UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 204$o
April 15, 1988
SAB-EEC-88-029
OFFiCt OF
, THE ADMINISTRATOR
Honorable Lee M. Thosas
Administrator
U. S. Environmental Protection Agency
401 M Street, S, W.
Washington, D. C. 20460
Dear Mr. Thcmas:
The Science Advisory Board's (SAB) Environmental Engineering Committee
has completed its review of the Underground Storage Tank (UST) Release
Simulation Model developed by the office of Underground Storage Tanks for
the purpose of developing a Regulatory Impact Analysis of the requirements
proposed to regulate underground gasoline storage tanks* The Coranittee
reviewed the model at a public meeting on May 11, 1987.
Ihe Cccnmittee's major conclusions and recomnendations include the
following:
o The overall structure and design of the model is sound, but only
in the context of substantiating regulatory decisions on underground
gasoline storage tanks that have been made by other means.
o Because the UST model involves such a ccpplex calculation of tank
failures and impacts, it would be useful to compare the model results to
simpler, order-of-magnitude estimates based on a first-order characteri-
zation of tank ages and failure probabilities. The simplified and full
models should each be compared to data bases on tank failure that are
currently becoming available.
O The documentation of the model is not clear, and many of the
model's assumptions are not explicit. Ihe model code should be
documented to facilitate a wider use.
-------
- 2 -
Ihe Committee appreciates the opportunity to conduct this evaluation
and acknowledges the cooperation o£ EPA staff in its review, Wfe request
that the J^gency formally respond to the scientific advice provided in
th is _ report.
Sincerely ,
Norton Nelson
Chairman
Executive Committee
Raymdrid loehr
Chairman
Environmental Engineering Ccttmittee
-------
SAB-EEC-88-029
REVIEW OF THE
UNDERGROUND STORAGE TANK (UST)
RELEASE SIMULATION MODEL
ENVIRONMENTAL ENGINEERING COMMITTEE
SCIENCE ADVISORY BOARD
U.S. ENVIRONMENTAL PROTECTION AGENCY
Washington, D.C.
-------
NOTICE
This report has been written as part of the activities of
the Science Advisory Board, a public advisory group providing
extramural scientific information and advice to the Administrator
and other officials of the Environental Protection Agency. The
Board is structured to provide a balanced expert assessment of
scientific matters related to problems facing the Agency. This
report has not been reviewed for approval by the Agency and,
hence, the contents of this report do not necessarily represent
the views and policies of the Environmental Protection Agency,
nor of other agencies in the Executive Branch of the Federal
government, nor does mention of trade names or commercial
products constitute endorsement of recommendation for use.
-------
Table of Contents
I. EXECUTIVE SUMMARY 1
II. INTRODUCTION 3
III. REVIEW OF THE TJST RELEASE SIMULATION MODEL
A. General Comments 3
B. Responses to Specific Technical Questions 5
C. other Potential Uses of the Model 8
IV. APPENDICES
A. Specific Questions for SAB
B. Rosters
C. Rough Calculations of UST Leak Volatilization
D. References
-------
EXECUTIVE SUMMARY
This report presents the scientific review of EPA'S Under-
ground Storage Tank (UST) Release Simulation Model conducted by
the Science Advisory Board's Environmental Engineering
Committee [1] . EP&'s office of Underground storage Tanks developed
this model to support its regulatory decisions. More
specifically, the model is the basis of the UST Regulatory
Impact Analysis of the requirements proposed to regulate under-
ground gasoline storage tanks [2,3].
EPA has not directly used the model to develop regulatory
requirements. Rather, it has been used only to substantiate
regulatory decisions that have been made through considering
other factors, and to conduct the MIA. For regulatory support,
the model has been designed to generate estimates of the areal
extent of plumes of gasoline in the unsaturated zone (i.e.,
floating plumes). To generate estimates of regulatory benefits
(i.e., risk reductions) in the RIA, the model also includes
saturated zone transport of benzene, linked to exposure and dose-
response assumptions. In this context, the Committee believes
that the overall approach and design of the modeling framework
are scientifically sound. However, the Committee does have
reservations concerning particular aspects of the current
implementation of the modeling framework, and was not able to
fully evaluate all aspects of the model. These reservations and
limitations are identified below and discussed further in this
report.
