
CORR
5.1
1.6
0.0
3.3
3.3
TANK
CORR
17.8
26.9
21.8
13.2
1.8
CAIAS-
SIROPII
RLS
0.0
0.1
1.2
0.1
0.0
0.1 | 60.5 | 12.9 1 10.5 1 13.6 I 87.5 1 2.0
Storage Tank Technology 11 and Accumulation Tank Technology 20
Underground, fiberglass reinforced plastic, 1,000 gallons
Baseline, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 1
HOSE | LOOSE | OVER
RUPTRl HOSEI FLOW
0.0
6.0
6.0
6.5
6.0
1.7
11.6
13.5
11.2
11.6
0.0
0.8
1.3
2.7
0.8
WELD
CASKET
INSTAL
0.1
0.0
0.0
0.0
0.0
PIPE
INSTAL
DEFIC
1.3
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
WELD/
CASKT
FAIL
20.3
15.9
9.5
8.7
1.7
PIPE
RUPTR
8.2
3.5
3.9
5.3
3.1
TANK
RUP1R
0.1
1.0
3.9
1.3
1.7
PIPE
CORR
0.0
0.0
0.0
0.0
0.0
TANK
CORR
0.0
0.0
0.0
0.0
0.0
CATAS-
STROPH
RLS
0.0
0.1
1.3
0.1
0.0
TOTAL | 21.5 I 55.6 | 5.6 I 0.1 I 1.3 I 0.0 | 59.1 I 21.3 I 17.3 I 0.0 | 0.0 | 2.1
-------
a ?<%:
i i i i
p pp
o r\j roac
- c e c c c
c c spr-
f s c o f
£5
T W5
CC
en en
-or
V>cnr
-TC
CZT
^
or
oa- -
coocc
£Jf WC
- > JC
or
r
£§?
O C.C
e s -
r i a
C.O.C
a> re
= ?&
» 3
»c
IffkO
* !
e »
n
e s o
e -. s
vi O C.
» r/
a n
o §
ai
R O
- 3
B)
§£
i0 re
9) O
O O
1C
c >
BZ
cpppc
b b f b b
s p c p p
ocbcb
r-i- c
i^
o >
S 2
XX
tf rs
VI C >
^s tn
c
r
8J
a
S
*
w
IV
c
J.
a
i-
e
«
a
Ul
*
>O
^
C
*
c
0
no
O
1 1 1 1
r\)
»»»»-
**»*»
S5C?J^
*0*aM
o oc
o**ac
ooccc
fi" O C fi" J^
c c s or-
j-c c o*-
oooc*-
coooc
S^,^WO
r^^QDfO
fcao*
coooc
CCfCC
ccsoc
ooocc
c c s e c
faf f c
c ~
3:
§5
X
Cft tn
-o
is
- or
Z > ri
to c/) r
>EC
S*2
cz -
^H en ^
>x
or
-or
r x o
c
35
0
X
82
il
X 7C
X
* l
^
s
^
8
*
c ~ -
T C
oe. =
e *
t) ' C
Hi
r> «) O
VI 1C
ft 3 t>
ACCIIUII
ircud p
i>
C ui
W1J1 W W0
oa:
z > "
wo-.
x o
cz-o
-. -. T
Or"
OZ-
"-Z
O r
wr
r - s
o c
£" r
5 ->
O - r.
z vr.
v> e r
10 C
» o
VI 1C
»<
3 a>
o a
R >
it r
e
o
p
c
c
c
c p p c p
b b c o c
p p c c p
.b b c o b
co c o
c >
or
39
O -O
5n
O -<
0>
az
SB X,
tn O
3B->
r 35
C
- = z x
in b ~
S5 ^ I
C 3
c r
ic a"
at n
c
V.
o
o
>
o o > c
^ r. a
if c o c.
IO C
«< X
no r
C >
-------
<_ i ouni mued )
FAILURE FREQUENCIES FOR SFORACE/ACCUMULATION TANKS
Storage Tank Technology 11 and Accumulation Tank Technology 20
Underground, fiberglass reinforced plastic, 1.000 gallons
Leak Testing with Ground-water Monitoring, Toluene
1
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL
FAILURE CATEGORY
HOSE
RUPTR
0.0
6.0
6.0
6.5
6.5
I (WELD I PIPE | 1ANK
LOOSE | OVER | GASKE T | 1 NSTAL | 1 NSTAL
HOSEI FLOW |l NSTAL | OEFICI OEFIC
5.5 1 0.0
14.1 I 1.3
13.7 I 1.3
11.5 1 1.2
9.8 I 1.3
0.0 | 0.1 I 0.0
0.0 I 0.0 | 0.0
0.0 | 0.0 | 0.0
0.0 | 0.0 | 0.0
0.0 I 0.0 | 0.0
WELD/
GASKT
FAIL
20.0
11.0
7.7
7.0
6.8
25.0 | 51.9 1 5.1 1 0.0 I O.U | 0.0 I 55.5
1 1
PIPE | TANK | PIPE
RUPTRI RUPTRt CORR
8.1
3.9
3.9
5.3
3.9
25.1
0.9
3.5
5.2
6.0
3.0
18.6
0.0
0.0
0.0
0.0
0.0
TANK
CORR
0.0
0.0
0.0
0.0
0.0
0.0 I 0.0
CATAS-
STROPH
RLS
0.0
0.1
1.3
0.1
0.0
2. 1
Storage Tank Technology 11 and Accumulation Tank Technology 20
Underground, fiberglass reinforced plastic, 1,000 gallons
Leak Testing with Ground-water Monitoring, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 1 IWELD
HOSE | LOOSE! OVER (CASKET
RUPTRt HOSEI FLOW (INS1AL
0.0 | 5.5
6.0 | 11.9
6.0 I 13.2
6.5 1 12.0
6.5 1 9.6
0.0
1.3
1.3
1.2
1.3
0.0
0.0
0.0
0.0
0.0
PIPE I TANK | WELO/I
1 NSTAL |l NSTAL I GASKT | PIPE
DEFICI DEFICl FAIL I RUPTR
0.1
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
20.0
11.0
7.7
7.0
6.8
8.1
3.9
3.9
5.3
3.9
TANK
RUPTR
0.9
3.5
5.2
6.0
3.5
PIPE
CORR
0.0
0.0
0.0
0.0
0.0
ICAIAS-
TANK JSIROPII
CORR | RLS
0.0
0.0
0.0
0.0
0.0
0.0
0.1
1.3
0.1
0.0
TOTAL | 25.0 I 55.1 I 5.1 I 0.0 I 0.1 | 0.0 I 55-5 I 25.1 I 19.1 I 0.0 I 0.0 I 2.1
Storage Tank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel, 1,000 gallons
Baseline, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL
FAILURE CATEGORY
HOSE
RUPTR
0.0
6.1
6.1
6.3
6.3
LOOSE
HOSE
3.3
13.1
17.5
7.9
10.1
OVER
FLOW
0.0
1.2
1.3
2.1
0.8
WELD
GASKET
1 NSTAL
0.1
0.0
0.0
0.0
0.0
PIPE
1 NSTAL
DEFIC
0.8
0.0
0.0
0.0
0.0
25.1 I 52.5 1 5.7 1 0.1 I 0.8
TANK
1 NSTAL
OEFIC
0.0
0.0
0.0
0.0
0.0
0.0
WELD/
GASKT
FAIL
19.7
15.1
12. 1
6.7
5.1
PIPE
RUPTR
1.6
1.6
1.6
2.8
2.1
TANK
RUPTR
0.1
2.0
2.9
1.6
2.5
1
PIPE I TANK
CORR j CORR
1.3 I 0.1
0.1 | 0.8
1.6 | 1.2
0.1 | 5. 1
0.8 | 1.6
CATAS-
STROPH
RLS
0.0
0.1
1.3
0.1
0.0
59.0 | 13.0 I 12.1 1 1.5 I 12.1 | 2.1
-------
EXHIBIT A-2 (continued)
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 12 and Accumulation rank Technology 21
Underground, stainless steel, 1,000 gallons
Baseline, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
6.1
6.1
6.3
6.3
1 (WELD | PIPE
LOOSEI OVER (GASKET HNSTAL
HOSEI FLOW IINSTALI OEFIC
3.3
13.1
17.5
7.9
11.3
0.0
1.2
1.3
2.1
0.8
0.1
0.0
0.0
0.0
0.0
0.8
0.0
0.0
0.0
0.0
TANK | WELD/
INSTALI GASKT
OEFICI FAIL
0.0
0.0
0.0
0.0
0.0
19.7
15.1
12.1
6.7
5.1
PIPE
RUPTR
1.6
1.6
1.6
2.8
2.1
1
TANK | PIPE
RUP1R) CORR
0.1
2.0
2.9
1.6
2.9
1.3
0.1
1.6
0.1
0.8
ICAFAS-
TANK ISTROPH
CORR j RLS
0.1
0.8
1.2
5. 1
1.6
0.0
0.1
1.3
0.1
0.0
TOTAL I 25.1 I 53.1 I 5.7 I 0.1 I 0.8 | 0.0 | 59.0 | 13.0 I 12.8 I 1.5 I 12.1 | 2.1
Storage Tank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel, 1,000 gallons
Corrosion Protection, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 1 IWELU
HOSE | LOOSEI OVER (GASKET
RUPTR | HOSEI FLOW I INSIAL
0.0
6.1
6.1
6.3
6.3
3.3
13.8
16.3
8.1
11.8
0.0
1.2
1.3
2.0
1.2
0.1
0.0
0.0
0.0
0.0
PIPE | TANK | WELO/I
INSTALIINSTALI GASKT | PIPE
OEFICI OEFICI FAIL | RUPTR
0,8
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
19.7
15.1
12. 1
6.7
5.1
1.6
1.6
1.6
2.8
2.1
TANK
RUPTR
0.1
2.0
2.9
3.3
2.9
PIPE
CORR
1.3
0.0
0.0
0.0
0.0
TANK
CORR
0.1
0.0
0.0
0.1
0.0
CATAS-
STROPH
RLS
0.0
0.1
1.3
0.1
0.0
TOTAL | 25.4 I 53.6 I 5.7 I 0.4 I 0.8 I 0.0 I 59.0 I 13.0 | 11.5 I 1.3 I 0.8 I 2.1
Storage Tank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel, 1,000 gallons
Corrosion Protection, Aqueous
1
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE
HOSE
RUPTR
0.0
6.4
6.4
6.3
6.3
LOOSE
HOSE
3.3
13.8
16.3
8.4
11.8
TOTAL I 25.1 1 53.6
OVER
FLOW
0.0
1.2
1.3
2.0
1.2
5.7
WELD
CASKET
INSTAL
0.1
0.0
0.0
0.0
o.b
0.1
CATEGORY
PIPE | TANK I WELD/
