United States Office of Pesticides EPA 560/5-86-013
Environmental Protection and Toxic Substances May 1986
Agency Washington, DC 20460
Toxic Substances
Underground Motor Fuel
Storage Tanks:
A National Survey
VOL. II. APPENDICES
-------
50273-101
RETORT DOCUMENTATION
PAGE
i._«Forr ho.
EPA' 560/5-86-013
«. TNI* and SuMltla
Underground Motor Fuel Storage Tanks: A National Survey
7. AuMwdi) Diet*, Stephen K. ;* Flora, Jairua D., Jr. ;b
Strenlo, Judith F;* Carmen J. Vincent* et *1.
». Pat«awwfan OnaiO«a<*an Mama M Wm»
Naacac, Inc., 1650 Raaaarch Blvd.. Rockvilla, HB 20850
ntldmac Raaaarch Inacicuea, 425 Volkar Blvd., Kanaaa Cicy, MB Mitt
Baccalla Coluabua Dlvlalon, Waahlngcon Oparaclona, 2030 N Scraac, Ml,
Waahlngcon, D.C. 20036
Waahlngcon Cenauldng Group, 1623 Kya Scraac, HW, Waahlngcon D.C. 20006
S. RaclpianTa Aeeaadan No.
S. Raport Data
May, 1986
*• Parfarmlng Organisation Rapt. No.
10. Preiact/TatkAWark Unit No.
Task 3
11. CantracttC) or QranttQ) No.
EPA No. 68-02-4243
and
-------
UNDERGROUND MOTOR
FUEL STORAGE TANKS:
A NATIONAL SURVEY
Appendices
Prepared by:
Westat, Inc.
1650 Research Boulevard
Rockville, MD 20850
Battelle Columbus Division
Washington Operations
2030 M Street, N.W.
Washington, D.C. 20036
Midwest Research Institute
425 Volker Boulevard
Kansas City, MO 64110
Washington Consulting Group
1625 Eye Street, N.W.
Suite 214
Washington, D.C. 20006
for the:
Exposure Evaluation Division
Office of Toxic Substances
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
Washington, D.C. 20460
May 1, 1986
-------
DISCLAIMER
This report was prepared under contract to an agency of the
United States Government. Neither the United States Government
nor any of its employees, contractors, subcontractors, or their
employees makes any warranty, expressed or implied, or assumes
any legal liability or. responsibility for any third party's use
of or the results of such use of any information, apparatus,
product, or process disclosed in this report, or represents'that
its use by such third party would not infringe on privately owned
rights.
-------
CONTENTS
Appendix Eaafi
A SAMPLE DESIGN AND ESTIMATION OF WEIGHTS
AND VARIANCES A-l
B SURVEY PROCEDURES AND ELIGIBILITY AND
RESPONSE RATES B-l
C DEVELOPMENT OF A TANK TEST METHOD C-l
D TANK TESTING DATA REDUCTION AND STATISTICAL
ANALYSIS LEADING TO LEAK STATUS DETERMINATION .. D-l
E INVENTORY RECONCILIATION METHODS E-l
F DATA COLLECTION FORMS AND MATERIALS F-l
G NATIONAL UNDERGROUND STORAGE TANK SURVEY
NATIONAL SAMPLE OF FARMS G-l
H ENVIRONMENTAL DATA COVERAGE H-l
I MULTIVARIATE ANALYSIS 1-1
-------
APPENDIX A
SAMPLE DESIGN AND ESTIMATION
OF WEIGHTS AND VARIANCES
I. TARGET UNIVERSE. OVERVIEW OF SAMPLE DESIGN
The target universe, or population of interest, for the
Survey of Underground Storage Tanks consisted of all underground
tanks which store motor fuel prior to dispensing it for use as
fuel, with exceptions as noted below. For example, in the retail
gasoline sector, this includes all underground tanks at service
stations but excludes large holding tanks at a distributor. In
sampling, we used a tank establishment, that is, a location with
eligible tanks, as the sample unit. Once a given establishment
was sampled, all its tanks were in the sample for the initial
data collection phase. For the physical tank testing stage, a
subsample of the sampled establishments was drawn, and all tanks
at the subsampled establishments were physically tested. For
purposes of list building, the target universe of establishments
was defined as a number of segments, with certain exclusions as
noted. The following types of establishments were identified as
potentially having underground motor fuel storage tanks:
o Gasoline service stations;
o Other establishments almost certain to have underground
storage tanks, including:
Transit and transportation fleets (such as taxi,
trucking, and bus companies; auto and truck rental
companies; railroads; and auto and truck dealers);
Marinas;
-------
Airports and other air transportation related
industries; and
Golf courses and country clubs;
o Government fleet service pumps, including:
Federal;
State;
Local — county, city, etc.; and
Military;
o Large companies with 20 of more employees in other (non
fuel-related) industries which have private fleet
service pumps; and
o Farms with underground motor fuel storage tanks.
Underground tanks containing motor fuels maintained by
private homeowners and tanks for private fleets maintained by
companies with fewer than 20 employees were excluded from the
scope of this survey. They were not estimated to account for a
large number of underground storage tanks. In addition, the cost
necessary to screen out businesses and residences with no
underground tanks was judged to be too great in comparison with
the anticipated low addition to the total universe from these
establishments.
A. Overview of Sample Design
The sample of establishments was drawn using a multi-stage
cluster design. The continental U.S. was divided into six
regions of interest. The sample was drawn to provide estimates
both at the national and regional levels. The first stage"of
sampling was Primary Sampling Units (PSUs) consisting of counties
or groups of contiguous counties with designated minimum
A-?
-------
estimated numbers of underground tank establishments. The sample
of PSUs was allocated to the regions and drawn within region
proportionally to their total estimated number of underground
tank establishments. Thirty-four PSUs were drawn.
Within each selected PSU, three establishment frames were
developed:
o Fuel tank establishments - consisting of gas stations,
establishments in other fuel-related Standard
Industrial Classification (SIC) groups, and government
tank locations;
o Large establishments - consisting of all businesses
with 20 or more employees not already listed as fuel
tank establishments; and
o Farms - consisting of all farms.
A national sample was drawn from each frame. For large
establishments and for farms, 600 establishments were selected
from each frame. For the fuel tank establishments, a national
sample size of 1,618 was allocated to the regions, and six
regional samples were drawn. In each case, the establishment
sample was drawn taking account of the PSU probabilities of
selection in such a way that the establishment samples were self-
weighting, nationally for the large establishments and farms, and
by region for the fuel tank establishments.
Once the three samples were drawn, the large establishment
and farm samples were telephone screened for the presence of
underground tanks. All large establishments and farms which have
underground fuel storage tanks became part of the field sample,
as did cases which could not be resolved over the telephone. No
substitutions were made for large establishments or farms with no
underground fuel storage tanks. The fuel establishment tank
sample consisted of establishments which were thought likely to
-------
have underground fuel storage tanks. Initial field work showed
that this list actually produced about a 50 percent survey
eligibility rate; that is, about half the sampled establishments
sampled were still in business and had underground motor fuel
storage tanks. Although lower than anticipated, this eligibility
rate indicates that the coverage of the target universe by the
selected SICs was probably quite good. In order to attain our
target sample size of 800 eligible establishments, the initial
sample sizes per region were doubled for the fuel establishment
segment, for a total sample draw of 1,618 cases.
B. Definition of Regions; PSU Sample Design
Table A-l lists the regions, giving the states included in
each. They are shown on a map in Figure A-l. The PSU frame was
developed for the entire continental U.S. as detailed in the
following paragraphs.
For each county, the following counts were developed:
o Number of gas stations based on the 1981 County
Business Patterns data (count for SIC 5541) ;
o Additional estimated number of gas stations allocated
to counties within states on a population basis to
bring the state totals up to the estimate provided by
Versar to the EPA; and
o Total number of establishments in the selected other
SICs (list in Table A-2) as given by the County
Business Patterns data.
These three counts were summed for each county to form the
estimated number of fuel tank establishments for the county.
The counties were grouped into initial PSUs by using the
Westat Master PSU Frame developed on a population basis, which
-------
Table A-l. Six region# for National Survey of Underground Fuel
Storage Tanks
1 — Northeast
Maine
New Hampshire
Vermont
Connecticut
Massachusetts
Rhode Island
New York
New Jersey
Pensylvania
Maryland
Delaware
Virginia
West Virginia
Washington, D. C.
2 — Southeast
Kentucky
Tennessee
Arkansas
Louisiana
Mississippi
Alabama
Georgia
North Carolina
South Carolina
Florida
3 — Midwest
Wisconsin
Minnesota
Iowa
Missouri
Illinois
Indiana
Ohio
Michigan
4 — Central
North Dakota
South Dakota
Nebraska
Kansas
Oklahoma
Texas
5 — Mountain
Montana
Wyoming
Idaho
Nevada
Utah
Colorado
Arizona
New Mexico
6 — Pacific
Washington
Oregon
California
-------
A
6-Pacif ic
1-Northeaft
c
Uam» *-1. Underground Survey Region.
-------
Table A-2. Selected SIC codes for fuel tank establishment frame
SIC code Descript ion
4010 Railroads, switching and terminal companies
4110+ Local and suburban passenger transportation
companies (includes airport transportation,
ambulance and limousine services)
4121+ Taxicab companies
4131+ Intercity highway transportation services
4140,+ Passenger transportation charter services
(includes bus charter, rentals and tours)
4151 School bus companies
4170 Passenger transportation terminal and service
facilities
4210+ Trucking companies
4231+ Motor freight terminals
4469A Marinas
4511 Air transportation, certificated carriers
4521+ Aircraft charter, rental and leasing —
non-certificated carriers
4 58 2A Airports
4582B+ Aircraft maintenance services
4583 Airport terminal services
5511+ Auto and truck dealers (new and used)
5521+ Used car dealers
5541+ Gasoline service stations
7512+ Passenger car rental and leasing agencies
7513+ Truck rental and leasing agencies
7519+ Utility and house trailer rental agencies
7992+ Public golf courses
7 997B+ Golf and country clubs
-------
follows the PSUs used by the Census Bureau in designing the
Current Population Survey. This initial list of PSUs was
transformed to a final list by splitting PSUs which had large
total counts into smaller sets of counties and combining PSUs
with insufficient counts, resulting in a set of PSUs which were
as small as possible while still containing a minimum number of
fuel tank establishments.
Once the PSUs were defined, the sample of PSUs was drawn as
follows. For each region, a target number of PSUs was
established. This was six PSUs per region, except in Region 5
(Mountain) where four PSUs were drawn. Within each region, the
PSUs were sorted by an urban versus rural designation, then by
state, and finally by size (total estimated number of fuel tank
establishments). The sample of PSUs was then drawn within each
region on a probability proportional to size basis.
C. Tank Establishment Frames Within PSUs: Sample of
Establishments
Once the thirty-four PSUs were selected, three establishment
frames were built for each PSU. A sample was drawn from each
frame, and eligible establishments in the three samples formed
the sample of establishments.
The first frame was the fuel tank establishment frame. It
consisted of establishments considered to be extremely likely to
have underground fuel storage tanks. The frame was constructed
from several sources. A list of business establishments with one
of the target SICs (refer to Table A-2) in the selected counties
was purchased from National Business Lists (NBL). This was
supplemented by any establishments found to have one of the
selected SICs in the large establishments list (see below).
-------
Lists of Federal, state, and local government establishments in
the sampled counties with underground fuel storage tanks were
developed by extensive telephone contacts with government
officials. In addition, a list of military establishments with
underground fuel storage tanks was provided by the military to
EPA. These lists were keypunched and added to the fuel tank
establishment frame.
The sample of fuel tank establishments consisted of 1,618
establishments in the country (in order to achieve a target of
800 eligible establishments). Based on the regional totals of
number of such establishments developed in the PSU frame-building
effort, the total sample size was allocated to the six regions.
Within each region, the establishments were sorted by PSU and
SIC, and a self-weighting sample was drawn. Since the PSUs were
sampled proportionately to the estimated number of
establishments, this resulted in an approximately equal number of
establishments per PSU within each region. There was not a
precisely equal number per PSU for two reasons: the PSUs were
sampled based on CBP counts and the establishments were sampled
based on actual frame counts; and the PSU sample measure of size
did not include an estimate for number of government
establishments.
The second frame to be developed was the large
establishments frame. This frame consisted of a list of business
establishments in the sampled counties with 20 or more employees
purchased from Dun's Marketing Identifiers (DMI). The
establishments on this list with the fuel tank SICs (Table A-2)
were clerically compared with NBL lists, county by county, to
eliminate duplication between the two frames. Duplicates .were
deleted from the DMI list, and any establishment on the DMI list
with one of these SICs not found on the NBL list were moved to
-------
the NBL list. The resulting DMI list was the frame for large
establishments not in fuel tank SICs.
The sample of large establishments was drawn by first
sorting the frame by region, PSU, and number of employees. Then
a self-weighting sample of 600 establishments was drawn across
the whole country. These establishments were contacted by
telephone to determine whether they had underground fuel storage
tanks. Those that did were part of the sample for initial data
collection; no substitution was made for establishments with no
tanks.
The third frame was farms. This was constructed by
obtaining a list of all farms in the sampled counties from the
U.S. Department of Agriculture, through EPA. The list included
crop acreage for each farm. Any establishment on the DMI list
with an agricultural SIC code was deleted from the DMI list and
added to the farm frame if it did not already appear there.
The farm frame was sorted by region, PSU, and acreage. A
national self-weighting sample of 600 farms was selected. These
were screened by telephone to determine the presence of
underground tanks. As with large establishments, no substitution
was made for farms with no tanks.
II• PRIMARY SAMPLE UNIT fPSUl SAMPLE
This subsection discusses the first stage sample of Primary
Sampling Units (PSUs). Appendix H discusses the sample of farms
from PSU selection though the final sample of farms. Thus> this
subsection and the following ones concentrate on the fuel
establishments and large establishments, although some data on
farms are included for completeness.
-------
This subsection begins with a statistical description of the
six survey regions based on data gathered in the construction of
the Primary Sampling Unit (PSU) frame. It goes on to describe
the PSU sampling process and concludes with a discussion of the
sample of PSUs drawn.
A. Survey Regions
The six survey regions are defined in A-I, above, which
includes a list of states in each region (Table A-l) and a map of
the regions (Figure A-l). Here we describe the regions
statistically in terms of characteristics important to the
present study. Table A-3 gives several characteristics by
region, both the amounts and the percent distributions.
The number of states and counties in each region is simply
based on the definitions of the regions. The number of states
ranged from three states in the Pacific Region (Region 6) to 14
states in the Northeast Region (Region 1). Alaska and Hawaii are
not included, and the District of Columbia is counted as a state,
making the total 49. In these 49 states there are 3,111
counties. The number per region ranges from a low of 133, again
in the Pacific Region, to a high of 874 in the Southeast Region
(Region 2).
The first step in constructing the PSU frame was to define
PSUs, a process described in Subsection A-I. These consist of
counties or groups of counties with a minimum estimated number of
fuel establishments. The minimum was set separately for e^ch
region based on the expected number of establishments to be
sampled per PSU in each region. The resulting PSU definition
groups the 3,111 counties into 1,362 PSUs. The number per region
-------
Table A-3. Statistical description of Underground Storage Tank Survey Regions [percent distribution in parentheses (4)]
Survey
region
Number
of
states
(Incl. DC)
Number
of
counties
Number of
primary
sampling
units
1980
Population
(1,000's)
U.S. Census
Land area
(sq. mi.)
Number of
gas stations
Versar
report (1)
Number of
facilities,
selected
other SICs
1981 CBP (2)
Sampling
measure
of size (3)
Number
of large
establishments
(>20 empl.)
1981 CBP (2)
Number of
farms. 1982
Census of
Agriculture
1
14
436
219
61.881
238.400
46.616
34.829
81.445
157.843
222.099
Northeast
(278)
(88)
(248)
(318)
(278)
(278)
(108)
2
10
874
348
45.371
466.678
59.576
19.403
78.979
108.367
548.926
Southeast
(208)
(168)
(318)
(178)
(268)
(198)
(248)
3
8
738
333
53.589
448.419
35.935
27.124
63.059
138.742
725.699
Midwest
(248)
(158)
(198)
(248)
(218)
(248)
(328)
4
6
650
259
22.531
634.346
24.634
11.738
36.372
61.756
464.680
Central
(108)
(218)
(138)
(108)
(128)
(118)
(218)
5
8
280
117
11.373
855.193
8.755
5.273
14.028
29.144
121.777
Mountain
(58)
(298)
(58)
(58)
(58)
(58)
(58)
6
3
133
86
30.433
318.994
18.142
13.843
31.985
87.461
152.630
Pacific
(148)
(118)
(98)
(128)
(108)
(158)
(78)
Total
49
3.111
1.362
225.178
2.962.030
193,658
112.210
305.868
583,313
2.235.811
(1) Versar Corp. report to EPA. 1984
(2) County Business Patterns. 1981
(3) Sum of Gas Stations and Facilities with Other SICs
¦)
(4) Percent distributions may not add to 1008 due to rounding.
-------
ranges from a low of 86, again in the Pacific Region, to a high
of 348 in the Southeast Region.
Two further statistics help set the stage for the survey in
describing the regions: the number and percent of 1980
population in each region; and the square miles and percent of
continental land area in each region. In terms of population,
Regions 1-3 (the eastern block of regions) contain 27, 20 and 24
percent of the population, respectively, for a total of 71
percent of the population. Regions 4-6 have 10, 5 and 14 percent
of the population, respectively. For land area the situation is
reversed, though not as dramatic. Regions 1-3 contain 39 percent
of the land area, while Regions 4-6 contain 61 percent.
The next three statistics form the basis of the PSU
selection. The number of gas stations was estimated per state by
Versar.1 The distribution by region ranged from 5 percent in the
Mountain Region (Region 5) to 31 percent in the Southeast Region
(Region 2). Regions 1-3 contain an estimated 73 percent of the
gas stations. The number of establishments with a Standard
Industrial Classification (SIC) code among those selected as
likely to have underground motor fuel storage tanks (see list in
Table A-2) was found as counted in the 1981 County Business
Patterns data.2 Seventy-three percent of these other fuel
establishments are in Regions 1-3. The percent by region ranges
1Leaking Underground Storage Tanks Containing Engine Fuels,
draft, March 1984, prepared by Versar, Inc. The gas station
estimates were based on figures given in the 1983 Petroleum
Marketing Hews Fact Book and include all retail outlets for
branded gasoline.
2At the time of PSU sample selection, the 1982 CBP data were not
yet available. They became available in time to use for final
weights, as discussed in Subsection A-V.
-------
from a low, again in the Mountain Region, of 5 percent to a high
of 31 percent in the Northeast Region. These two figures (gas
stations and other fuel establishments) are summed to form the
sampling measure of size. The distribution of gas stations and
other fuel establishments follows that of the population.
Although the PSUs were sampled based on the number of fuel
establishments, a sample of large establishments (with 20 or more
employees) and of farms was also to be drawn from the sample
PSUs. The region statistics show that large establishments
follow the same general pattern as population and fuel
establishments: 5 percent are found in the Mountain Region and
27 percent in the Northeast Region; Regions 1-3 contain 69
percent of the large establishments as reported by the 1981
County Business Patterns data. Farms are found mostly in Regions
2-4, which have 78 percent of farms as shown in the 1982 Census
of Agriculture. Looking at the East versus West breakdown we
have been considering, the Eastern regions (Regions 1-3) contain
67 percent of the farms.
In Table A-4 some of these statistics are shown for the
urban/rural breakdown. Each PSU is designated as urban or rural
according to whether it is part of a Statistical Metropolitan
Area or not. The majority of PSUs and constituent counties are
designed as rural (65 percent of PSUs, 77 percent of counties),
but the majority of the fuel establishments plus gas stations are
found in urban PSUs (69 percent). The large establishments are
even more concentrated in urban PSUs, with 85 percent found there,
B- Sampled PSUs
The sample of PSUs was drawn as stated in Section A-I, using
the number of fuel establishments as a sampling measure of size.
-------
Table A-4. Summary of PSU sampling frame, urban versus rural PSUs
(percent distributions in parentheses)
Urban/
Rural
Number
of
counties
Number
of
PSU's
Sampling
measure
of size (1)
Large
establishments
(>20 empl.)
1981 CBP (2)
Urban
722
482
212,164
(69%)
479,461
(85%)
Rural
2,389
• 880
93,704
(31%)
103,852
(15%)
Contin*"*'1
Total
3,111
1,362
305,868
583,313
(1) Number of gas stations (Versar) plus other fuel-related
establishments (CBP)
(2) County Business Patterns data
-------
Thirty-four PSUs were drawn — six from each region, except
Region 5 where four were drawn. Tables A-5 and A-6 give
estimates of frame counts that would result from weighting the
PSU sample data by inverse of the sampling probability. This
gives an indication of how closely the sample reflects the frame
from which it was drawn. Not surprisingly, the sampling measure
of size (number of fuel establishments) tracks the population
very closely, with the same percent distribution by region and
only one percentage point different for the urban/rural
breakdown. The large establishment counts are reproduced fairly
well by the weighted sample. The percent distribution by region
is within one or two percentage points of the population
distribution, but the urban/rural breakdown is not as close.
While 85 percent of large establishments were in the urban PSUs
nationally, in the weighted sample PSUs, 79 percent are in the
urban PSUs.
Tables A-7 and A-8 give unweighted counts for the sampled
PSUs. In Table A-7, we see that the 34 PSUs are composed of 76
counties. The number of fuel establishments plus gas stations as
estimated from the Versar and CBP sources for the sampled PSUs is
27,753, and the estimated number of large establishments is
74,768. Table A-8 shows that 11 of the 34 PSUs are rural, with
36 of the 76 counties. The rural PSUs tend to have more counties
in order to contain the minimum number of fuel establishments.
The vast majority of both fuel and large establishments in the
sampled PSUs are in the urban PSUs (95 and 98 percent,
respectively). In the sample, one county, Los Angeles, was large
enough to be self-representing. This PSU accounts for the large
unweighted counts for Region 6 (Pacific) throughout the tables.
• *
Overall, the PSU universe appears to be well reflected in
the sample of PSUs. Figure A-2 shows the location of the sampled
PSUs to indicate their geographic representation, as well. The
* x * J
-------
Table A-5. Weighted data from sampled PSUs, region summary
(percent distributions In parentheses
Survey
region
Number
of
counties
Number
of
PSU's
Sampling
measure
of size (1)
Large
establishments
(>20 empl.)
1981 CBP (2)
1
Northeast
561
210
81,364
(27%)
148,906
(25%)
2
Southeast
635
. 341
78,974
(26%)
123,360
(21%)
3
Midwest
912
328
63,139
(21%)
135,842
(23%)
4
Central
1,660
327
36,374
(12%)
57,475
(10%)
5
Mountain
344
120
14,030
(5%)
29,440
(5%)
6
Pacific
114
73
31,988
(10%)
89,358
(15%)
Total
4,227
1,399
305,868
584,381
(1) Gas stations plus other fuel establishments
(2) County Business Patterns data, 1981
(3) Percentages may not add to 100 due to rounding.
* _ 1 -7
i i —
-------
fable A-6. Weighted data from sampled PSUs, urban versus rural summary
(percent distribution in parentheses)
Urban/
Rural
Number
of
counties
Number
of
PSU's
Sampling
measure
of size (1)
Large
establishments
(>20 empl.)
1981 CBP (2).
Urban
613
364
207,558
(68%)
462,468
(79%)
Rural
3,614
1,036
98,309
(32%)
121,913
(21%)
Total
4,227
1,399
305,867
584,381
(1) Gas stations plus other fuel-related establishments
(2) County Business Patterns data
-------
Table A-7. Unweighted PSU sample data, region summary
Survey
region
Number
of
counties
Number
of
PSll's
Sampling
measure
of size (1)
Large
establishments
(>20 empl.)
1981 CBP (2)
1
Northeast
13
6
5,453
9,051
2
Southeast
12
6
3,321
5,888
3
Midwest
14
6
2,317
6,555
4
Central
19
6
5,074
12,573
5
Mountain
10
4
1,144
3,058
6
Pacific
8
6
10,444
37,643
Total
76
34
27,753
74,768
(1) Gas stations plus other fuel-related establishments
(2) County Business Patterns data
-------
fable A-8. Unweighted PSU sample data, urban versus rural summary
Urban/
Rural
Number
of
counties
Number
of
PSU's
Sampling
measure
of size (1)
Large
establishments
(>20 empl.)
1981 CBP (2)
Urban
40
23
26,62?
73,305
Rural
36
11
1,126
1,463
P.-knfttimr>* j|
Totals|
76
34
27,753
74,768
(1) Gas stations plus other fuel-related establishments
(2) County Business Patterns data
-------
Figure A-2.
Sample PSU locations
-------
establishment sampling frame construction and establishment
sample draw are described in the next section.
III. ESTABLISHMENT SAMPLE
Once the 34 PSUs were drawn, lists of all establishments in
the three sampling sectors were constructed for the 76 counties
which comprise the 34 PSUs. These lists are known as sampling
frames. The initial sample of 2,818 establishments was drawn
from these frames and screened for eligibility. Since so little
was known initially about what type of establishment would have
underground motor fuel storage tanks, the eligibility rates
themselves were an early finding of the survey. The 896 eligible
establishments form the final sample for the survey. This
process is described in detail below for the fuel establishment
and large establishment sectors (which account for 2,218 initial
sample cases and 876 eligible cases). Appendix H reviews the
process for the farm sector (600 initial cases and 20 eligible
cases).
A. Sample Frames for Fuel-Related Establishments and
Large Establishments
The sample frames were constructed as described in Section
A-I, above. For the fuel-related establishments, several methods
of list-building were combined to result in a single list. A
list of government agencies with eligible tanks was developed for
each PSU by a telephone search. Federal, state and local
government officials were contacted to generate lists of all such
civilian agencies, and a list of military establishments with
eligible tanks in the sampled counties was provided to EPA by the
Department of Defense (DoD). A list of the fuel-related business
-------
establishments (gas stations and other industries, see list in
Table A-2) was purchased from National Business Lists (NBL) and
supplemented by any additional establishments with one of the
selected SICs that appeared on the purchased DMI list of large
establishments. The constructed government and military lists
were appended to the purchased establishment list to form the
fuel establishment sampling frame.
The large establishment sampling frame was purchased from
Dun and Bradstreet's list of business establishments, the Dunn's
Market Identifiers (DMI). This list source is more expensive
than NBL but was required since it contains the number of
employees for each establishment, which NBL does not. A list of
all establishments in the sampled counties with 20 or more
employees was purchased. The establishments on this list with
any of the fuel-related SIC codes were selected from the large
establishment frame and printed out. They were clerically
compared with the fuel establishment frame, county by county, and
any such establishment not already on the fuel establishment
frame was added to it.
Table A-9 shows the resulting frame counts by survey region
for these two frames. The counts show fairly good (by no means
perfect) agreement with the counts in Table A-7, based on CBP and
Versar data. For large establishments not in fuel-related
industries, the frame count is about 10 percent lower than the
CBP count. Region 6 (Pacific) shows a higher percent deficit,
about 15 percent, and also the bulk of the amount, 5,000 cases.
For the fuel establishment sample, the total measure of size in
Table A-7 (27,753 establishments) does not include any allowance
for government and military cases, of which there were 3,139 on
the frame. Subtracting these from the frame total leaves 30,583
establishments, or about 10 percent more than the sampling
measure of size. Table A-10 shows the frame counts broken down
* ... O *>
-------
Table A-9. Number of establishments on the frames
for 34 sampled PSUs (unweighted), by
survey region
Survey
region
Fuel
establishment
frame count
Large, non-fuel
establishment
(>20 employees)
frame count
1
Northeast
5,403
8,472
2
Southeast
3,023
4,811
3
Midwest
3,355
6,193
4
Central
6,027
13,227
5
Mountain
1,650
2,698
6
Pacific
14,264
32,677
Total
33,722
68,078
-------
Table A-10. Number of establishments on the
frames for sampled PSUs (unweighted),
by urban versus rutal
Type
of PSU
Fuel
establishment
frame count
Largenon- fuel
establishment
(>20 employees)
frame count
Urban
33,208
66,935
Rural
1,723
1,143
Total
34,931
68,078
A-2 5
-------
by urban versus rural PSUs, which agrees well with the breakdown
found in Table A-8.
\
B. Establishment Sample Draw
As described in Section A-I, above, the fuel establishment
and large establishment samples were drawn separately.
For the large establishments, a single national self
weighting sample of 600 establishments was drawn. The frame was
sorted by PSU and by number of employees within PSU. Each case
was given a measure of size in inverse proportion to the sampling
probability of the PSU it was in. A systematic sample (based on
a random start) of 600 establishments was drawn using probability
proportional to this measure of size.
The fuel establishment sample was drawn one region at a time
so that sampling could begin before all frames were completed.
The target number of 800 eligible establishments was allocated to
the six survey regions based on their sampling measure of size.
Based on early results for eligibility rates of government and
gas station establishments, and based on the relative proportion
of the frame in each region that fell into these two categories,
the target number of eligibles was inflated to an allocated
initial sample size for each region. The net result was an
approximate doubling of the sample size. The detailed figures
appear in Table A-ll.
-------
Table A-ll. Target sample size, by region, for
fuel establishment sample
Survey
region
Target number
of eligible
establistraents
Allocated
size for
sample draw
1
Northeast
213
449
2
Southeast
206
415
3
Midwest
165
325
4
Central
95
1^4
5
Mountain
37
75
6
Pacific
84
160
Total
800
1,618
A-2 7
-------
C. Eligibility Rates for Fuel and Larcre Establishment
gapple
Once the samples were drawn, they were screened for
eligibility. Table A-12 shows tjie initial sample draw and number
of eligible cases, by region, for both samples. There were
several possible reasons for a sampled establishment being ruled
out of the scope of the survey. Some establishments were found
to be not actually located in the sampled county (48 cases for
these two samples), out of business (85 cases), or ineligible for
other similar reasons (22 cases). Six were duplicates of another
sampled listing. Of establishments found to be in the survey
counties and in business, 97 had only abandoned tanks and 1,084
had no underground storage tanks, or stored only non-motor fuel
substances, leaving 876 eligible establishments.
Table A-13 shows weighted eligibility rates by type of
establishment for the survey regions and overall. It shows that
about 80 percent of sampled gas stations were survey-eligible.
Ineligible gas stations were generally out of business. Eighty
percent of government and military were eligible. Some had been
mistakenly included on the frame. Ineligible government cases
were generally out of area or storing non-motor fuel substances.
The other fuel-related industries category shows about one-
quarter eligible. Here, the out of business rates were lower
than for gas stations, and most ineligible cases had abandoned
tanks or no tanks. For large establishments the overall
eligibility rate was 13 percent. Almost all of the ineligibles
in this sample were establishments which simply had no tanks.
These varying eligibility rates show that although^
underground motor fuel storage tanks are concentrated in certain
industries, they occur in establishments in a broad range of
industries.
-------
Table A-12. Sample eligibility, by region, unweighted counts of sampled cases
Fuel establishments
Large establishments
Survey
region
Total
sample draw
Number of
eligible
establishments
Total
sample draw
Number of
eligible
establishments
1
Northeast
447
225
158
21
2
Southeast
413
197
116
18
3
Midwest
324
161
142
13
4
Central
193
92
68
7
5
Mountain
75
42
29
4
6
Pacific
160
83
87
13
Total
1,612
800
600
76
^"1,618 cases were drawn, but 6 were found to be duplicates during the
screening process.