The Committee recommends several ways of evaluating the
results of the UST model. First, because the model involves such
a complex calculation of tank failure and plume movement, it
would be useful to compare the model results to simpler, order-
of-magnitude estimates based on a first-order characterization
of tank ages and failure probabilities. Second, the simplified
and full models should each be compared to the data bases on tank
failure that are currently becoming available. Third, to aid in
the comparison of the "UST model to simpler approaches, the
composite, system-wide hazard function which results from all the
individual failure probabilities in the UST model should be
computed and plotted. These aggregate plots will help
illuminate the overall structure and effect of the model's
assumptions. And last, because the theoretical basis for
modeling gasoline flows in the unsaturated and saturated zones is
relatively new, examples of laboratory and field validation of
the models should be included as part of the model presentation.
The incorporation of two-phase flows in a regulatory model
such as the UST model represents an innovative attempt to use
state-of-the-art science. Full review of the technical details
of the model's equations (particularly the plume formation and
the transfer process to the aquifer) requires technical expertise
beyond the members of the present Science Advisory Board UST
-------
Subcommittee* The Committee believes that uses other than
support of the RIA would require a more detailed peer review from
scientists currently working in the area of multiphase flow.
As indicated, however, the Committee does have some reser-
vations about the Model, even in its present context. These include;
1» The documentation of the model is not clear, and many
the model's assumptions are not explicit. The code, as
published in the Appendix of the report, is unusable.
It is a long and complex code, and contains no comment
cards or other explanatory statements that would make
it useful to anybody but the developers of the model.
The code should be documented so others can run the
model.
2. The air pathway is inadequately considered.
Volatilization of constituents, including benzene,
may well affect UST releases. Unless a rationale
exists for discounting volatilization, such releases
and their movement should be considered.
3. other potential pathways are also discounted without
explanation, including any surface water effects and
use of ground water for crop irrigation. Unless a
rationale exists for discounting other pathways, they
should be evaluated. If such a rationale does exist,
it should be presented and discussed.
4. The qualitative review of the uncertainties is a good
beginning for characterizing uncertainties. However,
it provides no insight into the magnitude of the
uncertainties and no indication of which model inputs
and assumptions most influence the model's results.
A quantitative sensitivity analysis of the model should
be performed to determine the critical parameters and
uncertainties. One should also know whether any of
the assumed inputs could take on values that would
result in a change in the cost-benefit rank ordering of
the options considered, and whether the selected Option
II is sensitive to particular parameter uncertainties.
Until such an uncertainty analysis is undertaken, we
are unable to determine the degree of confidence that
should be placed in the current results.
The program office anticipates using the UST model for
similar analyses in future regulatory processes when new
regulations are written for presently exempted USTs, This is an
appropriate use of the model. Benzene will not be an
appropriate surrogate for most chemical USTs, however.
More site-and area-specific uses of the model should not
be made until there is better documentation and validation.
-------
The specific assumptions incorporated into the logic and step-by-
step approach need to be clarified for other potential users of
the model,
II. INTRODUCTION
In November 1986, J. Winston Porter, Assistant Administrator
for Solid Waste and Emergency Response, requested that the
Science Advisory Board (SAB) review the Underground Storage Tank
(UST) Release Simulation Model in mid-1987. The SAB Executive
Committee accepted the request and assigned the review to the
Environmental Engineering Committee (EEC).
On March 5, 1987, staff from the Office of Underground
Storage Tanks (OUST) introduced the EEC to the UST model and to
the UST regulatory program, then under development. At the EEC's
May 11 meeting, the Agency presented additional details on the
model methodology and results and requested that the Committee
address several specific issues in the review (see Appendix A)*
The EEC formed a Subcommittee to draft a report. The
membership of the Subcommittee and the EEC appears in Appendix B.
The Subcommittee *s findings were discussed and accepted by the
EEC and subsequently reviewed and approved by the SAB Executive
Committee.
III. REVIEW OF THE UST RELEASE SIMULATION MODELS
A. GENERAL COMMENTS
The Committee believes that the overall approach and design
of the modeling framework is sound, but that limitations in the
current implementation are such that it should- be used only in
the context of substantiating regulatory decisions on underground
gasoline storage tanks that have been made by other means. For
regulatory support, the model has been designed to generate
estimates of the areal extent of plumes of gasoline in the
unsaturated zone (i.e., floating plumes). To generate estimates
of regulatory benefits (i.e., risk reductions) in the Regulatory
Impact Analysis (RIA), the model also includes saturated zone
transport of benzene, linked to exposure and dose-response
assumptions. The model's components are logically structured and
linked, in general. Section B, below, discusses some of the
calculations in more detail.