INSTALIINSTALI GASKT
OEFIC) DEFIC) FAIL
0.8
0.0
0.0
0.0
0.0
0.8
0.0
0.0
0.0
0.0
0.0
0.0
19.7
15.1
12.1
6.7
5.1
59.0
PIPE
RUPTR
1.6
1.6
1.6
2.8
2.1
TANK
RUPTR
0.1
2.0
2.9
3.3
2.9
PIPE
CORR
1.3
0.0
0.0
0.0
0.0
TANK
CORR
0.1
0.0
0.0
0.1
0.0
CAIAS-
STROPM
RLS
0.0
0.1
1.3
0.1
0.0
13.0 111.51 1.3 | 0.8 2. 1
-------
iAIM nil A-2 (continued)
I All 111(1 IIUQULNCIIS I OK SIOKACf/ACCUMUL AT ION IANKS
SUM .HJC I .ink technology 1? and A<:ciiiiuil-IO
1 1- 15
16-20
lAILOKf CAIICOKY
IIOSI
KOCIK
1.6
5.2
6.8
6.M
6.8
i oosr
HOSt
5.2
15.6
10.8
10.8
7.6
|WI 1 1)
OVIR IGASKI i
now
0.0
0.8
1.2
1.2
0.0
INSIAl
O.M
0.0
0.0
0.0
0.0
PI PI
INSIAl
Ut» 1C
2.0
O.O
0.0
O.O
O.M
IANK | WflD/|
INSIAl | CASKT
OLMCI fAIL
2.8
O.O
O.O
0.0
0.0
0.0
17.2
15.2
12. O
9.6
PIPE
RUPIR
0.0
1.6
3.6
7.2
7.2
IANK
RUPIR
O.M
3.2
6.8
2.8
2.0
PIPE
CORH
0.0
0.8
O.M
O.M
1.2
ICAIAS-
IANK ISTROPII
CORR
o.o
0.0
1.2
1.6
2.0
KIS
0.0
O.M
O.M
0.8
O.M
IOIAI | 26.8 I 50.0 I 3.2 I O.M I 2.M I 2.8 I 5M.O | 19.6 I 15.2 | 2.8 I M.8 I 2.O
SioiiKjo lank technology 12 and Accumulation lank technology 21
Onduiground, stainless steel, M.OOO gallons
l«;ak testing with Ground-water Monitoring, Toluene
Yl AK
Ol
(All OKI
1
2- 5
6- 10
1 1 - 1 4>
16-20
IOIAI
IIOSI
RUPIR
0.0
6.3
6.2
6.2
6.2
2M . 'J
1
LOOSE I OVfR
IIOSII FLOW
3.3 1 0.0
1M.O I 1.2
15.3 1 1.2
15.8 I 1.2
6.7 1 1.2
1 55 . 1 1 M.8
WE 10
GASKf 1
INSIAl
0.0
0.0
0.0
0.0
0.0
0.0
1 All UKt
PI PI | TANK
INSIAl | INSIAl
Of MCI DEI 1C
0.0 | 0.0
0.0 | 0.0
0.0 I 0.0
0.0 1 0.0
o.o 1 o.o
0.0 I 0.0
CA 1 1 OUt
WELD/
CASK1
FAIL
22.7
10. 7
16.6
6.6
5.7
62.3
-------
EXHIBIT A-2 (continued)
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel, 1,000 gallons
Leak Testing with Ground-water Monitoring, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL
FAIIURE CATEGORY
1 | IWELD
HOSE | LOOSE | OVER (CASKET
RUPTRI HOSEI FLOW I INSIAL
0.0
6.3
6.2
6.2
6.2
3.3
11.0
15.3
16.5
7.1
21.9 1 56.2
0.0
1.2
1.2
1.2
1.2
1.8
0.0
0.0
0.0
0.0
0.0
0.0
PIPE
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
-0.0
0.0
0.0
0.0
0.0
0.0
WELD/
GASKT
FAIL
22.7
10.7
16.6
6.6
6.1
62.7
PIPE
RUPTR
1.5
2.0
1.6
2.8
2.1
1
TANK | PIPE
RUPTRI CORR
0.1 I 1.2
1.6 1 0.8
1.6 I 0.8
3.2 | 0.1
3.2 | 0.1
13.3 I 10.0 1 3.6
ICATAS-
TANK ISTROPM
CORR RLS
0.0
0.0
1.6
5.8
6.9
11.3
0.0
0.1
1.2
0.1
0.0
2.0
Storage Tank Technology 13 arid Accumulation Tank Technology 22
In-ground, concrete, 2,100 gallons
Ba seIi ne, To Iuene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 1 IWELD
HOSE | LOOSE) OVER I GASKET
RUPTRI HOSE) FLOW I INSTAL
0.8
2.1
2.8
2.1
3.6
TOTAL | 12.0
6.6
18.1
8.5
10.7
9.0
3.3
17.2
6.5
9.0
6.8
52.9 1 12.8
0.0
0.1
0.0
0.1
0.0
PIPE
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.0
0.0
0.0
0.1
0.0
WELD/
GASKf
FAIL
15.6
30.3
20.1
9.9
1.8
0.8 I 0.0 I 0.1 I 80.7
PIPE
RUPTR
0.0
3.2
2.0
1.6
3.2
10.0
1ANK
RUPTR
0.0
0.0
0.1
1.6
9.0
PIPE
CORR
0.8
0.8
2.1
0.8
3.2
ICAIAS-
TANK JSIROI'II
CORR | RLS
0.0
0.0
0.0
0.0
0.0
0.0
0.8
1.6
1.6
0.8
11.0 I 8.0 1 0.0 I 1.8
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, concrete, 2,100 gallons
Baseline, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.8
2.1
2.8
2.1
3.6
LOOSE
HOSE
6.6
18.1
8.5
10.7
9.0
OVER
FLOW
3.3
17.2
6.5
9.0
6.8
WELD | PIPE
GASKET) INSTAL
INSTALI DEFIC
0.0 | 0.0
0.1 I 0.0
0.0 I 0.0
0.1 1 0.0
0.0 I 0.0
TANK
INSTAL
DEFIC
0.0
0.0
0.0
0.1
0.0
WELD/I
GASKF) PIPE
FAIL | RUPTR
15.6
30.3
20.1
9.9
1.8
0.0
3.2
2.0
1.6
3.2
TANK
RUPTR
0.0
0.0
0.1
1.6
9.0
PIPE
CORR
0.8
0.8
2.1
, 0.8
3.2
TANK
CORR
0.0
0.0
0.0
0.0
0.0
ICAIAS-
STROPH
RLS
0.0
0.8
1.6
1.6
0.8
TOTAL I 12.0 I 52.9 I 12.8 I 0.8 I 0.0 I 0.1 | 80.7 | 10.0 | 11.0 I 8.0 I 0.0 I 1.8
-------
EXHIBIT A-2 (continued)
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, concrete, 2,100 gallons
Partial Containment with Ground-water Monitoring, Toluene
I
I
FAILURE CATEGORY
I YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL
HOSE
RUPTR
0.0
3.2
8.0
3.2
1.0
18.1
LOOSE
HOSE
1.5
15.0
15.3
11.0
6.8
55.6
OVER
FLOW
5.3
11.6
11.3
12.0
8.8
19.0
WFLO
CASKET
INSTAL
0.0
0.1
0.0
0.1
0.0
0.8
PIPE
INSTAL
DEFIC
0.0
0.0
0.0
0.1
0.0
0.1
TANK
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
0.0
WELD/
CASKT
FAIL
8.1
10.8
18.9
7.6
1.8
80.2
PIPE
RUP1R
0.1
2.0
2.1
3.2
2.0
10.0
1
TANK | PIPE
RUP1RI CORR
0.0 I 0.0
0.0 j 1.6
0.0 | 1.8
2.1 | 3.6
5.6 I 3.2
8.0 I 16.1
TANK
CORR
0.0
0.0
0.0
0.0
0.0
0.0
CA1AS-
STROPH
RLS
0.0
1.2
2.0
2.8
0.0
6.0
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, Concrete, 2,100 gallons
Partial Containment with Ground-water Monitoring, Aqueous
1
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE
CATEGORY
| | (WELD | PIPE | TANK I WELD/
HOSE I LOOSE! OVER (CASKET | INSTAL | INSTALI GASKT
RUPTRI HOSE) FLOW I INSTALI DEFIC) DEFIC! FAIL
0.0
3.2
8.0
3.2
1.0
TOTAL I 18.1
1.5
15.0
15.3
11.0
6.8
5.3 I 0.0
11.6 I 0.1
11.3 | 0.0
12.0 | 0.1
8.8 I O.O
55.6 I 19.0 | 0.8
0.0
0.0
0.0
0.1
O.O
0.0
0.0
0.0
0.0
0.0
0.1 I 0.0
8.1
10.8
18.9
7.6
1.8
PIPE
RUP1R
0.1
2.0
2.1
3.2
2.0
1
TANK | PIPE
RUPTRI CORR
0.0 I 0.0
0.0 I 1.8
0.0 I 1.8
2.1 | 3.6
5.6 | 3.2
ICATAS-
TANK JSTKOPII
COKR 1 RLS
0.0
0.0
0.0
0.0
0.0
0.0
1.2
2.0
2.8
0.0
80.2 | 10.0 I 8.0 1 16.1 I 0.0 1 6.0
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, concrete, 2,100 gallons
Secondary Containment, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.8
3.2
3.6
3.2
1.0
1
LOOSE j OVER
HOSE) FLOW
1.1 1 2.0
16.1 | 8.0
16.0 | 8.8
12.8 I 8.1
11.6 I 5.6
WELD
GASKET
INSTAL
1.2
0.1
0.0
0.1
0.0
PIPE
INSTAL
DEFIC
0.1
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.8
0.0
0.0
0.0
0.0
WELD/
GASKT
FAIL
0.0
26.0
20.1
16.1
11.2
1
PIPE | TANK
RUPTRI RUPTR
0.8
2.0
2.1
1.6
2.8
0.0
0.0
0.0
0.0
0.8
PIPE
CORR
0.8
2.1
1.0
2.8
2.1
ICATAS-
TANK JSTROPH
CORR | RLS
0.0
0.0
0.0
0.0
0.0
0.0
0.1
2.0
1.6
1.2
TOTAL I 11.8 I 61.2 I 32.8 I 2.0 I 0.1 I 0.8 I 71.0 I 9.6 I 0.8 I 12.1 I 0.0 I
-------
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, concrete, 2,100 gallons
Secondary Containment, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1
HOSE | LOOSE
RUPTRJ HOSE
0.8 | U.(|
3.2 | 16. H
3.6 I 16.0
3.2 | 12.8
1.0 | 11.6
OVER
FLOW
2.0
8.0
8.8
8.