A-2 9
-------
Table A-13. Weighted eligibility rates (percent eligible), by region and type of establishment
Type of establishment
Survey
region
Gas stations
(%)
Other
fuel-related
industries
(%)
Government
and
military
(%)
Fuel
establishment
sample combined
<%)
Large
establishments
(%)
1
Northeast
83
27
70
53
13
2
Southeast
81
19
85
51
16
3
Midwest
83
21
81
/
53
9
4
Central
79
23
89
54
10
5
Mountain
84
27
100
60
14
6
Pacific
86
30
82
59
15
Tptal
83
24 >
80
54
13
-------
IV. SUBSAMPLS OF ESTABLISHMENTS FOR TANK TIGHTNESS TESTS
The eligible sampled establishments had approximately 2,000
\
underground motor fuel storage tanks or manifold systems. A
subsample was drawn for physical tank testing. For the survey at
large, the target number of tank tests was 500. Fifty were set
aside for farms (during the planning stage, it was not known how
many farm tanks would be found), leaving 450 tank tests for the
subsample of fuel-related and large establishments.
At the time the subsample was drawn, it was assumed that a
manifolded system of two or more tanks connected by piping would
always be physically tested as one unit and therefore would count
as one test. During the process of doing the testing it was
found that, in fact, some systems were relatively simple to break
apart for testing, and this was done where possible. In this
section, we count tanks or manifolded systems; but in the
sections reporting on tightness tests, thq counts of individuals
tanks or of separate tests are generally given.
Table A-14 shows the allocation of the 450 tank tests by
survey region. This allocation is the estimated number of tanks
or tank systems to be tested for each category; some variation
occurred in the final sample since establishments rather than
tanks were the sampling unit. For the farms, the number of tank
tests depended on what was found during the interviewing and tank
test scheduling.
The allocation was made as follows. Of the 450 tanlj: tests,
40 were allocated to Region 5 to assure a minimum sample size for
that region. The remaining 410 tank tests were allocated to
Survey Regions 1-4 and 6 in approximately the same proportion as
-------
Table A-14. Subsarapling establishments for tank tightness testing (fuel and large establishments combined
List of eligible
establishments
Subsample for tank tightness testing
Survey
region
Number of
eligible
establishments
(at time of
subsampling)
Number of
tank systems
at eligible
establishments
Target Number
of tank systems^"
to subsample
Number of
establishments
subsampled
Number of
tank systems
at subsampled
establishments
1
Northeast
248
587
115
51
112
2
Southeast
214
544
110
47 _
111
3
Midwest
175
426
90
38
86
4
Central
100
231
50
23
52
5
Mountain
46
116
40
17
43
6
Pacific
96
207
45
22
46
Total s
879
2,111
450
198
450
i
^¦In allocating and drawing the subsample of establishments for tightness testing, a manifolded tank system was
counted as one unit. Some such systems were separated for physical tes ting.
-------
the fuel establishment sample allocation. Allocating the sample
in advance permitted us to draw the sample on a region by region
basis as the final eligibility results came in from the field
interview phase of the survey.
For each region, a sampling frame was created, containing
eligible fuel and large establishments at which tanks were found
(including establishments that refused to be interviewed). The
frame construction waited until all cases had reached a final
status and preferably had a known number of tanks or manifolded
systems. The frame contained the establishment ID, the number of
tanks or manifolded systems, and the establishment sampling
weight. This list was then sorted by number of tanks, then by
PSU (from ID), and then by fuel establishment versus large
establishment (also part of ID). The weights were cumulated down
the entire list. The number of facilities to select, Mj, was
based on the allocated number of tanks, Nj, and the weighted
average number of tanks per establishment, Tj, as shown in the
following equation:
Mj = Nj/Tj
The sampling interval, Slj, was the grand total of the
weights divided by Mj (Mj was not rounded). The sample was drawn
in systematic fashion, beginning with a random start between 0
and Slj. The establishments selected in each survey region have
a total number of tanks or manifolded systems close to Nj (see
Table A-14). Within each survey region, all underground fuel
storage tanks or manifolded systems have an equal probability of
selection for physical tightness testing.
A-i J
-------
v. FINAL SAMPLE WEIGHTS
A. Questionnaire Weights for Business and Government
Establishments
t
1. Other Fuel-Related SICs (Other Than Gas Stations)
The final questionnaire weights for establishments sampled
with fuel-related SICs other than gas stations were based on a
ratio adjustment of the initial sample weights for all such
screened establishments to 1982 CBP counts of these SICS,
followed by a nonresponse adjustment among the eligible other
fuel-related establishments to account for the few
nonrespondents. The adjustments were made by survey region. The
ratio adjustment served to put the initial sample on a known
basis, the number of establishments with one of the fuel-related
SICs in each region. Then the eligible cases weight up to an
estimate of the number of such establishments with eligible
tanks, by region. The nonresponse adjustment assures that the
weighted results based on questionnaires received will equal the
estimates based on screening results.
2. Gas Stations fSIC 5541)
The gas stations were weighted in the same way as other
fuel-related SICs. First, the initial sample was ratio-adjusted
by region to CBP totals for gas stations. The eligible cases
then weight up to an estimate of the number of gas stations"with
eligible tanks, by region. A nonresponse adjustment again
assures that the weighted results based on questionnaires
received will equal the estimates based on screening.
-------
3. Other Industries (Establishments With 20 or
More Employees^
The sample sector of establishments with 20 or more
employees in industries not otherwise sampled (the large
establishments) was weighted the same way as the gas stations and
other fuel-related industries. The~CBP totals of establishments
of this size in all but the selected fuel-related SICs (which
include SIC 5541, gas stations) were used for a region by region
ratio adjustment of the initial sample. The weighted eligible
large establishments then estimate the number of such
establishments with eligible tanks in the country, by region.
Since all eligible large establishments participated in the
interview phase of the survey, no nonresponse adjustment was
needed.
Table A-15 shows the totals based on 1982 County Business
Patterns data which were used as the fixed totals the initial
sample weights were adjusted to sum to.
4• Government Agencies
No national statistics are currently available to estimate
the number of individual government agencies with underground
motor fuel storage tanks, which is the universe our frame was
built to cover. Therefore, no ratio adjustments can be made.
Nonresponse adjustments were made to account for the small amount
of nonresponse.
* -*>c;
-------
Table A-15. Known totals from 1982 County Business Patterns data
base used for ratio adjustment
Type of establishment
Survey
region
Gas station
(SIC ¦ 5541)
Other selected
fuel-related
industries
Large
establishments
(> 20 employees)
not in selected
industries
1
Northeast
28,212
42,173
158,320
2
Southeast
22,623
29,825
109,137
3
Midwest
27,551
37,391
131,769
4
Central
12,473
17,786
67,150
5
Mountain
6,100
7,881
30,129
6
Pacific
13,840
18,565
84,998
Total
110,799
153,621
581,503
-------
B. Physical Test Result Weights for Business and
Government Establishments
After calculating final questionnaire weights for all
responding establishments as described above, the sampling
weights for establishments chosen for physical testing were
adjusted to sum to the estimatfed totals for four establishment
types (government, gas station, other fuel-related, and other
industry) by region. This adjustment was made by an iterative
rating procedure in which the weights were adjusted first to
\
regional totals, then to establishment type totals, then
readjusted to regional totals, and so forth, until no further
adjustment was needed. This took five and a half iterations to
achieve.
A final adjustment was made for tank test result weights.
If all selected tanks had been tested, the weight for an
individual tank or tank system test would be equal to the
establishment physical test weight. However, some tanks were not
tested. Thus a "tank nonresponse" adjustment was made to the
tank/tank system weights to account for the untested tanks. A
single tank counted once (added its weight) in the count of tanks
selected and once in the count of tanks selected. A manifolded
tank system which was not tested counted once for each tank in
the count of tanks selected. A manifolded tank system which was
broken apart and tested as separate tanks also counted once for
each tank in each count. A manifolded tank system which was
tested as one system counted once for each tank in the count of
tanks selected and once for each tank in the count of tanks
tested. The ratio of the weighted count of tanks selected to the
weighted count of tanks tested was used to form the final -
adjustment to tank weights. This was done over the sample as a
whole rather than by region.
J /
-------
C. Farm Questionnaire and Physical Test Weights
Due to the distribution of farms within the survey regions
(both overall and in our sample).and the low yield of eligible
farms from the screening, for weighting and any regional analysis
I
purposes the survey regions have been consolidated into three
areas for farms (see Appendix H). These are:
o East (combines Northeast, and Southeast Survey Regions);
o Midwest; and
o West of the Mississippi (combines Central, Mountain and
Pacific Survey Regions.
Total counts of farms for these areas were obtained from the 1982
Census of Agriculture and used to form ratio adjustments for
eligible farms. Due to one refusal among farms, a nonresponse
adjustment was also made.
Since so few farm tanks were tightness tested, no weighted
estimates will be presented for that data, and hence final
weights were not prepared for physical test results for farm
tanks.
VI. VARIANCE ESTIMATION
A. Jackknife Approach to Variance Estimation
In a complex survey such as this one, it is difficult or
impossible to estimate the variance of survey estimates directly
from algebraic formulas. An alternative approach often used, and
A-3d
-------
adopted for this survey, is the so-called jackknife method of
variance estimation through replication. The idea behind the
method is to draw a collection of subsets of the sample, called
replicates, and use the subsets to form national estimates of the
statistic whose sampling variance is being estimated. The
variability among these estimates is used to estimate the
sampling variance of the estimate based on the full sample. [See
Sampling Techniques. 3rd Edition, y.B. Cochran, J. Wiley & Sons,
1977 for a brief discussion of the principles of this method.]
B. Replicate Formation
To form the replicates, the sampled PSUs were paired and one
PSU dropped from each pair in turn. Since there were 34 PSUs,
there were 17 pairs and 17 replicates. The pairs were formed as
follows. Thirty-four PSUs were drawn in six survey regions.
Except for one certainty PSU in Region 6, they were paired into
strata in straightforward fashion — PSU 1 with 2, PSU 3 with 4,
and so on. Region 6 required some special consideration. The
sample in the region consisted of PSUs 29 through 34, with PSU 31
being a certainty PSU. PSUs 29 and 30 were paired.
Establishments in PSU 31 were separated into "odds" and "evens"
and these sets were treated as a pair of PSUs. This left PSUs
32, 33, and 34 to consider. These three PSUs were grouped into
one stratum; PSU 33 was randomly paired with 32, giving the
paired PSUs 3/4's their initial weight; and PSU 34 was given
3/2's its initial weight. Then either the singleton or the
paired PSUs are randomly selected to be dropped for one
replicate.
The resulting strata and random selection of which i>SU to
drop from each stratum, in turn, to form a replicate (17
replicates in all) are shown in Table A-16.
-------
Table A-16. Definition of strata and replicates for
jackknife estimation of variance
Stratum
PSU 1
PSU 2
1
PSU
to drop
1'
1
2
1
2
3
4
4
3
5
6
6
4
7
t
8
8
5
9
10
9
6
11
12
12
7
13
14
14
8
15
16
16
9
17
18
17
10
19
20
19
11
21
22
22
12
23
24
23
13
25
26
26
14
27
28
27
15
29
30
29
16
31, odds
31, evens
31, odds
17
(32 & 33) (3/4 's)
34 (3/2's)
(32 & 33)
a-40
-------
C. Jackknife Replicate Weights and Variance Estimates
Seventeen replicates were formed by dropping a randomly
selected PSU from each stratum, in turn. Weights were calculated
for each replicate as follows. As an initial sampling weight for
the replicate, establishments at the selected PSU of a pair were
assigned twice their initial weight, while establishments in the
dropped PSU were assigned zero. Establishments in all other PSUs
kept their initial sampling weight. Then the ratio adjustment to
CBP totals by industry type and region and the nonresponse
adjustment by the same categories were done as described above
for the full sample final weights. For tank test replicate
weights, the subsampled establishments in the replicate had their
weights adjusted by raking to the replicate total by region and
industry type, and replicate tank test nonresponse adjustments
were made. Repeating all steps of final weight adjustment in
calculating the replicate weights ensures that the variance
estimates will reflect the impact of weight adjustments on the
variance.
Subscripting the 17 replicates by r = 1 ..., 17, the
A
variance of a national estimate, X, of a statistic X is given by:
a2 17 A A 2
sx " *
-------
all tanks that leak) but any statistic that can be estimated from
the full sample and from each replicate in turn.
-------
APPENDIX B
SURVEY PROCEDURES AND ELIGIBILITY AND RESPONSE RATES
I. IN-PERSON INTERVIEW PRETEST
In July and October of 1984, survey packages (including
introductory letter, questionnaire, general instruction booklets,
and inventory forms) were mailed to a pretest group of 10
establishments which were previously determined to have
underground storage tanks in use. They were selected through
liaison with local government and military officials rather than
by random sampling or from developed survey listings. The July
pretest group consisted of seven "fuel-related" establishments
and the October group included three military installations.
Using government-operated establishments in the pretest allowed
us to prepare for problems not normally encountered in non-
government situations. The purpose of the pretest was to
evaluate the format and wordings of the questions in the
interview for clarity and administerability; to determine the
length of administration time for the interview; and to assess
specific and overall response to the flow of the interview and
individual items in the interview. In addition, several on-site
procedures were tested including meter testing, tank sticking,
site diagraming and soil sampling. Several revisions to
materials and adjustments to on-site procedures were made prior
to the field period. No results from the pretest are included in
the final estimates of the survey.
B-l
-------
II. WESTAT TELEPHONE PRSSCREENING AND LIST CONSTRUCTION
PROCEDURES
Since lists of establishments with underground motor fuel
storage tanks do not exist, it was necessary to develop
establishment frame lists for each of the 34 PSUs. As described
in detail in Appendix A, the universe of all establishments with
underground motor fuel storage tanks was divided into three
segments: the fuel-related establishments, large establishments
(with more than 20 employees), and farms. Lists of fuel-related
establishments, large establishments and farms were purchased or
obtained for the 34 PSUs in the survey. Since a list of
government establishments and locations was not available, a
telephone list construction procedure (described in
Section B-II.B below) was used to construct government tank
establishments lists in the 34 PSUs. In the 34 PSUs a sample of
1,618 fuel-related establishments (including government and
military establishments), 600 large establishments, and 600 farms
was drawn to be surveyed. Since eligibility rates were expected
to be low (less than 50%) telephone screening procedures were
implemented using the Westat Research Telephone Center in order
to determine which farms and large establishments were "eligible"
(had underground storage tanks) for the survey. (Fuel
establishments, including government and military establishments
were screened in the field.)
A. Government Tank Establishment List Construction
Because there is no central listing source for government
establishments with underground motor fuel storage tanks,
federal, state, county, and city lists were developed using
extensive telephone research. Initial contacts with officials at
. b-z
-------
different government levels (i.e., state and county Fire
Marshall's Office, Public Works Department, local Police
Department) provided the telephone interviewer with the location
of underground storage tanks or referrals to other contacts who
could furnish information on underground storage tank locations.
Hard-copy listings were accepted by mail if the data was too
extensive to be given over the phone. After all leads were
exhausted, using a minimum number of calls, and the lists were
determined to be complete. They were then added in as part of
the fuel-establishment sample frame.
B. Farm and Large Establishment Screening
Using the farm and large establishment sample lists,
telephone interviewers contacted the owner or operator of the
establishment and asked whether the farm or business had any
underground storage tanks in use to store motor fuel. For those
establishments which were eligible, a contact name was obtained
to assist the field interviewer. All establishments that could
not be located by phone (19%) or refused the screening interview
(1%), were included in the field screening efforts. All but two
percent of the farms and large establishment that could not be
located or screened by telephone were located and screened in the
field screening effort.
III. UST SURVEY MAILOUT
The mailout for the UST Survey began on November 26, 1984
with survey Region 6 (West coast) and continued in phases working
through Region 4 (Southwest), then Region 2 (Southeast), Region 1
-------
(Northeast) and Region 3 (central U.S.)- (See survey region map
in Figure B-l.) The last phase of the mailout was completed on
May 3, 1985, with packages being sent to Region 5 (Midwest).
Survey packages were sent certified mail to a sample of 1,965
establishments. Included in this sample were those farms and
large establishments which could not be located through the
initial Westat telephone screening. Survey packages were mailed
according to the schedule of the field interviewers, so that the
respondents received the survey materials approximately two weeks
prior to the interviewer's arrival at the site. The purpose of
the survey mailout package was to allow the respondent time to
prepare for the in-person interview.
Because the packages were sent certified mail, the date the
package was received and the name of the recipient was available
for the interviewer. The field interviewer used this information
to trace those establishments which could not be located by
phone. Each day, certified mail receipt cards returned were
keyed into an automated receipt control system (discussed in
Section 5-V.B). For survey packages returned by the post office
to Westat, a log was kept indicating establishment identification
numbers and reason for return. This information was passed on to
the interviewer, who then took responsibility for getting the
survey materials to the respondent. Eleven percent of the
packages were returned by the post office, and less than one
percent were refused. However, field interviewers were able to
contact nearly all of the establishments for which the package
was returned.
D-**
-------
Figure B-l
-------
A. UST Survey Package
Every establishment received the same package of survey
materials, which were labeled with the establishment survey I.D.
number, establishment name, and address. The package consisted
of the following items, which are included as Exhibits in
Appendix F:
o Open Letter to Owners and Managers of Underground Motor
Fuel Storage Tanks —tAn introductory letter that
informed respondents of the need and purpose of the
survey;
o "Certification Statement for Establishments without
Tanks" — A labeled form for the respondent to sign and
return to Westat if there were no underground motor
fuel storage tanks located at the establishment?
o "Reporting Responsibilities of Tank Owners and
Operators" letter — A one-page information sheet
quoting the amended RCRA regulation that requires
respondents to participate in the study;
o General Instruction Booklet — A booklet describing
procedures for completing the questionnaire and
inventory forms. A "Request for Confidential Treatment
of Business Information" form was included in the
instruction booklet for the respondent to sign if
necessary;
o Underground Storage Tank Survey Establishments
Operator's Questionnaire — One labeled copy was
included to be reviewed by the respondent prior to the
in-person interview;
o Inventory Sheet for Tanks with Metered Dispensing Pumps
and Dispenser Meter Recording Sheet — Six labeled
copies were included in the package so that the
respondent could begin to keep inventory prior to the
interview;
o Manifolded Tank System Recording Sheet — One labeled
copy was included in the package; and
o Inventory Sheet for Tanks without Metered Disposal
Pumps — One labeled copy was included in the package.
-------
A toll-free Westat "hot line" number was included in the
introductory letter as well as in the General Instruction Booklet
to provide survey assistance for the respondent.
IV. FIELD PROCEDURES
Fieldwork for the UST Survey began December 2, 1984. A
staff of seven field interviewers was trained to collect data
from the sampled establishment^. Between one and three
interviewers were assigned to cover a PSU depending on the
numbers of establishments sampled per PSU. The interviewer's
tasks in each PSU included eliminating ineligible establishments
using field screening techniques, and scheduling and conducting
on-site interviews. These procedures are discussed below in
Section B-IV.A and B-IV.B. On the average, work in each PSU was
completed in 15 days. The field phase of the UST survey
concluded on June 29, 1985. However, data collection efforts
through the mail and by telephone for incomplete cases continued
until November 18, 1985.
Field Screening
An interviewer's assignment list for a PSU consisted of a
call record folder for each establishment to be screened and
interviewed. (See Appendix F for a copy of the UST call record
folder). These lists included the farm and large establishments
which could not be located through the Westat Telephone Research
Center screening procedure. As a part of the initial
appointment-making telephone call or visit, the interviewer
determined whether the establishment did indeed have underground
-------
motor fuel storage tanks on site. In most cases, the interviewer
was able to determine whether or not the establishment was
eligible through an initial phone contact. Where phone contact
was not possible, the interviewer traveled directly to the
establishment site to speak with the respondent. Once
eligibility for an establishment was determined, the interviewer
then scheduled appointments for the in-person interview. Those
establishments that sent the signed "Certification for
Establishments without Tanks" prior to the beginning of fieldwork
in a PSU were taken off the interviewer's assignment lists, and
were not field-screened.
I
1. Statistics on Eligible Establishments
Table B-l shows the number of establishments which were
sampled, screened, and eligible for the UST Survey.
Approximately three percent of all farms and 13 percent of all
large establishments sampled were eligible for the survey (had
underground motor fuel storage tanks). Almost 50 percent of all
fuel-related establishments sampled were eligible. Reasons for
ineligibility are discussed in Section 5-IV.A.2.
-------
Table B-l. Number of sampled, contacted, and survey-eligible
establishments, by sample stratum
Farms
Large
establish-
ments
Fuel-
related
establish-
ments
Total
Number sampled
5981
600
1,6122
2,810
Number contacted
596
596
1, 608
2,800
Number of establish-
ments contacted
that have tanks
("eligibles")
20
(3.4%)
76
(12.8%)
800
(49.8%)
896
1600 farms were sampled. Two farms were found to be duplicates
in the telephone pre-screening.
21,618 fuel establishments were sampled. Six were found to be
duplicates in the field screening.
2. Statistics on Ineligible Establishments
When a sampled establishment was determined to be ineligible
for the survey the interviewer assigned an appropriate status
code on the establishment's call record, and notified the Westat
field director. Table B-2 contains the reasons for ineligibility
and their frequency of occurrence by type of establishment. The
majority of establishments were found to be ineligible because
they had no tanks. Approximately 95 percent of all ineligible
farms and large establishments fall under this category. Among
the fuel-related establishments ineligible, 73 percent h,ad no
underground storage tanks. All establishments in Regions 1
through 5 found to have no underground motor fuel storage tanks
through field screening procedures were instructed to sign and
-------
Table B-2. Statistics on ineligible establishments
Farms1
Large
Estab-
lishments1
Fuel
Estab-
lishments
Total
A.
No. of establishments
contacted
596
596
1,608
2,
800
B.
No. of ineligible estab-
lishments
576
520
808
1,
904
C.
Percent of establishments
contacted that were ineli-
ble
97%
87%
50%
68%
D.
No. of establishments with
no underground tanks
544 (94.4%)
495 (95.2%)
589 (72.9%)
1,628
(85.5%)
E.
No. of establishments with
abandoned tanks
13 (2.3%)
5 (1.0%)
92 (11.4%)
110
(5.8%)
F.
No. of establishments out
of business
11 (2.0%)
10 (1.9%)
75 (9.3%)
96
(5.0%)
G.
No. of establishments out
of PSU
6 (1.0%)
7 (1.3%)
41 (5.1%)
54
(2.8%)
H.
No. of establishments out
of scope of the survey
2 (.3%)
3 (.6%)
11 (1.3%)
16
(.9%)
^Statistics for farms and large establishments are a combination of the Telephone Research
Center and field screening results.
-------
return a statement certifying their establishment has no tanks
(See Appendix F). Of the 745 establishments in the survey
Regions 1-5 with no underground motor fuel storage tanks,
82 percent returned the "No Tank" form. For establishments in
Region 6 with no underground tanks, the interviewer went directly
to the site, observed there were no tanks, and picked up the
signed form from the respondent. This was a quality control
measure.to check the accuracy of the certification.
Establishments which had abandoned tanks, were out of
business, out of PSU, or out of the scope of the survey accounted
for about 15 percent of the ineligible establishments.
It should be noted that if an establishment moved from the
site sampled to a different location within the PSU, the
establishment was considered eligible and the interviewer
followed the establishment to the new location to conduct the
interview. Also, if the owner of the establishment had sold the
business, the current owner/operator was interviewed.
On-site Procedures
Once at the establishment the interviewer had several types
of data to collect. On-site procedures included an in-person
interview using the EPA Underground Storage Tank Survey
Questionnaire, a discussion on keeping inventory records,
checking the accuracy of the fuel dispenser meters, making fill-
pipe and drop-tube measurements, preparing or obtaining a site
sketch map, and locating the establishment on topographical maps.
The respondent was to gather the necessary data prior to the
-------
interview to prepare for the on-site visit as instructed in the
survey package.
1. The Call Record Folder
All information and associated material gathered from the
on-site visit of each establishment were kept in an individually
labeled call record folder (Appendix F) for that establishment.
The call record folder became the case jacket for the
establishment and was preprinted with forms for address and name
updating interview status reporting, contact and call recording,
interview procedures guidelines, and an interviewer debriefing
form. All materials, such as questionnaires and inventory
information, collected at an establishment were labeled with the
establishment identification number and filed in the
establishment's call record folder.
I
For each PSU worked, the interviewer received a package of
pre-labeled call record folders, each call record folder
representing a sampled establishment. The label placed on the
outside of each call record folder contained the establishment
name, survey I.D. number, mailing address, tank location address,
contact name, contact telephone number, and the county and state
the establishment was located in. Below this label, in the Label
Verification area, the interviewer noted any changes in the
original information on the label. These changes were entered
into the automated receipt control system described in
Section B-V.B. Also on the front of the call record folder, the
interviewer indicated the completion status of each on-site _
procedure. Printed inside each folder was a script the
interviewer followed which led him/her through the interview.
Also printed inside the folder were a set of debriefing questions
-------
which asked how willing and prepared the respondent was for the
on-site visit. A record of all calls to the establishment or the
respondent was kept on the back of the folder. Each call record
folder had additional survey identification labels stapled inside
to be used for labeling any materials or records received during
the interview.
2. The Questionnaire
The questionnaire body is divided into eight sections, with
each section focusing on a particular topic or concern.
o Section A: Establishment Descriptive
Information
\
Section A has two purposes. The first purpose of the
section was to describe the type of establishment that was being
interviewed. (Question Al was an industrial classification, for
example.) The second purpose of the section was to find and
"screen out" any remaining "out-of-scope" cases. Question Al had
a screening-out route for bulk fuel plants and private
residences, for example. (Private residences were completely out
of scope. Bulk fuel plants were only in scope if they had motor
fuel storage that was non-bulk, for use directly by motor
vehicles. Private residences and bulk fuel plants were asked to
call the Westat home office for instructions on how to proceed.)
Question A6 was another screening question. Naturally,
given the nature of the survey, establishments that did not have
-------
underground motor fuel storage tanks were not to be interviewed.
(Also, in Question A6 any underground storage tanks that were
permanently out of service or that were used only to store non-
motor fuels such as chemicals of heating oil were excluded.)
Question All was used as a lead-in to the Tank Description Sheet
(which is described below) and also asked the respondent to
provide or draw a map of the establishment. The primary purpose
of the map was to help the field interviewer establish the
location and linkages between the tanks, pumps, and meters at the
establishment. The tank testing crews also used the map to help
identify the tanks to be tested, as well as to correctly number
the tanks on their data forms.
o The Tank Description Sheet
The Tank Description Sheet is a two-page sheet containing
specific questions about each tank at the establishment. A total
of 44 items about each tank include questions on the amount of
fuel held in the tank, the materials of construction, year of
installation, safety features, leak history, etc.
Tank Description Sheet information is used in conjunction
with tank test results in order to learn more about the factors
and features of tanks that are associated with leaking. The
information from the Tank Description Sheets was also used by the
tank testing crews. For these reasons it was of great importance
that the tank identification number of the Tank Description Sheet
and the tank identification number on the map and the inventory
were all the same.
j_> ~ Jl *
-------
o Section B: Operating Practices
The particular focus of Section B is on the establishment's
typical inventory record-keeping and inventory management
practices. The interest here is in the establishment-kept
records, in factors associated with the accuracy of those
records, and in the kinds of records that were kept.
o Section C: Operating History
This section contains questions that fill in the
establishment's past tank history. The Tank Description Sheets
provide basic historical information about the tanks currently in
use. In Section C information is obtained on tanks that have
been replaced, removed without being replaced, or abandoned in
place, and in the number, the date and the reason for each of
these three actions.
° Section D: Permits and Licenses
Section D comprises two questions about permits and licenses
a respondent has to install and operate his tank.
° Section E: Installation
*
Section E is a short series of questions about the methods
by which the tanks were installed at the establishment.
-------
o Section F; Protection
Section F asks about the types of leak-protection,
corrosion-protection, and leak-detection devices that have been
installed at the establishment, and the kinds of operating and
maintenance practices for the devices.
o Section G: Information Needs and
Availability
Section G focuses on the kinds of information and training
relating to tank operating and monitoring that were available to
the respondent. Also included were questions which asked the
respondent about types of liability insurance held by the
establishment to cover sudden and non-sudden spills (and leaks)
of motor fuel. k
Interview responses varied depending on how knowledgeable
the respondent was and how willing he/she was to participate.
Often, it was necessary for the interviewer to speak with more
than one respondent to get enough information to complete the
questionnaire. In some instances, the interviewer was unable to
get any information from the on-site respondents at all.
Operators of establishments owned by multi-establishment
corporate structures occasionally referred the questionnaire to
their home office, which was always off-site and generally
outside the PSU where the interviewer was located. In these
cases, followup calls from Westat were made to obtain the
B-16
-------
completed questionnaire. Interview response rates are discussed
in Section B-V.F.
3. Reviewing the Inventory Sheets
After completing the interview with the respondent, the next
step for the interviewer was to review the inventory forms.
Included in the survey package the respondent received were four
kinds of motor fuel inventory sheets, a schematic diagram of the
seven most common tank and dispenser hookup systems (in the
General Instruction Booklet), and an Inventory Recording Table
(in the General Instruction Booklet) to help him choose the
correct inventory sheets to use for his establishment. The
respondent should have started keeping inventory on these forms
prior to the interview. Because of the complexity of the data
being gathered, the interviewer was instructed to always review
the inventory sheets with the person responsible for keeping
them. This was not always the same respondent who answered the
questionnaire. Depending on the size and type of establishments,
several people were sometimes involved in keeping the inventory
records. It was the interviewer's ^ob to make sure the
respondent understood the inventory process and was filling the
forms out correctly. If the respondent chose to provide 3 0 days
of previously collected inventory, the interviewer reviewed the
data carefully and made sure all the necessary information was
provided (or that the respondent knew what information was
necessary if previously collected inventory was to be mailed in
from another location, for example, a home office where all
records were kept).
Before reviewing the inventory forms, the interviewer had to
verify that the tanks and meters were numbered the same on the
-------
map drawn by the respondent in the questionnaire, in the Tank
Description Sheets, and on all inventory forms^ It was very
important to make sure these numbers corresponded in order to
link data from inventory forms and tank tests to the
questionnaire data. The interviewer used the Tank to Dispenser
Meter Fuel Line Connections Sheet (Appendix F) to cross-check the
linkage system. This was done at the actual physical location of
the tanks, where tank and meter numbers were positively
identified.
After the inventory review, respondents were told that
someone would be contacting them within the next two weeks to
check on the status of the inventory forms. They were given a
toll free 800 number to call if they had any problems or
questions with the inventory recording procedures. The
interviewer also gave the respondent a postage-paid pre-addressed
envelope for returning the completed forms. Inventory response
rates are discussed in Section B-II.F.
4. Checking Meter Accuracy
Once all tank and meter numbers were verified and inventory
sheets reviewed, the interviewer checked the accuracy of all
dispenser meters using a five-gallon certified meter calibration
can. For each meter tested, a calibration (adjustment) ratio was
recorded on the appropriate inventory form. Using this ratio,
the inventory records were adjusted by computer to account for
the meter error.
O-l Q
-------
The accuracy testing procedure was the same procedure used
by agencies that certify meter accuracy. The interviewer first
pumped approximately one gallon of fuel into the can to wet the
inside. This reduced the surface tension inside the can and
allowed for a more accurate measurement. After wetting the can,
the fuel was returned back into the appropriate tank and the
meter reset to the zero (0.0) reading. Next, the interviewer
pumped five gallons of fuel into the test can and read the level
of fuel according to the measuring gauge on the front of the can.