Because the UST model involves such a complex calculation of
tank failure and impacts, it would be useful to compare the model
results to simpler, order-of-magnitude estimates based on a first-
order characterization of tank ages and failure probabilities.
The simplified and full models should each be compared to the
data-bases on tank failure that are currently becoming available.
-------
To aid in the comparison of the UST model to simpler
approaches, the composite, system-wide hazard' function that
results from all the individual failure probabilities in the UST
Model should be computed and plotted. This would include both
the hazard rate (the probability of failure in a given year given
that a tank has survived to that time) and the survival
distribution (the cumulative failure probability as a function of
age). These aggregate plots will help illuminate the overall
structure and effect of the model's assumptions.
The Committee, however, does have some resrvations about the
model, even in its present context. These include;
1. The documentation of the model is not clear, and many of
of the model's assumptions are not explicit. The linkages
between components are not well discussed. The model's
code is impenetrable, as it is presented with no explana-
tions or comments. The references used to support the
risk analysis are too frequently drawn from unpublished
sources even though better published works exist.
2. The air pathway is inadequately considered.
Volatilization of constituents, including benzene, may
well affect UST releases (see Appendix C). Spills may
volatilize before they infilitrate to ground water.
Constituents may also volatilize from the unsaturated
and saturated zone plumes. Not only will this mechanism
affect ground water releases, but it also creates a new
pathway for risks. Unless a rationale exists for discounting
volatilization, such releases and their movement should
be considered.
3. Other potential pathways are also discounted without
explanation, including any surface water effects and use
of gound water for crop irrigation. Unless rationale
exists for discounting other pathways, they should be
evaluated.
4. Monte Carlo Methods; The sampling procedure simulating
multiple tank histories, whereby the tank population is
divided into cohorts representing tank types and
hydrogeologic settings, appears to be appropriate and
well thought out. It is not clear from page C~l,
however, whether the 2000, 1000, or 500 tank
replications tested are within each cohort or over the entire
tank population. Also, when testing the model at different
sample sizes, it is not clear which summary statistics are
considered. Presumably, the summary statistics relate to
the total plume acres and detection-replacement costs
for the entire tank population, but this is not stated in
the text. Finally, the convergance of the model at
"small" sample sizes (i.e., 500 tanks) should be
demonstrated graphically by plotting the summary statistics
as a function of sample size.
-------
5. Chapter 9 of the RIA presents a qualitative review of
the uncertainties in the UST model and their possible impli-
cations. The chapter provides a good beginning for
characterizing uncertainties. However, it provides no
insight into the magnitude of the uneertainities and no
indication of which model inputs and assumptions most
influence the results. A quantitative sensitivity analysis
of the model should be performed to determine the critical
parameters and uncertainties. One should also know whether
any of the assumed inputs could take on values that would
result in a change in the cost-benefit rank ordering of the
options considered, and whether the selected Option II is
sensitive to particular parameter uncertainities. With
the current results it is difficult to determine which of the
uncertainities identified in Chapter 9 are likely to be
most critical to the regulatory assessment. Comparison of
model results (i.e. the composite damage function, the
number of leak incidents predicted, etc.) to other avail-
able estimates would help in this assessment, in addition to
the recommended sensitivity analysis.
B. RESPONSES TO SPECIFIC TECHNICAL QUESTIONS fSee Appendix A^
1. Transport of gasoline in the unsaturate.d zone and
2. Transport of benzene from floating plume to aqueous
plume
The characterization of multiphase flow through ground water
systems is a new area of research in environmental science,
though some technical foundations are available from the field of
petroleum engineering, As such, the incorporation of two~phase
flows in a regulatory model such as the UST model represents an
innovative attempt to use state-of-the-art science. Because the
theoretical basis for modeling gasoline flows in the unsaturated
and saturated zones is relatively new, examples of laboratory and
field validation of the models should be included as part of the
model presentation.