-------
FAILURE FREQUENCIES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Corrosion Protection, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
1.0
2.5
3.0
2.5
3.1
LOOSE
HOSE
5.8
17. 1
7.1
10.7
8.7
OVER
FLOW
2.9
17. 1
6.8
9.1
(4.8
WELD
GASKET
INSTAL
0.0
0.0
0.0
0.5
0.5
PIPE | TANK
INSTALI INSTAL
DEflCl DEFIC
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.5
0.0
WELD/I
GASKT
FAIL |
11.5
36.6
11.8
9.6
6.8
PIPE
RUPTR
0.0
3.0
2.5
1.5
3.5
1
TANK | PIPE
RUPTR | CORR
0.5
1.0
1.5
3.9
1.9
0.0
0.0
0.5
1.0
0.0
TANK
CORR
1.3
0.5
3.0
3.0
3.5
CATAS-
STROPH
RLS
0.0
1.0
1.5
2.0
0.5
I
TOTAL | 12.14 | 19.7 I '11.0 I 1.0 I 0.0 | 0.5 I 79.3 I 10.5 I 8.8 I 1.5 I 11.3 I 5.0 I
Storage Tank Technology 1*4 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Corrosion Protection, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.9
2.3
2.9
2.8
3.3
I IWELO
LOOSE) OVER (GASKET
HOSE) FLOW I INSTAL
6.6
17.9
7.0
10. ll
8.14
2.8
16.9
6.6
8.8
5.2
0.0
0.0
0.0
0.5
0.5
PIPE
INSTAL
OEFIC
0.0
0.0
0.0
0.0
0.0
TANK | WELD/
INSTALI GASKT
DEFICl FAIL
0.0
0.0
0.0
0.5
0.0
11.3
37.1
15.1
9.3
6.6
1
PIPE | TANK
RUPTRI RUPTR
0.0
2.8
2.1
1.1
3.3
0.5
1.0
1.5
3.7
1.9
PIPE
CORR
0.0
0.0
0.5
1.0
0.0
TANK
CORR
6.6
0.5
2.8
2.8
3.3
CATAS-
SIROPH
Rl S
0.0
1.0
1.5
2.0
1.0
TOTAL I 12.2 I 50.3 I 10.3 I 1.0 I 0.0 I 0.5 I 79.1 I 9.9 I 8.6 I 1.5 I 16.0 I 5.5
i
K)
NJ
Storage Tank Technology Ml and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Partial Containment with Ground-water Monitoring, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
I
HOSE | LOOSE
RUPTRI HOSE
0.0 I 1.9
3.0 I 13.2
6.6 1 13.7
3.5 1 12.0
1.5 1 8.9
OVER
FLOW
5.8
11.5
11.9
12.5
8.1
WELD
GASKET
INSTAL
0.0
0.0
0.14
0.0
0.0
PIPE
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.0
0.0
0.1
0.0
0.0
WELD/
GASKT
FAIL
6.6
12.1
11.3
5.2
1.8
PIPE
RUPTR
0.1
1.7
2.6
3.5
1.6
TANK
RUPTR
0.1
1.0
2.2
3.0
2.2
1
PIPE | TANK
CORR | CORR
0.9
2.7
5.8
1.7
, 2.1
11.9
25.7
22.1
19.0
1.0
CATAS-
STROPH
RLS
0.0
1.3
1.7
2.1
0.0
TOTAL I 17.6 I 52.7 I 50.1 I 0.1 I 0.0 I 0.1 | 73.3 I 9.8 I 11.8 I 13.2 I 82.7 I 5.1
-------
- , ~~r.i I. I IIUC2U
FAILURE fREQUENCIES FOR S10RAGE/ACCUMULAT ION TANKS
Storage Tank technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Partial Containment with Ground-water Monitoring, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
3.0
6.9
3.1
1.3
LOOSE
HOSE
1.8
13.5
11.1
11.7
8.8
TOFAL | 17.6 | 52.9
IWELD
OVER ICASKET
FLOW | INSTAL
5.7
11.9
11.8
12.2
8.7
50.3
0.0
0.0
0.1
0.0
0.0
PIPE | TANK
INSTAL | INSTAL
OEFICI DEFIC
0.0
0.0
0.0
0.0
0.0
0.1 | 0.0
0.0
0.0
0.0
0.0
0.1
0.1
WELD/
GASKT
FAIL
6.6
12.3
11.0
5.2
1.8
PIPE
RUPTR
0.1
1.7
2.6
3.5
2.1
TANK
RUPTR
0.1
3.9
3.5
2.6
2.2
1
PIPE
CORR
0.9
2.7
5.7
1.7
2.6
72.9 1 10.3 1 12.6 1 13.6
TANK
CORR
13.1
25.3
21.9
18.8
1.0
83.1
CATAS-
S FRO Pit
RLS
0.0
1.3
1.7
2. 1
0.0
5.1
Storage Tank lechnology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Secondary Containment, Toluene
I FAILURE CATEGORY
YEAR |
OF | HOSE
FAILUREI RUPTR
1 I 0.8
2- 5 1 3.2
6-10 I 3.6
11-15 1 3.2
16-20 I 1.0
LOOSE
HOSE
2.8
15.6
11.0
11.1
10.0
OVER
FLOW
3.2
11.0
12.8
12.0
8.1
WELD
GASKET
INSTAL
1.2
0.1
0.0
0.0
0.1
PIPE | TANK
INSFALI INSTAL
DEFICI DEFIC
0.1
0.0
0.0
0.0
0.0
1.0
0.0
0.0
0.0
0.0
WELD/
GASKT
FAIL
0.0
20.1
25.6
18.8
8.1
PIPE
RUPTR
0.8
2.0
2.1
1.6
2.8
TANK
RUPTR
0.1
2.8
3.2
2.8
2.1
PIPE
CORR
0.0
1.6
3.2
5.6
3.2
TANK
CORR
0.0
1.0
19.6
10.1
11.8
CATAS-
STROPH
RLS
0.0
0.1
2.0
1.6
1.2
TOTAL I 11.8 I 56.8 I 50.1 I 2.0 I 0.1 I 1.0 I 73.2 I 9.6 I 11.6 I 13.6 I 18.8 I 5.2
i
K)
OJ
Storage Tank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel. 2,100 gallons
Secondary Containment, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.8
3.2
3.6
3.2
1.0
LOOSE
HOSE
2.8
15.6
11.0
11.1
10.0
OVER
FLOW
3.2
11.0
12.8
12.0
8.1
WELD
GASKET
INSTAL
1.2
0.1
0.0
0.0
0.1
PIPE
INSTAL
DEFIC
0.1
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
1.0
0.0
0.0
0.0
0.0
WELD/
GASKT
FAIL
0.0
20.1
25.6
18.8
8.1
1
PIPE | TANK
RUPTR I RUPTR
0.8
2.0
2.1
1.6
2.8
0.1
2.8
3.2
2.8
2.1
1
PIPE | TANK
CORR | CORR
0.0
1.6
3.2
5.6
3.2
0.0
1.0
19.6
10.1
11.8
CATAS-
STROPH
RLS
0.0
0.1
2.0
1.6
1.2
TOTAL | 11.8 I 56.8 | 50.1 I 2.0 I 0.1 I 1.0 I 73.2 | 9.6 I 11.6 I 13.6 I 18.8 I 5.2
-------
FAILURE FREQUENCIES FOR SQG TANKS
Small Quantity Generator lank Technology 15
Above-ground, carbon steel, 200 gallons
Baseline, Aqueous and Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.0
0.0
O.ll
0.0
0.1
LOOSE
HOSE
2.1
8.1
8.9
8.5
6.2
OVER
FLOW
2.1
12.6
15.1
9.6
8.9
WELD | PIPE
CASKET | INSTAL
INSTALI DEFIC
0.0
0.0
0.0
O.ll
O.ll
0.0
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.0
0.0
0.0
0.8
0.0
WELD/I
GASKTI PIPE
FAIL | RUPTR
16.8
27.3
11.2
8.0
8.0
0.0
2.1
3.2
2.1
2.1
1
TANK | PIPE
RUPTR | CORR
0.1
1.2
3.3
1.2
0.8
0.8
2.1
2.1
2.9
0.8
TANK
CORR
0.0
5.9
13.9
11.7
9.6
CATAS-
S1ROPH
RLS
0.1
1.2
1.6
0.0
0.8
0.8 | 31.1 I 18.3 I 0.8 I 0.0 I 0.8 I 71.3 I 10.1 I 6.9 I 9.0 I 11.1 I 1.0 I
Small Quantity Generator Tank Technology 15
Above-ground, carbon steel, 200 gallons
Secondary Containment, Aqueous and Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPTR
0.1
0.8
0.1
0.1
0.8
LOOSE
HOSE
0.0
8.8
8.1
7.2
1.8
OVER
FLOW
2.0
8.1
8.8
8.1
5.6
WELD
GASKET
INS1AL
0.8
0.0
0.0
1.2
0.1
PIPE
INSIAL
DEFIC
0.8
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.8
0.0
0.1
0.0
0.0
WELD/
GASK1
FAIL
0.0
20.8
26.8
17.2
8.0
PIPE
RUPTR
0.8
2.0
2.1
1.6
2.8
TANK
RUP1R
0.8
1.6
1.6
2.0
3.2
PIPE
CORR
0.8
1.2
1.1
6.0
1.6
TANK
CORR
0.0
0.1
3.2
11.6
12.0
CAIAS-
STROPH
RLS
0.0
0.1
2.0
1.6
1.2
TOTAL I 2.8 I 29.2 | 33.2 | 2.1 | 0.8 I 1.2 I 72.8 I 9.6 I 9.2 | 11.0 I 27.2 | 5.2
Sma I I Quantity Generator Tank Technology 16
Underground, carbon steel, 200 gallons
Ba scIi ne, To Iuene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
HOSE
RUPIR
0.0
0.1
0.1
1.2
0.1
LOOSE
HOSE
0.9
8.0
7.5
1.8
8.5
(WELD
OVER j GASKET
FLOW I INSTAL
0.0
1.3
0.9
2.7
0.1
0.1
0.0
0.0
0.0
O.M
PIPE
INSTAL
DEFIC
0.1
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.0
0.1
0.0
0.0
0.1
WELD/
GASKT
FAIL
15.2
13.1
13.8
8.0
9.3
PIPE
RUPIR
1.5
1.7
1.2
3.0
2.5
1
TANK | PIPE
RUPTR | CORR
0.9 1 1.9
2.6 1 1.6
2.1 I 1.7
3.5 I 3.1
3.0 |i 3.1
ICATAS-
TANK ISTROPH
CORR | RLS
5.8
21.5
23.3
18.3
10. 1
0.0
0.1
1.3
0.1
0.0
TOTAL I 2.1 | 29.7 I 5.3 I 0.8 I 0.1 I 0.8 | 59.? I 12.9 I 12.1 | 11.1 I 79.0 I 2.1
-------
. . . .. j i milt i ruied )
FAILURE FREQUENCIES FOR SQG TANKS
Small Quantity Generator Tank Technology 16
Underground, carbon steel, 200 gallons
Baseline. Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY |
HOSE
RUPTR
0.0
O.U
O.U
1.2
O.U
IO1AL | 2.U
1
LOOSE! OVER
HOSEI FLOW
0.9
7.8
7.7
U.U
8.2
29.0
0.0
1.3
1.3
2.7
O.U
5.7
WELD
GASKET
INSTAL
O.U
0.0
0.0
0.0
O.U
PIPE | TANK | WELD/
INSTALI INSTALI GASKT
DEFICI DEFIC) FAIL
O.U
0.0
0.0
0.0
O.U
0.8 | O.8
0.0 I 15.3
O.U I 13.1
O.U | 13.9
O.U I 8.2
O.U I 9.9
1.6 I 60. U
1
PIPE | TANK
RUPTRI RUPTR
U.8
1.7
1.2
3.0
2.5
0.9
2.6
2. 1
3.0
2.6
PIPE
CORR
K6
1.7
3.1
3.1
TANK
CORR
8.7
21 .0
22.3
17.9
10.1
13.2 | 11.2 | Hi. 3 1 80.0
,
CA1AS-
STROPH
RLS
0.0
O.U
1.3
O.U
0.0
2. 1
Small Quantity Generator lank Technology 16
Underground, carbon steel. 200 gallons
Corrosion Protection, Toluene
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1 I (WELD
HOSE j LOOSE) OVER (GASKET
RUPTRI HOSE) FLOW I INSTAL
0.0
O.ll
0.4
1.2
O.ll
0.9
8.9
7.5
6.6
8.5
0.0
1.3
0.9
2.2
0.8
0.14
0.0
0.0
0.0
0.0
PIPE
INSTAL
DEFIC
O.ll
0.0
0.0
0.0
0.0
TANK
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
WELD/
CASKT
FAIL
15.2
13.14
114.2
8.5
9.2
PIPE
RUPTR
U.5
1.7
1.2
3.0
2.5
TANK
RUPTR
0.9
2.6
0.9
3.9
3.5
PIPE
CORR
«4.0
0.0
0.9
0.0
0.14
TANK
CORR
1.8
0.8
0.8
2.6
2.6
CATAS-
S1ROPII
RLS
0.0
O.ll
1.3
O.ll
0.0
I
TOTAL I 2.l| | 32.l| | 5.2 | O.ll I O.ll I 0.0 I 60.5 I 12.9 I 11.8 I 5.3 I 8.6 I 2.1 I
to
Ui
Small Quantity Generator Tank Technology 16
Underground, carbon steel, 200 gallons
Corrosion Protection, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
FAILURE CATEGORY
1
HOSE | LOOSE
RUPTRI HOSE
0.0
O.U
O.U
1.2
O.U
0.9
8.7
7.U
5.6
8.6
(WELD
OVER IGASKET
FLOW | INSTAL
0.0
1.3
1.3
2.2
0.8
O.U
0.0
0.0
0.0
0.0
PIPE | TANK
INSTALI INSTAL
DEFICI DEFIC
O.U
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
WELD/I | |
GASKTI PIPE I TANK | PIPE
FAIL | RUPTRI RUPTRI CORR
15.2
13.0
1U.2
8.6
10.0
U.8 I 0.9 I 3.9
1.7 I 2.6 I 0.0
1.2 | 0.9 I 0.9
3.0 I 3.9 1 0.0
2.5 1 3.51' O.U
ICATAS-
IANK ISTROPH
CORR 1 RLS
U.8
0.8
0.8
2.6
2.6
0.0
O.U
1.3
O.U
0.0
I
10TAL I 2.U I 31.2 | 5.6 I O.U I O.U I 0.0 | 61.0 I 13.2 | 11.8 | 5.2 I 11.6 I 2.1 |
-------
IAIIUK1 IIUQUINCIfS M)R SQG IANKS
:>m.i I I Uii.uitiiy Crneraior lank leclmoUxjy
IliMliM <|i uunil. cat hun steel, 2t)O gallons
SiM:oii
(>' 111
ii - r>
I6-^O
lAllUKl CAIEGOitV
HOSf
RUPIH
II. U
0.8
O.ll
O.ll
0.8
LOOSt
nose
3.2
7.2
7.6
6.l|
6.l|
OVER
(LOW
0.0
0.8
0.8
1.2
0.0
IOIAI | 2.B | 30.8 1 2.8
WHO
GASKI 1
INSIAl
O.H
0.0
0.0
O.O
0.0
PI PI
INSIAl
on ic
2.0
0.0
0.0
0.0
O.ll
IANK
INSFAI
on ic
2.l|
0.0
0.0
0.0
o.o
WELD/
GASKI
FAIt
0.0
11.1
16. l|
11.1
12. i|
PIPE
KUPIR
0.0
1.6
3.6
7.2
7.6
TANK
RUPIR
0.1
2.1
6.0
2.l|
2.0
PIPE
CORR
O.ll
2.8
1.6
1.6
0.0
IANK
CORR
0.1
3.2
13.6
11.1
10.8
CAIAS-
SIROPH
Rl S
O.O
O.ll
O.ll
o.a
O.ll
O.ll f 2.1 | 2.i| | 57.6 I 20.0 | 13.2 1 6.l| I 12.1 1 2.O
Sin.i I I Un.iiniiy (UMIIM alor I.'ink lecliriology 16
dude i <|i iiiiiu). cat l>un slue I, 2OO gallons
'><! ond.i i y Containment, Aqueous
Yt Al<
Of
(All ORE
1
2- 5
<>- l(»
1 1 - \'>
I6-2O
FAILURE CA II GORY
nosr
HUP IK
O.ll
0.8
O.ll
O.ll
0.8
IOOSE
HOSE
3.2
7.2
7.6
6.l|
6.M
OVER
HOW
0.0
0.8
0.8
1.2
0.0
WELD
GASKE 1
INSIAl
O.ll
0.0
0.0
O.O
0.0
PIPE
INSIAL
I)E E 1 C
2.0
O.O
0.0
O.O
O.ll
IANK | HELD/
INS1ALJ CASK1
OFEICI EAIL
2.l|
0.0
O.O
0.0
O.O
0.0
11.14
16.1
11.1
12.1
PIPE
RUPIR
0.0
1.6
3.6
7.2
7.6
TANK
RUPIR
0.1
2.1
6.0
2.1
2.0
PIPE
CORR
0.1
2.8
1.6
1.6
O.O
IANK
CORR
0.1
3.2
13.6
11.1
10.8
CAIAS-
SIROPII
RLS
O.O
0.1
0.1
o.a
0.1
IOIAI | 2.8 I JO.8 I 2.8 | 0.1 I 2.1 | 2.1 | 57.6 I 20.0 | 13.2 I 6.1 I 12.1 I 2.O
Small (Ju.unity Generator lank Technology 16
Untli: 11) round, carbon slue I, 200 gallons
Irak IKSting wild Giound-waler Monitoring, Toluene
Yl Al(
01
(All UKI
1
2- '>
(,- U)
1 1 - 1 ')
i6-;'u
IOIAI
FAILURE CAICGORY
MOSl
KUPIR
0.0
0.1
0.1
1 .?