The can was used to measure error in liters or gallons. A
"calibration ratio," which equaled the gauge reading divided by
the amount pumped into the can, was recorded for each meter
tested. The ratio was recorded in "cubic inches" (in3) if the
fuel was dispensed in gallons or in "milliliters" (ml) if the
fuel was dispensed in liters. A negative (-) or positive (+)
sign was always recorded with the ratio, to indicate whether the
pump was dispensing less or more than the amount indicated by the
meter.
After recording the calibration ratio, the interviewer
returned the fuel to the tank from which it came. The
calibration of all meters associated with the same tank were
checked before going to the next. If the respondent had already
started keeping inventory records, the amount of fuel returned to
the tank was recorded as a "delivery" on the inventory sheet, in
order to balance with the meter readings in the inventory
records.
5. Measuring the Fill Pipe/Drop Tube
The next procedure after checking meter accuracy was to
measure the diameter of the tank fill pipe. The interviewer also
-------
had to determine whether or not a drop tube was present inside
the fill pipe and, if present, whether the drop tube was
permanent or removable. This was done for each underground
storage tank and the data recorded on the Site Observations
Recording Sheet (Appendix F). This information was collected by
the interviewer to help prepare the MRI crew for tank tightness
testing. Certain factors, such as the size of the fill pipe or
the presence of a permanent drop tube hindered or prevented a
tank test. Knowing this beforehand, the crew was prepared to
solve the problem once on site for the test.
6. Map Reading
The interviewer was provided with topographical maps of each
PSU, which were included in the package with the call record
folders for establishments to be interviewed. These are U.S.
Geological Survey maps and are graphic representations of
selected man-made and natural features of the earth's surface
plotted to definite scales. Such maps record physical
characteristics of the terrain as determined by precise
engineering surveys and assessments. Using a standard symbol
guide to help read the maps, the interviewer located the tanks on
the map, circled the location, and identified it using the survey
I.D. number for that establishment. The interviewer returned the
unused maps to EPA. The maps with tanks located on them were
returned to Westat, where they were reviewed to make sure all
establishments for that PSU were mapped, copied, then sent to
EPA. Using the precise longitude and latitude of the tanks from
the map, soil characteristics and other physical characteristics
of the site could be matched to the tanks specific for that
location. There were fewer than 20 sites for which USGS
topographic maps could not be obtained, and these were covered to
-------
the extent possible by local street or road maps. The data
obtained through the map linkage are discussed in Appendix H.
C. Interviewer Evaluation
Immediately after leaving the site, the interviewer
completed the debriefing questions printed inside the call
record. These eight questions were used to evaluate the overall
character of the interview and the cooperation and knowledge of
the respondents. Table B-3 shows the debriefing statistics for
the 890 establishments surveyed.
-------
Table B-3. Debriefing statistics
Percent
Percent of respondents who had
questionnaire completed prior to interview 28%
Percent of respondents who had
inventory sheets started 12%
Percent of respondents who had problems
or errors in completed parts of inventory 31%
Percent of respondents who
understood inventory process 98%
Percent of respondents who understood
most/all questions in questionnaire 99%
Percent of respondents who were cooperative 94%
Percent of respondents who were hostile 3%
Percent of respondents who were guessing
a lot in answering interviewer's questions 4%
Percent of establishments where it was
necessary to talk to more than one
person to obtain all required information 29%
Less than one-third of the respondents had prepared for the
on-site interview by completing the questionnaire prior to the
interviewer's arrival on site. Only 12 percent had started
keeping inventory records prior to the interview. Of those
respondents who had started keeping inventory records, the
interviewers found that 31 percent had errors in the completed
parts of the inventory. Almost 100 percent of the respondents
*>
understood the inventory process and the questions in the
questionnaire. In approximately 30 percent of all cases it was
necessary to talk to more than one respondent to obtain all
-------
required information. Even though most respondents were
unprepared for the survey, 94 percent were willing to cooperate.
After completing the debriefing questions, the interviewer
made necessary name and address changes to the label in the Label
Verification section of the call record. If it was necessary to
talk to more than one respondent, a contact name and phone number
for each respondent interviewed was written on the front of the
call record. The interviewer then assigned a questionnaire
completion status for the case and circled the appropriate
completion status codes for the inventory record keeping, the
meter accuracy test, the site mapping, the debriefing, and the
confidentiality request form. After checking to make sure that
all materials in the call record were properly labeled and
editing the questionnaire for completeness, the interviewer
returned the completed case to Westat, where it was reviewed and
entered into the receipt control system (discussed in Section
B-V) .
D. Refusals
Each sampled establishment received a survey package
containing a copy of the Resource Conservation and Recovery Act
(RCRA) amendments to Section 9005(a) stating that the
responsibility of the tank owners and/or operators to furnish
information for the UST Survey. Nevertheless, a small number of
respondents still refused to participate. When an interviewer
encountered a refusal to participate either over the phone or in
person, he/she told the respondent that the EPA legal office
would be informed of the refusal. The interviewer then contacted
the Westat field director immediately. The field director
notified EPA's Office of Enforcement and Compliance Monitoring of
^3
-------
the refusal by phone and by letter. In most cases, the
respondent agreed to participate after a phone call from an EPA
attorney. In other cases, a warning letter from the Waste
Enforcement Division was sent to the respondent when a phone call
did not result in cooperation.
Some respondents refused to participate in any part of the
interview, while others only refused to keep the inventory
records. The number of interview and inventory final refusals is
shown in Table B-4, lines F and J respectively. Overall, less
than one percent of respondents refused to complete the interview
and less than two percent refused to complete the inventory
recordkeeping.
When a respondent who had initially refused the interview
decided to participate (either as a result of a phone call or
enforcement letter) the Westat field director was notified. If
the field interviewer was still on site in that PSU, an interview
was set up with the respondent. If the interviewer had already
left the PSU, the person assigned to "clean-up" (see
Section B-IV.E) these special cases made the appointment and
completed the interview.
E. Interview "Clean-Up"
It was necessary to use a "clean-up" interviewer who
followed behind the field teams, to handle special circumstances
when all on-site procedures could not be completed during the
W- 4L +
-------
Table B-4. Field interviewing and inventory status statistics
Farms
Large
Estab-
lishments
Fuel
Estab-
lishments
Total
A.
Number sampled
5981
600
1,6122
2,810
B.
Number contacted
596
596
1,608
2,800
C.
Number of establishments contacted
that have tanks ("eligibles")
20
76
800
896
D.
Number of interview responses
19
76
795
890
E.
Response Rate (percent of elidible
respondents who completed interview
95%
100%
99.4%
99.3%
F.
Number of interview refusals
1
0
5
6
G.
Refusal rate (percent of elidible
respondents who refused interview)
5%
0%
0.6%
0.7%
H.
Number of inventory responses (includes
both complete and partial complete)
7
60
630
697
»
Response rate (percent of eligible
respondents who returned inventory)
35%
79%
78.8%
77.8%
J.
Number of inventory refusals
6
1
8
15
K.
Refusal rate (percent of elidible respon-
dents who refused to record inventory)
30%
1.3%
1%
1.7%
It.
Number of delinquent inventory responses
4
9
131
144
M.
Number of establishments for which
inventory measurements are impossible
3
6
31
40
1 . *
600 farms were sampled. Two farms were found to be duplicates in the telephone
pre-screening.
21,618 fuel establishments were sampled. Six were found to be duplicates in the
field screening.
-------
time the interview team was working in a particular PSU.
Some of these special circumstances included the following:
o The respondent most knowledgeable of the underground
storage tanks was unavailable during the time the
original interviewer was in the PSU.
o The respondent had refused to participate and decided
to participate after the original interviewer left the
PSU.
o The business was closed due to seasonal operation when
interviews were being conducted in the PSU.
o The establishment was remodeling its underground
storage tank systems and could not provide all
necessary data at the time interviews were being
conducted.
o A calibration check could not be done due to adverse
weather conditions or seasonal operation of the
establishment.
o An establishment could not be located by the original
interviewer.
Work done by the "clean-up" interviewer accounted for five
percent of all completed interviews.
F. Field Interview Data Collection Statistics
Table B-4 contains data collection statistics for the field
interview portion of the survey. It covers statistics on
interview and inventory response and refusal rates.
-------
1. Interview Response Rate
The interview response rate for this mandatory survey is
nearly 100 percent overall, as well as for each sample segment.
Out of 2,800 establishments contacted, 896 had underground motor
fuel storage tanks, and were therefore eligible for the survey.
Of those, 890 or 99.3 percent completed interviews. The highest
response rate among the sample segments was among the large
establishments, where 100 percent of the eligible establishments
provided interview data.
2. Inventory Response Rate
Nearly 78 percent of the eligible establishments have
furnished complete or partial inventory data. Even this low
response rate was achieved only after extensive edit and followup
efforts by Westat's survey staff, sixteen percent of the
eligible establishments have not yet provided inventory records.
It was impossible for 4.5 percent of the eligible establishments
to keep inventory records. These reasons are discussed below in
Section B-IV.F.3.
3• Problems Preventing Inventory Record Keeping
Of the 896 eligible establishments, 40 were unable to
provide inventory records for any of their tanks using the
designated record keeping procedures. The reasons are listed
below.
-------
o No conversion chart — Twelve establishments were
unable to obtain conversion charts needed to convert
inches to gallons for their tanks because they did not
know the dimensions of the tanks or the company which
installed them.
o Bent fill pipes — Nine establishments were unable to
stick their tanks because the fill pipes were installed
with a bend to prevent pilferage.
o Facility closed — Seven establishments have closed
down since the time of the interview.
o Tanks abandoned/removed — Five establishments have
either removed or abandoned their tanks since the time
of the interview.
o No inactive period — Inventory analysis procedures for
tanks without meters to record the total product
dispensed consists of an analysis of volume measurement
changes for inactive periods. Four establishments,
which have tanks without meters, dispense fuel 24 hours
a day, so there is no inactive period to analyze.
o No wav to record deliveries — Two establishments
pumped fuel at irregular intervals from an above-ground
tank into the underground storage tanks with no means
of measuring the amount pumped into the tanks.
o No kev to tank — The locked tank of one establishment
was inaccessible due to delay caused by probation of
the estate of the tank operator, who died with the only
key in his possession.
V. DATA PREPARATION
Data preparation for the UST Survey began with a development
phase involving questionnaire layout and code manual design.
Inventory recording forms were developed by EPA. The coding
format, however, was designed by Westat. Operational phases
included document handling (including receipt control), <»
coding/editing, data entry, and machine editing. Location coding
from the topographic maps is discussed in Appendix H.
i>-/. <>
-------
A. Questionnaire and Code Manual Design
The questionnaire layout was designed for ease of data
preparation/data processing, as well as for ease of respondent
understanding and recording. Many items were designed as
•'precoded1' questions, that asked the respondent to answer by
circling a code to indicate his/her response. This eliminated
the need for a coder to translate check-marks or other non-code
symbols into coded answers. Computer field positions were
printed in the questionnaire for most data items. These field
positions were useful as reference locations for coders, machine
editors, and data entry staff.
A detailed code specification manual using an automated code
book formatting program for the UST Survey Questionnaire was
developed. This manual described the data to be encoded from the
questionnaire, item by item. Figure B-2 lists the item
characteristics by which each data item was described in the code
manual. Figure B-3 is an example data item description from the
Underground Storage Tank Survey Establishment Operators
Questionnaire.
-------
Item characteristics described in code manuals
a. Field position and record number
b. Item name (the name by which the item was called in all
computer programs and other documentation)
c. Quotation of the item from the questionnaire
d. List of all code values and their definitions
e. List of reasons for legitimate item nonresponse (the
"inapplicable" definition)
f. List of all missing value codes
g. Flags indicating logical relationships between the item and
subsequent items.
Figure B-2. Item characteristics
described in code manuals
F2A 020-022 EBEtUEbCI-CE.IHSEECIIQb
~ ~ * INAPPLICABLE# COOED 2# 8, C« 9 IN MA,
CCL 016, REC 09; OR C00EO 1/ ( OM IK
B0XF2, COL 19, «EC 09
001-365 ¦ FREQUENCY CF INSPECTION
* 998 ¦ DCK'T KNOW
* 999 * NCI A5C EST»INED
* SHIP P2UC» COL C23-C24, REC 0»
Figure B-3. Code manual data item description
-------
B. Inventory Data Editing
Inventory forms were designed to include "worksheets" for
respondents to record individual meter and manifolded tank
readings, and then to record the sums of the individual readings.
Both the individual readings and the summary readings were edited
and key punched. The raw data collected from the inventory
recording process was entered or key punched (Section B-V.D)
directly from the edited inventory forms. A code manual and
editing instructions detailing the layout and the valid code
ranges for the inventory forms was prepared to assist the editors
and the data entry operators.
C. Receipt Control
All returns were tracked by Westat's automated receipt
control system. Each day, documents received, including
certified mail cards and "No Tank" Certification statements, were
keyed into the system. All documents from an establishment were
linked by a survey identification number specific to that
establishment (discussed in Section B-V.B.l). Using this I.D.
system, returns were tracked by type of document, and reports on
the survey status and on an individual establishment status were
produced.
For each document received, the date of receipt, a status
code and "batch" number (Section B-V.B.2) were entered into the
receipt control system using the procedure specific for that
document. In addition, any name or address changes from ,the-call
record were also entered upon receipt of a questionnaire from the
field.
r»_ •> 1
-------
1. Survey Identification Number
The survey I.D. number is a ten-digit number which shows the
sampling frame from which the establishment was selected, the PSU
in which it is located, and a sequential number. The survey I.D.
uniquely identifies the establishment within the survey and links
all documents and data records for the establishment.
2. Questionnaire and Inventory "Batching"
Questionnaires and inventory forms were "batched" into
groups of 10 documents for coding, editing, and filing purposes.
Each batch was given a number, which was written at the top of
the Batch Control Sheet (Figure B-4), as well as on the
questionnaire or inventory form. Questionnaires and inventories
were batched separately. Listed on the Batch Control Sheet were
the survey I.D. numbers of all the questionnaires (or
inventories) and their statuses for that specific batch.
Questionnaire and inventories remained in their batches until
they were coded and sent to data entry. If they were removed
from the batch for any reason, the date, person taking the
document, and reason were noted on the front of the Batch Control
Sheet. A copy of each Batch Control Sheet was kept in a log for
quality control purposes for both questionnaire and inventory
batches.
-------
batch control sheet
BATCH
ID LABEL
STATUS
CHECK OUT
TO/ON
DATE
RETURNED
VERIFIED
BY
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
COOERs
DATE:
VERIFIER:
DATE: % VERIFIED:
a lire B - 4
B-33
-------
D. Coding/Editing
A staff of six coder-editors was trained to code the
questionnaire and inventory. The initial training session
covered procedural matters as well as specific coding of the UST
Survey Operators Questionnaire and the four types of inventory
recording forms. It included an item-by-item discussion of the
coding of the documents, practice coding examples, and group
review of the coding of practice examples. Training materials
included code manuals, practice inventory and questionnaire
examples, and a marked-up version of the questionnaire that
linked the questionnaire to the code manual and the general
coding instructions.
Coders were trained to edit questionnaire responses and
inventory records for consistency and completeness as they were
coding them. Coders flagged any problems they discovered during
coding, and referred the problem documents to the coding
supervisors. Some problems required the development of new codes
— such as when different units of measure than those specified
in the questionnaire were specified for quantity questions.
Other problems required that the respondent be called to verify a
response or provide missing information (a process called "data
retrieval"). In some instances, decisions could be made based on
the evidence available, by the Project Officer or by other EPA
staff. Decisions, both general and case-specific, were recorded
in a Decision Log for future reference.
All coding was 100 percent sight verified by a senior coder
or the coding supervisor prior to being sent for data entry.-
a-34
-------
E. Data Entry
Questionnaire and inventory data were entered ("key
punched") by highly trained data entry operators, using a key-to-
tape entry system. This key-to-tape system is computer-driven
and provides a formatted entry keying program that minimizes many
types of data entry errors. All data entry was 100 percent key
verified by a different operator from the entry operator.
The questionnaire booklets and inventory records were sent
to data entry in "key batches." Lists of the survey I.D. numbers
associated with each key batch of inventory records were made and
put into a Key Batch Control Log. All questionnaires sent to
data entry were checked off against a list of completed
interviews, which was generated by the receipt control system.
This enabled the coders to make sure that all questionnaires were
keyed.
F- Machine Editing
Machine editing is a means of data quality control that uses
a computer program to test item ranges, skip patterns, and
logical consistencies in a data file. Such a machine edit
program was prepared for the questionnaire and for the inventory
forms.
Machine editors were selected from among the trained coders
available from the coding staff. The training consisted of
procedural instructions, and a walk-through using an example edit
problem.
-------
The machine edit programs provided a list of test errors for
each edited case, as well as a listing of each case in error.
Each of the errors was checked, and often the hard copy of the
case was reviewed. Updates to the data files were written on
update sheets, key-entered and run against the data file to
produce a new master file. Then the edit cycle was rerun to make
sure that the update corrections had been made correctly.
Because of the complexity of some of the data files (particularly
inventory data files), it was necessary to rerun edit cycles
several times: updates to some fields tended to unexpectedly
impact consistencies with other fields.
After the final machine edit cycles, frequency distributions
for all items of the data files were reviewed by supervisors to
spot problems not captured by the machine edit programs.
VI. DATA RETRIEVAL
Data retrieval is the term used to refer to recontacting
respondents for the purpose of verifying or clarifying responses
to completed questionnaires for interviews. For this study, it
was necessary to recontact respondents for problems found in the
inventory records as well as questionnaires. These questionnaire
and inventory data retrieval procedures are discussed separately
below in Section B-VI.F.1 and B-VI.F.2. Part of the coding staff
was trained for the data retrieval process.
-------
A. Questionnaire Data Retrieval
Recontact of respondents for questionnaire problems
generally took the form of a telephone call, though occasionally
it was necessary to mail a list of questions to a respondent.
Approximately 60 percent of all respondents were recontacted for
questionnaire data retrieval.
B. Inventory Data Retrieval
Because of the complexity of the inventory record-keeping
procedure, each respondent received a "prompt" call by Westat
approximately two weeks after the field interviewer left the
site. The purpose of the call, which was made by a staff member
trained for inventory data retrieval, was to inquire about he
status of the inventory and when the records would be completed.
The prompt caller also assisted the respondent with any questions
or problems that may have occurred about keeping the inventory.
A large proportion of the inventory records received from
the respondents contained errors or inconsistencies ranging from
minor to major. When these problems were spotted by coder-
editors or coder-verifiers, the inventory form was flagged for
inventory data retrieval. The inventory data retrieval process
began with a phone call to the respondent with a discussion of
the problem. Some problems were resolved on the telephone.
Often, an explanatory letter and copies of the returned inventory
with problem areas marked were sent to the respondent. The
respondent then sent corrected inventory records back. It was
sometimes necessary to send multiple letters explaining the
problem before usable data was returned. Of the 697 inventory
B-37
-------
responses received to date, approximately 85 percent needed data
retrieval, four percent of which needed multiple data retrieval
efforts. At the writing of this report, there are still
establishments which have not yet responded to the data retrieval
efforts. They account for 25 percent of all cases needing data
retrieval.
C. Followup of Inventory Nonrespondents
After multiple prompt calls were made to inventory
nonrespondents, EPA sent a formal warning letter (Figure B-5) and
Status Report form (Figure B-6) to respondents who were
delinquent in returning inventory records. Of the 300 letters
sent, 25 percent did not respond and two percent refused.
As a result of all data retrieval efforts made by Westat and
EPA, 78 percent of all establishments have sent in inventory
records, but approximately 50 percent of all inventory records
received are complete enough for inventory reconciliation
analysis. Of the 896 eligible respondents, 16 percent have not
yet returned inventory records.
-------
£ A \
V^-vy ? UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
CERTIFIED MAIL
RETURN RECEIPT REQUESTED
office or;
PCSTlCtDCS ANO TOXIC »UifTANCK3
Dear
The Environmental Protection Agency (EPA) has been informed
by Westat, the Agency's contractor for the National Survey of
Underground Motor Fuel Storage Tanks, that as of July 31, 1985
the 30 days of motor fuel inventory data you are required to
provide EPA had not been received.
As was explained in the survey instructions mailed earlier,
Congress passed and President Reagan signed into law in 1984,
amendments to the Resource Conservation and Recovery Act (42
U.S.C., Sec. 6901) that require EPA to conduct this study. This
law also requires that you, as an owner or operator of an
underground motor fuel storage tank, provide EPA with the
information requested in this survey.
I wish to stress that the evaluation of inventory data is an
essential part of this National study, and EPA is requiring this
information from all establishments selected for the survey.
Failure to comply with this requirement may result in an
enforcement action.
Enclosed is a form for reporting the status of your 30-day
inventory data collection. We ask that you complete and return
the form within 24 hours of receipt to verify that you are
complying with this requirement. Simply check and complete the
correct inventory status block, sign and date the form, and mail
it in the enclosed self-return envelope.
Thank you for your cooperation.
Figure B-5
-------
; A i UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
MOTOR FUEL INVENTORY STATUS REPORT
Please complete this form and mail in self-return
envelope wihtin 24 hours.
I have completed and mailed my 30-day inventory data
Westat.
I am still collecting my 30-day inventory data and wi
mail it to Westat by ,
(date)
I have not yet begun my 30-day inventory data collect
but will do so immediately and mail it to Westat
by .
(date)
I need further instructions to complete and submit my
30-day inventory data collection.*
Other situation (please describe).
(Signature)
(Date)
~
*~hor.o toll free (800) 633
-89S5
B-40
-------
APPENDIX C
DEVELOPMENT OF A TANK TEST METHOD
This appendix is a summary of the report, "Development of a
Tank Test Method for a National Survey of Underground Storage
Tanks." The work was conducted under EPA Contract No. 68-02-
3938, Work Assignment No. 25.1
The appendix first summarizes the search for a suitable
tightness testing method and the reasons for the final selection.
Then the field procedures developed in the pilot test are
described. A more detailed description of the field tightness
test plan may be found in the test and analysis plan.2
I. SELECTING A METHOD
In preparation for the field tightness testing, MRI first
searched for a suitable test method. Their objectives were to
evaluate potential test methods to be used for the national
survey, to conduct a pilot survey using the test method selected,
and to develop a test plan for the national survey. The research
was conducted in five stages. The first stage consisted of a
1,1 Development of a Tank Test Method for a National Survey of
Underground Storage Tanks," H.K. Wilcox, J.D. Flora, C.L. Haile,
M.J. Gabriel, and J.W. Maresca, April 1986.
2"Test and Analysis Plan for the Tank Testing Program of the
National Survey of Underground Storage Tanks," H.K. Wilcox, J.W.
Maresca, Jr., J.D. Flora, C.L. Haile, June 10, 1985.
-------
review of current methodology for detecting leaks in underground
tanks. Second, field observations were made of several methods
in use. Third, of the several methods observed, five were
selected to be evaluated by conducting tests of these methods on
a single tank system at a closed service station. Three of these
five methods were selected for further evaluation in the fourth
stage by testing tank systems at four military installations and
at an operating service station. In the final stage, the method
chosen for use in the national survey program was tested in a
pilot study of 17 tank systems.
II. GENERAL METHOD SELECTION CRITERIA
The main criteria used to select a method for the national
program were:
1. Quantitative measurements were desired. However, this
did not preclude consideration of other approaches.
2. A detection level of 0.05 gal/h as established by the
National Fire Protection Association, Inc., was taken
as the target detection limit.
3. Minimal disruption to the station operation was
considered to be important.
4. The method and equipment had to be rugged for use on
the national survey.
5. The test should be applicable in a wide variety of tank
system configurations.
6. The method should allow a reliable assessment of
accuracy, precision, and sensitivity.
7. Costs for testing and data analysis had to be-within
the available budget.
8. sufficient equipment and manpower to conduct the
national survey were required.
v- -1
-------
The scope of the method selection research and pilot study
did not permit exhaustive method evaluation of all available test
methods in order to select a procedure with optimum
characteristics for all criteria. Hence, some compromise was
necessary to proceed expeditiously with the survey.
III. PRELIMINARY REVIEW AND TESTING
The methods reviewed in the first stage are shown in Table
1. Those for which further evaluations were conducted are also
indicated. The methods were classified into groups according to
their measurement characteristics.
Five methods were selected for further testing at a closed
service station in Kansas City. Brief descriptions of each are
provided below. A more complete review of tank testing methods
can be found in EPA's report.3
° The ARCO method utilizes a photo optical sensor to
monitor the level of a partially filled tank. If the
test conditions are set up properly, the device is self
compensating for temperature changes. Only the portion
of the tank containing the product is tested.
o The Certi-Tec method uses pressure transducers which
are located just below the surface of the liquid to
measure level changes. Seven thermistors are used to
measure temperature at various levels in the tank
during testing. The tank is overfilled during the test
by adding an extension to the fill pipe. Both the tank
and lines are tested at the same time.
3"Underground Tank Leak Detection Methods: A State of the Art
Review," EPA/600/2-86/001, January 1986.
-------
Table 1. Leak Detection Methods Reviewed
liter-
ature
Detection method review
Field Prelim- Devel-
site inary opment Pilot
visits testing study study
Volumetric
ARCO tank test X
Certi-Tec test X
Ethyl Tank Sentry X
Ezy-Chek X
Heath Petro-Tite tank X
and line testing system
Hydrostatic (standpipe) X
testing
Lasar interferometry X
Leak Lokator test X
Mooney tank leak detector X
Pald-2 leak detector X
Pneumatic testing (air X
test method)
Non volume trip.
Dye method X
Vacutect method X
Helium leak detection X
method
Tracer Research X
Inventory monitoring
Manual methods X
Automated X
External monitoring
Pollulert X
Remote infrared sensing X
Ground water and soil X
core samples
Underground radar X
XXX
X X
X X X X
XXX
X X
x
X X
X
-------
o Leak Lokator uses a buoyancy probe to monitor level
with a single thermistor located at the midpoint of the
tank. The method can be used to test a partially
filled tank (with lower sensitivity) or an overfilled
tank. Either the tank or the tank and lines can be
tested.
o The Petro-Tite method monitors level visually in an
extended fill pipe. The product level is returned to
the reference level at 15 minute intervals during the
test. The product is stirred continuously during the
test to achieve a uniform temperature. Temperature is
monitored with a single thermistor located at the inlet
to the pump near the top of the tank. The tank and
lines are all tested at the same time.
o The Varian helium leak detection method, a
nonvolumetric method, is based on the detection of
helium outside a tank which has been slightly
pressurized with helium. The tank should be empty
during the test if the entire tank is to be tested. It
is also helpful to drill a number of small holes in the
surface above the tank to assist in the location of the
leak. Pressure can be monitored simultaneously to
provide a quantitative estimate of the leak rate. The
lines are also tested at the same time.
A. Experimental Procedures
Each method was tested over a 2- to 3-day period. A leak
simulation system was designed and fabricated by MRI and used to
draw product from the tank at a known rate. The precision of the
leak simulator was at least an order of magnitude better than
that of the test methods. In testing the tank, the objective of
each test group was to estimate different simulated leak rates.
The leak rates measured by each method were compared with the
rates used in the simulation.
The data from the quantitative tank tests were analyzed to
determine the precision and accuracy of the tests. For these
analyses the accuracy of the test (or bias) was estimated by the
-------
mean of the (signed) differences between the leak rate reported
and the leak rate simulated. A paired t-test was used to test
the hypothesis that the method was unbiased; that is, that the
mean signed difference was 0. A linear regression of the
reported leak rate on the simulated leak rate was calculated. An
ideal regression equation in a tight tank would be y = 0 + l.Ox.
The scatter of the data about the regression line (correlation
coefficient, R) was used as an estimate of the precision of the
method. The bias and precision were combined to obtain an
estimate of the root mean squared (RMS) error.
B. Results
A summary of the statistical analysis for the quantitative
methods as a group is presented in Table 2.
1. ARCO Underground
The ARCO method was used for 15 different simulated leak
rates, including one zero rate. An average difference of 0.01
gal/h was observed between the rates reported by ARCO and those
calculated by MRI. This estimated bias in the results was not
significantly different from 0 (t = 0.21, 14 degrees of freedom).
The intercept did not differ significantly from 0 and the slope
did not differ significantly from 1. The R for the regression
was 94.3 percent, indicating that most of the variability of the
data was explained by the regression. The RMS error estimated
for the method under the conditions of the Kansas City test was
0.05 gal/h. The tests averaged just under an hour (55.7> min) in
length. In order to reduce the variability estimated with the
method, either repeated determinations or a longer test time
would be needed.
C-6
-------
Table 2. Summary of Statistical Analyses of Quantitative Methods
Tested at Kansas City Site
Standard
Method
na
Bias
Intercept
Slope
error
RMS
R2
ARCO
15
0.01
0.005
0.95
0.049
0.050
94.3%
Certi-Tec
12
-0.25
-0.30
0.71
0.166
0.302
38.9%
Leak
22
-0.01
-0.01
0.94
0.020
0.021
98.9%
Lokator
Petro-Tite 18 0.05 0.06 1.05 0.101 0.113 75.9%
an « number of simulated leaks.
-------
2. Certi-Tec Method
The Certi-Tec method was used for 12 simulated leak rates,
of which two were set at 0 and so represented the normal
condition of a tank test. The leak rates reported by the Certi-
Tec method took slightly over an hour (average 64.3 min) for each
rate. The estimated bias in the results (difference between the
reported rate and the simulated leak rate) averaged -0.25 gal/h.
This bias was quite large and was significantly different from 0
(t = -5.23, 11 degrees of freedom). The intercept differs from C
at the 5 percent significance level and the slope differs from 1
at the 5 percent significance level as well. The standard error
of the regression was 0.167 gal/h. The R of the regression was
only 38.9 percent, indicating that slightly less than 40 percent
of the variability in the reported leak rates was explained by
the simulated leak rates used in the test.
Thus this method, as implemented during this test, appears
to have substantial bias and relatively low precision. Even
though taking several repeated determinations of the leak rates
and averaging them would reduce the random error, the bias would
remain a problem.
3. Leak Lokator Method
The Leak Lokator method was used on 22 tests simulating leak
rates. Of these, three were zero simulated leak rates and so
represented tests of the tank without any simulated leak? Three
simulated leaks into the tank were also used. Using the method,
the average reported leak rate was 10.8 min.
C-8
-------
The bias in the determinations was estimated to be -0.005
gal/h, which was not significantly different from 0 (t = -0.23,
21 degrees of freedom). Although the estimated slope and
intercept agree closely with the ideal, both differed from the
ideal values significantly at the 5 percent level although not at
the 2 percent level.
These data showed a very small scatter about the regression
line, resulting in small estimated values for the standard error
of the slope, intercept, and regression. These small standard
errors led to the borderline significances of the difference
between the regression parameters and their theoretical values.
In light of the nonsignificance of the other test for the bias
and the small magnitude of both the intercept (-0.012 gal/h) and
the bias (-0.005 gal/h), these are probably not of major
importance.
4. Petro-Tite Method
The Petro-Tite method was tested under 18 simulated leak
rates, of which three were zero rates, corresponding to a tight
tank situation. While the usual Petro-Tite test takes an average
of four leak rates each reported over a 15-min period, only five
of these determinations were based on an hour's data. The
remaining leak rates reported were each based on a 30-min test.