The general approach and components included in the model
appear to be appropriate. However, full review of the technical
details of the model's equations requires technical expertise
beyond the members of the present Science Advisory Board UST
Subcommittee. Peer review from scientists currently working in
the area of multiphase flow is recommended.
The plume formation and the transfer process to the aquifer
should be more fully described and subsequently reviewed, possibly
by specialists in these areas. The current descriptions of these
processes left the Committee with questions about the mass
balance in the modeling framework. How is the mass discharge
-------
from a ruptured tank accounted for in the formation of the
floating plume which is defined by equations that yield
volumetric values? Are these consistent with the mass discharge
rates?
The description of mass transfer from the immiscible to the
dissolved phase on page 230 of the report (I), is inadequate.
Note that on page 232, the value of 5.5 x 10~13 kg/m2/ sec is
a mass transfer rate, not a "diffusion coefficient" as stated in
the text.
The degree to which a point source is a reasonable
approximation of the transfer of benzene from the lens to the
ground water should be further examined, as well as the benzene
transfer mechanism itself. The areal extent of the benzene layer
(or lens) provides a basis for determining the cost of remedial
action. In the RIA, the benzene is transferred from the
immiscible layer to ground water, In reality, this input to the
transport model is an area source rather than a point source.
The implications and errors introduced by this approximation
should be evaluated.
3, Transport of aqueous plume to well
The equation used to estimate the concentration of
miscible (dissolved) benzene at downgradient monitoring wells
uses an accepted advective dispersive model with sorption and
decay. The transport components are described by three-
dimensional advective flux. The differential equation ({!}),
p.232) is solved in the usual fashion for a point source under
steady conditions in an infinite medium,
The final working equation is appropriate for a slowly
leaking underground tank, the rate of release from which is
assumed to exist for a sufficiently long period to achieve a
steady-state condition. The steady-state equation, however, is
not appropriate for time-variable discharges and particularly not
for relatively rapid releases, e.g. a "catastrophic" event.
Since the time .variation may be significant in such conditions,
it would be appropriate to make available the solutions for this
case and discuss the situations in which it may be considered.
The basic equation is solved for a conservative substance,
for which case the retardation effect is eliminated in a steady
state solution. For the analysis of non-conservative
constituents and time-variable releases, the retardation
coefficient is retained in the final solution. Although it is
recognized in the description of the model that these cases may
be important, they are not specifically addressed in the
documents, and the degree to which they can be included in the
model is not specified.
Fundamentally, the plume equation and the ground water
equations compose a two-phase systen and should be solved
-------
simultaneously. The Committee appreciates that the system is
complex and all the mechanisms not fully understood, much less
quantified. In spite of these recognized limitations, the
fundamental relations should be explicitly expressed in
differential form including both state, as well as mass,
equations. The necessary approximations and empiricisms may then
be introduced,
The Committee recognizes the difficulty of assigning
dispersion coefficients representative of regional areas based on
soil classification. To evaluate the effect of a plume on a well
supply in relatively close proximity to the source/ however, it
may not be necessary to incorporate three demensional dispersion.
The analysis may be greatly simplified, yet remain equally valid
using the one-dimensional-dispersion condition. In some
instances (short distances and higher ground water velocities),
no significant error is introduced if dispersion is eliminated in
all dimensions.
4. Assumptions about well locations
The rationale presented (in chapter 3 {Section D) and
Appendix F) for determining well locations and populations at
risk from leaking USTs appears to be sound and clearly shows the
association between USTs and population density. The inverse
association between population centers and shallow ground water
use, especially private wells, is also fully considered. For the
purposes of generating supporting evidence for the RIA, the
methodology employed should suffice since, on a community basis,
adequately conservative estimates are generated. Such generic
assumptions, however, are not applicable to site-specific
analysis,
5. Calculations of benzene risk
The standard method used for calculations of benzene
risk is sufficiently comprehensive and conservative for a RIA.
Benzene toxicity is largely characterized by carcinogenic effects
having typically long latency periods requiring lifetime
exposures. The exposure times modeled for leaking USTs seem to
be unlikely to approach those needed to create carcinogenic
results. It may be worthwhile to select another compound with
acute short-term effects, if possible, for a check on the
exposure risks,
6. Use..of benzene as a surrogate for gasoline
Use of any single compound as a surrogate for a
mixture as complex as gasoline is an oversimplication raising
some concern. Gasoline is made up of a variety of compounds of
highly variable chemical nature and behaviors aliphatic and
cycloparaffinic hydrocarbons; benzenoid compounds like benzene,
-------
toluene, xylene, and crnnene,* compounds with two or more ring
structures and a wide variety of subsequent groups; and a broad
and variable assortment of other compounds containing sulfur,
nitrogen, or oxygenated groups.