0.1
IOOSL
HOSE
1.3
9.2
9.2
7.9
8.3
(Will)
OVIR JCASKII
FLOW 1 INSIAl
0.0
1.3
0.9
1 .2
1.3
0.0
0.0
O.O
O.O
0.0
Pll'l
INSIAl
OFF 1C
0.0
0.0
0.0
0.0
0.0
IANK
INSIAl
OEFIC
0.0
0.0
0.0
0.0
0.0
WELD/
GASKI
FAIL
22.9
13.7
8.7
9.3
5.7
PIPE
RUPIR
1.8
2.2
1 .2
3.0
2.5
i'.'l 1 3'j.9 1 I./ 1 0.0 1 0.0 | 0.0 I 60.3 I 13.7
TANK
RUPIR
0.1
2.1
2.5
2.6
3.0
PIPE
CORR
5. 7
1.3
0.0
3. 1
3.5
ICAIAS-
FANK JS1ROPH
CORR j RLS
6.2
22.9
22.0
19.1
12.1
0.0 1
0.1 I
1.3 1
0.1 I
0.0 I
10.6 1 1'3.6 1 82.9 1 2.1|
-------
FAILURE FREQUENCIES FOR SQG TANKS
Small quantity Generator Tank Technology 16
Underground, carbon steel, 200 gallons
Leak Testing with Ground-water Monitoring, Aqueous
YEAR
OF
FAILURE
1
2- 5
6-10
11-15
16-20
TOTAL
FAILURE CATEGORY
HOSE
RUPIR
0.0
O.ll
0.«4
1.2
0.»4
LOOSE
HOSE
1.3
10.14
9.0
8.2
8.3
OVER
FLOW
0.0
1.3
1.3
1.2
1.3
2.M 1 37.2 I 5.1
WELD
GASKET
INSTAL
0.0
0.0
0.0
0.0
0.0
PIPE
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
0.0 | 0.0
TANK
INSTAL
DEFIC
0.0
0.0
0.0
0.0
0.0
0.0
WELD/
GASKT
FAIL
22.9
13. «4
8.6
9.6
6.1
60.6
PIPE
RUPTR
14.8
2.2
1.2
3.0
2.5
1 1
TANK | PIPE | TANK
RUPTR j CORR | CORR
0.14 I 5.6
1.7 I 1.3
2.1 | 0.0
2.6 I 3.1
3.0 | 3.5
7.8
22.5
21.7
19.1
12.14
13.7 I 9.8 I 13.5 1 83.5
CATAS-
STROPH
RLS
0.0
0.«4
1.3
O.U
0.0
2.1
-------
REPRESENTATIVE RELEASE PROFILES FOR TREATMENT TANKS
Treatment Tank Technology 1
Above-ground, on cradles, carbon steel, closed, 2,300 gallons
Basel ine. Toluene
PROF 1 LE WF lf*HT
VOLUME RELEASED (cubic meters)
YRS. 1- 5 YRS. 6-10 | YRS. 11-15 YRS. 16-20
1 .103 18.116
2 .277 10.801
3 .121 11.660
1 .100 10.622
5 .100 7.060
WEIGHTED
AVERAGE 1.001 13.172
26.980 | 8.956 16.070
11.352 13.850 19.272
15.609 7.157 15.536
25.310 10.136 16.300
15.911 21.010 16.950
1
20.028 I 11.156 17.019
TOTAL
70. 122
55.275
19.962
62.668
60.931
61.976
Treatment Tank Technology 1
Above-ground, on cradles, carbon steel, closed, 2,300 gallons
Secondary Containment, Toluene
1
rtr>f\rttr~ 1 t tt~ t r*itr
VOLUME RELEASED (cubic meters)
\ YRS. 1- 5 YRS. 6-10 | YRS. 11-15 YRS. 16-20
1 I .821 0.000
2 .152 0.000
3 .008 | 0.000
1 .008 t 0.000
5 .001 10.100
6 .001 0.000
WEIGHTED | I
AVERAGE I 1.000 | 0.012 |
0.000 | 0.000 0.000
0.000 | 0.000 0.113
10.100 | 0.000 0.101
0.000 | 10.100 0.000
0.000 0.000 0.000
0.000 0.000 10.1OO
1
0.083 1 0.083 0.110
TOIAL
0.000
0.113
1O.501
10.100
10.100
10.100
0. JIB
Treatment lank Technology 2
Above-ground, on cradles, carbon steel, open, 2,300 gallons
Baseline, Aqueous
rtr*f-\ r i t r itrioiiT
VOLUME RELEASED (cubic meters)
YRS. 1- 5 YRS. 6-10 YRS. 11-15 YRS. 16-20
1 .837 0.537
2 .063 0.167
3 .011 0.659
1 .011 1.811
5 .018 15.277
WEIGHTED
AVERAGE 1.000 0.856
1:183 0.111 0.838
9.127 0.611 0.288
2.261 2.239 8.798
2.210 0.816 8.131
2.176 2.276 1.135
1
1/305 0.301 1.500
TOTAL
5.669
10.526
1 3 . 960
13.361
21.161
6.963
>
00
-------
REPRESENTATIVE RELEASE PROFILES EOR TREATMENT TANKS
treatment Tank Technology 2
Above-ground, on cradles, carbon steel, open, 2,300 gallons
Secondary Containment, Aqueous
PROF 1 LE
1
2
3
t|
5
6
WEIGHTED
AVERAGE
UF IHHT
.8(10
.132
.008
.008
.008
.OOM
1.000
YRS. 1- 5
0.000
0.000
6.960
0.000
0.000
0 . 000
0.056
VOLUME
YRS. 6-10
0.000
0.000
0.000
1 1 . 100
0.000
0.000
0.089
RELEASED (cubic
YRS. 11-15
0.000
0.708
0.000
0.000
11.100
0.000
0.182
: meters)
YRS. 16-20
0.000
0.000
0.000
0.000
0.000
11.100
O.OW
TOIAL
0.000
0.708
6.960
1 1. 100
11.100
11.100
0.371
Treatment Tank Technology i|
Above-ground, ongrade. carbon steel, 60,000 gallons
Baseline, Aqueous
PROP 1 1 F
i
2
3
14
5
WEIGHTED
AVERAGE
ur i PUT
.S'lO
.0714
.066
.012
.008
1.000
YRS. 1- 5
16.600
2145.000
0.076
0.162
1123.000
Ml. 065
VOLUME
YRS. 6-10
17.350
291 . 500
133.630
2.1408
1251.000
55.223
RELEASED (cubic
YHS. 11-15
17.<470
285.900
17.23M
292.070
12<48.000
50.1458
: meters)
YRS. 16-20
16.110
256. 100
O.M23
301.7014
1113. 000
45.036
IOIAL
67.530
1081.500
151.363
596. 3«43
'4735.000
191.782
>
ro
Treatment Tank Technology (4
Above-ground, ongrade, carbon steel, 60,000 gallons
Corrosion Protection, Aqueous
rKOr 1 Lt
1
2
3
14
5
WEIGHTED
AVERAGE
.926
.033
.029
.008
.0014
1.000
YRS. 1- 5
2M.710
0.189
351.500
962.815
102.113
141.192
VOLUME
YRS. 6-10
28.230
0.871
387. 100
1527.000
0.896
M9.615
RELEASED (cubic
YRS. 11-15
27.590
0.005
391. MOO
1520.000
562.000
51.3914
: meters)
YRS. 16-20
214.1480
1140.690
3M8.300
1370.000
10614.000
, 52.628
TOTAL
105.010
1141.755
H481.299
5379.812
1729.009
1914.829
-------
REPRESFNIAfIVE RELEASE PROFILES FOR TREATMENT TANKS
Treatment Tank Technology 14
Above-ground, ongrade, carbon steel, 60,000 gallons
Partial Containment with Ground-water Monitoring, Aqueous
PROF 1 LE
1
2
3
o
Treatment Tank Technology 5
In-ground, concrete, 3,700 gallons
Baseline, Aqueous
1
2
3
14
5
WEIGHTED
AVERAGE
.979
.008
.0014
.0014
.0014
0.999
YRS. 1- 5
5.810
12.517
150.600
332.200
92.195
8.088
VOLUME
YRS. 6-10
5.9W
1.655
165.600
3614.000
3.0H4
'
7.959
RELEASED (cubic
YRS. 11-15
6.14140
18.237
165.700
3614.600
0.961
8.576
: meters)
YRS. 16-20
8.803
187.300
H49.700
329.000
1.068
'
12.036
TOTAL
26.993
219.709
631.600
1389.800
97.238
36.658
-------
REPRESENTATIVE RELEASE PROFILES EOR TREATMENT TANKS
Treatment Tank Technology 5
In-ground, concrete, 3,700 gallons
Partial Containment with Ground-water Monitoring, Aqueous
PKOr 1 Lt
1
2
3
'1
5
WEIGHTED
AVERAGE
WE IGII1
.979
.008
.001)
.001
.001
0.999
YRS. 1- 5
0.795
150.100
17.120
0.795
330.700
3.377
VOl UME
YRS. 6-TO
0.795
165.500
19. 100
0.795
361.000
3.638
RELEASED (cubic
YRS. 11-15
0.795
165.500
273.000
202.172
36<4.000
5.160
: meters)
YRS. 16-20
3.155
119.000
282 . 200
290 . 300
328.100
7.881
T01AL
5.510
630.100
591.719
191.362
1387.099
20.360
Treatment Tank Technology 5
In-ground, concrete, 3,TOO gallons
Secondary Containment, Aqueous
PROf ILE
1
2
3
1
5
6
WE IGHTED
AVERAGE
WEIGHT
.901
.072
.008
.008
.001
.001
1.000
VOLUME RELEASED (cubic meters)
YRS. 1- 5
0.000
0.000
0.000
0.000
0.000
0.000
0.000
YRS. 6-10 1 YRS. 11-15
0 . 000
0.092
17.900
0.000
0.135
0.000
0.000
0.000
0.000
1 7 . 900
63.500
0.000
1
0. 150 I 0.397
YRS. 16-20
0.000
0.000
0.000
o.ooo
0 . 000
1 7 . 900
0.072
1OIAI
0.000
0.092
1 7 . 900
17.900
63.635
1 7 . 900
0.619
I
co
Treatment Tank Technology 6
In-ground, carbon steel, 3,700 gallons
Baseline, Aqueous
PROFILE
1
2
3
1
5
WEIGHTED
AVERAGE
1
WEIGHT
.967
.017
.008
.001
.001
1.000
VOLUME
YRS. 1- 5
10.310
368.900
796 . 000
2.121
10.150
.