From all the tests, the bias was estimated at -0.05 gal/h
but was smaller (0.040 gal/h) when restricted to the hour-long
tests. The bias from the complete set of tests is significantly
different from 0 at the 5 percent level but not at the 1 percent
level. If attention is restricted to the 1-h tests, the bias is
not significantly different from 0. The intercept is not
significantly different from 0, suggesting that the bias is not
-------
statistically significant. The slope does not differ
significantly from the ideal or theoretical value of 1 at the 5
percent significance level. The R for the regression was 75.9
percent and the standard error of the regression was 0.101. This
standard error is interpreted as the precision of a single leak
rate determination. It should be noted that the normal test with
four 15-min rate determinations should be somewhat more precise
than what was reported, and that precision could be improved
further by testing for a longer period of time and averaging more
individual leak rates reported.
5. Helium Detection Method
Two tests were conducted using the helium detection method.
In the first test the tank was tested in its original state.
Several large leaks were discovered during the first day's
testing, which were repaired. The next day's test revealed
substantial reduction in helium loss.
While some helium was detected around the tank, the amounts
were generally very small and could have come from pipe fittings
or the tank bungs. Low levels were, however, encountered in one
area. The concrete was removed for inspection purposes to see if
a line was located in that area. None was found, but helium
levels in the excavation were moderate.
The basic problem encountered using the helium detection
method is that helium can escape in measurable quantities through
threaded connections which have been poorly coated with sealer.
Gasoline will not normally pass through these poorly sealed _
connections at measurable rates under normal operating
conditions. This can lead to results which are hard to
interpret. In addition, no quantitative results can be produced.
-------
C. Conclusions
As a result of the preliminary testing in Kansas City, the
ARCO, Leak Lokator and Petro-Tite methods were selected for
further evaluation. The helium method was dropped because of the
decision that a quantitative method presented a better option for
the national survey. The Certi-Tec method was dropped because of
the prototype state of development and its relatively lower
performance.
IV. DEVELOPMENT STUDY TESTING
A. Experimental Procedures
Five facilities were selected by the EPA for tank testing.
A total of 13 tanks were tested. The initial plan was for each
tank to be tested by all three methods. Difficulties in
scheduling and plumbing problems at some sites, however,
precluded a complete round of testing.
Two types of tests were conducted at each sites: baseline
tests which were conducted in the same manner as if no
evaluations were being conducted, and leak simulation tests which
consisted of measuring leaks under a variety of simulated leak
rates (usually four). The process was nearly identical to that
described for the preliminary testing.
Three sets of data from the development study were analyzed:
baseline test data; leak simulations; and time series analysis of
the ambient volume fluctuations after the simulated leaks were
removed.
-------
The baseline data for each method was tabulated and compared
for each tank where more than one method was used to test the
same tank. Where differing conclusions regarding the tightness
of the tank were obtained, the data and conditions of the test
were further examined in an effort to resolve the conflict.
The data from the leak simulations were analyzed by fitting
a linear regression to the data from each tank and method by
regressing the reported leak rate on the simulated leak rate.
The intercept of this regression represents an estimate of the
leak rate of the tank or tanks system when there is no simulated
leak. The difference between the intercept of the regression
line and the test result from the baseline test provides an
estimate of bias or accuracy of the test. The variability of the
data about the regression line provides an estimate of the
precision of the test. Combining these two measures yields an
estimate of the mean square (or root mean square error)
associated with the testing method.
The third analysis consisted of a time series analysis of
the ambient volume fluctuations after the simulated leaks were
removed.
B. Results
1- ARCO Method
The ARCO method was used to test seven tanks during the
development study. Of these seven tanks, one tank had onlj£ the
baseline test run, one tank test resulted in the baseline test
and one simulated leak rate, and the other five tank tests all
had the baseline leak rate and several simulated leak rate tests.
-------
The baseline test results for ARCO are summarized in
Table 3. The ARCO result disagreed with the conclusion for three
tanks. However, it must be noted that the ARCO system tested
tanks approximately 75 percent full, under no additional head
pressure. Thus, the ARCO system provides a test most
representative of the usual operating conditions of the tank. If
a tank system has a hole in or near the top or fill pipe, or if
there is a leak in the lines, this would not result in product
leaking under normal operating conditions. While it may be
unlikely that all of the leaks encountered during the study are
in the top of the tank, it is a possible explanation.
A summary of the results from the leak simulation tests
using the ARCO method are summarized in Table 4. By this method
of testing, none of the tanks tested were reported to be leaking.
However, other test methods gave different results for some
tanks.
The data indicate, however, that the ARCO test method
performed well at the Damneck and Pitstop North test locations.
If a single data point that appears to be an outlier is removed,
the method also does reasonably well at the Langley facility.
One of the sites (Scott Tank 18) showed essentially no regression
of the reported leak rates on the simulated leak rates. This is
disturbing because for that test the method could not quantify
leak rates under the simulation. One other test, at Fort Lewis,
gave a slope substantially different from 1, which indicates that
an (unknown) interfering factor is present.
The ARCO method gave a precise determination of a leak rate
under some operating conditions. In other cases, it failed to
give valid results for reasons that were not understood. In
other cases, it failed to give valid results for reasons that
-------
Table 3. Summary of Baseline Results and Tank Tests Attempted
Facility and
tank
ARCO
Leak lokator
Petro-Tite
MR I
conclusion
Damneck +0.02 C®
Pitstop
1 (south) +0.02 C
2 (north) 0.0
Scott
1 (17)
2 (18) +0.02C C
Out of time
Ft. Lewis
1 (8C25 -0.04 C
north)
2 (8C25 0.0C C
south)
3 (4194)
4 (10E10)
Langley
1 (HS tank 3) —
2 (HS tank 5) Physical problem
with tank
3 (MoGas)
4 (Golf
course)
-0.03
-0.077 N
-0.741 N
(Poor sensi-
tivity)
-0.012 C
-0.299 N
-0.178 N
Problem,
possibly mani-
folded
Leak about
gasket-could
not test
-0.027 C
-0.172 N
(Poor sensi-
tivity)
-0.191d N
-0.448
•3.0d
N
N
+0.003
-2.892
-0.05
+0.004
-0.812
C
N
C
N
-0.342 N
-3.0 N
-0.024 C
-2.540 N
Tight
Leak
Tight
Tight'
Leakc
Tight
Leak
Leak
T i ght
Leak
Leak
T i ght
Leak
.Certifiable.
Noncertifiable.
.Test OK, but leak (possibly in upper part or piping) not found.
Test appeared OK, but data are inconsistent,
interactive effects between Tanks 17 and 18 were observed by Leak Lokator
- (negative sign) indicates leak out.
indicates testing was not conducted at that tank by the test company
indicated.
C-14
-------
Table 4. Results of Leak Simulation Tests Using arco Method
Tank
Baseline
rate
Intercept
Bias
Slope
SE
RMS
Damneck
0.02
-0.023
-0.003
1.049
0.022
0.023
Pitstop south
0.02
-
-
-
-
-
north
0.0
-0.092
-0.092
0.809
0.041
0.101
Scott 18
0.02
-0.145
-0.165
-0.044
0.099
0.192
Fort Lewis
8C25 south3
8C25 north
0.0
-0.04
-0.005
-0.094
-0.005
-0.054
1.140
0.493
0.047
0.072
Langley
MoGas
-0.03k
-0.03d
-0.336
-0.027
-0.306
0.003
0.419
1.167
0.367
0.118
0.478
0.118
Negative = Leak out
Positive = Leak in
Bias = Intercept - base
?Two points only.
Outlier removed.
c-15
-------
were not understood. It can detect inflow or outflow, but would
be defeated if the water table were at a level that approximately
balances the hydrostatic pressure of the product. It is also
subject to interference from wind and is sensitive to vibration.
It has the advantage of not requiring an overfilled tank, but
this is counterbalanced by the disadvantage of not being able to
detect potential leaks in the upper quarter of the tank.
The ARCO method was not recommended for use on the national
survey program for several reasons. The primary reason was the
decision to test the entire tank. Secondary reasons were the
sensitivity of the method to interference from vibration and the
relatively high frequency of tests that did not adequately
quantify the simulated leak rates.
2. Leak Lokator Method
The Leak Lokator method was used to test 10 tanks during the
development study. Of these, two tanks had only baseline tests
and no simulated leak tests conducted. The baseline test results
are summarized in Table 5. The Leak Lokator test conclusions
agreed with MRI's conclusion in 6 of the 10 tank tests. Of the
other four, the Leak Lokator test failed to certify three tanks
that had been concluded to be tight based on data from all test
methods and certified one tank that had been determined to be
leaking.
A summary of the results from the leak simulation tests
using the Leak Lokator method is presented in Table 5. The RMS
errors ranged from about 0.02 gal/h to 0.44 gal/h. The standard
errors ranged from 0.015 to 0.304. Among the tanks judged to be
tight, the standard errors ranged from 0.015 to 0.165 and the RMS
error ranged from 0.021 to 0.437. The large values for the upper
C-16
-------
Table 5. Results of Leak Simulation Tests Using Leak Lokator Method
Tank
Baseline rate
Intercept
Bias
Slope
SE
RMS
Oamneck
(@ 120")
-0.0775 @ 125
(+0.008 @ 118)
-0.0825
(-0.005)
-0.005
(-0.13)
0.786
0.025
0.0255
(0.028)
Pitstop south
-0.524
-
-
-
0.209
-
north
-0.012
-0.026
-0.014
0.879
0.015
0.021
Scott 17
-0.299
-0.366
-0.067
0.839
0.048
0.082
18c
-0.178
-
-
-
0.047
-
Fort Lewis
8C25 south
4194
10E10 NTC
TC
-0.027
-0.171
-0.191
-0.191
-0.010
-0.159
-0.596
0.069
0.017
0.013
0.405
0.260
0.734
0.749
0.541
0.835
0.097
0.026
0.165
0.098
0.099
0.029
0.437
0.278
Langley
HS 3
HS 5
-0.448
-3 or more
-0.641
0.126
-0.193
0.126
-1.78
2.43
0.048
0.304
0.199
0.329
Negative = Leak out
Positive = Leak in
Bias = Intercept of their (adjusted for base) regression
Intercept = Bias plus base
?NTC - not temperature corrected.
TC - temperature corrected by Leak Lokator.
cLeak Lokator observed interactive effects between Tanks #17 and #18 during
the testing of #18. The reasons for this are not understood.
-------
end of the range are from a test that had problems. If that data
point is excluded, the upper end of the ranges becomes 0.048 and
0.082. With the ability of Leak Lokator to obtain multiple leak
rate determinations fairly rapidly (about one every 10 to 15
min), one could presumably reduce these error estimates by making
several leak rate determinations at a tank and averaging them.
The Leak Lokator method gave valid estimates of leak rates
in most cases. The variability of a single leak rate measurement
tends to be somewhat large relative to a 0.05 gal/h criterion,
but the ability of the system to obtain leak rate determinations
in about 10 min once the test is running would allow multiple
determinations and averaging to reduce this variability. The
method has the advantage that its level monitoring system can be
used at any desired level (head pressure). Thus, if line leaks
are a problem, the testing could, in principle, be conducted
using a level below the piping to determine the location of the
leak.
The hydrostatic pressure from a water table could pose a
problem for this test. Testing did not appear to be standardized
to any specific product level. Since the leak rate through a
given aperture would change with head pressure, testing different
tanks at different levels makes leak rate determinations
difficult to compare and quantify.
3• Petro-Tite Method
The Petro-Tite method was used to test nine tanks during the
development study. The locations of these tank systems and -
reported leak rates were given in Table 3. Three of the systems
tested had leak rates so large (in excess of 5 gal/h) that
simulation of additional leak rates on the order of 0.2 gal/h was
C-18
-------
not feasible. Simulated leak rate testing was performed on five
tank systems.
The baseline tests conducted by Petro-Tite agree with the
conclusions reached by MRI based on analysis of all the data. It
should be noted that in some cases (e.g. Ft. Lewis #1) where
other testers experienced difficulties, Petro-Tite would have
also had difficulty.
A summary of the results from the leak simulation tests
using the Petro-Tite method is presented in Table 6. The RMS
errors ranged from 0.036 to 0.193 for tanks judged to be tight.
The 0.193 is rather large, but that tank posed special problems,
leading to the conclusion that the 0.193 is not representative.
Error estimates on tanks judged to be leaking were larger,
ranging up to 0.24 gal/h. Larger errors are to be expected for
systems with large leaks because large leaks make it difficult to
maintain product level and so therefore to obtain an accurate
volume. However, the errors remained acceptably low relative to
the associated leak rates.
As a result of the more detailed analysis of Petro-Tite
data, several suggestions for improving the errors involved in
the Petro-Tite method were developed. None of these involve
significant procedural changes. Improved algorithms could likely
result in better test results.
The Petro-Tite method seems capable of identifying and
successfully dealing with many types of interferences in tank
testing. Although there are situations that can lead to invalid~
test results, for the tanks tested in this study all tests but
one were believed to be valid. However, difficulties were
encountered that increased the error associated with the
estimated leak rates beyond that which is desirable. In
c-i^
-------
Table 6. Results of Leak Simulation Tests Using Petro-Tite Method
Tank
Baseline
rate
Intercept Bias
Slope
SE
RMS
Damneck
+0.003
-0.009
-0.012
1.01
0.052
0.054
Pitstop south
-2.89
-
-
-
0.240
-
north
+0.050
+0.069
+0.019
1.26
0.078
0.075
Scott 17
+0.004
+0.002
-0.002
1.075
0.036
0.036
18
-0.812
-0.774
0.038
0.608
0.109
0.115
Fort Lewis
8C25 south
4194
10E10
-0.342
-3.0
-0.024
-0.038
(Could not fill tank)
-0.014 1.50
0.107
0.193
0.193
Langley
golf course
-2.54
-
(Could not keep filled)
-
-
Negative = Leak out
Positive = Leak in
Bias - Intercept - base
-------
difficult cases, the error rates were such that one could not
reliably detect leak rates as small as 0.05 gal/h. Most of the
situations with large error estimates were cases where a
substantial leak was present, and hence the loss in precision did
not interfere with the detection of the leak.
4. Time Series Analysis of Ambient Noise Data
Because the data obtained from ARCO was not sufficient, time
series analyses were performed only on the Leak Lokator and
Petro-Tite data.
a. Description of Ambient Noise Analysis
The second analytical approach was to remove the simulated
leaks from the data to produce volume, temperature, and
temperature compensated volume time series that were longer than
normally used during a tank test. These data were analyzed to
determine whether the results obtained during a standard tank
test period (i.e., a baseline test) were consistent with longer
test times and to determine whether the temperature-estimated
volume changes required for compensation adequately accounted for
the total volume changes in a non-leaking tank.
Petro-Tite Method
Continuous time series of the change in volume and the
*'
change in temperature (converted to volume using the product
volume and the coefficient of thermal expansion) for an entire
day of Petro-Tite testing were generated from the data collected
every 15 min by subtracting the simulated leak volume from the
-------
measured volume. The volume change used for this 15-min interval
was an average of the volume changes observed before and after
this period. Cumulative time series of volume, temperature, and
temperature-compensated volume were then generated for analysis.
The temperature-compensated time series were generated by
subtracting the temperature (expressed in volume) from the
measured volume on a point-by-point basis. This is the same
method used by Petro-Tite. A least squares line was then fit to
each of the three time series to estimate the mean rate of change
of volume, temperature, and temperature-compensated volume. The
temperature-compensated volume was compared to the baseline test
results. The standard Petro-Tite data analysis method was used
to estimate the temperature-compensated volume rate for the
baseline tests (i.e., sum of the temperature-compensated volume
computed for four 15-min periods).
Leak Lokator Method
Time series of the cumulative volume and cumulative
temperature were generated for each simulated leak sequence of
the Leak Lokator data. Each time series ranged from a total of
40 min to over 100 min and included four to nine of the standard
Leak Lokator volume rate measure periods. The simulated leak
rate was subtracted from the uncompensated volume rate
measurements made by Leak Lokator and converted to volume using
the reported measurement time. These volume measurements were
then summed to obtain the cumulative volume time series. The
mean volume rate for each simulated leak sequence was taken from
the Leak Lokator data sheets. A continuous time series of
temperature was generated each day of testing from annotated
readings of temperature made every 5 to 10 min and placed on the
strip chart of temperature by Leak Lokator personnel. Those
sections of the temperature time series which bracketed the
r*_T>
-------
volume data for each simulated leak sequence were used in the
analysis to compensate for temperature. The temperature data was
converted to a volume time series and a least squares line was
fit to the data to estimate the average rate of change of volume
caused by the rate of change of temperature over an hour. A mean
temperature-compensated volume rate was then computed for each
simulated leak period by subtracting the mean rate of change of
temperature from the mean rate of change of volume and compared
to the results from the baseline test and the other simulated
leak test sequences.
b. Petro-Tite Ambient Noise Analysis Results
A summary of the mean and 95 percent confidence
intervals on the mean volume rate, temperature rate, and
temperature-compensated volume rate estimated from the long
Petra-Tite time series is presented Table 7. The rates were
obtained by fitting a least squares line to each time series.
The confidence intervals are based on the standard deviation of
the ordinate about the regression line. The site, tank number,
number of 15 min data points, and the test result using Petro-
Tite' s 0.05 gal/h detection criterion are also given. For
comparison, the baseline test result is added to the table.
Agreement between the baseline test results and the long time
series results is good, except for Pitstop Tank Ho. 2. The time
series from the Fort Lewis Tank No. 4 indicate that a potential
leak began several hours after the test had begun.
The time series of volume, temperature, and temperature-
compensated volume were generated by removing the simulated leaks
from the Petro-Tite volume time series. The time series are 3 to
6 times longer than the standard 1 h Petro-Tite test. The first
hour of each time series contains the baseline data. Several
-------
Table 7 . Summary of the Petro-Tite Analysis
Temperature
Temperature-
compensation
Baseline test results
Volume rate
(gal/!)
volume rate
(«al/h)
volume rate
(flal/h)
Temperature-
compensatec
Location
Tank
N
X
95% CI
X
95% CI
X
95% CI
Test result
volume rate
(gai/h)
Test
results
Damneck
1
20
0.043
0.008
0.064
0.008
-0.021
0.003
Tight
+0.003
Tight
Fort Lewi r
2
7
-0.28?
0.036
0.084
0.038
-0.371
0.050
Leaking out
-0.0342
Leaking out
Fort Lewis
k
4
22
9
13
-0.025
0.023
-0.104
0.022
0.035
0.027
0.017
-0.051
-0.006
0.025
0.133
0.040
-0.042
0.074
-0.098
0.027
0.117
0.435
Tight
Leaking in
Leaking out
-0.024
Tight
Pitston
2
19
0.151
0.010
0.013
0.009
0.133
0.011
Leaking in
-0.05
Tight
Scott AFB
1
24
0.184
0.017
0.190
0.014
-0.006
0.005
Tight
0.004
Tight
Pitstop
-1
8
-2.493
0.139
0.279
0.061
-2.773
0.089
Leaking out
-2.892
Leaking out
Scott AFB
2
16
-0.714
0.024
0.009
0.005
-0.722
0.028
Leaking out
-0.812
Leaking out
•First 2.25 h of the test,
last 3.5 h of the test.
Direction of flow only; not statistically significant from zero.
-------
observations about the strengths and weaknesses of the method can
be made from the data.
First, the time series for Damneck Tank No. 1 and for Scott
Air Force Base Tank No. 1, tanks declared to be tight,
illustrates the high correlation between the low frequency trends
of the temperature and volume data required for temperature
compensation. This suggests that the method of temperature
compensation, circulation of the product and measurement of the
rate of change of temperature with one temperature sensor, is a
reasonable approach.
Second, negative, high-frequency correlations were observed
between the temperature and temperature-compensated volume rate
time series for some of the tests. This suggests that the method
is overcompensating for temperature effects. These high-
frequency temperature fluctuations are probably caused by
inadequate resolution of the Petro-Tite temperature sensor. This
increase in the high-frequency fluctuations in the temperature-
compensated volume data can be a problem if the test time is too
short.
Third, inspection of the temperature-compensated volume rate
time series for each test suggests that a one-hour test is too
short to reliably detect small leaks. Within a test,
fluctuations with period of 30 to 90 min are observed which are
sufficiently different from the low frequency trend exhibited by
the entire time series.
Fourth, the time series for the tests conducted on Fort -
Lewis Tank No. 2, Scott Air Force Base Tank No. 2, and Pitstop
Tank No. 1 indicate that the tanks are leaking. The measured
temperature changes are too small to account for measured volume
changes.
-------
c. Leak Lokator Ambient Noise Analysis Results
A summary of the mean and 95 percent confidence intervals on
the mean volume rate, temperature rate, and temperature-
compensated volume is presented in Table 8. The site, tank
number, duration of the test sequence, the number of Leak Lokator
volume rate measurementss in the test sequence, and the test
result based on Leak Lokator's 0.05 gal/hr criterion are also
given. For comparison, the baseline test results are also shown.
Several observations about the data presented in Table 8 are
noteworthy. First, the test sequence results for each tank
tested are internally inconsistent. The results from five of the
six tanks tested could be declared tight or leaking depending on
which data sequence was used. The results of the other tank test
(Ft. Lewis, Tank #3) indicate that the tank is leaking but cannot
determine whether the flow is into or out of the tank. Second,
temperature, volume, and temperature-compensated volume rate data
exhibit a large range of variability compared to 0.05 gal/hr.
The high variability in the temperature compensated volume rate
suggests that the test time is too short and a single thermistor
is not adequate for measuring the mean temperature change in the
tank. These conclusions are based on the raw Leak Lokator data
and an analysis similar to that used by Leak Lokator except (1)
an average of four to nine standard Leak Lokator volume rate
measurements were used instead of one and (2) the average rate of
changes of temperature over one hour was determined by fitting a
least squares line to 5 to 10 temperature values over the hour
instead of the two end points. The uncertainty in the Leak -
Lokator temperature-compensated volume rate results presented in
Table 8 is about a factor of five smaller than the uncertainty of
a single 10 min volume rate measurement and a two-point
temperature rate measurement.
-------
Table 8. Summary of Leak. Lokator~-lO-im'n Weighted Sample Analysis
Total
Volume rate
(qal/h)
Temperature
volume rate
(qal/h)
Temperature-
compensation
volume rate
(qal/h)
Location
Test
Tank sequence
time
(min)
N
x 95% CI
x 95% CI
x 95% CI Test result
Damneck
2
Baseline
1
-0.
077
Leaking out
1
101
8
-0.051
0.004
-0.011
0.007
-0.040
0.008
Tight
2
112
9
-0
015
0.007
-0.052
0.011
0.037
0.013
T i ght
«.t
93
8
0
043
0.027
-0.041
0.005
0.084
0.027
Leaking in
4
59
5
-0
028
0.010
-0.023
0.011
-0.005
0.015
Tight
Fort Lewis
2
Baseline
-0
027
Tight
1
44
4,
0
017
0.014
0.215
0.059
-0.198
0.061
Leaking out
2
25
3
-0
044
0.028
0.096
0.096
-0.140
0.100
Leaking out
"5
3
40
4
0
019
0.020
0.001
0.026
0.018
0.032
Tight
j Fort Lewis
3
Baseline
-0
172
Leaking out
j
1
37
5
-0
130
0.012
0.043
0.004
-0.173
0.013
Leaking out
2
35
5
-0
089
0.031
-0.001
0.010
-0.088
0.032
Leaking out
3
43
7
-0
042
0.020
-0.104
0.041
0.062
0.046
Leaking in
Fort Lewis
4
Baseline
-0
191
Leaking out
1
56
6
-0
203
0.052
-0.251
0.041
0.048
0.066
Tight
2
41
6
0
157
0.068
-0.031
0.014
0.188
0.070
Leaking in
3
41
5
0
158
0.034
-0.003
0.010
0.161
0.036
Leaking in
Pitstop
2
Baseline
-0
012
T i ght
1
48
7
0
096
0.009
0.079
0.033
0.017
0.035
T i ght
2
44
4
0
053
0.005
0.006
0.078
-0.003
0.079
Tight
3
47
6
0
054
0.007
0.057
0.078
-0.003
0.079
Tight
4
56
5
0
053
0.005
0.221
0.045
-0.168
0.045
Leaking out
Scott AFB '
1
Baseline
-0
299
Leaking out
1
55
6
-0
323
0.029
-0.262
0.075
-0.061
0.080
Leaking out
f
2
44
6
-0
225
0.008
0.032
0.008
-0.193
0.011
Leaking out
3
56
5
-0
241
0.005
-0.008
0.024
-0.233
0.024
Leaking out
4
36
5
-0
206
0.0374
0.008
0.014
-0.214
0.040
Leaking out
-------
The time series plots of temperature (converted to volume)
and uncompensated volume were generated for each of the 21
sequences of Leak Lokator data. These cumulative time series
plots illustrate the reasons for the inconsistent test results
and the high variability. The volume and temperature time
series, and the least squares line fit to the temperature data
are presented in the report, "Development of a Tank Method for a
National Survey of Underground Storage Tanks."4
Some difficulty is evident in using a two-point analysis
approach. Depending on which two points are taken, a positive,
nearly zero, or negative slope can be determined because of the
large fluctuations in temperature.
C. Recommendations for the National Survey Testing
The findings of the development study have resulted in
several recommendations concerning the method of tank testing to
be used in the national survey program. These recommendations
are summarized below.
o The tank testing method should include putting a head
of pressure on the tank. There are two reasons for
this. First, proper compensation for water table
effects are necessary if the proper conclusion is to be
reached under high water table conditions. Second,
this process enhances the flow of product through small
holes, making them more likely to be detected,
particularly if they are near the top of the tank.
4See Footnote 1.
-------
o The tank test method should provide freq^*"*-
temperature measurements with a precise thermistor and
adequate temperature compensation. The product should
be circulated or mixed during the test. Adequate
temperature compensation is a key to successful
interpretation of tank test data. Such data must
consist of accurate temperature measurements at
frequent intervals. The judgment to mix is a choice of
techniques which is associated with the better
performance achieved by the single thermistor approach
used by Petro-Tite over the single thermistor approach
used by Leak Lokator.
o Data on temperature and level changes must be collected
frequently. This is necessary to minimize aliasing of
the high frequency fluctuations (out of the signal
band) into the lower frequencies (in the signal band).
This conclusion is based large on data analysis
performed by Vista Research, Inc.
o Data collection must continue for an adequate period of
time so that sufficient data for a precise analysis can
be provided. A minimum of 4 to 6 hours with frequent
temperature and tank level change intervals is needed.
While a test length of 4 to 6 hours with frequent
temperature and level readings is desirable, the
practical considerations of cost and disruption to an
establishment are also factors.
o The test method must incorporate an adequate
statistical analysis of the data to draw supportable
conclusions about the leak rate. None of the
techniques were found to collect either sufficient test
data or to provide adequate analysis algorithms.
Improved analysis protocols will be required.
5,1 Analysis of the Pilot Study Tank Test Data," Vista Research,
Inc., July 1985.
-------
v. PILOT STUDY
A. Objectives
The results from the earlier stages led to the
recommendation that a test using modified Petro-Tite equipment
and procedures be adopted for the national survey. The major
objective of this final stage was to modify and evaluate the
performance of the Petro-Tite method as it was to be used on the
national survey. This process included:
o Determining the best sampling interval for collecting
the data; that is, the time interval at which product
in the standpipe should be re-leveled and data readings
made;
o Determining the best length of the test;
o Developing and testing the analysis algorithm;
o Implementing the procedures operationally in the field
to identify operating difficulties and correct them;
o Field testing the entire survey data collection effort
including scheduling, data collection, and analysis;
o Estimating the detection performance of the method; and
o Finalizing the test protocol.
B. Overview
A sample of 25 tanks was selected from two primary sampling
units (PSUs) on the west coast for use in the pilot study. The
owners and operators of these tanks were contacted to arrange for
the tanks to be tested and to schedule the tests. Timing of the
contacts and arrangements for fuel delivery, payments, and
scheduling presented difficulties. Recommendations for mitigating
c-io
-------
these on the national survey were developed. Notifying owners
earlier of the test and giving a longer lead time to arrange and
schedule the tests were found to be necessary to expedite
testing.
Data were collected at three different time intervals and
for three different total time periods. The resulting data were
analyzed by various methods to select the most practical and
effective data collection interval and test length. A standard
data analysis protocol was developed for use when no testing or
data problems are identified. Data management procedures for the
national survey were developed which included the use of on-site
computers to collect data. Data and test review procedures were
developed to check each tank test for validity and to ensure that
the standard analysis was adequate. A simplified analysis that
can be used in the field to visually inspect the data and
identify potential testing problems was developed and
implemented. The tank test data were analyzed and a data report
prepared and submitted to EPA.
C. . Data Collection
Data identifying the tank, size, location, product, etc.,
were entered onto the top of a spreadsheet data file utilizing a
portable computer. Then test data are entered as each data point
becomes available. This provided a preliminary analysis and
estimated volume change rate that can be obtained on the scene.
D. Data Analysis
The data from the pilot study tank tests were analyzed with
two objectives. One was to determine the best sampling interval,
-------
and the second was to determine the best total test duration.
Sampling intervals of 1, 5, 10, and 15 minutes were considered.
Data collection at 1-min intervals was found to be impractical
for the large scale survey. Both the 5- and 10-minute intervals
provided improvements in the precision of the test data, but the
5-minute interval resulted in better precision. Thus, data
collection at 5-minute intervals was selected as the standard.
This analysis is presented in detail in Vista Research, Inc.1 s
report.6
Selection of the total time of the test was not so clear-
cut. Longer test times were desirable from a data quality
standpoint, but practical limitations were also considered. A
compromise of 2 hours of data at the low level was selected as
providing sufficient data while still proving to be practical for
the field data collection.
The test protocol used the same equipment as for a standard
Petro-Tite test. There were no changes in the test procedures
except for the sample interval and length of the test.
The analysis algorithm was modified to include smoothing of
the temperature data before applying the temperature correction.
A regression line was then fitted to the corrected data to obtain
the leak rate.
Seventeen tanks were tested in the pilot study. A summary
of the test results is presented in Table 9.
A family of performance curves was generated for the latge
and small tanks to estimate detection performance for a given
leak rate as a function of probability of detection, probability
6Ibid.
c-??
-------
Table 9 Summary of Pilot Study Results
Site no.
Tank
no.
System
leak rate
(gal/h)
Standard
Error (2 h)
(gal/h)
OHM rate
(gal/h)
Conclusion
Fuel
Tank size
(gal)
1
0.036
0.0074
+0.037
C
UNL
11,907
2
T1
-0.036
0.0098
-0.038
C
UNL
1,034
2
T2
-1.381
0.0490
-1.518
N
D
7,896
2
T3
-0.263
0.0138
-0.367
N
D
7,896
2
T4
0.009
0.0107
+0.015
C
0
10,152
3
-0.012
0.0242
+0.032
C
PUNL
1,036
4
+0.294
0.0601
+0.256
N(I)
UNL
10,152
5
T1
0.026
0.0114
0.042
C
0
10,152
5
12
-0.107
0,0041
-0.115
N
D
10,152
5
T3
+0.054
0,0047
-0.005
C
UNL
10,152
6
-0.008
0.0137
-0.024
C
UNL
8,000
7
T1
0.036
0.0245
+0.016
C
LR
6,006
7
12
0.042
0.0307
+0.096
N(I)
LP
6,006
7
T3
0.013
0.0348
+0.031
C
UNL
6,006
8
-0.056
0.0067
-0.029
N
0
10,383
9
T1
-0.010
0.0098
+0.028
C
UNL
1,036
9
12
-0.015
0.0130
-0.008
C
LR
1,036
C = Certifiable by NFPA standard.