The chemical behaviors of each of these groups greatly
affect their transport through the soil! their solubility in
water, and thus their transport in ground water. The range of
properties is so great that representative substances from each
group ought to be evaluated in the model to at least establish
the range of exposures that could result.
C* OTHER POTENTIAL USES OF THE MODEL
The program office anticipates using the UST model in
future regulatory processes when new regulations are written for
presently exempted USTs. This is an appropriate use of the
model. Benzene will not be an appropriate surrogate for most
chemical USTs, however.
More site™ and area-specific uses of the model should not
be made until there is better documentation and validation. The
specific assumptions incorporated into the logic and step-by-step
approach need to be clarified for other potential users of the
model.
The code, as published in the Appendix of the report, is
unusable (1). It is a long and complex code, and contains no
comment cards or other explanatory statements which would make it
useful to users of the model. The code should be documented so
others can run the model. The Committee suggests that the model
be run by an independent contractor who can evaluate the code
itself. This independent evaluation may also point out
weaknesses of the model in the support'of the regulations.
The present Subcommittee did not feel competent to provide
an in-depth review of all aspects of this very complex model. If
uses other than support of the RIA is made of the model, we
suggest that a more detailed peer review be conducted.
-------
APPENDIX A
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 204<0
APR 3 1987
Off»C£ Of
SOLID WASTE AND EMERGENCY RESPONSE
MEMORANDUM
SUBJECT! tlST Release Simulation Model - Areas foe Science Advisory
Board (SAB) Review
YV
FROM: Sammy K. Ng
Office of Underground /Storkq* Tanks
TO 5 Eric Malls
Science Advisory Board
As requested by the Environmental Engineering Committee of the SAB, we have
considered the areas of the Underground Storage Tank (UST) Release simulation
Model ("Model") that might be appropriate for the Committee'* review.
The (1ST Model is composed of three main routines: the failure routine* the
release routine, and the transport routine. In the failure routine, the model
determines the tine and location of failure within an OST facility) the release
routine calculates the time to detection of the release, the total volume of
product released, and the oast of repairing or replacing the facility* The
transport routine determines the travel time of the release from the facility to
its point of detection* It then calculates the are* of the floating plume that
results if the release reaches groundwater and computes the discounted cost of
any corrective action necessary to clean up the release and the plume*
We believe that the most productive manner in which the Committee might
participate in the review of the Model would be to focus on one or more fairly
broad, but technically complex and sensitive areas of the analysis in which the
special expertise of the Committee reviewers are particularly strong. Our
suggestion is that the Committee focus on the transport-* to-exposure aspect of
the Model, We believ* that the Committee's review of the assumptions,
computational procedures, and data associated with the estimated risks of UST
failures would be a particularly helpful conterpart to the review that we
are currently conducting.
-------
page 2
The following list provides some of the issues in which the Committee may
be interested in pursuing}
o modelling of transport of gasoline in the onsaturated zone,
o modelling of the transfer of benzene from the floating plum* to the
aqueous plume And its transport through groundwater to the wells,
o well locations and the population exposure to benzene in drinking
water, and
o risks resulting from exposure to the benzene component of gasoline,
If you have any questions, please give me a call at 392-7903* I
forward to working with you and the Committee On this project.