22.690
YRS. 6-10
10.310
106.100
89 1 . 000
210.365
12. 190
21.921
RELEASED (cubic meters)
YRS. 11-15
1 1 . 790
106.000
907.000
971.000
311.710
30.702
YRS. 16-20
10.700
366.000
801 . 800
871.700
585.100
28.816
TOTAL
13.170
1516.999
3398.800
2058.189
922.150
107. 158
-------
REPRESENIATIVE RELEASE PROFILES FOR TREATMENT TANKS
Troatmont lank Technology 6
In-ground, carbon steel, 3,700 gallons
Corrosion Protection, Aqueous
PROF 1 LE
1
2
3
4
5
WEIGHTED
AVERAGE
ur i r.in
.983
.004
.004
.004
.004
0.999
YRS. 1- 5
2.140
817.400
0.120
585.700
368.800
9.192
VOLUME
YRS. 6-10
6.447
900.000
624 . 300
645.000
406 . 000
16.639
RELEASED (cubic
YRS. 11-15
15.080
900.000
925.000
652.000
406.000
26.356
: meters)
YRS. 16-20
22. 110
812.900
832.600
583.500
365.400
32.112
T01AL
45.777
3430.300
2382.020
2466.200
1546. 199
84.298
Treatment Tank Technology 6
In-ground, carbon steel, 3,700 gallons
Partial Containment with Ground-water Monitoring, Aqueous
PROP 1 1 r
i
2
3
4
5
6
WE IGHTED
AVERAGE
III IT 1 PUT
.954
.029
.004
.004
.004
.004
0.999
YRS. 1- 5
0 . 000
166.300
0.000
368.800
813. 000
545.100
11.730
VOLUME
YRS. 6-10
0.036
183.000
0.000
406 . OOO
895.000
600 . 000
12.945
RELEASED (cubic
YRS. 11-15
0.498
183.000
0.000
406.000
895.000
600 . 000
13.386
: meters)
YRS. 16-20
1. 172
164.700
0.000
365.400
808.500
540.200
12.751
10IAL
1. 706
69 7 . 000
0.000
1546.199
3411.500
2285. 300
50.812
Treatment Tank Technology 6
In-ground, carbon steel, 3,700 gallons
Secondary Containment, Aqueous
1
2
3
4
5
6
WEIGHTED
AVERAGE
.844
.136
.008
.004
.004
.004
1.000
YRS. 1- 5
0.000
0.000
1 7 . 900
0 . 000
0.000
0.000
0.143
VOLUME
YRS. 6-10
0.000
0.000
0.614
23.100
0.000
0.188
0.098
RELEASED (cubic
YRS. 11-15
0.311
0.000
0.000
0.010
70.100
44.520
0.721
: meters)
YRS. 16-20
0.009
0.000
0.307
0.000
0.004
0.170
1
0.011
TOTAL
0.320
0.000
18.821
23.110
70.104
44.878
0.973
-------
REPRESENTATIVE RELEASE PROFILES FOR TREATMENT TANKS
Trea mient lank Technology 7
In-ground, stainless steel, 3,700 gallons
Baseline, Aqueous
PROFILE
T
2
3
1
5
WEIGHTED
AVERAGE
WE IGHT
.912
.011
.008
.001
.001
0.999
YRS. 1- 5
6.800
1 36 . 900
369.300
515.100
816.000
20 . 1 1 7
VOLUME
1 YRS. 6-10
6. 160
150.600
106 . 900
600.000
895.000
21.212
RELEASED (cubi(
YRS. 11-15
6.580
150.000
106.600
607.000
897.000
21.617
: meters)
YRS. 16-20
5.203
110.000
365.500
513.100
806 . 600
18.961
TOTAL
21.713
577.500
1518.299
2295 . 200
3111.600
82.211
Treatment Tank Technology 7
In-ground, stainless steel, 3,700 gallons
Corrosion Protection, Aqueous
PROF ILE
1
2
3
1
5
6
WEIGHTED
AVERAGE
WE IGHT
.963
.012
.008
.008
.001
.001
0.999
YRS. 1- 5
7.280
516.100
155.025
828 . 000
0.000
0.217
21.129
VOLUME
1 YRS. 6-10
0.317
601 . 000
631.000
915.000
0.000
621 . 300
22.117
RELEASED (cubic
YRS. 11-15
3.627
602.000
621.000
906 . 000
0.000
926.000
1
26.661
: meters)
YRS. 16-20
0.206
510.100
559.000
815. 100
0.000
832.700
2 1 . 006
IOIAI
1 t . 160
2292.200
1969.025
3161.100
0.000
2383.217
1
91.513 I
I
OJ
OJ
Treatment Tank Technology 7
In-ground, stainless steel, 3,700 gallons
Partial Containment with Ground-water Monitoring, Aqueous
PROF 1 LE
1
2
3
1
5
6
WEI GH1 ED
AVERAGE
WE IGHT
.550
.380
.050
.008
.008
.001
1.000
YRS. 1- 5
0.000
0.000
110.300
111.700
368.800
813. 000
12.875
VOLUME
YRS. 6-10
O.'OOO
0.000
122.600
620.000
106.000
895.000
t
17.918
RELEASED (cubic
YRS. 11-15
0.000
0.000
126.500
620.000
106 . 000
895.000
18.113
: meters)
YRS. 16-20
0.165
0.000
113. 900
558.000
365.100
805.600
.
16.395
TOTAL
0.165
0.000
173.300
1912.700
1516.199
3108.600
65.301
-------
REPRESENIAIIVE RELEASE PROFILES FOR TREATMENT TANKS
rreatment Tank Technology 7
In-ground, stainless steel, 3,700 gallons
Secondary Containment. Aqueous
PROF 1 1 P
1
2
3
1
5
6
WEIGHTED
AVERAGE
ur i PUT
.812
.152
.016
.008
.008
.001
1.000
VOLUME
YRS. 1- 5 1 YRS. 6-10
0,000 1 0.315
0.000 | O.OOO
17.900 0.000
0.000 1 0.000
O.OOO I 0.000
0.000 1 0.003
1
0.286 1 0.256
RELEASED (cubic
YRS. 11-15
0.072
0.000
0.191
0.000
1.650
70.112
0.355
: meters)
YRS. 16-20
0.035
O.OOO
0.088
1 7 . 900
32 . 900
0.117
0.137
TOTAL
0.122
0.000
18.182
1 7 . 900
3M . 550
70.232
1.331
I
OJ
-------
REPRESENTATIVE RELEASE PROFILES FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 8 and Accumulation Tank Technology 17
Above-ground, on cradles, carbon steel, 5,500 gallons
Baseline, Aqueous and Toluene
PROF 1 1 P
i
2
3
<4
5
6
WEIGHTED
AVERAGE
.759
.127
.051
.012
.017
.001
1 .000
VOLUME
YRS. 1- 5 1 YRS. 6-10
1.730 | 1.316
0.926 I 1.3/8
3.620 | 8.677
0.658 1.080
0.000 I 0.000
13.131 I 25.100
1
1.697 I 1.763
RELEASED (cubic
YRS. 11-15
0.000
2.551
3.500
0.000
0.000
6.211
0.528
: meters)
YRS. 16-20
2.620
5.750
2.199
6.923
0 . 000
0.511
3.139
101AL
5.666
10.608
18.296
8.661
0.000
15.589
7.127
Storage lank Technology 8 and Accumulation Tank Technology 17
Above-ground, on cradles, carbon steel, 5,500 gallons
Secondary Containment-, Aqueous and Toluene
1
2
3
1
5
6
WEIGHTED
AVERAGE
.612
.311
.016
.012
.008
.008
1.000
YRS. 1- 5
0.000
0 . (100
10.1(00
O.OOO
O.OOO
0.000
0.166
VOLUME
YRS. 6-10
0.000
0 . 090
0.000
1O.100
0.021
0 . 000
0.156
RELEASED (cubic
YRS. 11-15
0 . 0(10
0 . 000
0.000
O.OOO
0.000
10.100
0.083
: meters)
YRS. 16-20
0.000
0.000
0.325
O.OOO
10.100
0 . 000
0.088
IOIAI
0.000
0.090
10.725
10.100
10.121
10.100
0.191
Storage Tank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210,000 gallons
Baseline, Toluene
PROF 1 LE
1
2
3
1
5
WEIGHTED
AVERAGE
Wl_ 1 On 1
.800
.117
.027
.022
.001
1.000
YRS. 1- 5
0.000
910.000
19580.000
1016.000
6662.000
777.130
VQLUME
YRS. 6-ilO
1.780
1911.000
19860.000
6870.000
13070.000
t
1026.390
RELEASED (cubic
YRS. 11-15
52.150
3361.000
20180.000
10510.000
21620.000
1107.518
: meters)
YRS. 16-20
287.800
5170.000
21630.000
11960.000
6521 . 398
1
1929.157
I TOTAL
311.730
11385.000
81550.000
36386.000
17876.398
5110.821
-------
REPRESENTATIVE RELEASE PROFILES EOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210,000 gallons
Baseline, Aqueous
PROE 1 LE
1
2
3
i|
5
WEIGHTED
AVERAGE
WE IGHT
.886
.066
.031
.013
.001
1.000
YRS. 1- 5
1.575
1116.000
16950.000
25980.000
9551.898
998.906
VOLUME
YRS. 6-10
. 16.653
3150.000
16950.000
26030.000
26310.000
1238.111
RELEASED (cubic
YRS. 11-15
279.100
6390.000
16960.000
26060 . 000
26160.000
1639.101
: meters)
YRS. 16-20
616 . 600
10210.000
17200.000
26110.000
26620.000
2227.837
TOTAL
976.927
21196.000
68060.000
101180.000
88911.875
6101.258
Storage Tank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210,000 gallons
Corrosion Protection, Toluene
PROF 1 LE
1
2
3
1
5
6
WEIGHTED
AVERAGE
vir i nil T
.810
.062
.053
.027
.013
.001
0.999
YRS. 1- 5
185.100
2/78.000
O.OOO
19550.000
3827.000
6662.000
931.969
VOLUME
YRS. 6-10
253.900
1326.000
0.000
19550.000
8120.000
1 3060 . 000
1167.137
RELEASED (cubit
I YRS. 11-15
321.100
6220 . 000
O.OOO
19560.000
11010.000
21620.000
1155.255
: meters)
YRS. 16-20
396.700
8170.000
0.000
19550.000
21600.000
6522.719
1693.108
101AL
1 160. 100
21791.000
0.000
78210.000
17587.000
17861.719
5217.169
Storage Tank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210,000 gallons
Corrosion Protection, Aqueous
PROFILE
1
2
3
1
5
6
WEIGHTED
AVERAGE
WEIGHT
.861
.053
.035
.031
.013
.001
1.000
VOLUME RELEASED (cubic meters)
YRS. 1- 5
215.600
0.000
2723.000
15250.000
25980.000
9551.898
1130.281
YRS. 6-10 I YRS. 11-15
308.900
0.000
5760.000
15250.000
26030.000
26310.000
f
1381.869
131.600
0.000
992O.OOO
15250.000
26060.000
26160.000
1637.171
YRS. 16-20
631.000
0.000
15250.000
15250.000
26110.000
26620.000
1
2000. 186
1
TOTAL |
1590.100
0.000
33653.000
61000.000
101180.000
88911.875
6152.801 |
-------
REPRESENTATIVE RELEASE PROI i i ., FOR STORAGE/ACCUMUI A!ION 1ANKS
Storage Tank Technology 9 arid Accumulation Tank Technology 18
Above-ground, ongrade. carbon steel, 210,000 gallons
Partial Containment with Ground-water Monitoring. Toluene
PROF 1 LE
1
2
3
14
5
6
WEIGHTED
AVERAGE
.
WE IGHT
.591
.320
.0<40
.040
.004
.OOM
0.999
YRS. 1- 5
0.000
0.000
1 7600 . 000
21495.000
114070.000
21650.000
958.679
VOLUME
YRS. 6-10
182.480
O.OOO
17600.000
5307.000
19575.000
21650.000
1201.025
RELEASED (cubit
YRS. 11-15
599.400
O.OOO
17600.000
8470.000
0.000
24650.000
1495.645
: meters)
YRS. 16-20
1439.000
0.000
17600.OOO
12690.OOO
0.000
13100.000
2114.448
TOTAL
2220.880
0.000
70400.000
28962 . OOO
33645.000
87O50.000
5769. 797
Storage Tank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210,000 gallons
Partial Containment with Ground-water Monitoring, Aqueous
PROF 1 1 f
i
2
3
4
5
6
WEIGHTED
AVERAGE
\jc i H14T
.583
.316
.061
.026
.009
.004
0.999
---._-._-.-___.