N = Not certifiable (I) Inconclusive test
0 = Diesel
UNL = Unleaded
PUNL = Premium unleaded
LR = Leaded regular
LP = Leaded premium
-------
of false alarm, and test time. Detection performance for 0.05
gal/h leaks was unacceptable. A test period of 1 hour or less is
too short to achieve reasonable detection performance. For the
small tanks, test times of 1, 2, and 3 hours result in the
detection of 0.10, 0.075, and 0.05 gallon per hour leak rates
with a PD = 95 percent and a PFA < 5 percent. For the large
tanks, test times of 1 and 2 hours result in the detection of
0.25 gal/h leak rates with a PD = 95 percent and a PFA - 2
percent and 5 percent, respectively.
Of the 17 tanks tested, one resulted in a clearly invalid
test. One test was problematical, but the system is probably
tight. Three tanks appear to have significant leaks, and the
remainder appear to be tight. Due to the fact that the Petro-
Tite method places a higher head pressure on the tank than is
found in normal operation, the reported rates are overestimates
of product loss or leakage in operation.
Since the pilot study data available for analysis was
somewhat limited, the determination of the detection limit of the
Petro-Tite method could not be established as well as hoped.
Further data from the national survey will need to be examined.
VI. RECOMMENDATIONS FOR NATIONAL SURVEY
The recommendations for the national survey are:
1. Use a modified Petro-Tite test method;
2. Data should be collected at 5-minute intervals for
2 hours at each tank; and
-------
3. Data analysis should use improved algorithms to fit
data which exhibit curvilinearity in the test results.
The final proposed equipment configurations and data
collection, environmental measurement, and data analysis
procedures which resulted from the development and pilot studies
were specified in a separate document.7 The actual procedures
and methods which were followed in the field are documented in
Sections 6 and 7 and Appendix D of this report.
7MNational Survey of Underground Storage Tanks: Draft Test and
Analysis Plan," Midwest Research Institute, June 10, 1985.
-------
APPENDIX D
TANK TESTING DATA REDUCTION AND STATISTICAL
ANALYSIS LEADING TO LEAK STATUS DETERMINATION
I. INTRODUCTION
This appendix contains additional detail and in some cases a
more technical presentation of topics covered in Sections 7 and 8
of the report. Parts II and III of this appendix provide further
details on the tightness test raw data and the initial reduction
steps which produced the basic volume change rate estimates and
the estimated within-test standard errors for these estimated
rates. Part IV provides further detail on the retest results,
which is summarized in Part V of Section 7. Part V of this
appendix provides a more technical description of the estimation
of total test variance than is given in Section 8. Part VI
provides a more technical description of the leak status
determination rule than appears in Part III of Section 8, and
Part VII gives more details on the adjustment to test pressure
than appears in Part II of Section 8.
I- DATA COLLECTION AND MANAGEMENT
A. Data Collected
The tank testing data collected consist of several data
elements. A sample of a typical Petro-Tite data sheet is dis-
played as Figure D-l. Identifying information about the site,
tank system, and product were determined and entered as header
-------
11 turn TO rur
JS>_
ii. uMcm
i vum vvmrr
17. flU UT FOR TUT
A.
ItLm
liA
n^tnci m*o«i m* mm* u*u comnunmni oaei> oa «acm Mtnaio mumhw outturn
JdAL
6ft" 3tJ
accfe
. 5QLQ.
it sm* tmmm *m memmt i* rw m* i*m °^T
>M ¦-»<»*¦—«»»—«¦* O— »*»«—<_ , -» ,
«KM iCCOSSRV StSTIM
il tank mjtammm m
tstt wawu
Mi w *m r t • * «¦*
_aa_
m.
\*A
a. UTWSIOH HOM StTTM
_3A
2i. immmmimmm men* m ?• tirr tho tuh
<1 GHbrti) '» INMHti* ~# ««NM'
22. IIm#4
f«Mi .„• * t«M Qwm'
*n
» _____
iMniutimn
Miiittm
ij
il
•
M
eSw
MM
It
m
a.
aSmwlMMlil
il
to
"T
mumi
^ NMa
-¦—-
M.
IM
*.
T
M
ii.
m • m-
tSmSm
'SSSS
¦ IfltMIW*
MM MM
m
mm
>1 II
*
lam*
taMl
mSm
rxyrti
r 3
f»<-Vo
r-fjiJ
.cr>Sfc
am
I
.915
V\.
a
.5HS
.*;iO
-.03S
-an
3iSC>_
GanV*A
a
Ht,?l
.SO
Mia
&no
-,C.\T 1
lV»
\
\
*M,T
uftO
-,CA^
(ena
HI
4.CJ0CS
-,.0lS
r?o
41S
.Hkfk
-.0V5
>!i10
*0
KOCQ
-¦G\n
3*5]
<*
.^5&
-.ton
G.1Q
vft
HClCEi
-,0\Ci
3HC6
T J
n
>»s>0
.a^n
U,c\
QCBIj
oc*n
m •«¦ wt-r ™ J
?>
ITTl
13
.ain
-snn
~ rrri
t.nn
v \
ocn"
_iQQ4t.
-.oaoiu-
-.00 (o
ocao_
Q
v"^
no
"MO
v.oon
e„nri
¦tO
V.QCiO
*.Oan
Y>Y1
\\
V5
.•vio
•and
YO
v mr»
JrSOCL.
-^SQSu
- mi.
u
d ' s*
no
.Trio
f..no
*n
1.CW3
v.CCO
Tm,i
^ 1
•yto
.¦via
VO
Y.0CT1
-crt,
y.acn
Cuir,
vrt
v.ocsn
¥.CCJQ
-QCMr,
CCfiH
1 Me
-ino
•»-!&
c.iri
VO
%cc\r>
t.dCsn
T
im n j
;in
^ns
vrrvci
Lilt
vO
*.
6«a.i
!**
.ItlA
V.C5ft<\
^T>
V.oe.fi
- OCia.
\-ZL>
a4n
.TSO
y.OCO
-------
information. Additional data needed to set up the test were
recorded. These included the diameter of the tank, the depth
from grade to the bottom and top of the tank, and the depth of
the water table. An initial thermistor reading was taken and tl
internal check of the thermistor unit was performed. The spe-
cific gravity of the product was measured and used to determine
the coefficient of expansion. The tank volume was determined.
Presence of water in the tank was checked. If water was present
the volume of water in the tank was calculated and subtracted
from the tank volume to determine the volume of product. A fine
adjustment to product volume was to add the volume in the test
equipment (usually 2 to 3 gallons).
After the preliminary data had been entered in the header,
the actual test data were taken and entered. The time of readir
was entered. The reference level was noted. The volume in the
graduated cylinder before releveling was found and entered.
After releveling, the volume in the graduated cylinder was founc
and entered as "volume after." The fuel temperature in terms oi
the digit reading on the thermistor unit was found and entered.
The actual test data used to calculate leak rates consist of the
time, the volumes before and after, the temperature, the tank
product volume, the digits per degree Farenheit, and the
coefficient of expansion.
B. Data Management
The test data collected as described above were recorded or
a Petro-Tite data sheet by the test crew. The MRI technician at
the site keyed these data into a Lotus 123 worksheet file ,that
had been configured to receive the data and perform preliminary
calculations. An example printout of the data portion of this
file is shown in Figure D-2. The MRI technician entered the
u-3
-------
19-Aug-85
Page 1
Survey ID
Tank Test Firm DBL CHK
2 of Test Crew
2 MRI Crew STEVE
Time
Hr Min
Level
(div)
Fuel Type UNLEADED
Date
T digits
T digits/F
V Before
(gal)
Tank Vol 3010
API Dens 58.6
Exp Coef 0.00060366 Leak Rate
Std. Err
V After Fuel Temp
(gal) (digits)
AUG 7 1985
16676
322
0.001
0.0058976
Tcorr dV Leak Rate
(gal) (gal/h)
0
17
12
N/A
N/A
16669
N/A
N/A
0
22
12
0.270
0.270
16670
-0.006
-0.068
0
27
12
0.27
0.27
16670
0.000
0.000
0
32
12
0.27
0.27
16670
0.000
0.000
0
37
12
0.27
0.27
16670
0.000
0.000
0
42
12
0.27
0.27
16670
0.000
0.000
0
47
12
0.27
0.27
16670
0.000
0.000
0
52
12
0.27
0.27
16670
0.000
0.000
0
57
12
0.27
0.27
16670
0.000
0.000
1
2
12
0.27
0.275
16670
0.005
0.060
1
7
12
0.275
0.275
16670
0.000
0.000
1
12
12
0.275
0.275
16671
-0.006
-0.068
1
17
12
0.275
0.275
16671
0.000
0.000
1
22
12
0.275
0.275
16671
0.000
0.000
1
27
12
0.275
0.275
16671
0.000
0.000
1
32
12
0.275
0.275
16671
0.000
0.000
1
37
12
0.275
0.28
16671
0.005
0.060
1
42
12
0.28
0.28
16671
0.000
0.000
1
47
12
C.28
0.28
16671
0.000
0.000
1
52
12
0.28
0.28
16671
0.000
0.000
1
57
12
0.28
0.28
16672
-0.006
-0.068
2
2
12
0.28
0.28
16672
0.000
0.000
2
7
12
0.28
0.28
16672
0.000
0.000
2
12
12
0.28
0.28
16672
0.000
0.000
4*
17
12
0.28
0 .'29
16673
0.004
0.052
Figure D-2. LOTUS data sheet
D-4
-------
header data including the date, test crew, testing company, MRI
person, and the time, level, volumes before and after, and fuel
temperature (digits). The program calculated the leak rate,
standard error, and other intermediate values.
After the data were entered into the computer on site, they
were stored on a diskette. In order to facilitate expeditious
data analysis, the data were transmitted to MRI via telephone
using a modem. The diskettes containing the data files were
shipped to MRI on a weekly basis. The original Petro-Tite data
sheets were also shipped to MRI.
Upon receipt of the electronically transmitted data files,
they were printed and the volume trends plotted. Figure D-3
shows an example of such a plot. The calculations of the leak
rate and standard error were checked. Any unusual features of
the data such as outliers or curvilinearity were noted. The
computer file was archived as received and the hard copy was
placed in an archive file. A copy of the computer file was
placed in a working directory.
When the disk containing the data file was received, the
disk file and the telephone file were compared using the IBM DOS
utility file compare program to determine whether the data
transfer was complete and accurate. If the files were found to
differ, a new hard copy of the data and graph were printed.
Upon receipt of the Petro-Tite data sheets, the printed data
from the computer file were checked against the raw data sheets.
Any discrepancies were corrected in the computer file. If the
final file differed, another hard copy of the data and graph was
printed. The final form of the computer file was archived.
-------
0.26
0.24
0.22 H
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
-0.02
-0.04
-0.06
o Uncor V6\
o
o»
m
E
3
O
©
>
3
E
3
a
:t=§:
i 1 1 1 1 1 1 1 1 r
0.2 0.4 0.6 0.8 1
+ Temp eff
Elapsed Time (h)
~ Cor Vol
41
i—i—i—i—i—i—i—r
1.2 1.4 1.6 1.8 2
Smth Temp
Figure D-3. Plot of data using Lotus Data Pile
-------
After the data had been checked against the original sheet,
the final data analysis was done for the tank or system. When
the analysis was completed, a final copy of the data was printed,
incorporating any special analysis with the final leak rate and
standard error estimates. The final computer file was archived.
III. DATA REDUCTION ANALYSIS METHODS
A. Statistical Methods Considered and Choice
Several methods of statistical analysis of the tightness
test data were considered for use on the national survey. This
section presents a discussion of the advantages and disadvantages
of each and gives the reasons for the selection of those used.
The test method produced a volume change measurement at 5
minute intervals. This change was measured directly by bringing
the standpipe to a reference level and collecting the product
recovered or measuring the additional product needed. The other
measurement recorded at 5 minute intervals was a temperature
measurement. This measurement was taken by means of a thermistor
probe and box. To make this reading, a resistance bridge was
balanced by means of a dial. The instrument reading was con-
verted to a temperature by means of the calibration chart for the
instrument. The readings—after conversion to temperature—were
the temperature of the product in the tank at 5 minute intervals.
The temperature record of the product as measured by the thermis-
tor must be converted to an equivalent volume change using the
volume of the tank and the thermal coefficient of expansion. _One
essential difference between the volume and temperature readings
should be noted. The temperature was recorded as a cumulative
-------
reading—the tank temperature—while the volumes were recorded as
differences.
In order to make the temperature and volume data comparable,
they must be put in the same form. Either both must be changes
or both must be cumulative. Several approaches can be used for
the analysis. The standard Petro-Tite approach to the analysis
of the data is to take differences in the temperature readings.
The time interval used by Petro-Tite is 15 minutes rather than
the 5 minute intervals selected for the national survey testing.
After taking differences in the temperature readings, the change
in temperature is multiplied by the volume of the tank and the
thermal coefficient of expansion for the product to produce a
volume change due to temperature. This is subtracted from the
observed volume change at each point. The resulting differences
are temperature-adjusted volume changes. The standard Petro-Tite
analysis adds up four of these 15 minute readings to obtain the
hourly leak rate that they report. An advantage of this method
is its simplicity. A disadvantage is that no estimate of
variability is provided. An additional disadvantage is that four
15 minute data points do not provide sufficient data to ensure
that the test is valid.
A similar approach could be followed for analysis of the
survey data. Consecutive temperatures could be differenced to
obtain temperature changes for each 5 minute interval. This
would provide a set of observed volume changes and temperature
changes. The temperature changes would be converted to volume
changes by use of the coefficient of expansion. At this point
two different approaches to the analysis could be used.
One approach is to regard the observed volume changes and
the temperature volume changes as a paired sample. In this
analysis, one would calculate differences in each pair. These
-------
differences would be averaged to obtain an estimated leak rate.
The variability of the differences would be used to obtain an
estimate of the variability measured by the standard deviation.
The variation of the mean would be estimated by the standard
error of the differences (the standard deviation divided by the
square root of the number of terms in the average). This would
result in n-1 degrees of freedom for the standard error. Both
the mean and standard error (or standard deviation) would be
rescaled to an hourly leak rate.
There are a number of advantages to this approach. It is
directly comparable to the standard Petro-Tite tests. It is
relatively simple and should be easily understood. It does pro-
vide an estimate of variation. If the volume change and tempera-
ture changes are dependent, it accounts for this by pairing the
data. In addition, if the differences were less variable than
the original data, it would provide a more precise estimate than
other approaches. A disadvantage is that if the data are not
dependent, it sacrifices degrees of freedom unnecessarily. In
addition, if pairing does not reduce variability, then this anal-
ysis would lose precision.
A slightly different approach is to regard the volume data
and the temperature-volume data as two samples rather than as a
paired sample. With this approach, the mean volume change would
be calculated as would the mean temperature-volume change. The
difference in these two means would be calculated. This would
result in the same estimate of the leak rate or volume change as
with the paired data. However, there would be a difference in
the estimation of the variability. Each set of data—volume and
temperature-volume—would have its variability estimated
separately by the sample variance. If it were assumed that these
variance estimates were estimating the same quantity, a pooled
variance estimate could be calculated from these two. This would
-------
have a total of 2n-2 degrees of freedom, where n is the number of
data points of each type. This approach has an advantage if
there is no inherent dependence in the two types of readings. It
also is advantageous if pairing does not reduce variability
enough to offset the loss in the number of degrees of freedom.
If it were concluded that the variation of the two types of
data is different, then the sample variances should not be
pooled. In this case, the variance of the difference in sample
means would be the sum of the two variances of the means (the
variance of the mean is the sample variance divided by n). The
assumption would be that n is large enough so that the sample
mean would be approximately normally distributed. After the
variance of the difference in the means is calculated, the square
root of this number would be taken. Finally, the estimated leak
rate and the standard error of it would be rescaled to an hourly
leak rate as before. Thus, while the estimate of the leak rate
would be the same, the estimate of the variability would differ.
This approach has the same advantages of the previous approach.
The essential difference is in the calculation of the variabil-
ity. The choice between these two approaches should be based on
whether the assumption that the temperature-related volume
changes and the observed volume changes have the same variability
is valid. Consideration of the precision of the two measuring
instruments and of the rounding errors involved in the two
measuring processes suggests that the temperature-related volume
changes and the observed volume changes do not have the same
variance in general. Consequently, this latter approach would be
preferred.
The result that the variability in the temperature-felaJted
volume data is larger than the variability in the observed volume
changes suggests that it may be advantageous to smooth the
temperature data before adjusting the observed volumes for
" _ J
-------
temperature. Basically, this approach would use some degrees of
freedom to smooth the temperature data by fitting a curve of some
sort to them prior to making the volume adjustments for tem-
perature. It would use the fitted curve in the adjustments in
order to reduce variability.
Since the temperature data as recorded represent the
temperature of the tank over time, one approach is to fit a curve
to these temperatures and use the expected or predicted values
from the fitted curve for adjustment. In the typical test, the
temperature increased smoothly in a nearly linear fashion over
the period (about 2 hours) of the test. In this case, a linear
regression through the origin (or the starting temperature)
provides an adequate smoothing. The predicted values from the
regression can be used to adjust the volume changes. In some
cases, the temperature displayed a curvilinear form so that the
straight line fit was inadequate. In these cases, adding a
quadratic term to the regression provided a satisfactory fit.
Occasionally, the temperature was not monotonic or displayed some
other unusual behavior. In this event, moving averages were used
to smooth the temperature-related volumes prior to adjusting the
volumes.
The advantage of smoothing is that it may reduce the
variability of the estimate and so improve the precision of the
test. A disadvantage is that it is somewhat more complicated
than a linear or quadratic fit. An additional potential
disadvantage is that it may require a different form of analysis
to be used depending on the temperature data. On the other hand,
any method of analysis should allow for diagnostics to ensure
that the data from the test meet the assumptions adequately. It
should be anticipated that some tests will give data that do not
meet the standard assumptions. Such tests will either be judged
invalid or will require specialized analysis.
D-ii
-------
A rather different approach can be taken by cumulating the
volume differences. This would provide two sets of cumulative
data (one for volume, one for temperature-related volume) that
can be viewed as time series. With this approach, time series
models could be fit to both series. A transfer function could be
used to relate the two series and form a third series of the
temperature-adjusted volumes. The estimate of the temperature-
adjusted volume change rate could be made from the parameters of
the time series model of the derived series. This approach would
have an advantage if the volume measurements and temperature
measurements showed common forms of serial correlation that would
leak to a particular form for a time series model in the majority
of cases. There are some disadvantages of this approach. One is
that a large number of data points is required in order to fit
the time series models and have a sufficient number of degrees of
freedom. A second is that the analysis is much more complicated
and time consuming. A third is that the analysis must estimate
the appropriate model form for each series. The major drawback
is that time series analysis requires more data than was
available from the tests in the national survey.
A spectral analysis of the data from a long test during the
pilot study led to the conclusion that for test times exceeding
one hour, a sophisticated time series algorithm was not
necessary.
B. Standard Analysis
As a result of the considerations of the types of,analyses
available and the advantages and disadvantages of each, a
standard analysis was designed. For the standard analysis, the
temperature-related volume change and the observed volume change
u-^.Z
-------
were both expressed in cumulative form, beginning at zero for the
start of the low level (4-psig) test. A straight line through
the origin was fit to the temperature-volume data by least
squares. The predicted values of this line were calculated and
used as a smoothed temperature correction. The data were plotted
and inspected visually for outliers or deviations of the tempera-
ture data from linearity. Any questionable data were checked in
detail or considered for special analysis.
If no problems with the data were found, the predicted
values from the smoothed temperature line were used as the tem-
perature correction. This smoothed temperature correction was
subtracted from the observed volume data for each time point.
The resulting differences were divided by the time interval to
obtain a series of volume change rates expressed in gallons per
hour, typically based on a 5 minute interval. The arithmetic
mean of these rates was calculated and used as the estimate of
the leak rate. The standard error of this mean was calculated
and presented as the standard error of the estimate. In the
variance computation, n-1 was used as the divisor, where n is the
number of terms in the mean. The result was divided by n to form
the variance of the mean. The square root of this is the within-
test standard error reported before adjusting for between-test
variation. (See Section D.V, below, for. discussion of total
variance.)
The question of the appropriate number of degrees of freedom
was considered. It was possible that the terms in the mean might
be correlated, implying that the actual degrees of freedom would
be less than n-1. Spot checks of the serial correlation of the
terms showed generally no significant (at the 10% level) correla-
-% **
tions. For a few data sets some of the lag correlations were
significant. However, this occurred in only about 20% of the
data sets. Those where one or more significant correlations were
-------
found showed no consistent pattern of which serial correlations
were significant. Consequently, this was interpreted as being
likely to be due to chance. No adjustment of the degrees of
freedom is thought necessary.
C. Special Analyses
A number of data set features called for a different or more
detailed analysis than that described above. The most obvious
case was that of a manifolded tank system. Within the set of
manifolded systems, a slightly different analysis was needed for
different numbers of tanks, and a different analysis was needed
for systems tested together as opposed to those with tanks tested
separately.
Manifolded tanks that were separated and tested separately
provided two or more individual tank tests. As individual tank
tests, these were subjected to the standard analysis (or special
analysis if needed). This provided volume change rate estimates
and standard errors for each tank (and its associated lines).
These needed to be combined to estimate a system volume change
rate. In the descriptive data presented in the first part of
Section 9, the individual test results for tanks in a manifolded
system were used separately when available. The multivariate
analyses were restricted to single-tank systems. Thus, creating
system volume change rates was done for completeness in the
deliverable data file. This was done by summing the two
estimates of volume change rates. The variability of this
combined rate was estimated by taking the variances of the
individual volume change rates and adding these. Taking the
square root of this gave the standard error of the combined rate.
This extends to any number of tanks in a manifolded system tested
separately.
-------
Manifold tanks tested together provided slightly different
data. A single standpipe (or two connected by a siphon) was
used. A single volume change was recorded for the system every 5
minutes. However, each tank had a circulation pump and the
associated thermistor unit to measure temperature. In general,
each tank could have a different volume, although the usual case
was for tanks of the same volume to be manifolded.
A temperature-related volume change was calculated for each
tank. These were summed. The result represented the total
temperature-related volume change. This was used as the tempera-
ture effect. It was smoothed as before with a least squares line
through the origin, and the temperature adjusted volume change
rates calculated as before.
A number of other special cases were found and were dealt
with on an individual basis. Occasionally apparent outliers were
found. These were checked against the raw data and the test log
to see if there was any physical reason for them. A few tests
had thermistor boxes fail during the test for some reason (rain,
FM interference). These generally gave temperature data that
appeared, as outliers. When outliers were found and a physical
reason identified, the aberrant data were removed from the
analysis. This generally required smoothing over the missing
data by interpolation. If errors were identified, they were cor-
rected and the analysis redone.
The typical data showed a monotonically increasing tem-
perature, generally linear. A smaller proportion of the data
sets showed linearly decreasing temperature. Some data s^ts -
showed evidence of temperature increase that was curvilinear. If
this curvilinearity appeared or was suspected, a test for curvi-
linearity was done by fitting both a linear and quadratic to the
-------
temperature data by least squares (through the origin). If the
quadratic improved the fit significantly, the curvilinear fit
(using both linear and quadratic terms) was used for smoothing.
A few cases were found where both temperature and volume
were not only non-linear, but also non-monotonic. Provided that
they showed the same pattern, analysis proceeded. In this event,
a five point moving mean was used to smooth the temperature data.
Equal weights were used. This resulted in the loss of four data
points; two at the start and two at the end of the test. The
moving mean smoothed temperature volumes were subtracted from the
volume changes to obtain temperature-corrected volumes. These
were divided by the time intervals and expressed as gallons per
hour. The arithmetic mean and standard error of these
temperature corrected volume rates were calculated and used as
the estimates of the volume change rate and its standard error,
respectively.
Some tests showed volume change rates that were initially
increasing rapidly and curvilinear, while the temperature changes
were quite linear. The volumes typically increased rapidly for
the first few times, then slowed. This was interpreted as
relaxation of tank deformation. The apparent relaxation appeared
to follow an exponential curve and to approach the temperature
change rate as an asymptote. However, the constant of this
asymptote differed by tank. The rate of relaxation may be
related to the nature of the soil in backfill and water
conditions. When this was identified, the initial points
exhibiting this relaxation of the tank deformation were deleted
before analysis.
-------
D. Criteria for Invalid Data
A few of the data sets from the tank tests were judged
invalid based on the analysis of the data. This occurred quite
infrequently.
There were a number of criteria for declaring a data set to
be invalid. The most common was that the data showed a volume
increase even after adjusting for temperature. Since the test
method places pressure on the tank, a volume increase cannot
occur from inflow of water. Data that showed volume increases
after temperature adjustment that exceeded levels that could be
reasonably attributed to the variability of the measurement proc-
ess were judged to be invalid tests. The reason for this is that
such an apparent volume increase with no explanation could be
eclipsing a small actual volume loss or leak. Generally any tank
that showed a volume gain rate of more than 0.1 gallons per hour
after temperature adjustment was judged to be an invalid test.
The most likely explanation for such tests is that those tanks
had trapped vapor pockets.
A variety of other data features led to the conclusion that
the test was invalid. Some of these may also have been caused by
trapped vapor. A few instances were found where the temperature
as recorded fluctuated erratically during the test while the
volume measurements were relatively stable. If the temperature
data were so erratic as to preclude a temperature adjustment,
then the test was declared to be invalid. One or two tests
showed both temperature and volume measurements that were erratic
and did not appear to track together. These tests were also
judged invalid. Such behavior may have been caused by incomplete
tank deformation, followed by relaxation, combined with mixing
problems. No valid volume change rate could be estimated.
-------
IV. RETEST RESULTS
Three types of retests were conducted as part of the
national survey of underground storage tanks. One was a back to
back retest, conducted immediately after the original test used
to estimate the leak rate. The second was a leak simulation test
also conducted immediately after the original test. The third
type was a complete retest conducted on a different day and
generally by a different crew. Each of these types estimates a
different source of variation possible in the tank tests. A
tabulation of all of the retests appears as Table D-l. (Note
that a negative volume change is a leak, while a positive volume
change represents net inflow. In the body of the report, leaks
are reported without minus signs.) The simulated leak retests
are tabulated in Table D-2. A table summarizing the estimates of
bias (accuracy) and standard deviation (precision) based on each
type of test is presented as Table D-3. It should be noted that
the three types of retests estimate different sources of
variation and so are not directly comparable to each other.
A. Leak Simulations
The leak simulation tests were conducted after the original
test was concluded. Generally they were only conducted when the
original test indicated that the tank was tight or had a small
estimated volume change. The volume rate used for leak
simulation was on the order of 0.1 gallons per hour, so a large
volume change would overwhelm it.
The purpose of the leak simulation tests was to document
that the testing method could detect leaks of known size in tanks
that appeared to be tight. In addition, use of the leak
1 q
-------
Table D-l. Retest Data Summary
Survey ID
Volume
Fueltype
Type
Initial
oc
Initial
SE
Retest
SE
Date
Date
Rate
Rate
N02784A
2007
DIESEL
BTB
0730
0731
.015
.007
-.009
.005
N131078A
3985
UNLEADED
BTB
0822
0822
.102
.018
-.079
.013
N171261A
3979
GAS0H0L
BTB
0804
0804
•
049
.019
.040
.010
N21581B
3973
PRE UNLD
BTB
0731
0801
.822
.038
-1.315
.059
N281389B
11988
REGULAR
BTB
0806
0807
.025
.019
-.032
.020
L01034A
1039
UNLEADED
RT
0709
0812
•
013
.014
-.005
.009
L01036B
2005
DIESEL
RT
0712
0826
. 055
.049
-.009
.008
L01037A
4013
SUP UNLD
RT
0724
0828
•
019
.013
-.028
.022
L01037B
4013
REGULAR
RT
0724
0828
•
036
.016
.017
.012
L02068A
3989
REGULAR
RT
0809
0810
•
039
.014
-.019
.012
G03018A
3010
DIESEL #1
RT
0731
0827
.194
.01
-.226
.005
G03018B
3010
REGULAR
RT
0731
0827
•
060
.009
-.005
.009
L03095A
6049
DIESEL
RT
0802
0826
.036
.011
-.117
.006
L03095B
6048
DIESEL
RT
0802
0826
-
.032
.013
-.047
.007
G06013A
6018
REGULAR
RT
0724
0828
-
.153
.018
-.097
.016
G06013B
6018
DIESEL
RT
0724
0828
.089
.011
-.325
.017
G06028A
2964
REGULAR
RT
0721
0829
•
053
.016
.049
.008
G06028B
2964
DIESEL
RT
0721
0829
.708
.018
-.613
.015
G07010A
277
DIESEL
RT
0628
0826
.007
.054
-.001
.009
G07010B
566
REGULAR
RT
0628
0826
.005
.027
-.017
.012
G10020T1
10155
REGULAR
RT
0625
0816
1
.189
.322
.175
.022
G10020T2
10155
UNLEADED
RT
0626
0816
•
584
.028
.109
.018
N141107A
1035
GASOHOL
RT
0817
0831
•
006
.007
-.013
.010
N141107B
1033
DIESEL
RT
0817
0831
.327
.010
-.377
.027
N151141A
10576
DIESEL
RT
0817
0824
.621
.023
-.411
.015
N151141C
21154
DIESEL #1
RT
0817
0824
.129
.008
-.009
.008
G16005AR
1003
unleaded
RT
0722
0828
.006
.021
.025
.011
G16005BR
295
REGULAR
RT
0722
0828
•
046
.015
. 022
.011
L16394A
1023
REGULAR
RT
0728
0831
.021
.012
-.030
.008
L16394B
1039
UNLEADED
RT
0728
0831
~
018
.012
.011
.014
N171261G
576
DIESEL
RT
0804
0810
.014
.007
-.010
.004
N181323C
1005
UNLEADED
RT
0721
0825
•
025
.04
-.013
.008
N181323D
1005
REGULAR
RT
0721
0825
«
034
.013
-.015
.008
N181326B
1033
UNLEADED
RT
0722
0829
-
.076
.018
-.032
.007
G19068A
1005
REGULAR
RT
0715
0828
.614
.014
-.559
.018
G19068B
4032
DIESEL
RT
0715
0828
•
070
.008
-.002
.011
G19068C
1038
UNLEADED
RT
0715
0828
.068
.014
-.076
.011
G19101A
566
DIESEL
RT
0712
0827
-
.078
.02
-.100
.008
N19525A1
8060
UNLEADED
RT
0710
0822
•
032
.05
.044
.030
N34128A
6262
SUP UNLD
RT
0617
0828
-
.034
.009
.044
.003
N34128B
8000
UNLEADED
RT
0615
0828
.080
.03
-.053
.014
-------
Table D-2. Simulated Leak Summary
Survey ID
Test
Volume
Fuel
Bckgrd
SE
Obsrvd
SE
Sialtd
Diff*
Sin-Diff
Date
Type
LR
LR
LR
Obe-Sack
L01036A
0712
10154
UNLEADED
0.002
0.055
-.095
0.011
-.101
-.097
-.004
L05131B
0718
5955
REGULAR
-.082
0.200
-.055
0.012
-.049
0.027
-.076
N141107BL
0831
1036
DIESEL
-.377
0.027
-.200
0.008
-.059
0.177
-.236
N161191B
0731
10575
AV JET
-.485
0.010
-1.455
0.030
-.875
-.970
0.095
N181317A
0806
565
REGULAR
0.021
0.031
-.067
0.017
-.048
-.088
0.040
N181317A1
0806
565
REGULAR
0.021
0.031
-.084
0.019
-.113
-.105
-.008
N181326B
0829
1033
UNLEADED
-.032
0.007
-.184
0.006
-.154
-.152
-.002
G19068A
0828
4033
DIESEL
-.002
0.011
-.044
0.007
-.037
-.042
0.005
G19101A
0827
566
DIESEL
-.100
0.008
-.096
0.003
-.056
0.004
-.060
N19525A1
0710
4032
UNLEADED
0.032
0.050
-.079
0.020
-.059
-.111
0.052
N261347A
0820
10146
DIESEL
-.005
0.025
-.016
0.033
-.051
-.011
-.040
F271172A
0813
1037
REGULAR
0.019
0.012
-.070
0.019
-.090
-.089
-.001
N271375B
0817
1039
UNLEADED
0.017
0.010
-.094
0.012
-.115
-.111
-.004
-------
Table D-3. Retest results
Type
Mean
difference
(gallons per hour)
N
Variance
(gph)2
Mean
squared
error
(gph)2
Standard
deviation
(gph)
Root mean
squared
error
(gph)
Leak simulation
-0.00891
11
0.00066*1)
0.00074
0.0257*1)
0.0272
Back to back
0.00629
14
0.00053^
0.00057
0.0231W
0.0239
Retests
0.00297
34
0.00254 <2)
0.00255
0.0504 (2^
0.0505
(^For the leak simulation and back to back retests, the variance of the simulated minus the
differenced observed rates and the initial minus the retest rates is an estimate of twice
the underlying within-test variance plus any variance due to testing at successive 2 hour
periods. The corresponding estimated variance is reported here.