ccs Louise wise, OUST
-------
APPENDIX B
EEC UST Subcommittee
Dr. leros Cartwrigtit (Chair)
IL State Geological Survey
Dr. William Haun
General Mills (retired)
Dr. Donald O'Connor
Environmental Engineering Science Program
Manhattan College
Dr. Thomas Shen
NY State Dept. of Environmental Conservation
Dr. Mitchell Small
Department of Civil Eagineering
Carnegie-Mellon University
Dr. Herb Ward
Dept, of Environmental Science and Engineering
Rice University
Remainder of EEC
Dr. Maywond Loehr (EEC Chairman)
Civil Engineering Department
University of Texas (Austin)
Dr. Joan Berkowitz
Risk Science International
Mr. Richard Conway
Union Carbide Corporation
Dr. Ben Ewing
Institute for Environmental Studies
University of Illinois (Champaign-Urbaaa)
Dr. William Glaze
Dept. of Environmental Science and Engineering
University of California (Los Angeles)
Mr. George Green
Public Service Conpany of Colorado
Dr. Joseph Ling *
3M Company
Dr. Charles O'Melia Executive Secretary
Dept. of Geography and Environmental Engineering
The Johns Hopkins University Mr. Eric Males
Dr. Paul Roberts Staff Secretary
Department of Civil Engineering
Stanford University Mrs. Marie Miller
-------
APPENDIX C
Rough Calculation of UST Leak Volatilization
Volatilization Estimation:
Ei - Di CSi A P 4/3 VL
Where i can be benzene, EDB, alcohol, or other constituents.
$1 * diffusion coefficient of benzene - 0.08708 em^/sec @ 20%
C * » saturated concentration * _p_M " 3.73 x 78 - 0.016 g/cm3 @ .20*0
RT 62.3 x 293
p - vapor pressure (mm tig)
M - molecular weight
T » temperature (K°)
R *• Universal gas constant (am Hg-cm^/lC^-iaole)
P - soil porosity - 0.4
Mj - weight of benzene in gasoline * 10%
L * depth of benzene to ground level LI - 10 a (to coae center of mass)
I»2 - 14 a
2 *-**
A - exposed area of benzene Ai - 1,037 m~ ( t( rs)
Aj - 7,500 a2
BA1 - 0.08708 x 0.016 x 1. 6 x 1Q7 x 0.4 4^3 ic 0.1/1000 •* 0.658 g/sec.
I12 - 0.08708 x 0.016 x 7.5 x 107 it 0.295 x 0.1/1400 - 0.220 g/sec.
Assume
Average rate of leaching downward -» 0.5 is/day
Average rate of spreading with ground water velocity - 1.0 in/day
(0.1 to 5.0}
ti - Lj/tj_ - 8/0.5 • 16 days
t2 - &2/t2 * 250/1.0 » 250 days
?olatlHzatlon Amount:
Case 1: 0.658 g/sec. x 8640 sec. /day x 16 days - 90,962 grams
(from the unsatu rated zone floating plume, spreading cone)
Case 2: 0.220 x 8640 x 250 - 475,000 grans
(from the ground water, dissolved plume)
-------
Case 1 loss to air: 90,962/1.01 x 10s - 0,068* of total
Case 2 loss to air; 475,000/6,75 x 106 - 7*0403! of total
7.1081 of gasoline loot to air
Note; The gasoline vapor can migrate to basements via pipeline trenches.
Assumptions:
vj^ « leaching downward velocity «• 0.5 m/day
vj - ground water flow velocity - 1.0 m/day (0.1 to 5.0 n/day)
AI - downward spreading area - iTrs - 1,037 a^ - 10*37 x 10^ cn^
2 A 9
Aj * plume spreading in ground water - 7,500 in - 75 x 10 en
VL -lTr2H/3 - <225)(l2)/3 - 2,827 a3 - 2.8 x 109 cm3
V2 - 7500 m2 x O.OOto - 75m3 - 75 x 106 ca3
Mx - 2.8 x 109 cm3 x 0.9 g/cm3 x 0.1 x 0.4 - l.Ql x 108 gran
(density) (conc'n) (porosity)
M - 75 x 106 cm3 x 0.9 x 0.1 - 6.75 x 106 gram
-------
Rough Calculation of UST Leak Volatilization
(not to scofe)
A1
Ground level
i i-mm
L < I vifi
/ _\l4m
2m '
L2=14m
12m
Drinking
Water Well
VI =2,827m
(side view)
V2=75m3
12m
A1= 1,037m
S=22m (curved)
250m
-------
APPENDIX D
REFERENCES
1. U.S. EPA, Office of Underground Storage Tanks* "Final Report Underground
Storage Tank Model," submitted by Pope-Reid Associates, December Ii86.
2. U.S. EPA, Office of Underground Storage Tanks. "Regulatory Impact Analysis
for Proposed Technical Standards for Underground Storage Tanks,* March 30,
1987.
3. "Underground Storage Tanks; Proposed Rules," 52 PR 12662, April 17, 1987,
------- |