YRS. 1- 5
0.000
0.000
9950.000
19750.000
8580.000
68. 100
1197.942
VOLUME
YRS. 6-10
185.900
0.000
9950.000
19750.000
24900.000
5192.598
1473.700
RELEASED (cubic
YRS. 11-15
518.700
0.000
9950.000
19750.0OO
24900.000
26710.000
1753.792
: meters)
YRS. 16-20
1197.000
0.000
9990.000
19750.000
24900.000
27730. OOO
2155.760
1
IOIAL
1901.600
0 . 000 1
39840.000
79OOO.OOO
83280.000
59700.695
1
6581.191 I
Storage Tank Technology 9 and Accumulation Tank Technology 18
Above-ground, ongrade, carbon steel, 210.000 gallons
Secondary Containment, Aqueous and Toluene
1
2
3
4
5
6
WEIGHTED
AVERAGE
.680
.292
.012
.008
.004
.004
1.000
YRS. 1- 5
0.000
6.300
0.000
0.000
0 . 000
0.000
1.840
VOLUME
YRS. 6-10
0 . 000
6.343
397.000
0.000
0.000
0.000
6.616
RELEASED (cubic
YRS. 11-15
0.000
0.159
O.OOO
397.000
397.000
397.000
6.398
: meters)
YRS. 16-20
0.000
0.000
0.000
0.000
0.010
0.000
0.000
1 TOTAL
0.000
12.802
397.000
397.000
397.010
397.000
14.854
-------
LAII1151 i A >ntinned)
RFPRFSfN1ATIVE RELEASE PROFILES I OR STORAGE/ACCUMULATION TANKS
Storage lank technology 10 and Accumulation lank Technology 19
Underground, carbon steel, 1,000 gallons
Baseline, Tolnene
PROF 1 LE
1
2
3
1
5
6
WEIGHTED
AVERAGE
\jr i p|JT
.366
.218
.206
.181
.021
.008
1.000
«_ .- *
YRS. 1- 5
10.500
0.225
0.256
16.510
19.M20
0.000
7.311
VOLUME
YRS. 6-10
0.191
5.180
0.280
16.120
3 1 . 600
0.000
1.892
RELEASED (cubic
YRS. 11-15
5.050
16.160
1.816
0.000
13.970
0.000
13.827
2 meters)
1 YRS. 16-20
0.000
1 1 . 000
37.230
0.002
50.360
0.000
11.125
T01AL
15.711
62.565
12.582
32.932
115.350
0 . 000
37.185
Storage lank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel, 1,000 gallons
Baseline, Aqueous
1
2
3
1
5
6
WEIGHTED
AVERAGE
.262
.221
.212
.201
.096
.001
0.999
YRS. 1- 5
0.000
0.000
0.188
10.710
15.350
0.000
6.578
VOLUME
YRS. 6-10
0.000
19.177
0.632
39.610
0.077
0.000
12.532
RELEASED (cubic
YRS. 11-15
0.003
52.370
19.310
0.000
0.005
0.000
15.669
: meters)
YRS. 16-20
5.315
0.000
19.630
1.911
0.396
0.000
12.357
10TAL
5.318
71.81?
69.760
52.291
15.828
0.000
17.136
>
O>
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel, 1,000 gallons
Corrosion Protection, Toluene
PROf 1 LE
1
2
3
1
5
6
WEIGHTED
AVERAGE
HL 1 UH 1
.806
.068
.055
.012
.017
.013
1.001
YRS. 1- 5
0.190
21 . 900
0.000
0.002
20.270
32.110
2.851
VOLUME
YRS. 6-10
0.190
0.000
0.000
0.005
30.670
59.020
1.681
RELEASED (cubic
YRS. 11-15
0.190
0.000
0.000
0.151
13.350
1 7 . 020
1.372
: meters)
YRS. 16-20
0.190
0.000
0.000
27.111
56.100
0 . 000
2.187
I TOTAL
1.960
21 . 900
0.000
27.569
150.390
108. 180
8.391
-------
REPRESENTATIVE RELEASE PROFILl
JR STORAGE/ACCUMULAlION TANKS
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel, 1,000 gallons
Corrosion Protection, Aqueous
poor 1 1 F
1
2
3
1
5
6
WEIGHTED
AVERAGE
V/F 1 PUT
.762
.092
.051
.051
.025
.012
0.999
VOLUME
YRS. 1- 5 1 YRS. 6-10
0.519 I 0.510
33.670 | 0.000
0.000 | 0.000
0.000 | 0.000
15.700 | 22.380
0 . 000 I 3 . 356
1
3.886 I 1.011
RELEASED (cubic
YRS. 11-15
0.565
1.570
0.000
1.121
31.170
37.790
1.885
; meters)
YRS. 16-20
0.587
0.000
0.000
25.210
10.060
0.000
2.812
TOTAL
2.211
35.210
0.000
26 . 661
109.310
11 . 116
9.593
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel, 1,000 gallons
Secondary Containment, Toluene
PROF 1 1 F
i
2
3
1
5
6
WEIGHTED
AVERAGE
UF 1 PUT
.700
.280
.008
.001
.001
.001
1.000
YRS. 1- 5
0.001
0.000
7.571
0.018
15. 160
0.020
0.122
VOLUME RELEASED (cubic
YRS. 6-10 | YRS. 11-15
0.000 | 0.020
0.000 I 0.000
0.002 0.162
7.577 I 0.000
0.000 1 0.000
0.000 I 0.000
1
0.030 I 0.018
: meters)
YRS. 16-20
0.000
0 . 000
0.000
0 . 000
0.000
7.570
0.030
TOTAL
0.021
O.OOO
8.038
7.595
15. 160
f.590
0.200
I
CO
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon steel, 1,000 gallons
Secondary Containment, Aqueous
rKOt 1 Lt
1
2
3
1
5
6
WEIGHTED
AVERAGE
Wt lv»nl
.700
.280
.008
.001
.001
.001
1.000
YRS. 1- 5
0.001
O.OOO
7.571
0.018
15.160
0.020
0. 122
VOLUME
YRS. 6-10
0.000
O.OOO
0.002
7.5/7
0.000
0.000
0.030
RELEASED (cubic
YRS. 11-15
0.020
0.000
0.360
0.000
0.000
0.000
0.017
: meters)
YRS. 16-20
0.000
0.000
0.000
0.000
0.000
7.570
0.030
TOTAL
0.021
0.000
7.936
7.595
15- 160
7.590
0.200
-------
i nucu ;
lUI'lUSENlAriVE RELEASE PROFILE fOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 10 and Accumulation Tank Technology 19
Under-ground, carbon steel, U.OOO gallons
leak Testing with Ground-water Monitoring, Toluene
PROf 1 1 F
i
2
3
14
5
6
WEIGHTED
AVERAGE
.937
.025
.017
.008
.008
.00i|
0.999
YRS. 1- 5
0.375
<4.'l6>4
2.8'IQ
0.000
15. 100
16.900
0.700
VOLUME
YRS. 6-10
0.000
1.050
10.100
0 . 000
0.000
0.000
0. 198
RELEASED (cubic meters)
YRS. 11-15 1 YRS. 16-20
0.525 1 0.000
7.231 I 0.787
O.M30 0.000
0.000 0.000
0.000 0.000
1.800 I 0.322
1
0.687 I 0.021
TOTAL
0.900
13.532
13.370
0 . 000
15. 100
19.022
1.606
Storage Tank Technology 10 and Accumulation Tank Technology 19
Underground, carbon stoel, i|,000 gallons
leak Testing with Ground-water Monitoring, Aqueous
1
2
3
t|
5
6
WEIGHTED
AVERAGE
.950
.012
.012
.008
.008
.008
0.998
YRS. 1- 5
0. I'lO
1.820
0.001
0.000
15.100
20.221
O.M37
VOLUME
YRS. 6-10
0.000
15.100
0.000
0.000
0.000
0.000
0.181
RELEASED (cubit
YRS. 11-15
0.273
O.OOO
0.966
0.000
0.000
0.85'!
0.278
: meters)
YRS. 16-20
0.000
0 . 000
11. 120
0.000
0.000
7.078
0.190
roiAL
0.1413
16.920
12.087
0.000
15. 100
28.153
1.086
I
*-
o
Storage Tank Technology 11 and Accumulation Tank Technology 20
Underground, fiberglass reinforced plastic, 14,000 gallons
Baseline, Toluene
PKUr 1 Lt
1
2
3
l|
5
6
WEIGHTED
AVERAGE
WL 1 On 1
.866
.052
.026
.022
.017
.017
1.000
YRS. 1- 5
0.150
0.000
5.062
0.«»85
0.000
7.588
0.661
VQLUME
YRS. 6-.10
1. 150
0.000
0.012
O.M55
0.000
0.005
1.006
RELEASED (cubic
YRS. 11-15
0.000
0.000
2.930
O.H55
15.100
0.005
0.3U3
: meters)
YRS. 16-20
0.780
0.000
0.000
17.191
0.000
0.005
I
1.05«4
TOTAL
2.380
0.000
8.001
18.586
15.100
7.603
3.06U
-------
_ .. ^ j \ <-UHL i iiuea j
Hi,. LSENTATIVE RELEASt PROFILES FOR SIORAGE/ACCi.-riULAT ION TANKS
Storage Tank Technology 11 and Accumulation Tank Technology 20
Underground, fiberglass reinforced plastic, 1,000 gallons
Baseline, Aqueous
ppnr i i F
i
2
3
1
5
6
WEIGHTED
AVERAGE
UF i r*iiT
.836
.052
.017
.026
.022
.017
1 . 000
YRS. 1- 5
0.226
0.000
7.581
5.062
0.366
0.085
0.686
VOLUME
YRS. 6-10
0.387
0.000
0.005
0.012
0.325
0.085
0.333
RELEASED (cubic
YRS. 11-15
0.390
0.000
0.005
2.930
0.325
15.168
0.667
: meters)
YRS. 16-20
0.390
0.000
0.005
0.000
17.135
0.000
0.703
TOTAL
1.393
0.000
7.599
8.001
18.151
15.338
2.390
Storage Tank Technology 11 and Accumulation Tank Technology 20
Underground, fiberglass reinforced plastic. 1,000 gallons
Secondary Containment, Toluene
rKOr lit
1
2
3
1
5
6
WEIGHTED
AVERAGE
Wt 1 l»n 1
.680
.300
.008
.001
.001
.001
1.000
YRS. 1- 5
0.020
0.000
0.015
0.000
15.16O
0.018
0.071
VOLUME
YRS. 6-10
0.003
0.000
0.000
10.093
0.000
7.578
0.073
RELEASED (cubic
YRS. 11-15
0.000
0.000
7.573
0.000
0.000
O.OOO
0.061
: meters)
YRS. 16-20 I 10TAL
0.000 I 0.023
0.000 I 0.000
0.007 7.595
0.000 10.093
0.000 15.160
O.OOO 7.596
1
0.000 1 0.208
Storage Tank Technology 11 and Accumulation Tank Technology 20
Underground, fiberglass reinforced plastic, 1,000 gallons
Secondary Containment, Aqueous
PROFI LE
1
2
3
1
5
6
WEIGHTED
AVFRAGE
WE IGHT
.676
.301
.008
.001
.001
.001
1.000
YRS. 1- 5
0.020
0.000
0.015
0.000
15.160
0.018
0.071
VOLUME
YRS. 6-10
0.003
O.OOO
0.000
10.093
0.000
7.578
0.073
RELEASED (cubic
YRS. 11-15
0.000
0.000
7.573
0.000
0.000
0.000
0.061
: meters)
YRS. 16-20
0.000
0.000
0.007
0.000
0.000
0.000
0.000
TOTAL
0.023
0.000
7.595
10.093
15.160
7.596
0.208
-------
REPRESENTATIVE REI EASE PROF 11 <
in. i nueui
OR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 11 and Accumulation lank Techno logy 20
Underground, fiberglass reinforced plastic. 1.000 gallons
leak Testing with Ground-water Monitoring, Toluene
PROF ILE
1
2
3
1
5
6
WEIGHTED
AVERAGE
WE 1 GH T
.787
.089
.031
.031
.030
.026
1 . 000
YRS. 1- 5
0.000
0.000
0.000
10.556
0.000
0.000
0.359
VOLUME
YRS. 6-10
0.000
0.000
7.570
0.000
6.830
0.000
0.162
RELEASED (cubic
YRS. 11-15
0.232
0.000
0.000
2.080
0.000
O.OOO
0.253
: meters)
YRS. 16-20
0.000
0 . 000
0.000
0 . 000
0.000
9.290
0.212
TOTAL
0.232
0.000
7.570
12.636
6.830
9.290
1.316
Storage Tank Technology II and Accumulation Tank Technology 20
Underground, fiberglass reinforced plastic, 1,000 gallons
Leak Testing with Ground-water Monitoring, Aqueous
PROF 1 LE
1
2
3
1
5
6
WEIGHTED
AVERAGE
WF IP.HT
.