(2)For complete retests, the variance of the initial rates minus the retest rates estimates
twice the total variance (within- and between-tests). The corresponding estimated total
variance is estimated here.
-------
simulation allows for an estimate of the accuracy of the test as
well as its precision. The accuracy refers to the ability of the
test to measure a known volume change, while the precision of the
test refers to its ability to reproduce measured rates.
Thirteen leak simulation tests were conducted. Two of these
were conducted on tanks that had estimated volume rates that
indicated that the tanks were probably leaking. These tests were
excluded from the analysis because variability is known to
increase for leaking tanks. The results from all of the leak
simulation tests are tabulated in Table D-2. Using the leak
simulation results from the tanks with small estimated volume
changes (less than 0.1 gallons per hour in absolute value) gave
the following results.
Three rates were calculated from leak simulations. The
first was a baseline rate for the tank. This was estimated dur-
ing the regular tank test. While the leak simulation was con-
ducted, a measured rate was estimated. This is the rate observed
by the testing method during leak simulation. It is presumed to
be composed of the tank rate plus the simulated rate. The simu-
lated rate is calculated by collecting product drawn from the
tank at a constant rate, weighing it on a triple beam balance,
and converting the weight to volume at the temperature of the
product in the tank. The difference between the observed rate
during the simulation and the baseline rate provides an estimate
of the simulated rate. The difference between this and the
actual simulated rate can be used to assess the accuracy of the
test.
The average difference between the measured rate and th^
simulated rate was -0.00891 gallons per hour, based on the 11
leak simulations where the tank was not estimated to be leaking.
If the other two simulations are included, this mean difference
-------
increases to -0.0184 gallons per hour. The difference between
the measured rate and the simulated rate is interpreted as an
estimate of bias. The variance of the differences about their
mean provides an estimate of twice the within-test precision plus
any variance due to taking successive 2 hour test periods.
Taking half the variance of differences estimates the variance
itself. The estimate was 0.00066 gallons per hour squared for
the 11 tests. (It was larger, 0.00291 gallons per hour, if all
13 tests were used.) A mean squared error (MSE) can be
calculated to incorporate both types of error—accuracy and
precision. The mean squared error is the sum of the bias squared
plus the within-test variance. In this case it was 0.00O74
gallons per hour squared (or 0.00325 gallons per hour squared for
all 13 tests).
The bias is clearly not significant, in that it does not
differ significantly from zero (t =» -0.347, 10 degrees of
freedom). As a result, the variance and the mean squared error
are nearly identical. A measure of variation often used is the
standard deviation (or root mean squared error if bias is
present), which is the square root of the variance (or MSE).
This measure has the advantage that its units are the same as the
measurement, gallons per hour. The standard deviation was
estimated to be 0.0257 gallons per hour for these data.
B. Back to Back Retests
Back to back retests were conducted on a total of 18 tanks,
which includes the 13 tanks with leak simulations. Five tanks
had back to back retests without leak simulation. The purpose of
the back to back retests was to estimate the stability of the
test method. That is, to ensure that the volume change estimate
-------
did not differ markedly if based on the succeeding 2 hours after
the test.
As with all of these tests, variability is expected to be
larger if the initial leak rate or volume change is larger. For
this reason, the results of the back to back retests are pre-
sented primarily for those tests with volume change rates less
than 0.1 gallons per hour in absolute value. Retest results for
tanks with larger volume rates were more variable but generally
consistent.
The average difference between the original and retest for
the 14 tests with small volume changes was 0.00629 gallons per
hour. The estimate of within-test plus change over 2 hour
periods variance was 0.00053 gallons per hour squared, giving a
mean squared error of 0.00057 gallons per hour squared. The
corresponding standard deviation was 0.02 31 gallons per hour and
the root mean squared error estimate was 0.0239 gallons per hour.
The mean difference was not significantly different from zero
(t - 0.272, 13 df).
If all 18 back to back retests are used, the estimates are
slightly larger. The mean difference was -0.0134 gallons per
hour, with the variance and MSE being 0.00893 and 0.00910 gallons
per hour squared, respectively. The mean difference did not
differ significantly from zero (t = -0.14, 17 df).
c* Complete Retests
The complete retests consist of revisits to the site on a
¦>
different day. Typically this includes a different crew and
involves rescheduling and refilling the tank. The complete
retests incorporate all of the features of a tank test and so
-------
include all the sources of error including potential difference
from crew to crew and differences due to weather conditions,
nearby traffic flow, day of the week, etc. In addition, there is
a possibility that the tank is different at the time of the
retest. In fact, two of the retests originally scheduled were
cancelled when it was found that the tanks had been repaired
between the initial test and the scheduled retest. In addition,
two retests were performed and it was then discovered that the
tanks had been repaired between the initial test and retest.
These data are also not included, as they would measure an
additional source of variation which is not of interest, i.e.,
repair. Two other retests were performed on tanks that were
initially determined to have large vapor pockets. These two
tanks were retested later and on retesting were again found to
have large vapor pockets. The results of the test and retest for
these tanks with vapor problems agreed qualitatively; however,
the numerical agreement was not close. The reason for this may
be that the vapor pocket trapped in the tank was of different
size. There were also different ambient conditions that would
affect the vapor differently. For these reasons, the vapor
retests were not included in the estimate of the variance from
the retests.
The mean difference from the subset of 34 good complete
retests was 0.00297 gallons per hour. For complete retests, the
variance of the differences between initial and retest rates
estimates twice the total variance; that is, the within-test plus
between-test components. We report here the corresponding
estimated total variance. The estimated total variance was
0.00254 gallons per hour squared, giving a mean squared error of
0.00255 gallons per hour squared. If attention is restricted to
initial tests with estimated volume change rates of less than 0.2"
gallons per hour in absolute value, the results change slightly.
For this set of 30 retests, the mean difference was 0.0137
ri~->s
-------
gallons per hour, while the variance was 0.00181 gallons per hour
squared. This resulted in a mean squared error of 0.00200
gallons per hour squared. Neither mean difference is
significantly different from zero (t = 0.059, 33 df, t = 0.322,
29 df, respectively). The cases with larger volume change rates
were somewhat more variable, however.
As noted above, there were two retests of tanks that had
vapor problems. The initial test results showed volume increases
of 1.189 gallons per hour and 0.584 gallons per hour,
respectively, based on very short test times. The retests based
on longer times gave volume increases of 0.175 gallons per hour
and 0.109 gallons per hour, respectively, with again the
conclusion of a trapped vapor pocket. Both of these retests
agreed on the presence of vapor. The difference in apparent
volume increase rates may be due to a number of factors. The
initial test was terminated quite early. The early termination
may have led to variable results. The size of the vapor pocket
may have differed between the initial and retest. The changes in
conditions—temperature, barometric pressure—that affect the
vapor pocket may have differed. All of these could lead to the
observed differences in apparent volume increase rates. However,
the consistency of the test and retest in identifying the tank as
having a problem with trapped vapor suggest that the test method
is consistent in identifying problem tanks.
There were two tanks that were retested after the tank was
repaired. One of these had an initial leak rate estimated to be
-0.057 gallons per hour with a standard error of 0.004 gallons
per hour. The rate estimated on the retest was -0.017 gallons
per with a standard error of 0.0094 gallons per hour. Although
the tank was considered to be leaking by the NFPA Standard 329
and the owner took corrective action, the volume change rate
estimated initially was fairly small. The second tank had an
T)-2G
-------
initial leak rate estimated as -0.137 gallons per hour with a
standard error of 0.009 gallons per hour. On the retest, the
estimated volume change was -0.132 gallons per hour with a
standard error of 0.007 gallons per hour. Little change was
observed. However, on the retest, the testing company certified
the tank as tight based on the last hour of data, where they
estimated a rate of -0.044 gallons per hour. The data from this
test showed little difference from the initial test. Except for
the known fact that some repairs were done to the tank, there
would be no reason to exclude it from the retest data. Even the
former retest would not be viewed as suspect from the change in
estimated leak rates.
The retest data analysis showed no evidence of bias in the
test methods. Both the back to back retest and the leak sim-
ulations estimated within-test (plus variation from one 2 hour
period to the next) standard deviations on the order of 0.025
gallons per hour. The complete retest data gave a total standard
deviation estimate of 0.05 gallons per hour.
V. ESTIMATION OF TOTAL VARIANCE
The various types of retests offered not only a means of
estimating both within- and between-test variation, but also
evidence that the between-test variation is sizeable compared to
the observed variance of a single test result. In order to use a
statistical hypothesis testing approach to determine whether the
observed leak rate in a given test is evidence of a leak rather
than due to measurement fluctuation, the total variance must be
estimated for each test. This was done by estimating the
between-test variation from all the data taken together ^and""
adding this to the estimate of within-test variance generated by
the data from each test. The within-test standard error was
n-77
-------
squared, the overall between-test variance added, and the square
root of the sum was taken as the estimate of total standard error
used in the leak status decision process.
Two sources of information were used to estimate the
between-test variance. The two sources agreed fairly well, which
served as a validity check on the results. The two estimates
were then averaged (using relative weights based on the number of
cases each estimate was based on) to form the needed estimate of
between-test variance. Table D-4 summarizes this process.
The complete retests provided one data base from which to
estimate between-test variance. For a retested tank i, let k
index the test (1 or 2) and j index the 5-minute volume change
measurement for a given test. Then a given 5-minute volume
change measurement, x^^j, can be written:
xikj " Li + dik + eikj [Equation D-l]
where
= tank i*s true leak rate under test conditions;
dik = random measurement error of L* due to differences
from one test occasion to another; and
5ikj
random measurement error of the individual
5-minute volume change measurement for this test.
Since the various quality assurance double-testing methods showed
no evidence of bias, it is reasonable to assume that
E(eikj) = 0
E(dik) - 0.
D-28
-------
Table D-4. Estimates of between-test variance (based on observed volume change rates not
adjusted for test pressure)
Data Base
N
Estimated
total variance
(gph)
Estimated
within-test
variance
(gph)2
Estimated
between-test
variance
(gph)2
Estimated
total
standard error
(gph)
Complete retest
34
0.00254
0.00033
0.00222
0.0504
Tank tests with
measured volume
change between
0.0 and 0.2 gph
133
0.00267
0.00073
0.00193
0.0517
Combined estimate
167
0.00264
0.00065
0.00199
0.0514
-------
We also assume that
E(dj^<*i2) ¦ 0,
E(®il®i2) m
E " °'
and that the d^ and e^j each have a constant variance, denoted
as
a2 - between-test variance ¦ E(d^]c2)
and
a2 - within-test variance ¦ Efe^2),
where the mean of the e^j is taken over all measurements for the
k-th test of the i-th tank, usually 24.
Starting with Equation D-l, an estimate of total variance
can be based on the two estimated leak rates, (the initial
rate) and x±2 (the retest rate) as follows:
E(*ii - *i2>2
" E
-------
Thus
B((l/2)(l/n) E (5U - xi2)2) - (J2 + 02.
and
n 2
E((lA(2) (n)))T: Z S2 ) - 6*
i-1 k-1 ik w
where
si* " 1/ni(ni-1B?r, (5i* " xiW>2
i-1
Therefore, letting
Sb,r.t..t " (VJXVnl^r (2u -Ii2)2
we have
n 2 _ 1
^ 21 sf V
i-1 k-l 1KJ
[Equation D-2]
2(sJ . . ) - <5j.
b,retest b
The 34 retest leak rates and their within-test standard errors
were used in Zquati
the retest results.
w.r. used in Equation D-2 to conput. an ..ti«at. of eg b«..d on
D-31
-------
Tests on tanks which can be assumed not to be leaking
provide a second estimate of 6^. Here, the true leak rate is
zero, and we have
xij Bdi + eij
[Equation D-3]
with assumptions on d^ and e^j as stated above. (We suppress k
since only one test was done on these tanks.) In this case we
have
Defining tanks which can be assumed not to be leaking requires
some decision-making. By limiting this group to tanks with
measured average volume change between 0.0 and 0.2 gallons per
hour, the tanks which may be leaking (negative measured volume
change) are eliminated as are the test results which are likely
due to vapor pockets (high positive measured inflow).
The results of applying Equation D-3 to the 34 retests and
Equation D-4 to the 133 measured volume changes between, 0.0- and
0.2 gallons per hour are shown in Table D-4. It can be seen that
the two approaches yield similar estimates and in particular
indicate the importance of the between-test component of the
E(x^2) - 6^ + 62
v i b w
and
[Equation D-4]
Clearly
E^Sb,tight tanks^ °b*
-------
total variation in x^. It should be noted that these figures are
all as measured. and not as adjusted for test pressure. The
adjustment deflates the measured leak rate by about half (the
factors range from 0.395 to .608), but is applicable only to
actual leaks, since it adjusts the rate from test pressure to an
assumed operating pressure.
To get one estimate of between-test variance to use in
adjusting within-test standard error up to total standard error,
the two estimates described above were averaged with relative
weights based on the number of cases each was based on:
34/167 (0.00222) + 133/167 (0.00193) - 0.00199
Thus, to estimate the total standard error for a given observed
leak rate, 0.00199 was added to the reported (within-test)
standard error squared, and the square root taken. This total
standard error was used in the statistical hypothesis test method
for determining leak status described in section 8 of this
report.
VI. DETERMINATION OF LEAK STATUS
The physical tightness test for each tank system provided an
unbiased estimate of volume change rate and an estimate of the
within-test variability of that rate. The complete retest data
provided an estimate of the between-test variability of the
measured rates. However, the test itself did not provide a
definitive leak status determination, that is, an unequivocal
"yes" or "no" to the questions "Is this tank tight?" or "Is this
tank leaking?" in order to estimate the number of tanks in the
country that are leaking and to look at the subset of leaking
U~ -J ^
-------
tanks to investigate factors associated with leaking, such a
determination must be made (or the test result ruled
inconclusive) for each tested tank system. Two approaches were
considered for making this determination: a cut-off rule,
comparing the observed volume change rate to a pre-determined
cut-off; or declaring a system leaking or not by a hypothesis
testing approach. The latter approach was chosen for the study
determination of leak status. Two drawbacks of the cut-off
approach were that there was no scientific basis for establishing
a specific level for the cut-off at the time of the survey, and
that it did not take into account the differences in precision
achieved by the individual tests.
The null hypothesis to be tested in determining leak status
is:
H0 : H - 0
where is the true leak rate of the tank. The alternative is
HA : ^ > 0.
As shown in Part V, above, we model the test result, x^, as
having a total variance composed of a within-test and between-
test component. This total variance is estimated as
where the first term is the within-test variance measured from
the i-th tank test data and the second term was estimated*as ~
described above (Part V). The test statistic is therefore
z - Vst
-------
and is compared to one-tailed tables of the Normal distribution
to determine whether HQ can be rejected at a certain level of
significance. If HQ is rejected, we say the tank system is
judged to be leaking.
Several significance levels were examined, as was the trade-
off between significance and power. The power was estimated for
a specific leak rate after adjusting the leak rates and their
associated standard errors for test pressure (see Part VII,
below, for this adjustment procedure). A significance level of
OC= 0.05 was used for the survey determination of leak status.
VII. ADJUSTMENT OF TEST LEAK RATES
The Petro-Tite test places increased hydrostatic pressure or
the tank system for the test. As a consequence of this, any leak
or flow through an orifice in the tank will be increased over
what would occur under the (smaller) pressure encountered in
operation, similarly, the line test places a higher pressure on
the delivery line and so the leak rates estimated under the test
will be higher than what would occur in operation.
For systems, tanks, or lines that are determined to be
leaking, it is useful to adjust the leak rates estimated under
the test conditions to a standard set of operating conditions.
It should be noted that the basis for the adjustment is the
assumption that the leak is a flow of a liquid through an orifice
or hole. Such flows are more rapid under higher pressure than
under low pressure. However, if there is no orifice, no flow
would occur under high or low pressure. Thus, it is not
U 3 Z*
-------
logically consistent to adjust test volume change rates for pres-
sure in the event that the system was judged to be tight.
The adjustments are based on Bernoulli's law. More
specifically, adjustments are based on Torricelli's form of the
Bernoulli equation. In order for the adjustments to be reason-
able, the assumptions for these physical laws must hold. It
should be noted that the assumptions for Torricelli's and
Bernoulli's law assume that the flow is through an orifice with
neither resistance nor turbulence. In practice, this is not the
case. While the flow rate will be generally small enough so that
the assumption of a turbulence is reasonable, and so that the
head change is slow enough to be neglected, in most cases, leaks
will probably be through corroded sections and will be into soil
which may present some resistance. The effect of resistance
would be to lower the flow rate. However, how much the flow rate
would be lowered under the different pressures is not known.
Consequently, the effect of violation of these assumptions on the
adjustment to the leak rates is not known. It is assumed to be
negligible. There are some other, implicit assumptions. These
include that the orifice is constant, that the temperature and
density do not change, and that the product is not viscous.
Torricelli's form of Bernoulli's law can be used to
calculate adjustments to the flow rates. In order to do this,
several assumptions must be made. The set of assumptions used in
these calculations is detailed below. A step by step procedure
for the adjustments is given first. These are the adjustments to
be made in the ideal situation where the tank system leak was
quantifiable and a valid line test with quantifiable leak rate
was done. In our data base, among tank systems judged „to be
leaking with quantifiable leak rates, only 39 percent had valid
line test leak rates. Since the majority of cases had no valid
line data and the separate analysis described in Section 8 of the
D-3o
-------
report showed that line leaks accounted for a very small
proportion of system leaks when they were done, leak status and
leak rate as reported in the Major Findings are based on measured
tank system leak rates adjusted directly to operating conditions,
without adjusting for line test results. We present the line
test adjustment procedure since it was used for the analysis in
Part V of Section 8 and for future use in analyzing data
collected in the national survey.
A. Adjusting the Line Leak Rate to the System Leak Rate
Since the line test is conducted at higher pressure than the
system test, the leak rates estimated from the line test are not
directly comparable to those estimated from the system test.
This adjustment accounts for the difference in pressure and
adjusts the line test rates to be comparable with the system test
rates. These adjustments are calculated differently for pressure
systems and suction systems and for gasoline and diesel fuels.
The assumptions made for this adjustment are the following.
These are in addition to the assumptions needed for the use of
Bernoulli's equation to adjust the flow rates.
o The orifice where the leak (if any) occurs is where the
line joins the top of the tank.
o The tank is assumed to be buried to a depth of 3 feet
to the top of the tank.
o The water table is assumed below the bottom of the
tank.
o Three tank diameters are assumed: 48 inches, 64 inches,
and 96 inches.
-------
Table D-5 gives the adjustment factors to adjust the rates
estimated from the line test to the conditions assumed for the
system test. The factors as presented are multiplicative. To
convert a rate estimated from the line test to the equivalent
system rate, multiply the estimated line rate by the factor in
the table.
The difference by type of delivery system results from the
fact that the line test is conducted at 15 PSIG for suction lines
and at 50 PSIG for pressure lines.
B. Subtracting Line Rates From System Rates When Valid
Line Results are Present
After adjusting the line test results by the factors in
Table D-5, the line test rates would be comparable to the system
test results. The line test rates could be subtracted to obtain
an approximate tank rate. This is the rate for the tank system
excluding delivery lines, but still including any other plumbing
such as fill pipes, vent pipes, etc.
If a system has more than one delivery line, each line test
rate would be adjusted, then all line test rates subtracted from
the system rate. For the tank systems for which the line was
found to be untestable, the line rate cannot be separated from
the system rate.
-------
Table D-5. Adjustment factors for line test rates
Tank diameter
48 inches
(0 - 1,000 gallons)
64 inches
(1,101 - 7,000 gallons)
Suction Pressure
0.431 0.236
0.395 0.216
96 inches
(7,001 - 15,000 gallons)
0.317
0.174
-------
C. Adjusting the Tank Rate for System Rate) to Assumed
Operating Rate
Since the test is conducted at elevated pressure, flow rates
through any orifices will be larger under the test conditions
than they would be under actual tank operation. The magnitude of
the difference depends on a large number of variables. In
particular, flow rates would vary by location of the hole in the
tank (distance from the bottom), amount of fuel in the tank, and
pressure of a water table part way up on the tank. The
adjustment factors would also vary with diameter of the tank.
Since diesel tanks were tested at the same pressure (hence at a
lower head-distance) as gasoline tanks, the adjustment also
varies with fuel type because of the density difference.
The standard assumptions for calculating the adjustment
factors presented in Table D-6 are as follows. These are in
addition to the basic assumptions of Bernoulli's law.
o The water table is assumed to be below the bottom of
the tank.
o The tank is assumed to be buried to the depth of 3 feet
from grade to top of tank.
o Three tank diameters are assumed (48, 64, and 96
inches).
o The average operating level of the tank is assumed to
be half full.
o The orifice or hole is assumed to be in the bottom of
the tank.
Table D-6 then gives adjustment factors to adjust the
estimated tank system leak rate to the assumed standard set of
operating conditions. The factors should be multiplied by the
leak rate estimated under the system test to obtain the adjusted
-------
Table D-6. Adjustment factors for tank (system) rates*
Adjustment factor
Tank diameter Gasoline Diesel
48 inches 0.395 0.430
(0 - 1,000 gallons)
64 inches 0.456 0.496
(1,101 - 7,000 gallons)
96 inches 0.558 0.608
(7,001 - 15,000 gallons)
~If a standard height had been used for both fuels, the
gasoline column would apply to both.
D-41
-------
leak rate. Note that this adjustment can be done to the system
test leak rate, or to the leak rate remaining after any relevant
line leak rates have been adjusted to test conditions and
subtracted off.
Multiplying the rates estimated under the system test by the
adjustment factors given in Table D-6 will give adjusted rates
for the assumed standard set of operating conditions described in
the assumptions above.
D-42
-------
APPENDIX E
INVENTORY RECONCILIATION METHODS
i. EPA lyvgyroRY pecqnciliatiqn ksthqp
EPA has developed a simple method1 for monitoring
underground motor fuel storage tank inventory records to detect a
systematic deficit which may be attributable to a leak. The
method is based on counts of the number of daily underages found
in the inventory record and is simple enough to be implemented by
a tank operator without excessive calculation or burdensome
record-keeping. As originally formulated, the method is intended
for application as the "first line of defense against leaks" in
an on-going monitoring program. Thus, the approach is sequential
in nature and involves making a decision on the presence or
absence of an inventory deficit at the end of each 30-business-
day period, based on a comparison between the cumulative count of
daily underages and certain statistically-derived "action
numbers"1. A cumulative number of underages in excess of the
appropriate action number was to be interpreted as evidence of a
deficit. The statistical model and calculations underlying the
method were detailed in the report from Battelle Columbus
Laboratories to EPA2. The basic method required modification for
application to the inventory data collected in the survey because
each sampled facility provided only a single, one-time record of
U.S. EPA, Office of Toxic Substances, "More About Leaking -
Underground Storage Tanks: A Background Booklet for the
Chemical Advisory," (October 1984).
2David c. Cox, "Performance of the Chemical Advisory Inventory
Analysis Method Under Various Scenarios," Report from Battelle
Columbus Laboratories to EPA under contract No. 68-01-6721
(April 1984).
-------
30 days' inventory for analysis. The purpose of this section is
to describe the statistical model on which the modified EPA
method is based.
The decision rule for the proposed method will be defined by
considering a well-run station where the only sources of
discrepancy in the inventory records are (i) a daily leak of
magnitude L and (ii) unavoidable random error in the daily stick
measurement of the tank. Successive daily errors are assumed
independent and identically normally distributed with mean zero;
this assumption is supported by the research of Warren Rogers3'4.
Hence, we can write:
+ e^
where is the ith daily stick measurement, is the true
quantity of gasoline in the tank at the close of the ith day, and
e^~N(0, O2) is the stick measurement error. Now consider a
period of n days, assuming for simplicity that the station is
open every day. The process of balancing inventory at the end of
each day, as described in the literature5 and assuming that there
is no metering error at the pump6, leads to a set of daily
variances (discrepancies),
dj_ = -L + e^ - e^_1, i = 1, ... n.
3"Inventory Reconciliation system," Warren Rogers Associates.
4Warren Rogers, personal communication.
5American Petroleum Institute: "Recording Practices for Bulk
Liquid Stock Control at Retail Outlets," (1977).
6Metering error, if present, can be estimated and removed from
the record, see American Petroleum Institute, "Recommended
Practice for Bulk Liquid Stock Control at Retail Outlets,"
(1977) .
E-2
-------
Let N be the total number of negative daily variances,
N - #{i|l < i < n, < 0>.
Clearly, large values of N suggest that there is a leak,
i.e., L > 0. The exact probability distribution of N is, in
general, very difficult to derive. However, of the special case
of no leak, i.e., L « 0, the calculation has been carried out7.
Table E-l shows the distribution for the case n = 30 of most
interest. In general, we must rely on a normal approximation to
the distribution. This is derived as follows. We first find the
mean E(N) and variance V(N) as follows. Define:
p » Pr(dj < 0)
- Pr(di < 0, di+1 < 0)
li « 1 , If di < 0
0 , else
Then E(1^) ® p, E(1^Ij) * p2 if |j-1| > 1 (because then Ij, Ij are Indepen-
dent), E(IiI1+1") * Pr Thus
n
E(N) « E(1|l 1^) « np. Also,
E(N2) - E< j, if ~ 2^. 1,1,)
• E(,£, l] * $ I,I1+, * I,Ij)
« np + 2(n-l)p^ + [n(n-l) - 2 (n-l)]p2
Therefore
V(N) » E(N2) -(E(Nf [1]
• np(l-p) - 2(n-l)(p2 - Pj)
so(L)2
7Warren Rogers, "The Exact Null Distribution of the Number of
Negative Daily Variances," Report from Warren Rogers Associates
to EPA, (September 1984).
-------
Table 1-1. Probability distribution of the number of negative
daily variances, N, for the no-leak case, based on
30-day inventory
No. of
negative variances
Probability of
occurrence
< 10
0.0024
11
0.0121
12
0.0456
13
0.1161
14
0.2022
15
0.2432
16
0.2022
17
0.1161
18
0.0456
19
0.0121
> 20
0.0024
3-4
-------
We approximate N by a normal distribution with mean np +0.5 and
variance a(L)2. The mean is taken as np +0.5 to provide an
approximate continuity correction for use in the upper tail of
the distribution, in which our greatest interest lies.
To check the accuracy of the approximation, consider the
case L = 0. Then,
Pi * pr n)
(exact)
Pr(N > n)
(approximate)
15
0.6216
0.6217
16
0.3784
0.3783
17
0.1762
0.1788
18
0.0601
0.0630
19
0.0145
0.0162
20
0.0024
0.0029
-------
Clearly the approximation is sufficiently accurate over the
range of n reported. For L / 0, the exact distribution of N has
not been derived. We will rely on the normal approximation in
such cases. The mean and standard deviation of the approximating
distribution have been calculated and are shown in Table E-3.
Table E-3. Mean and standard deviation of normal approximation
to the distribution of N, the number of negative
daily variances, for various values of the daily
leak rate L, for a 30-day inventory
L
Standard
(gallons)
Mean
deviation
2
16.46
1.636
3
16.93
1.641
4
17.41
1.647
5
17.88
1.654
6
18.34
1.665
7
18.81
1.678
8
19.27
1.684
9
19.72
1.699
10
20.16
1.707
The final feature for which we must account before we can
determine the decision rule is round-off error. In practice,
inventory values are typically reported to the nearest gallon so
that an exact inventory balance, i.e., a zero variance, can occur
due to round-off. This is fairly common in actual inventory
data. We will assume that a zero variance is reported if the
actual variance is less than 0.5 gallons in absolute value.
Thus, a negative variance is reported only if the actual variance
is less than -0.5 gallons. Let N* be the number of negative
variances actually reported and assume a/2 » 25 gallons. "Then
the distribution of N* should be approximated by a normal
distribution with mean and standard deviation shown in Table E-4.
j
-------
Table E-4. Mean and standard deviation of normal approximation
to the distribution of N*, the number of negative
daily variances accounting for round-off error, for
various values of the leak rate L, for a 30-day
inventory
L
Standard
(gallons)
Mean
deviation
0
15.26
1.633
1
15.74
1.634
2
16.22
1.635
3
16. 69
1.638
4
17.17
1.644
5
17.64
1.650
6
18.11
1.660
7
18.58
1.672
8
19.04
1.681
9
19.49
1.687
10
19.94
1.703
Now suppose we have 30 days1 inventory and there is no leak.
Using the approximating distribution from Table E-4 the number of
daily variances observed should have the distribution shown in
Table E-5.
Table E-5. Probability distribution of the number of negative
daily variances, N*, observed when no leak is present
n = number of
negative variances
(Pr(N* > n)
15
0.564
16
0.326
17
0.142
18
0.047
19
0.011
20
0.002
T?_*7
-------
Thus, if we make 18 or more negatives our criterion for deciding
that a deficit is present, there is approximately a five percent
false-positive rate. That is, a tank with no leak and no source
of error in inventory other than random measurement error due to
sticking has approximately a five percent chance of being
erroneously classified as a leaker. Note that false-positives
due to other factors such as theft are not accounted for here.
The detection capability of this version of the EPA inventory
analysis method can now be calculated using the values given in
Table E-4. Results are shown in Table E-6.
Table E-6. Probability of detection of leaks of various sizes
using the modified EPA inventory method based on
30 days' data
Actual
leak
Detection
probability
Gallons/day
Gallons/hour
1
.04
0.08
2
.08
0.14
3
.12
0.21
4
.17
0.31
5
.21
0.41
6
.25
0.53
7
.29
0.64
8
.33
0.73
9
.37
0.81
10
.42
0.87
Thus, leaks of at least nine gallons per day or more have better
than 80 percent chance of detection. It should be noted that the
detection capability of the simple inventory method based on only
30 days' data would be expected to be poor. The method was
designed, as explained previously, for use as a tool for -on-going
monitoring programs.
-------
II. WARREN ROGERS ASSOCIATES' INVENTORY RECONCILIATION METHOD
Warren Rogers Associates (WRA) has developed a computerized
system for analyzing daily inventory data from underground
storage tanks in order to identify leaks8. The details of the
method are proprietary. This section provides a brief
description of publicly-available information on the model and
should not be interpreted as an evaluation or endorsement by EPA.
The WRA system was developed in response to the perceived
inadequacy of conventional, routine inventory accounting in
detecting small or moderate leaks. Typically, such leaks are
masked in the data by a variety of errors. For example, a single
delivery error of 300 gallons could mask a 10 gallon-per-day leak
based on 30 days' inventory. The purpose of the model is to
isolate, identify, and quantify these errors.
Errors accounted for include:
Delivery errors;
Unexplained additions;
Pump meter error;
Temperature effects;
Stick error; and
Tank or line leaks.
Occasionally, other, rarer, errors will appear, e.g., use of an
incorrect tank conversion chart, or theft. The data required by
the model include only daily stick readings, deliveries, and
sales.
8Warren Rogers Associates, Inc., "Inventory Reconciliation
System," (undated).
E-9
-------
The basis for the model is that the major errors and
discrepancies in the inventory data are very distinct in their
characteristics and thus in the way they contribute to the total
record. Thus, for example, an unrecorded over-delivery or an
unrecorded removal will cause a permanent shift in the record
which remains as a fixed component in all future observations.
This effect can be estimated and removed from consideration when
evaluating the possibility of a continuing day-to-day trend
indicative of a leak. By contrast, a large stick error caused by
a mistake in reading the stick or conversion chart will typically
cause a large discrepancy in that day's inventory which will be
followed the next day by a discrepancy of similar size in the
opposite direction. The two discrepancies will tend to cancel
out in the cumulative inventory record. The "signature" of a
pump meter error is different: such an error will induce day-to-
day errors of constant sign proportional to the through-put of
the tank.
WRA's report to clients includes a record of day-to-day
variances and the cumulative variance between book inventory and
stick measurement for the period. It also provides:
Over- or under-deliveries by date of occurrence and
amount. That is, the discrepancy between the amount of
product actually delivered as opposed to the amount
reported;
Unexplained one-time gains or losses also by date and
amount;
Meter errors at the pump;
Trends which are indicative of either a tank or line
leak; and
"> **
Effects of possible disparities between the ambient air
temperature and underground temperature.
E-10
-------
As a special contribution to this study, WRA also provided a
"data quality code" based on professional interpretation and
experience. The data quality code is explained in Table E-7. A
sample WRA inventory report is shown in Figure E-l. Based on a
discussion with the developers of the WRA model, the false-
positive rate is five percent, comparable to the modified EPA
method.
Table E-7. WRA data quality code
Category
Definition
1
Confident of the result
2
The trend could have been
delivery-induced
3
The trend is noisy but believable
4
No confidence in the trend due
to the data
5
Data is questionable and requires
further investigation.
HI- ENTROPY LIMITED INVENTORY RECONCILIATION SYSTEM
Entropy Limited has developed the Precision Tank Inventory
Control (PTIC) system9. The analysis is based on principles
similar to the WRA system and accounts for the same types of
errors and discrepancies. Entropy appears to consider thermal
effects and vapor losses more comprehensively than does WRA.
However, additional input data to the system is required for
these analyses.
9Entropy Limited, "Precision Tank Inventory Control," (1984).
-------
PAGE 1
10/27/85
TANK 10 2 I
PRODUCT * UNLEADED
TANK SUES
10000
I
h
MONTH
DAY
^AlES
4
3
1150
4
4
1163
4
5
628
4
6
1444
4
7
902
4
8
466
4
9
1055
4
10
856
4
11
524
4
12
1285
4
13
1195
4
14
1158
4
15
630
4
16
921
4
17
784
4
18
1453
4
19
532
4
20
1028
4
21
881
4
22
992
4
23
488
4
24
1169
4
25
528
4
26
827
4
27
786
4
28
661
4
29
842
4
30
827
5
1
1401
5
2
1051
DELIVERIES STICK BOOK DAILY VARIANCE
0. 2926. 2901. 25.
0. 1733. 1738. -30.
l098- lll0» -1-
4100. 3759.- 3766. 5.
0. 2864* 2864. 7.
0. 2373. 2398. -25.
0. 1301. 1343. -17.
, 0. 437. 487. -8.
4230. 4152. 4193. 9.
2060. 4948. 4968. 21.
0. 3759. 3773. 6.
0. 2616. 2615. 15.
3750. 5744. 5735. 8.
0. 4815. 4814. -8.
0. 4021. 4030. -10.
.0. 2554. 2577. -14.
5ir0. 7102. 7155. -30.
0. 6073. 6127. -I.
0. 5214. 5246. 22.
0. 4218. 4254. -4.
0. 3693. 3766. -37.
4100. 6594. 6697. -30.
0. 6073. 6169. 7.
0. 5214. 5342. -32.
0. 4416. 4556. -12.
0. 3759. 3895. 4.
5240. 8125. 8293. -32.
0. 7289. 7466. -9.
0. 58 76. 6065. -12.
0. 4815. 5014. -10.
END OF PERIOD CUMULATIVE VARIANCE
-199.
NUMBER OF NEOAllVE DAILY VARIANCES
19
DELI VERY DISCREPANCIES
CUM,
MONTH
4
DAY
12
AMOUNT
46.
VAklANCL
25.
-5.
-12.
-7.
0.
-25.
-42.
-50.
-41.
-21).
-14.
1.
9.
1.
-9.
-23.
-53.
-54.
- 32.
-36.
-73.
-10 3.
-96.
-12U.
-140.
-136.
-Iu8.
-177.
-109.
-199.
Figure E-l. WRA Inventory Report
-------
WAKREN ROGERS ASSOCIATES, INC
65 BELLEVUE AVENUE
NEWPORT, RI 02840
(401) 846-4747
INVENTORY ANALYSIS
TANK ID ! 1
PRODUCT I UNLEADED
TANK SIZE; 10000
MONTH DAY AMOUNT'
4 29 -17.
UNEXPLAINED ONE TIME GAINS AND LOSSES
MONTH DAY AMOUNT
4 14 39.
4 21 19.
4 23 -36.
4 26 -19.
PUMP METER ERROR
NONE
DAILY TREND
-8.37
AVERAGE STICK ERROR
6.95
LARGE STICK ERRORS
MONTH DAY
4 19
TEMPERATURE DIFFERENTIAL
—2*66
SIGNIFICANCE OF TREND
0.00
i
-------
The PTIC system reports its leak findings as an estimated
leak rate, in gallons per day, and as a "probability of leak"
(see the sample inventory report in Figure E-2). According to
the model's developers, the probability of leak is based on a
Bayesian-type analysis which accounts for various factors
including the quality of the inventory data. Details are
proprietary. Typically, the decision rule is phrased in terms of
the leakage probability as follows:
The 50 percent cutoff point corresponds to a false-positive rate
of approximately two percent. To obtain a more typical five
percent false-positive rate, a cutoff of 30 percent leak
probability should be used to decide that the tank is leaking.
Leak probability
Decision
< 10%
10% - 50%
> 50%
Tank is tight
Inconclusive
Tank is leaking
-------
TAlfK * « 2 STAGE II VAP COtTOOL LEVEL (9=HO VAP LOSS) » 1
TANK SIZE « 4*M)0. 8 sIIK'3 UNLDi INSTYn* 1973
INVENTORY KF.CORni.H (.WLM'WS: n = GJi'.KSS, I = HET
0 l)F,L 1 Vfclll I.S
0 STICK IKVEM'.i'-V
U DISPENSE!! i UOJUOT
M.B. II STICK-HEADING AND TOTALIZING OCCURS OH THE H0IU11RC AFTER TOE LISTED DATE
1
DATK
DISPENSED
sio;*v:e
DEL 1V
U1 SC11F.P
1
1/ 2/85
1-17.8
13^4.0
o.o
o.o
2
1/ 3/35
147.8
12? 1 .0
0.0
-5.2
3
U 4/85
180.4
921.0
0.0
-113.6
4
1/ 5/85
230.8
581.0
y.o
-110.2
5
1/ 7/85
202. B
1816.0
1100.0
364.8
6
1/ 0/85
10O.4
176-.V.0
0.0
76.4
7
1/ 9^8i>
li*. 0
136'). 0
<).«
- 1 1 1.0
a
1/10/85
i:m .5
iU". r. 0
¦ >.o
-7:1.5
9
1/11/85
17'). 8
II0U.0
0.0
-86.2
10
1/12/05
134.4
923.0
0.0
-48.6
11
1/14/85
192.9
64'). 0
0.0
-8*1. 1
12
1/15/85
230.3
28 >. 0
0.0
-13*1.7
13
1/16/85
232.5
1569.0
110).0
412.5
14
1/17/85
160. 1
12') .'. <»
0.0
-111.9
15
1/18/85
81. 1
1221.0
0.0
8. 1
16
1/19/85
ino.it
*H. 1 .0
0.0
- 7') . 2
17
1/21/85
174.7
716.0
n.o
-70.3
It)
1/22/85
111.7
5'!I .0
o.o
-83.3
19
1/23/85
29 1.5
1087.0
1100.0
35 . 5
20
1/24/85
177.5
1451.0
0.0
-58.5
21
1/25/85
201.5
114*.0
o.o
-104.5
22
1/26/85
13<).f>
V<> 1 .0
o.o
-5:1.8
23
1/20/85
181.6
6R2.0
0.0
-07.4
24
1/29/85
Kit.2
42). 0
0.0
-6U.I1
23
1/30/85
2P.>.9
15.10.0
i ioo. <;
281 .9
26
1/31/85
224.5
1297.0
o.o
-8.5
27
2/ l/£5
46. 1
1221.0
0.0
-•¦It. 9
20
2/ 2/85
1 J2.0
1031.O
0.0
-73.0
29
2/ 3/05
194.5
7« t. 0
0.0
-55.5
30
2/ 4/05
239.7
2005. o
1100.0
3f;«.7
31
2/ 5/85
120.5
1806.0
0.0
-78.5
TOTALS
EVCT
30.0
DISPENSED
5191.7
STOHAOF.
~32.0
URL IV
5500.0
DIPCREP DISTRIBUTION
NUMtEll OK DAYS Willi (~) DISVM^P
NUMBER OF DAYS WITH <-) DfSCUEP
Disciwr.p
123.7
7 ( 23.3*5)
23 < 76.7K)
UV STATS
DV - OISCRKP CCAL) COWIECTED FOR TE*T. VAPOR LOSS AID VATFJl INFLOW
I)V COUNTS * »(>.«>«>(:>i(>'>00
DV VARIANCE (li 1 AS ED) « 27212.47
LAG-I AUTOCOKn * -0. I9-.35
Figure E-2. Entropy Limited Inventory Report
-------
DV DISTRIBUTION
HUMBER OF DAYS WITH (+) DV * 7 C 23.3S)
HUHBEit Of DAYS WITH <-) DV - 23 < 76.7X)
DIPSTICK READ ERRORS
HOWE
DELIVERY DISCREPANCIES
S
1/ 7/OS
389.?n
• DELIVERY OVi:?l(+)
OR
UKI»ER(~>
13
1/-16/BS
4:>3.84
*ik;mverv ovkiu+j
Oil
UNI'K'H -)
19
1/23/B3
411.26
* DEP. 1 VERY OVE1U + )
OK
UNDKHC-)
23
1/30/HS
347.07
« OKI, 1 VERY OVT.IK+)
OR
HN0E1K-)
SO
2/ 4/U3
401.91
*DELI VERY OVER<+»
OR
UNWER(-)
UNEXPLAINED ONE-TIME DISCREPANCY
NONE
UNMODIFIED DISCREPANCIES
DISCREPANCIES CORRECTED FOR TEMP. VAP, CHART CALIB AND METER ERR
SOURCES Of INVENTORY DlSCllEPAI.CIKJJ
ESTIMATE ONCER
1
LEAK RATE (CAI^DAY) «
01. I6B
6.B7B
2
OALIB CHART ERR ( I09K*CAL/G\1,-THU!» *
3. IS
J. 31
9
1VJAL METER KRR t «0O?:*CAi^OAL-1HRU) »
M.M
ou.oe
4
T1IKRMAL HKtllliKAGE LOSS (GAL) *
17.3
8
VAPOR LOSS (GAL) «
15.5
9.8
6
NET DE..IVEHY DISCREPANCY «
0.®
>*.9
8
WATER INFLOW TO TANK (GAL) *
O.O
0.9
9
WATER OUTFLOW FROM TANK < GAL) «
9.0
9.9
I RECORIiKKKPlNC FlAC9:
H* EXCESSIVE UELl VKIlY DISCtlEPANCIRS
<*' EXCESSIVE VARIANCE IN STICK HEADINGS
PROBABILITY OF TANK LEAKAGE*
FR0QAULE TANK LEAKAGE
199.ees
tv
to
-------
APPENDIX r
DATA COLLECTION FORMS AMD MATERIALS
F-l
-------
Intentionally Blank Page
F-2
-------
APPENDIX F
TABLE OF CONTENTS
Page
Open Letter to Owners and Managers of Underground
Motor Fuel Storage Tanks; W.D. Ruckelshaus, United
States Environmental Protection Agency, October 15,
1984 F-ld
General Instructions: U.S. Environmental Protection
Agency Underground Storage Tank Survey F-5
Instructions for Completing the Inventory Sheet
for Tanks with Metered Dispensing Pumps F-24
Instructions for Completing Dispenser Meter
Recording Sheet F-27
Instructions for Completing the Manifolded Tank
System Recording Sheet F-31
Inventory Sheet for Tanks Without Metered Dispensing
Pumps F-40
Reporting Responsibilities of Tank Owners and Operators... F-43
Certification Statement for Establishments Without Tanks.. F-44
Establishment Operator's Questionnaire: United States
Environmental Protection Agency Underground Storage
Tank Survey F-45
Tank to Dispenser Meter Fuel Line Connections F-76
Site Observations Recording Sheet F-77
Call Record/Case Folder: United States Environmental
Protection Agency Underground Storage Tank Survey F-79
Instructions: Preparation for Tank Testing F-83
Environmental Conditions Data Sheet F-84
Temperature Profile Data Sheet F-85
-------
APPENDIX F
TABLE OF CONTENTS (Continued)
Page
Site Diagram and Detail Diagram Sheet F-86
Picture Description Data Sheet F-87
Critical Features Data Sheet F-88
Edit Checklist F-89
Simulated Leak Test Volume Data F-90
Simulated Leak Data Form: Leaking Underground
Storage Tank F-91
Petro-Tite Data Sheet F-93
Line Test Data Sheet F-94
4
-------
* tin \
I UN! t ED £ i A i ES ENVIRCNMEN . AL PROTECTION AGENCY
WASHINGTCN. D C 20460
Tnt ACMINlSTRATOn
OCT 15 m*>
OPEN' LETTER TO OWNERS AND MANAGERS OF
UNDERGROUND MOTOR FUEL STORAGE TANKS
The Environmental Protection Agency (EPA) is conducting a
national survey to learn more about the problem of leaking
underground motor fuel storage tanks ana piping. The purposes cf
the study are to -find out how widespread the leakage problem is,
and to collect information on factors that cause tanKs to leak.
The study will help the Agency assess the impact 6f leaking tanks
on the economy and the environment, ana the need for Federal
regulations to prevent leaking tanks.
I am writing to personally ask for your participation in this
vital project, the results of which could have a major impact as
to how we deal with this potential environmental threat.
Let me assure you that EPA is not conducting this survey to
locate owners of leaking tanks to take legal action against
them. To do so would defeat the purpose of the survey. In the
case of leaking tanks, however, EPA will request that the owner
report any leak to the proper local authority and take corrective
action such as tank repair, replacement or removal from use.
In order to conduct this study, EPA has selected a random
sample of about 1,000 establishments nationwide including far-ns,
gasoline service stations, transportation-related businesses,
businesses with private gas pumps, and government facilities.
The sample of 1,000 estaolishments was selected to represent as
many types of underground storage tank facilities as possible in
order to develop national estimates of leakage on a scientific
basis. Your establishment is one of the 1,000 selected to
participate in this important study.
F-5
-------
Within Che next 2 weeks, an interviewer from Westat, Inc., a
privace contractor conducting the Survey for E?A, will be
contacting you to schedule an appointment for an interview with
you at your place of business. A copy of the interview form is
enclosed. We would appreciate it if you would take the time to
fill out the questionnaire before the interviewer arrives, but do
not mail the Questionnaire back to EPA. The interviewer will
review your answers with you during the visit.
In addition to the interview, the interviewer will be making
a sketch map of your facility layout, and will want to know where
each of your tanks is located. It would be helpful if you have a
map of your tank and dispenser layout ready to show the
interviewer..
As part of the survey,' we will be asking you to provide
product inventory records for a 30-day operating period, so it is
necessary that we know the accuracy of your punm readings. If
the calibration of your pump (or dispenser) meters has not been
checked and certified within the past three months, the
interviewer will need to check the meter calibration with a
certified 5-gallon metering can.
Your inventory data for each tank system will be analyzed by
computer to identify and explain any shortages or overages.
Results of the analyses will be provided to you at no cost and
will be confidential if you so request. Later, we will want to
conduct professional tightness tests on some fraction of the
tanks inventoried in the survey. All tests will be provided free
to the participant, and, if requested, results will be treated as
confidential by the Agency.
The enclosed booklet of General Instructions will provide you
with definitions of key terms, answers to questions you might
have about the survey, and directions on completing the
questionnaire and providing inventory information. If you have
any further questions about this questionnaire, or need any other
assistance, please call Westat at the toll-free survey assistance
number 800/638-8985, and ask for the EPA Specialist.
-------
You may claim confidentiality for ail or any part of your
response under 40 CFR Part 2. You should do this when you
provide the information to the interviewer. A confidentiality
request form is included in the instructions booklet.
Although EPA is conducting the survey.under Federal
authority, we are seeking your full and active participation on a
cooperative basis. I hope we can count on your help.
Enclosures
Sincerely,
William 0. Ruckelshaus
F-7
-------
Intentionally Blank Page
F-8
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
UNDERGROUND STORAGE TANK SURVEY
GENERAL INSTRUCTIONS
Prepared by:
WESTAT
An Employee-Owned Research Corporation
i6SOnm«f-cnBiva « Rockv.iib MD S08S0 •-301 231 -1SOO
F-9
-------
Intentionally Blank Page
F-10
-------
GENERAL INSTRUCTIONS FOR COMPLETING THE
ESTABLISHMENT OPERATOR'S QUESTIONNAIRE
PLEASE READ THE FOLLOWING INSTRUCTIONS BEFORE YOU BEGIN TO
FILL OUT THE ENCLOSED QUESTIONNAIRE. IF YOU SHOULD NEED
FURTHER ASSISTANCE, CALL WESTAT AT THE TOLL FREE SURVEY
ASSISTANCE NUMBER, (800) 638-8985, AND ASK FOR THE EPA
SURVEY SPECIALIST.
PURPOSE OF THE SURVEY
The Environmental Protection Agency (EPA) is conducting
this study to learn more about the problem of leakage in under-
ground storage tanks. The purposes of this study are to find
out how widespread the leakage problem is, and to collect infor-
mation on factors that cause tanks to leak. The study will help
the Ayency assess the impact o£ leaking tanks on the economy and
the environment, and the need for Federal regulations to prevent
leaking tanks.
HOW ESTABLISHMENTS WERE SELECTED
Establishments were selected to participate in this survey
from a preliminary listing of facilities that are likely to have
underground storage tanks. This list was compiled by EPA from a
variety of sources, including government agencies, federal program
rosters, and private and telephone directories. Your facility
rfas not purposely chosen from this listing, but sampled on a
probability basis using scientific random selection procedures.
The purpose of the probability selection procedures is to obtain
i broad representation of kinds of establishments with underground
notor fuel storage tanks.
If your company operates more than one establishment that
las underground motor fuel storage tanks, the establishment you
ire to respond for can be identified by the facility's name and
tddress on the questionnaire label. If the questionnaire label
oes not provide you with enough information to know which estab-
ishment to respond for, please call the EPA Survey Specialist
t the toll free hot line number, (800) 638-8985.
-------
HOW THIS SURVEY WILL BE CONDUCTED
Within the next two weeks, an interviewer from Westat, Inc.
will be contacting you to arrange an appointment for an in-person
interview with you at the establishment location. (Westat, Inc.
is a survey research company that is assisting the EPA in con-
ducting the Under-ground Storage Tank Survey. ) Enclosed with
this instruction booklet is a copy of the questionnaire, so that
you will know what questions the interviewer will ask. In order
to answer some of the questions, you may need to consult your
records, so you should prepare your answers to the interview
before the interviewer calls. Since the interviewer will record
your answers in a separate copy of the interview, the enclosed
copy is yours to keep.
AUTHORITY
This survey is being conducted under authority of Sections
9005 and 9009 of the Resource Conservation and Recovery Act
(RCRA), as amended by the Hazardous and Solid Waste Amendments
of 1984. Subsections (a) and (b) of Section 9005 detail EPA's
authority for conducting the survey and the conditions under
which EPA will treat information provided by owners and opera-
tors as confidential business information (see CONFIDENTIALITY).
Section 9009 details EPA's responsibilities in conducting studies
of underground storage tanks.
REIMBURSEMENT
Section 9009(f) specifies that owners or operators of under-
ground storage tanks shall be provided "fair and equitable reim-
bursement" for "costs, including the loss of business opportunity,
due to closure or interruption of operation of an underground
storage tank solely for the purpose of conducting studies author-
ized by this Section." Under Section 9009(f)(2), claims for
reimbursement must be "filed with the Administrator [of EPA] not
later than 90 days after the closure or interruption whic"h gives
rise to the claim."
-------
CONFIDENTIALITY
Section 9005(b) of RCRA, as ammended requires EPA to make
survey information available to the public upon request, unless
you have requested that the information be treated as confiden-
tial business information under 40 CFR, Part 2 and Section 1905
of Title 18 of the United States Code. As explained in the
Administrator's open letter, you can request that all of the
information you provide be treated as confidential business
information, or that certain items be treated as such. Informa-
tion that has been determined by EPA to be confidential business
information cannot be made available to the public by EPA, but
can be made available to authorized officers, employees and
representatives of EPA, and to the Congress, if requested.
Although EPA is conducting this survey under Federal
authority, we are seeking your participation on a cooperative
basis. Be assured that the contractor and staff conducting the
survey are pledged not to disclose the name or address of any
participant. The contractor provides survey data to EPA identi-
fied only by a participant code number. Only if an establishment
refuses to participate will the name and address be given to
EPA. Should this occur, the Agency may be required to take
legal steps to obtain data necessary to the survey. However, we
would use legal action as a last resort and would strive to
avoid its use.
If you want to request that some or all of the information
you provide will be treated as confidential business information,
please read and complete the "Request for Confidential Treatment
of Business Information" form enclosed with this package. You
should give the completed, signed request form to the interviewer
at the time of the interview.
T7-1 ">
-------
Intentionally Blank Page
F-14
-------
REQUEST FOR CONFIDENTIAL TREATMENT
OF BUSINESS INFORMATION
I hereby request that information I have provided to the
Environmental Protection Agency in response to (certain/all)
the questions in the "Underground Storage Tank Establishment
Operator's Questionnaire" or the "Inventory Record Form" be
treated as confidential business information under 40 CFR
Part 2, and Section 1905 of Title 18 of the United States
Code.
LIST THE QUESTION NUMBERS OF THE RESPONSES FOR WHICH YOU
ARE REQUESTING CONFIDENTIAL TREATMENT:
PLEASE PRINT OR TYPE:
ESTABLISHMENT NAME:
MAILING ADDRESS:
Street
I I ! I t
City State Zip
TELEPHONE: Mil " I I I I - I 1 I I I
Extension
ESTABLISHMENT OWNER/
OPERATOR:
(Print or type) (Signature)
DATE: / /
Month Day Year
-------
DEFINITION OF TERMS
Cathodic Protection - Used to reduce or eliminate corrosion
of a metallic structure which is in contact with corrosive
soil by applying an electric current to the structure
which is greater in strength and opposite in direction to
the current that is causing corrosion.
Passive (galvanic) Cathodic Protection - The required
current is generated by the corrosion of sacrificial
anodes, such as Magnesium or Zinc, which are attached
to the surface of the protected material (tank or
pipe) in the soi1.
Impressed Current Cathodic Protection - The required
current is provided by an external source and is
passed through the system using non-sacrificial
anodes, such as Carbon or Platinum, which are buried
in the ground.
Continuous Electronic Monitoring System - This system could
include the following:
. thermal conductivity sensors;
¦ electrical resistivity sensors;
. gas detector; and
¦ interstitial monitoring in double-walled tanks.
Establishment - The term establishment is used to mean a
commercial or non-commercial location that is used for
any purpose other than just a residence. That is, any
location that is used for a nonresidential purpose (even
if it is also used as a residence) is considered to be an
establishment. Examples of establishments include gaso-
line service stations, farms, schools, factories, fire
stations, highway maintenance facilities, parks, stores,
offices, delivery services, military installations,
airports, etc. (If you believe that your facility does
not fit the definition of an establishment, please call
the toll free survey assistance number, (800) 638-8985,
and explain your situation to the EPA Survey Specialist.)
-------
External Corrosion Protection System - This system could include
the following special equipment or materials:
• cathodic protection;
• electric isolotion;
• polyethylene wrappings;
• coatings; and
• paints.
Inventory Reconciliation - The balancing of "book" inventories
against observed inventories (meter/dipstick readings).
Manway - A means of entrance into an underground storage tank
allowing internal inspection.
Motor Fuel - Any substance that is used to power a motorized
vehicle (such as an automobile, boat, airplane, truck,
etc.). For example, motor fuels such as:
• leaded gasoline;
• unleaded gasoline;
• diesel fuel;
• aviation gas;
• jet fuel; and
• gasohol.
Pressure Pump Delivery System (also called submerged pump delivery
system) - This system works on the principle of positive
pressure to push the liquid from a low point to a high
point using a submerged pump (coupled with an electric
motor) mounted inside the tank.
Remote Gauge - A measuring device that indicates the quantity of
fuel stored in a tank on an external scale or dial.
Secondary Containment - A secondary enclosure or barrier intended
to contain any spills or leakage from the primary storage
tank or from pumps, piping and other equipment. These may
include:
• concrete vaults or basins;
• plastic or clay lined basins;
• soil sealants (soil cement or bentonites); or
• double-walled tanks or pipes.
-------
Siphon Pump Delivery System (also called suction pump delivery
system) - This system works by drawing liquid from a low
point because of a vacuum at a high point, using a suction
pump. This pump is located at grade (i.e., ground level),
either directly above the storage tank or, as in the case
of some dispensing operations, at some distance from the
storage tank (at the pump islands).
Underground Storage Tank - A large vessel or container placed
beneath the surface of the earth used for storing and
handling of liquids (such as petroleum products) or waste
materials (such as used or waste oil).
Used or Waste Oil - Oils (whether used or unused) that are no
longer fit for their intended use because of contamination
or degradation. These oils include, but are not limited to:
• automotive engine oils;
• gear lubricants;
• diesel engine oils;
• railway diesel oils;
• oil storage and treatment residuals (such as bottoms);
• hydraulic oils;
• metal working oils;
• transformer oils; and
• oils contaminated with water.
Water Finding Paste - A paste applied to the bottom of the
dipstick which changes color when it comes in contact with
water.
Water Table - The upper limit of the portion of the ground
(soil) wholly saturated with water.
-------
ORGANIZATION OF THE QUESTIONNAIRE
The Establishment Owners/Operators Questionnaire is designed to
obtain data on your establishment's underground fuel and waste
oil storage operation, including such items as tank design,
operating and installation characteristics, tank corrosion pro-
tection and tank leakage monitoring: The questionnaire is
divided into seven sections, as follows:
A. Screening Information
This section of eleven questions asks for information about
the establishment itself, including questions about the
type of establishment, the owner and operator of the estab-
lishment, and the number of tanks at the establishment.
Question A.11 provides instructions for completing Tank
Description Sheets for the establishment.
Tank Description Sheets - A Tank Description Sheet
must be completed for each underground tank. Ques-
tions asked will include information on specific tank
characteristics, such as reported age, size and typi-
cal fill volume, manufacturer, installer, materials of
construction, inspections or leak tests, and other
design characteristics.
B. Operating Practices
This section asks questions about practices such as taking
tank inventories using a dipstick, checking and recording
dispenser meter readings, inventory procedures after a
delivery and inventory reconciliation or "balancing"
between stick readings, dispenser meter readings, and
delivery records.
C. Operating History
In this section you will be asked about any tanks that have
been replaced, removed without being replaced, or abandoned
in place, and in what year and why this occurred.
D. Permits and Licenses
This is a short section about any special permits or licenses
needed for tank installation or storage of flammable materials
-------
Installation
Section E includes overall questions about how the tank
was installed.
Protection
This section asks questions about any protection systems
in use against external corrosion, and any monitoring
systems used to detect tank leakage.
Information Needs
Section G is about the kinds of information and services
relating to tank monitoring that are currently available
to you.
-------
USE OF THE QUESTIONNAIRE
The questionnaire has been designed to minimize the
effort required for it's completion. "Skip patterns" have been
incorporated to enable respondents to by-pass sections of the
questionnaire which are not relevant to them. The following
section describes how you are to complete the questionnaire in
preparation for the call from a Westat interviewer.
EXAMPLES OF QUESTIONS
Most of the questionnaire items are straightforward
and require only the circling of the correct code(s) or the
completion of short answers on the 1ines which are provided.
The following examples illustrate the use of other question
formats found throughout the questionnaire.
Example A
Some questions require that you indicate a distance or
frequency and also circle the correct unit of measurement or
time as indicated in the sample questions below. Different
units have been specified for your convenience. Please do not
neglect to circle a unit code (as shown} or to write in an
appropriate unit of measurement or time. This question, as with
all questions, includes its own instructions printed in capital
letters and enclosed in brackets.
£4, What is the shortest distance between any of your tanks and any neighboring underground
tan* or other solid underground structure (such as a basement wail, sewer, or utility
vault}? [ENTER DISTANCE AND CIRCLE UNIT CODE]
SHORTEST DISTANCE FROM /
UNDERGROUND STRUCTURE s U? /23-2£
[CIRCLE ONE]:
INCHES . 01
FEET <£> /29-3C
OTHER [SPECIFY]: 03
-------
f2. How often do you inspect your external corrosion protection system? [ENTER
FREQUENCY AND CIRCLE WIT COOE]
IF YOU NEVER INSPECT THE EXTERNAL CORROSION PROTECTION SYSTEM, CHECK HEREI I
AND SKIP TO F3.
/19
FREQUENCY OF INSPECTION: o2. /20-22
[CIRCLE ONE]:
PER DAY 01
PER WEEK. 02
PER MONTH /23-24
PER YEAR 04
OTHER [SPECIFY]! 05
Example B
Other questions require that you code a "yes or "no"
answer for each category listed, as indicated in the sample
question below. The "Other [SPECIFY]" line enables you to enter
an answer not covered by the preprinted response categories.