.787
.089
.031
.031
.030
.026
1.000
YRS. 1- 5
0.000
0.000
0.000
10.556
0.000
0.000
0.359
VOLUME
YRS. 6-10
0.000
0.000
7.570
0.000
6.830
0.000
0.162
RELEASED (cubic
YRS. 11-15
0.232
0 . 000
0.000
2.080
0.000
0.000
0.253
: meters)
YRS. 16-20
0.000
0.000
0.001
0.000
0.000
8.170
0.220
TOTAL
0.232
0 . 000
7.5/1
12.636
6.830
8.170
1.295
Storage Tank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel, 1,000 gallons
Ba seIi ne, T oIuene
1
2
3
1
5
6
WEIGHTED
AVERAGE
.820
.081
.071
.013
.008
.001
1.000
YRS. 1- 5
0.505
0.000
0.000
0.000
0.275
50.990
0.620
VOLUME
YRS. 6-10
0.505
0.000
0.000
5.270
0.275
66.000
0.719
RELEASED (cubic
YRS. 11-15
0.505
0.000
1.327
11.160
15.310
15.002
1.182
: meters)
YRS. 16-20
0.505
0.000
36. 100
1 7 . 900
0.000
0.010
3.210
1 TOTAL
2.020
0.000
10.127
67.630
15.860
132.002
6.061
-------
RLPRESFNTATIVE RELEASE PROI
JR STORAGE/ACCUMULAII ON TANKS
Storage Tank Technology 12 and Accumulation Tank t«..hnology 21
Underground, stainless steel, 1,000 gallons
Baseline, Aqueous
1
2
3
1
5
6
WEIGHTED
AVERAGE
.820
.081
.050
.029
.013
.001
1.000
YRS. 1- 5
0.380
0.000
0.000
0.005
0.000
36.1)30
0.157
VOL UME
YRS. 6-10
0.380
0 . 000
0.000
1.989
1.510
16.760
0.615
RELEASED (cubic
YRS. 11-15
0.380 1
0.000
0.979
20.O80
31.930
57.000
1.586
meters)
YRS. 16-20
0.380
0.000
23.760
38.810
58.190
67.100
3.651
TOTAL
1.520
0.000
21.739
60.881
91.630
207.590
6.3IO
Storage Tank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel, 1,000 gallons
Corrosion Protection, Toluene
PROF 1 LE
1
2
3
1
5
6
WEIGHTED
AVERAGE
WE IGHT
.883
.079
.017
.013
.001
.001
1.000
YRS. 1- 5
0.507
0.000
0.601
0.000
16.900
50.990
0.730
VOLUME RELEASED (cubic
YRS. 6-10 1 YRS. 11-15
0.510 I 0.510
0.000 0.000
0.000 7.570
10.300 0.381
0.000 I 10.100
66.000 I 15.002
1
0.818 1 0.681
: meters)
YRS. 16-20
0.510
0.000
0.000
0.180
0.322
0.010
0.151
TOTAL
2.037
0.000
8. 171
10.861
27.322
132.002
2.716
Storage Tank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel, 1,000 gallons
Corrosion Protection, Aqueous
PROF ILE
1
2
3
1
5
6
WE IGHTED
AVERAGE
WEIGHT
.866
.079
.029
.013
.008
.001
0.999
YRS. 1- 5
0.380
0.000
0.003
0.681
0.000
36.130
0.181
VOLUME
YRS. 6-10
0.380
0.000
6.123
0.010
1.350
16.760
0. 705
RELEASED (cubic
YRS. 11-15
0.380
0.000
0.005
31.972
10.100
57.000
1.093
: meters)
YRS. 16-20
0.380
0.000
0.005
7.110
0.001
67.100
0.695
TOTAL
1.520
0.000
6. 136
13.073
11.151
207.590
2.976
-------
HI I'M StNTATIVE RELEASE PROM
OR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 12 and Accumulation Tank technology 21
Underground, stainless steel, 1,000 gallons
Secondary Containment, Toluene
PROF i i r
1
2
3
1
5
6
WEIGHTED
AVERAGE
.6'lO
-3l|0
.008
.001
.OO'I
.OOU
1.000
YRS. 1- 5
0.001
0.000
0.020
0.018
7.571
15.160
0.092
VOLUME
YRS. 6-10
0.000
0.000
0.000
7.570
0.002
0.000
0.030
RELEASED (cubic meters)
YRS. 11-15 1 YRS. 16-20
0.020 I 0.000
O.OOO I 0.000
0.000 I 7.570
0.000 0.000
0.012 0.000
0.000 1 O.OOO
1
0.013 1 0.061
TOTAL
0.021
0.000
7.590
7.588
7.588
15.160
0.196
Storage lank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel, 1.0OO gallons
Secondary Containment, Aqueous
1
2
3
M
5
6
WEIGHTED
AVERAGE
.636
.311
.008
.OOU
.OO'I
.001
1.000
YRS. 1- 5
0.001
0.000
0.015
0.018
7.5/1
15. 160
0.092
VOLUME
YRS. 6-10
0.000
0.000
0.000
7.570
0.002
0.000
0.030
RELEASED (cubic
YRS. 11-15
0.020
0.000
7.573
0.000
0.012
0.000
0.073
: meters)
YRS. 16-20
0.000
0.000
0.007
0.000
0.000
0.000
0.000
IOIAL
0.021
0 . 000
7.595
7.588
7.588
15.160
0.195
Storage Tank Technology 12 and Accumulation Tank Technology 21
Underground, stainless steel. 1,000 gallons
Leak Testing with Ground-water Monitoring, Toluene.
1
PROF 1 LE I
1
1
2
3
1
5
6
WEIGHTED
AVERAGE
Wt 1 Un 1
.8'! 3
.070
.029
.025
.017
.017
1.001
YRS. 1- 5
0.000
0.000
0.000
0.000
0.000
0.001
0.000
VOLUME
YRS. 6-10
0.000
0.000
0.000
0.000
2.990
7.570
0.180
RELEASED (cubic
YRS. 11-15
0.116
0.000
0.000
5.050
0.000
0.000
0.219
: meters)
YRS. 16-20
0.000
0.000
2.780
0.328
7.560
0.000
0.217
1 TOTAL
0. 116
0.000
2.780
5.378
10.550
7.571
0.616
-------
Rtl > < MNIAT IVE RELEASE PROMS .. fOR STORAGE/ACCUMULAT ION TANKS
Storage Tank Technology 12 and Accumulation T.ink technology 21
Underground, stainless steel, 1,000 gallons
leak Testing with Ground-water Monitoring, Aqueous
PROf I LE
1
2
3
>l
5
6
WEIGHTED
AVERAGE
Ljr 1 f*L| T
.839
.070
.058
.017
.012
.001
1.000
YRS. 1- 5
1 .220
0 . 000
0.000
2.183
1.020
0 . 000
1.073
VOLUME
YRS. 6-10
0.000
0.000
0.160
0.010
0.000
15.100
0.070
RELEASED (cubit
YRS. 11-15
0.000
0.000
0.701
0.688
0.165
0.000
0.051
: meters)
YRS. 16-20
0.012
0.000
6.190
2.820
1 1 . 009
0.000
0.567
TOTAL
1.232
0.000
7.351
5.701
12. 191
15. 100
1.761
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-grounU, concrete, 2,100 gallons
Basel i tie, To I uerie
1
2
3
1
5
WEIGHTED
AVERAGE
.725
.098
.078
.053
.015
0.999
YRS. 1- 5
0.310
0.378
1 . 862
0.310
0.816
0.181
VOLUME
YRS. 6-10
1.332
1.651
0.310
0.310
1.212
1.227
RELEASED (cubic
YRS. 11-15
0.368
2.311
0.310
2.522
3.152
0.796
: meters)
YRS. 16-20
0.310
1.357
2.992
0.310
3.192
0.775
IOTAL
2 . 380
5.700
5.531
3.512
8.372
'
3.280
>
Oi
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, concrete, 2,100 gallons
Baseline. Aqueous
rKOr 1 Lt
1
2
3
1
5
WEIGHTED
AVERAGE
.725
.098
.078
.066
.033
1.000
YRS. 1- 5
0.569
0.230
0.263
0.535
1.155
0.529
VOLUME
YRS. 6-10
0.'230
0.279
0.230
0.160
0.213
0.-250
RELEASED (cubic
YRS. 11-15
1.116
0.230
2.936
0.259
3.331
1.188
: meters)
YRS. 16-20
0.231
0.813
0.230
2.831
0.230
0.163
TOTAL
2.116
1.582
3.659
1.088
1.959
2.130
-------
.,Lr,u otNlATIVE RELEASE
FOR STORAGE/ACCUMULATION TANKS
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, concrete, 2,100 gallons
Partial Containment with Ground-water Monitoring, Toluene
PROF 1 1 F
1
2
3
1
5
WEIGHTED
AVERAGE
WF 1 PUT
.943
.024
.012
.012
.008
0.999
YRS. 1- 5
0.3110
0.3140
0.340
0.310
0.340
0.340
VOLUME
YRS. 6-10
0.340
0.340
4.312
0.340
4.312
0.419
RELEASED (cubic
YRS. 11-15
0.340
2.992
0.340
4.312
0.340
0.451
: meters)
YRS. 16-20 1 TOTAL
0.340 I 1.360
0.340 4.012
0.340 | 5.332
0.340 | 5.332
0.340 | 5.332
1
0.340 I 1.549
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, Concrete, 2,100 gallons
Partial Containment with Ground-water Monitoring, Aqueous
1
2
3
4
5
WEIGHTED
AVERAGE
.943
.024
.012
.012
.008
0.999
YRS. 1- 5
0.230
0.230
0.230
0.230
0.230
0.230
VOLUME
YRS. 6-10
0.230
0.230
4.204
0.230
4.204
0.309
RELEASED (cubit
YRS. 11-15
0.230
2.884
0.230
4.204
0.230
0.341
: meters)
YRS. 16-20
0.230
0.230
0.230
0.230
0.230
0.230
TOTAL
0.920
3.574
4.894
4.894
4.894
1.110
Storage Tank Technology 13 and Accumulation Tank Technology 22
In-ground, concrete, 2,100 gallons
Secondary Containment, Toluene
I
rKOr 1 LL
1
2
3
4
5
6
WEIGHTED
AVERAGE
.848
.120
.016
.008
.004
.004
1.000
YRS. 1- 5
0.000
0.000
0.000
0.000
0.001
0.000
0.000
VOLUME
YRS. 6-10
0.000
0.015
3.970
0.011
0.000
0.000
0.065
RELEASED (cubic meters)
YRS. 11-15 I YRS. 16-20
0.000 | 0.000
0.010 | 0.000
0.008 0.000
3.970 0.000
0.000 3.970
0.000 3.970
1
0.033 1 , 0.032
TOTAL
0.000
0.025
3.978
3.981
3.971
3.970
0.130
-------
RLI'RFSFNIATIVF RELEASE PROF I It
H STORAGE/ACCUMULArI ON TANKS
Storago Tank Technology 13 and Accumulation Tank Technology 22
In-ground, concrete. 2,101) gallons
Secondary Containment, Aqueous
PROFILE
1
2
3
it
5
6
WEIGHTED
AVERAGE
WEIGHT
.818
.120
.016
.008
. 001
.OO'I
VOLUME RELEASED (cubic meters)
YRS. 1- 5 1 YRS. 6-10 I YRS. 11-15
0.000
0.000
0.000
0.000
0.001
0.000
1
1.000 I 0.000
0.000
0.015
3.970
0.011
0.000
0.000
0.065
0.000
0.010
0.008
3.970
0.000
0.000
0.033
YRS. 16-20
0.000
0.000
0.000
0.000
3.970
3.970
0.032
TOTAL
0.000
0.025
3.978
3.981
3.971
3.970
0. 13O
Storage lank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Ba seIi ne, T oIuene
pttnr i i F
i
2
3
1
5
6
WEIGHTED
AVERAGE
IL/F 1 f^MT
.538
.188
.159
.101
.010
.005
1.001
.._._.._...._..
YRS. 1- 5
2.650
0.085
0.000
6.897
7.150
0.000
2.213
VOLUME
YRS. 6-10
0.001
0.177
0.000
1.613
0.052
0.000
0.560
RELEASED (cubic
YRS. 11-15
2.650
6.636
0.183
1.100
3.350
0.000
2.817
: meters)
YRS. 16-20
0.905
0.000
6.100
1.171
25 . 600
O.OOO
1.909
TO1AL
6.2(16
7.198
6.583
11. Ill
36.152
O.OOO
7.528
Storage Tank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Baseline, Aqueous
I
1
2
3
1
5
6
WEIGHTED
AVERAGE
.179
.202
.178
.117
.019
.005
1.000
YRS. 1- 5
6.660
2.222
0.000
18.860
15.286
0.000
6.136
VOLUME
YRS. 6-10
3.790
0.000
0.000
0.902
28.281
0.000
2.158
RELEASED (cubic
YRS. 11-15
0.000
0.082
8.322
3.018
1.010
0.000
1.871
: meters)
YRS. 16-20
0.122
8.610
1.111
2.721
0.023
0.000
(
2.326
I TOIAL
10.572
10.911
9.163
25.531
11.603
0. 000
12.791
-------
>tniAHVt RELEASE PROt
OR STORAGE/ACCUMULATION TANKS
Storage Tank Technolt,jy 11 add Accumulation Tank technology 23
In-ground, carbon steel. 2,100 gallons
Corrosion Protection, Toluene
PROFILE
1
2
3
1
5
6
Wt (CHILD
AVERAGE
WEIGHT
.781
.063
.063
.018
.038
.005
YRS. 1- 5
0.211
0.261
0.087
5.372
0.000
16.069
1
1.O01 1 0.526
VOLUME RELEASED (cubic meters)
YRS. 6-10
0.205
0 . 000
0.000
0.026
0.000
7.909
0.202
YRS. 11-15
0.072
0.593
5.380
0.000
0.000
O.O11
0.133
YRS. 16-20 1 TOTAL
0.637
3.370
0.519
0.366
0.000
0.128
0.763
1. 125
1.227
5.986
5.761
0.000
21.117
1.923
Storage Tank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Corrosion Protection, Aqueous
1
2
3
1
5
6
WEIGHTED
AVERAGE
.711
.171
.038
.038
.033
.005
1.002
YRS. 1- 5
0.000
0.000
O.OOO
0.081
11.670
12.826
0.152
VOLUME
YRS. 6-10
0.000
1.117
0.000
0.000
0.000
18.970
0.317
RELEASED (cubic
YRS. 11-15
0.106
2.650
0.000
7.710
0.000
0.010
0.831
: meters)
YRS. 16-20
0.101
0.329
0.000
0.519
0.000
0.128
0.366
TOTAI
0.510
1 . 126
0 . 000
8.313
11.670
31.931
1.996
I
p~
Oo
Storage Tank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Partial Containment with Ground-water Monitoring, Toluene
PROF ILE
1
2
3
1
5
6
WEIGHTED
AVERAGE
Wt 1 On 1
.557
.127
. 109
.095
.090
.023
1.001
YRS. 1- 5
0.767
0.000
0.000
0.000
0.000
13.376
0. 735
VOLUME
YRS. 6-10
0.000
0.000
0.000
9.700
0.188
13.370
1.273
RELEASED (cubic
YRS. 11-15
0.000
0.000
1.810
0.000
12.610
0.000
1.332
: meters)
YRS. 16-20
0.000
0 . 000
7.120
0.000
0.000
0.000
0 . 809
TOTAL
0.767
0.000
9.230
9.700
13.098
26.716
1. 119
-------
REPKU.IN1A1IVE RELEASE PROEILE:
STORAGE/ACCUMULATION TANKS
Storage Tank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Partial Containment with Ground-water Monitoring, Aqueous
PROF 1 LE
1
2
3
1
5
6
WEIGHTED
AVERAGE
UF 1 HHT
.176
.118
.131
.122
.096
.026
0.999
YRS. 1- 5
0.000
8.060
0.000
0.000
0.000
9.211
1.135
VOLUME
YRS. 6-10
0.000
1.910
11.170
O.OOO
0.000
30.370
2.579
RELEASED (cubic
YRS. 11-15
1.160
O.OOO
0.000
0.000
7.110
8.980
1.197
: meters)
YRS. 16-20
0.000
0.192
O.OOO
0.000
10.130
0.000
1.030
TOTAL
1.160
10.212
11.17O
O.OOO
17.810
18.561
6.511
Storage Tank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel. 2,100 gallons
Secondary Containment, Toluene
PROF 1 1 P
1
2
3
1
5
6
WEIGHTED
AVERAGE
ur ir.ur
.732
.201
.020
.020
.016
.008
1.000
YRS. 1- 5
0.000
0.000
0.000
0.335
0.000
0.000
0.007
VOLUME
YRS. 6-10 I
0.000
0.193
0.000
2.650
1 1 . 780
3.970
1
0.313
RELEASED (cubi<
YRS. 11-15
0.000
0.000
2.650
0.000
2.580
0.000
0.091
: meters)
YRS. 16-20
0.000
0.000
0.000
0.000
O.OOO
0.000
O.OOO
IOIAL
0.000
0.193
2.650
2.985
11.360
3.970
1
0.111 |
I
4>
VO
Storage Tank Technology 11 and Accumulation Tank Technology 23
In-ground, carbon steel, 2,100 gallons
Secondary Containment, Aqueous
1
2
3
1
5
6
WEIGHTED
AVERAGE
.732
. .200
.028
.021
.012
.001
1.000
YRS. 1- 5
O.OOO
O.OOO
0.312
0.000
0.000
0.000
0.009
VOLUME
YRS. 6-10
0.000
O.OOO
2.650
2.650
9.210
11.120
0.293
RELEASED (cubic
YRS. 11-15
0.000
0.105
0.000
0.000
10.290
21.150
0.230
: meters)
YRS. 16-20
0.000
0.000
O.OOO
0.000
0.000
0.000
0.000
TOTAL
0.000
O.105
2.962
2.650
19.500
32.570
0.532
-------
PRESENTATIVE RELEASE PROFILES FOR SQG (ANKS
Small Quantity Generator lank Technology 15
Above-ground, carbon steel, 200 gallons
Baseline, Aqueous and Toluene
PROF 1 LE
1
2
3
l|
5
6
WEIGHTED
AVERAGE
WE ICHT
.672
.097
.076
.063
.059
.034
1.001
YRS. 1- 5
0.000
0.126
0.378
0.135
0.115
0.000
0.058
VOLUME
YRS. 6-10
0.252
0.224
0.279
0.379
0.378
0.000
0.258
RELEASED (cubic meters)
YRS. 11-15 1 YRS. 16-20
0.000 0.126
0.5011 0.126
0.378 O.379
0.000 0.379
0.128 I 0.000
0.000 I 0.000
1
O.085 1 0.150
TOTAL
0.378
0.980
1.111
0.893
0.651
0.000
0.551
Small quantity Generator Tank Technology 15
Above-ground, carbon steel, 200 gallons
Secondary Containment, Aqueous and Toluene
PROFILE
1
2
3
1
5
6
WEIGHTED
AVERAGE
WEIGHT
.668
.288
.016
.012
.008
.008
1.000
VOLUME RELEASED (cubic meters)
YRS. 1- 5 1 YRS. 6-10
0.000
0.000
0.000
0.000
0.000
0.378
0.000
0.000
0.378
0.378
0.000
0.000
1
0.003 1 0.011
YRS. 11-15 1 YRS. 16-20 I 1OTAL
0.000
0.033
0.000
0.000
O.OOO
0.000
0.000
0.000
0 . 000
O.OOO
0.378
0.000
0.000
0.033
0.378
0.378
0.378
0.378
1 1
0.010 | 0.003 I 0.026
I
I/I
o
Small Quantity Generator Tank Technology 16
Underground, carbon steel, 200 gallons
Ba seIi ne, To Iuene
PROFILE
1
2
3
4
5
6
WEIGHTED
AVERAGE
WE 1 GHT
.735
.099
.076
.067
.018
.004
_________
YRS. 1
0
0
0
0
0
3
1
0.999 1 0
VOLUME RELEASED
- 5 1 YRS. 6-10 I YRS. 11
.000
.734
.000
.000
.000
.920
0.000
0.758
1.450
0.000
0.000
0.114
0
0
0
0
0
1
1 1
.088 1 0.186 I 0
(cubic meters)
-15
.000
.093
.756
.000
.000
.340
.072
YRS. 16-20 I TOTAL
1
0
0
0
0
0
0
.200
.000
.000
.631
.000
.017
1.
1.
2.
0.
0.
5.
1
.924 | 1.
200
585
206
631
000
391
270
-------
I'RtSENTATIVE RELEASE PROFILES FOR SQ( TANKS
Small Quantity Generator *.. > technology 16
Underground, carbon steel, 200 gallons
Ba seIi ne, Aqueous
PROFILE
1
2
3
U
5
6
WE 1 CM 1 EO
AVERAGE
WEIGHT
.585
.135
.127
.105
.031
.017
1.000
VOLUME
YRS. 1- 5 1 YRS. 6-10
0.581
O.OOO
0.000
0.000
2 . 355
0.000
0.000
0.000
2.373
1.038
0.000
0.000
1
0.
-------
Small Quantity Generator Tank technology 16
Underground, carbon steel. 200 gallons
Secondary Containment, Toluene
I
PROFILE
1
2
3
1|
5
6
WEIGHTED
AVERAGE
WEIGHT
.892
.072
.012
.008
.008
.008
1.000
YRS. 1- 5
0.000
0.000
0.000
0.000
0.000
0 . 000
o.ooo
VOLUME
YRS. 6-10
0.000
0.012
0.000
O.OOO
0.000
0.506
0.005
RELEASED (cubic meters)
YRS. 11-15 1 YRS. 16-20
0.000 I 0.000
O.OOO 0.000
0.379 O.OOO
O.OOO 0.378
0.378 0.000
0.000 I 0.000
1
0.008 I 0.003
1 TOIAL
0.000
0.012
0.379
0.378
0.378
0.506
0.016
Small Quantity Generator Tank Technology 16
Underground, carbon steel, 200 gallons
Secondary Containment. Aqueous
1
PROFILE | WEIGHT
1
1
2
3
1
5
6
.892
.076
.008
.008
.008
.008
YRS. 1
OOCOOO
WEIGHTED | |
AVERAGE | 1.000 1 O
VOLUME RELEASED
- 5 1 YRS. 6-10 1 YRS. 11
.QUO
.000
.000
.OOO
.000
.000
.000
0.000
0.012
0.000
0.000
0.361
O.OOO
0
0
0
0
0
0
1
0.001 I 0
(cubic meters)
-15
.000
.000
.000
.378
.000
.379
.006
YRS. 16-20 I TOTAL
oooooo
0.
000
000
378
000
000
000
0.
0.
0.
0.
O.
0.
1
003 I 0.
000
012
378
378
361
379
013
I
Ol
N)
Small Quantity Generator Tank Technology 16
Underground, carbon steel, ZOO gallons
Leak Testing with Ground-water Monitoring, Toluene
ppnr 1 1 r
t
2
3
14
5
6
WEIGHTED
AVERAGE
ur i PUT
.692
.093
.088
.081
.031
.013
1.001
YRS. 1- 5
O.OOO
0.000
O.OOO
0. 116
O.OOO
0.000
0.010
VOLUME
YRS. 6-10
0.000
1.153
1.935
0.000
0.000
0.000
0.278
RELEASED (cubic
YRS. 11-15
0.108
0.910
0.000
0.000
0.000
0.000
0.159
: meters)
1 YRS. 16-20
0.000
0.000
O.M01
1.261
1.723
0.000
0.195
1 TOTAL
0.108
2.063
2.336
1.380
1.723
O.OOO
0.611
-------
. . .,., , ,.,. ui uunu-waier nunitoring. Aqueous
PROFILE
1
2
3
<4
5
6
WEIGH1CD
AVERAGE
WEIGHT
.6149
.126
.087
.065
.061
.013
1.001
VOLUME RELEASED (cubic meters)
YRS. 1- 5 1 YRS. 6-10
0 . 000
0.000
2.063
0.000
0.007
0.000
0.180
0.000
0.000
0.000
0.000
0.778
0.000
O.OH7
VRS. 11-15
0.000
2.859
0.000
0.000
1.915
0.000
O.H77
YRS. 16-20
0. 121
0.000
0.000
2.912
0.000
0.000
0.268
10TAL
0.121
2.859
2.063
2.912
2.700
O.OOO
0.972
I
ui
LO
------- |