*ucn 3f folla»inq fj«l types ««re icarra in C!u»
;»rw xirmq past 12 montn«? [CIRCLl ONE C00£ FOR
£.
0ie»ei f"ue: '
ruti J.
^asonoi
Otn.r UPECIF-i:
farm
-------
Example C
When a series of similar questions apply consistently
to a given category, they have been formatted into tables or
grids to facilitate the administration of the questions. Notice
also that Question C6b requests that all applicable response
categories be circled, not just the most prominent one, as indi-
cated in the sample question below.
C6. Please answer the following questions about each tank that has been removed without
being replaced. [SPACE HAS BEEN PROVIDED FOR UP TO FOUR TANKS. IF MORE THAN FOUR
TANKS HAVE BEEN REMOVED WITHOUT BEING REPLACED, WRITE THE ANSWERS FOR THE ADOITIO*Al
TANKS ON A PLAIN SHEET OF PAPER]
First Tank
Second Tank
Third Tank
Fourth Tank
C5a. In what year was the
(first/second/third)
tank removed?
74
81
(year)
/20-23
(year)
/J4-37
(year)
/48-51
(year)
/&2-65
C6b. Why was tne tank
removed?
[CIRCLE ALL THAT
APPLY FOR EACH TANK]
a. Because it
was leaking?. . . .
01
(9
©
01
b. Because other tanks
were being removed
at that time? . . .
©
02
02
02
c. Because it was no
longer needed/in
use?.
03
03
©
03
d. Or for some other
reason [SPECIFY]; .
04
04
04
04
(specify)
724-55
(specify)
738-47
(specify)
/52-61
tsplcifyT
Z66-74
-------
SKIP INSTRUCTIONS
Skip instructions indicate the next question to be
answered. They s.ave time by allowing you to ignore irrelevant
questions. The following is an example of a skip instruction
attached to an answer category.
B1. Do you (or another establishment employee) inventory the contents of your tank(s) by
¦easuring the depth of the contents with a dipstick? [CIRCLE ONLY ONE CODE]
YES [GO Oi TO 82] 1
NO [SKIP TO B5] (T)
Skip instructions are sometimes not attached to an
answer but are enclosed in a box, as shown below.
814. Ho# often is the accuracy of your dispenser weters checked? [CIRCLE ONLY ONE COOE]
/16
IF THE ACCURACY Of YOUR DISPENSER METERS IS NEVER CHECKED, CHECK HERE /32
AND SKIP TO 816.
DAILY 01
WEEKLY 02
EVERY TWO WEEKS 03
MONTHLY 04 /J3-34
ANNUALLY 05
OTHER [SPECIFY]: 06
F-24
-------
THE MOTOR FUEL INVENTORY SHEETS
Enclosed in the survey package are four kinds of sheets for
keeping daily motor fuel inventory records. The type of tank
and dispenser systems you operate will determine which inventory
sheet(s) you will need to use. You may need only one kind of
sheet or as many as three kinds.
In Figure 1 on the following page, you will find schematic
diagrams of the seven most common tank and dispenser hookup
systems currently in use. These seven hookup systems are listed
in Table 1, below. Use the diagrams in Figure 1 to determine
which tank and dispenser hookup system(s) you have. Then use
Table 1 to determine which kind(s) of sheet(s) you should use
for inventory recording.
Table 1. Inventory Recording
Possible tanlc/Dispenser Meter/
Dispenser Hookups
Appropriate Inventory Review Forms
Inventory
Sheet
for Tanks
without
Metered
Dispensing
Pumps
Inventory
Sheet for
Tanks with
Metered
Dispensing
Pumps
Dispensing
Meter
Recording
Sheet
Manifolded
Tank System
Recording
Sheet
Single tank, unmetered
X
Single tank with single
dispensing meter
X
X
Single tank with multiple
dispensing meters
X
X
Custom Blending:
2 tanks, 2 dispensing meters,
1 dispenser
X
X
Custom Blending:
2 tanks, multiple dispensing
meters and dispensers
X
X
Manifolded Tanks:
Multiple interconnected tanks,
multiple dispensing meters
X
X
,x
Manifolded Tanks, Custom
Blending
X
X
1
X |
1
-------
Figure 1
Schematic Diagrams of Possible Tank/
Dispensing lleter/Dispenser Hookup
Single tank. Single tank. Custom Blending:
unmetered sin9'8 dispensing multiple dispensing 2 tanks, 2 dispensing
meter meters meters, 1 dispenser
Single tank.
Custom Blending:
2 tanks, multiple dispensing
meters and dispensers
00
Manifolded Tanks:
multiple interconnected tanks,
multiple dispensing meters
(m) (w) (m) (w)
c
2
Manifolded Tanks, Custom Blending
T-^r NcLpJar V ^ Nr
-------
Regardless of which inventory sheets you use, you will need
to provide 30 complete inventory readings for each of your tanks.
It is preferable that each of these readings represents one
operating day. Many tanks and tank systems are inactive (not
used) for certain days during the week. If your tank(s) are
inactive on a particular day, you can use the inactive day as an
inventory day only if you take and record dipstick readings for
the tank(s) for that day. (You cannot carry down the closing
stick readings from the previous day.) You must provide actual
stick readings (or remote gauge readings, if available) for each
of the 30 inventory days. If your dispensers are metered, you
must also provide meter readings for each of the 30 inventory
days. If you do not have complete inventory information for a
day, do not use that day as an inventory day.
Instructions for using each of the four kinds of Motor Fuel
Inventory Sheets, along with example copies of the Sheets, are
provided on the following pages of this booklet. After you have
used Figure 1 and Table 1 to determine which inventory sheets
you will be using, please read the instructions on how to complete
the sheets.
If you have any questions about:
• Which sheets you should use for your tanks;
• How to complete the sheets; or
• Any recording problems you may have;
please call Westat at the toll-free survey
assistance number, (800) 638-8985, and ask for
the EPA Survey Specialist.
-------
INSTRUCTIONS FOR COMPLETING THE
INVENTORY SHEET FOR TANKS WITH METERED DISPENSING PUMPS
The Inventory Sheet for Tanks with Metered Dispensing Pumps
is used for any individual tank or system of connected tanks
(i.e., manifolded tanks) that has one or more metered dispensing
pumps. The sheet is used to record daily physical inventory
measurements (stick readings and deliveries) and volume of fuel
pumped from the tank, as calculated from dispensing meter read-
ings. You will need one Inventory Sheet for Tanks With Metered
Dispensing Pumps for each tank (or system of tanks) that has
metered dispensers.
You should fill out one line of the Inventory Sheet for
each day that inventory readings are taken. (Days for which
inventory readings are not taken should not be entered on the
sheet.)
• In Column 1, enter the date of the reading (day and
month).
. In Column 2, enter the opening dipstick reading, in
gallons. (On days 2 through 30, opening dipstick
reading will be the same as the closing stick reading
of the line above.)
• In Column 3, enter the day's deliveries to the tank,
in gallons.
• In Column 4, enter the sum of Columns 2 and 3. (This
is your "opening physical inventory."
• In Column 5, enter your closing dipstick reading to
the nearest quarter inch.
• In Colunn 6, enter your closing dipstick reading,
converted to gallons (using your conversion chart for
this tank).
• In Column 7, subtract the amount in Column 6 (your
closing stick inventory) from the amount in Column 4
(your opening physical inventory) and write the
remainder in Column 7. This column represents the
quantity gone from the tank, according to your physi-
cal inventory records.
-------
• In Column 8, enter the "meter sales" (the number of
gallons pumped from the tank according to your meter
readings). You must record the actual meter readings
and calculate the meter sales on a Dispenser Meter
Recording Sheet. Column 8 of the Inventory Sheet
should equal Line I of the Dispenser Meter Recording
Sheet for the same date.
The Inventory Sheet for Tanks With Metered Dispensing Pumps
is printed as a four-page booklet along with a Dispenser Meter
Recording Sheet. (The dispenser Meter Recording Sheet is the
last three pages of the booklet.) Six copies of the Inventory
and Dispenser Sheet booklet are included in the survey package.
If there are more than six tanks with metered dispensers at your
establishment, please photocopy as many additional sheets as are
required.
•F-?9
-------
INVENTORY SHEET FOR TANKS WITH METERED DISPENSING PUMPS
Tank Number:
Dispenser Meter Numbers:
(Name of Facility)
(Street Address)
Type of Fuel:
Size of Tank:
(City/Town)
(State)
(Zip)
Year Installed:
Dipstick* Inventory
Dav
Column 1
Column 2
Column 3
Column 4
Column 5
Column 6
Column 7
Column 8
Date
Opening
Dipstick*
Inventory
(qallons)
Deliveries
(in qallons)
[Column 2]
plus
[Column 3]
Closing
Dipstick*
Inventory
(inches)
Closing
Dipst ick*
Inventory
(qallons)
Gone from
Tank:
(Column 4]
minus
[Column 6]
Meter sales**
(gallons)
(from meter
sheet)
10
11
12
13 I
n
15
16
22—L
18
19 I
20
21
22
24
25
26
27
28
29
30
•If tank has remote gauge, check here |__J and use remote gauge readings instead of stick readings.
•• I of Dispenser Meter Reoordinq Sheet.
r'— JU
-------
INSTRUCTIONS FOR COMPLETING
DISPENSER METER RECORDING SHEET
The Dispenser Meter Recording Sheet is used to record daily
meter readings and to calculate volume of fuel pumped for all
dispenser meters connected to an individual tank or system of
tanks. One 30-day set of Meter Sheets is kept for each individ-
ual tank or connected tank system. On each day of inventory
readings, record each meter's closing reading (in gallons) on
Line G ("Today's Closing Meter"). Record "Yesterday's Closing
Meter" on Line H. (For Day 2 through Day 30, "Yesterday's
Closing Meter" will be the same as Line G ["Today's Closing
Meter"] from the day before.)
The gallons of fuel dispensed daily through a given meter
is calculated by subtracting "Yesterday's Closing Reading" (Line
H) for that meter from its "Today's Closing Reading" (Line G).
Enter the difference between the two readings in Line I for each
meter. This is the number of gallons dispensed (pumped) through
that meter during that day. After you have entered the gallons
dispensed by each meter in Line I, add up Line I for all meters
and enter that figure in the column marked "Line I Totals." The
"Line I Total" figure is the daily "gallons dispensed" for al1
meters. The "Line I Total" must also be recorded for the same
day in Column 8 of the Inventory Sheet for this tank.
Dispenser Meter Recording Sheets are printed in a four-page
booklet, along with an Inventory Sheet. Six copies of this book-
let are included in the survey package. Please photocopy extra
copies if needed.
-------
ni'J'tNSI II W If li HH IWKIM. Mil I I
[Kin IANKS HI III Ml If l!l 0 IHSI'l NSINi; PUMPS |
IriiU" No.s
lype of Iuel:
Mpler 1 Meter 1 Meter
Meter
Meter
Meter
Meter
Heter
line I*-
Day
Hate
Meter Recordinqs in (lallnon
1 11 I 12 I #J
14
16
#7
18
lotala
1
C.
today's Closinq Meter
I 1 1 1 1 1 1
i
1 1
H.
Yesterday's Closinq Meter
I 1 I i t 1 1
L
-
I
I.
Gallons Dispensed (C - M)
1 1 1 1 1 1
1
1
G.
Today's Closirtq Meter
1 1 1 1 1 I
1
2 I
H.
Vesterday's Closinq Meter
1 1 I
1 1 1
1
1.
Gallons Dispensed (G - H)
I I 1 1 I 1 i
1
C.
Today's Closinq Meter
1 I 1 I I 1 1
1
3 I
H.
Yesterday's Closimj Meter
1 1 1 1 ! 1 1
i
1
«.
Gallons Dispensed (G - 10
1 1 1 1 1 1 1
1
C.
Today's Ciosirtq Meter
1 1 1 1 1 1 1
« 1
1 H.
Yesterday's Closinq Meter
1 1 I I 1 1 1
. ¦
I
1.
Gallons Dispensed (G - II)
1 1 1
1 t 1 1
1
1
C.
Today's Closinq Meter
1 1 1 1 1 1 1
i
5 I
H.
Yesterday'a Closinq Meter
1 1 1 1 1 1 1
1
1
'•
Gallons Dispensed (G - H)
1 1 1 1 1 1 1
1
I
C.
Today's Closinq Meter
1 1 1 1 1 1 1
1
6 1
Yesterday's Closinq Meter
1 1 1 1 1 1 1
1
1 :
1
«•
Gallons Dispensed (C - H)
1 1 1 1 1 1 1
1
1
1
| c.
Today's Closinq Meter
1 1 1 1 1 1 1
1
^ 1
1 M.
Yesterday's Closinq Meter
i 1 1 1 1 1 1
1
1
1
I.
Gallons Dispensed (G - H)
1 1 1 1 1 1 1
1
1
i
| C.
Today's Closinq Meter
1 1 1 1 1 1 1
1
i-v\
8 1
1 H.
Yesterday's Closinq Meter
t 1 1 1 I 1 1
1
rO';.r
1
1 1.
Gallons Dispensed (G - H)
1 1 1
1 1 1 1
1
1
I C.
Today's Closinq Meter
1 1 1 1 1 1 1
1
i .
9 1
1 H.
Yesterday's Closinq Meter
1 1 1 1 1 1 1
1
i .
J
I »•
Gallons Dispensed (G - H)
1 1 1 1 1 1 1
1
i
I c.
Today's Closinq Meter
1 I 1 1 1 1 1
1
r
10 )
i H.
Yesterday's Closinq Meter
1 1 1 1 1 1 1
1
i ''. ;' .
1
1 I-
Gallons Dispensed {(; - II)
I 1 1 1 I
1 1
I
i
"For a manifolded tanl< system, list the rxmbers of all of the tanks In the system.
"Transfer line I totala to Column B of Inventory Sheet,
-------
DJSKNSIH Mi Tf« Rl CIIHDINt'i SIM I
Iftm IANKS Willi Ml URFI) OlSHLNfilNG PlIMI'S)
lank* No.*
lype of Fuels
Pay
Date
Heter Kcconlinqa in Gallons
Met er
ft
Meter T
»± L
Wider
Meter
Meter
Meter
Meter
Meter
Line !*•
#3
14
#5
H
#7
<8
Totals
t
1 c.
loday's Hasina Meter I
1
1
I
i
i
1
i
" 1
1 H.
Yesterday's Closinq Meter I
1
1
1
I
I
1
i
i
1 I.
Gallons Dispensed (G - H) |
)
1
1
I
i
i
i
!
i G.
today's Closinq Meter |
I
I
1
l
i
i
I
12 |
I H.
Yesterday's Closing Meter !
1
I
1
i
i
1
i
I
1 *•
Gallons Dispensed (G - H) |
1
1
i
I
1
i
I
1 C.
today's Closinq Meter I
1
I
t
I
i
1
l
u I
1 H.
Yesterday's Closinq Heter |
I
t
1
i
1
1
I
I
1 I-
Gallons Dispensed (G - H) I
i
1
1
t
i
1
i
I
1 G.
Today's Closinq Meter |
1
1
1
i
i
1
t
n |
1 H.
Yesterday's Closinq Meter |
1
1
1
]
I
1
f
I
I I.
Gallons Dispensed (G - M) |
I
I
1
i
[
I
i
I
1 c.
Today's Closinq Meter 1
1
1
1
i
l
1
¦¦
15 |
1 H.
Yesterday's Closinq Meter I
1
1
i
i
f
1
1
1
1 1.
Gallons Dispensed (G - H) 1
1
1
I
I
I
1
1
1
| C.
Today's Closinq Meter I
1
I
I
I
i
1
1
16 |
1 H.
Yesterday's Closinq Heter I
I
1
i
i
i
1
1
1 1.
Gallons Dispensed (G - M) 1
1
i
i
I
1
i
1
1 G.
today's Closinq Meter I
1
1
i
i
I
i i
17 |
I M.
Yesterday's Closinq Meter I
1
1
1
l
l
i
I
1 1.
Gallons Dispensed (G - H) |
1
1
i
i
!
i
I
1 c.
Today's Closinq Meter I
1
i
i •
1
i
h '•
t
18 |
| H.
Yesterday's Closinq Meter 1
1
1
i
I
1
i
i.»\
I
I 1.
Gallons Dispensed (G - M) ]
I
1
I
I
t
i
i
I
1 G.
Today's Closinq Meter 1
1
1
I
i
1
i
i
19 |
1 11.
Yesterday's Closinq Meter I
i
1
i
i
1
i
i
?
i 1.
Gallons Dispensed (G - H) |
I
I
i
l
I
i
i
1
1 c.
Today's Closinq Meter |
1
I
i
1
1
i
t *
zn |
1 M.
Yesterday's Closinq Meter |
1
1
i
i
1
i
I
1
1 1.
Gallons Dispensed (G - II) |
1
1
.. 1
I
1
1
i
•for a Manifolded tank ayate«, list tto> numbers of all of the tanks in the system,
••transfer tine I totals to CoIimti H of Inventory Stinet.
-------
Day
Dal c
oisimnsir mi in! mniiioiNi; situ i
(»I)R I ANKS Nl III M II RID OISI'I NSINU PUM'Sl
lank^ No.
1ype of FukIi
Muter Recordings in Callous
Meter T Iteter
12 I #5
Meter
IU
Meter
#*>
Meter
#6
Meter
17
Meter
fa
Line I" •
Totala
I
i
Today ' a OnBilKJ Meter
i
i
1
i
1 1
i i
i
21 |
I
Yesterday's Closinq Meter
1
1
! 1
! 1
1
1
I n
Galloon Dispensed G - II)
1
I
I
1
1 1
1 !
t
1
1 c.
Today's Closinq Meter
1
1
1
1
1 1
1 1
i
22 |
1 H.
Yesterday's Closinq Meter
1
1
t
1
1 i
1 1
1
1
1 I.
Gallons Dispensed (C - H)
1
1
1
1
1 i
1 I
1
1
1 C.
today's Closing Meter
1
I
1
i
1 i
1 I
1
|
I M.
Yesterday's Closinq Meter
I
1
1
1
1 1
1 1
i
I
1 I.
Callons Dispensed (C - H)
1
I
1
1
i 1
1 1
1
1
1 G.
Today's Cloainq Meter
1
t
1
1
i i
1 1
i
24 |
1 H.
Yesterday's Closinq Meter
i
1
1
1
i 1
i i
1
1
1 I-
Gallons Dispensed (C - M)
I
1
1
1
1 1
1 1
1
1
1 c.
Today's Closing Meter
1
1
I
1
i 1
I t
I
25 |
i H.
Yesterday's Closinq Meter
1
I
I
1
I 1
1 i
I
I
I 1.
Callons Dispensed (C - H)
i
1
1
I
i 1
1 I
1
i
i c.
Today's Closinq Meter
I
I
I
I
i i
i I
1
26 |
1 H,
Yesterday"® Closinq Meter
I
I
i
1
i I
I i
I
1
1 I.
Gallons Dispensed (C - H)
1
I
i
I
1 I
1 I
1
1
[ C.
Today's Closinq Meter
I
1
1
1
i I
I 1
t
27 |
I M.
Yesterday's Closinq Meter
1
1
1
1
1 1
1 1
f
1
1 I.
Gallons Dispensed (G - H)
1
1
I
1
I 1
i 1
i
1
1 c.
Today's Closinq Meter
I
I
i
i
1 i
1 I
i
28 |
1 H.
Yesterday's Closinq Meter
1
1
• 1
1
i i
1 1
i
1
i I.
Gallons Dispensed (G - H)
I
1
i
i
I t
i I
i
1
1 e.
Today's Closinq Meter
1
1
1
1
I i
I 1
i
29 |
1 H.
Yesterday's Closinq Meter
1
i
1
1
i i
1 1
i
1
1 «.
Gallons Dispensed (U - H)
1
i
i
I
i i
I 1
i
!
i c.
Today's Closinq Meter
1
1
1
1
i i
1 1
i
5(1 |
i «.
Yesterday's Closinq Meter
1
1
I
I
I i
1 1
i
1
i 1.
Gallons Dispensed (G - tl)
1
1
1
1
i i
1 1
1
•lor a manifolded tank system, list the lumbers of all of the tanl's in the system,
••transfer Line I totals to CoitMn B of Inventory Short.
-------
INSTRUCTIONS FOR COMPLETING THE
MANIFOLDED TANK SYSTEM RECORDING SHEET
The Manifolded Tank System Recording Sheet is an eight-page
booklet that is used whenever two or more tanks are connected by
piping to make a multiple or manifolded tank system. One Mani-
folded Tank System Recording Sheet booklet is to be used for
each manifolded tank system that is to be inventoried.
The purpose of the Manifolded Tank System Recording Sheet
is to provide a convenient way to keep individual daily stick
and delivery records for each tank in the system. At the end of
each day, you should add up and record each line of inventory
measurements (Lines A through F) for all tanks in the manifolded
system. These daily totals are entered in the "Tank System
Totals" column of the Manifolded Tank System Recording Sheet,
and then transferred to the appropriate columns of the Inventory
Sheet for the tank systems. The "Transfer to Inventory Sheet"
column on the righthand side of the sheet indicates that Inven^
tory Sheet column number to which the total should be transferred.
You must also complete a Dispenser Meter Recording Sheet for
the tank system.
-------
MAN If 01 Dt D TANK SYSltM RECORDING SI It E T
Tank Numbers of Tanks in This Manifolded System:
Type of fuel:
Day
Date
Meter Recordings in Gallons
Tank
#1
Tank
#2
Tank
#3
Tank
tu
Tank
15
Tank
16
Tank
#7
Tank
18
Tank
System
Totals
Transfer
to
Inventory
Sheet
1
A. Opening Stick (gals) (YesterdHy's Line E) | | | | | | | | |
Col 2
1
B. Deliveries (qals) I I I I I I I I I
Col 3
1 |
C. Total of Fuel in lank (A ~ B) | | | | | | | | |
Col 4
I
D. ('losing Stick (inches) I I I I I I I I I
Col 5
I
t. Closing Stick (qals) | | | | II I I I
Col 6
I
F. Fuel Gone from Tank (gals) (C-E) | | | | | | | |
Col 7
I
A. Opening Stick (gala) (Yesterday'a Line E) | | | | | | | | |
Col 2
B. Deliveries (qals) I I I I I I I I I
Col 3
2 |
C. Total of Fuel in Tank (A + B) | | | | | | | | |
Col 4
I
D. Closing Stick (inches) I I I I I I I I I
Col 5
E. Closing Stick (gals) I I I I I I I I
1
Col 6
I
F. Fuel Gone from Tank (gals) (C-E) | | | | | | | | |
Col 7
I
A. Openinq Stick (qals) (Yesterday's Line E) | | | | | | | | |
Col 2
I
B. Deliveries (qalo) I I I I I I I I I
Col 3
3 |
C. Total of Fuel in Tank (A + 0) | | | | | |
1 1 1
Col 4
D. Closing Stick (inches) I I I I I I I I I
Col 5
I
E. Closing Stick (qals) I I I I I I I I
1
Col 6
F. Fuel Gone from Tank (qala) (C-E) | | | | | | | | |
Col 7
I
A. Openinq Stick (qals) (Yesterday'a Line E) | | | | | | | | |
Col 2
I
B. Deliveries (qals) I I I I I I I I I
K\
O
U
4 |
¦J
C. Total of Fuel in Tank (A + B) | | | | | | | | |
Col 4
I
D. ! Closing Stick (inches) I I I I I I I I I
Col 5
I
E. Closing Stick (qals) I I I I I
1 1 1 1
1 Col 6
i
F. Fuel Gone from Tank (qals) (C - C) | |
I I I I III-
1 Col 7
* Transfer Tank Systen Totals to the indicated columns on the correct Inventory Sheet for this Tank System.
-------
MANIFOLDED TANK SYS1EM RECORDING SJILLI
Transfer
Tank
to
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Systea
Inventory
Day
Date
Meter Recordinqs in Gallons
11
12
13
»4
#5
#6
#7
f8
Totala
Sheet
1
i
5 I
1
1
i
A. Openinq Stick (qala) (Yesterday'a Line E) |
1 1 1
Col 2
0. Deliveries (qala) |
1 1 1
Col 3
C. Total of Fuel in lank (A + B) 1
1
1 1 I
Col 4
i D. Clooinq Stick (inches) 1
1 i 1
Col 5
i E. Cloainq Stick (oals) |
1
1 1 1
Col 6
F. Fuel Gone from Tank (qals) (C - E) |
I i I
Col 7
6
i A. Openinq Stick (qala) (Yesterday's Line E) |
1 1
1 Col 2
[ B. Deliveries (qala) |
1 1 I
Col 3
; C. Total of Fuel in Tank (A + B) |
1
1
I
1 Col 4
; D. Clooinq Stick (inches) |
1 I 1
| Col 5
E. Cloainq Stick (qals) |
1
I 1 I 1
1 Col 6
F. Fuel Gone from Tank (quia) (C - E) |
1
1 i I
1 Col 7
7
A. Openinq Stick (qals) (Yesterday's Line E) |
L 1
L_
1
L J
I Col 2
1 8. Deliveries (qala) |
L
L 1
till
1 Col 3
1 C. Total nf Fuel in lank (A + B) |
1
III)
1 Col 4
| 0. Cloainq Stick (inches) |
1111
1 Col 5
E. Cloainq Stick (qala) |
1
1111
I Col 6
F. Fuel Gone from Tank (qals) (C - E) |
1 I 1 1
I Col 7
0
A. Openinq Stick (qals) (Yesterday'a Line f) |
1
1111
1 Col 2
! B. Deliveries (qals) |
1
I
1 1 I 1
1 Col 3
-C» Total of Fuel in lank (A ~ B) |
1
)
1111
1 Col 4
; D. Clooinq St irk (inches) |
1
1
till
I Col 5
«
: E» Cloainq Stick (qals) |
1
I
till
Col 6
F. Fuel Gone from Tank (qala) (C - E) |
1 i
1 1 Col 7
•Transfer lank Syotem Totals to the indicated columns on the correct Inventory Sheet for this Tank System,
-------
MANIFOLDED TANK SYSfEM RECORDING StlEEI
Transfer
Tank
to
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Systea
Inventory
Oey
Date
Meter Recordings in Gal I oris
#1
#2
#3
14
#5
16
17
«B
Totals
Sheet
9
A. Openinq Stick (qala) {Yesterday's Line E) |
Col 2
8. Deliveries (qals) I
1
Col 3
C. Total of fuel in lank (A + B) |
1
Col 4
I D. Closinq Stick (inches)
Col 5
I E, Closinq Stick (qala) |
i
I
1 Col 6
F. Fuel Cone fro» lank (qala) (C ~ E) |
1
1
1
1
1 Col 7
1
1
to |
1
I
1
| A. Openinq Stick (qals) (Yesterday's Line E) |
1
I
1
1
1 Col 2
| B. Deliveries (qals) |
I
i
1
i
1
I Col 3
| C, Total of Fuel in lank (A + B) |
1
i
1
1
I
1 Col 4
; D. Closinq Stick (inches) 1
1
1
1
1
1
1
1 Col 5
i E. Closinq Stick (qals) I
I _
1
1
1
1
i
1
I Col 6
F. Fuel Gone fro* Tank (qals) (C - E) |
| |
1
| |
1
1
f
! Col 7
1
I
11 I
I
I
I
1 A. Openinq Stick (qals) (Yesterday's Line E) |
i
1
1
1
I
i
I Col 2
I B. Deliveries (qals) |
I
1
1
1
1
1
1 Col 3
I C. Total of Fuel in Tank (A + B) |
i 1
i
I
i
I Col 4
I D. Closinq Stick (inches) 1
I
1
1
1
i
I
1
1 Col 5
I E. Closinq Stick (qals) I
I
I
I
1
1
1
I
r
1 Col 6
1 F. Fuel Gone fro* Tank (qals) (C - E) I
I
I
I
1
1
1
i
i
i Col 7
12
I A. Openinq Stick (qals) (Yesterday's Line E) |
I
I
1
1
1
1
I
i
1 Col 2
1 B. Deliveries (qals) |
I
I
I
1
1
1
1
i
I Col 3
1 C. Total of Fuel in lank (A + B) |
I
I
I
1
I
1
f
i
I Col 4
J
I D. Closinq Stick (inches)
i
I
I
i 1
I
I
i
1 Col 5
1 E.» Closinq Stick (qals) |
I
I
1
1
I
f
i
t Col 6
I F. Fuel Gone front Tank (qala) (C - C) |
i
f
1
1
1
1
1
I
1 Col 7
•Transfer Tank Systea Totals to the indicated columns on the correct Inventory 9ieet for this Tank Syatea.
-------
MANIFOLDED TANK SYSTEM RECORDING SHEET
Transfer
Tank
to
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Syatea
Inventory
Day
Oats
Meter Recordinqs in Gallons
11
#2
#3
#4
#5
#6
#7
#8
Totals
Sheet
1
1
1) |
I
I
I
A. Openinq Stick (qals) (Yesterday's Line E) I
Col 2
B. Deliveries (qals)
I
I
I I Col 3
C. Total of Fuel in Tank (A ~ B)
I
I
I I
Col 4
D. Cloainq Slick (inches)
I
I
Col 5
i 1. Cloeinq Stick (qals)
i
i
Col 6
F. Fuel Gone frow Tank (qals) (C - E)
Col 7
I
I
14 J
I
I
I
A. Openinq Stick (qals) (Yesterday's Line E)
Col 2
B. Deliveries (qala)
t
I
I
I Col 3
! C. Total of Fuel in Tank (A ~ B)
I
i
I
I
I Col 4
D, Cloainq Stick (inches)
I
I
I
I Col 5
E. Cloainq Stick (qals)
i
I
I
I
I Col 6
F. Fuel Gone fro« Tank (qals) (C - E)
i
I
I
I Col 7
15
A. Openinq Stick (qals) (Yesterday's Line E)
I
I
I Col 2
i B. Deliveries (qals)
*
I
I
I
I Col 3
I C. Total of Fuel in Tank (A ~ B)
I
I
I
I Col 4
I D. Cloainq Stick (inches)
f
I
I
I
I
I I
I Col 5
I E. Closinq Stick (qals)
I
I
I
I
I
I
I
I
I Col 6
F, Fuel Gone fro* Tank (qals) (C - E)
I
I
I
I
I
I
I
I Col 7
16
1
i
I A. Openinq Stick (qala) (Yesterday's Line E) |
I
I
I
I
I
I
I
I Col 2
| B. Deliveries (qals)
I
I
I
I
I
I
I
I
I Col 3
1 C. Total of Fuel in Tank (A ~ B)
I
I
I
I
I
I
I
I Col 4
I D. Cloainq Stick (inches)
i
I
I
I
I
I
I
I
I
I Col 3
I E. Cloainq Stick (qala)
I
I
I
I
I
i
i
i
I Col 6
I F. Fuel Gone fro» Tank (qals) (C - E)
I
|
I
I
I
I
I
I
I
I Col 7
•Transfer Tank Syate* Totals to the indicated columns on the correct Inventory Sheet for this Tank System.
-------
MANIF011XD TANK SYSTEM RECORDING SHEET
Transfer
Tank
to
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Systea
Inventory
Day
Date
Meter Recordlnqa in Gallons
#1
#2
#3
#4
#5
16
17
18
Totals
Sheet
I
i
17 |
I
I
I
A, Openinq Stick (qals) {Yesterday's line E)
1
Col 2
B. Deliveries (qals)
Col 3
C. Total of fuel in Tank
-------
MANIfOLDED TANK SYSTEM RECORDING SHEET
Transfer
Tank
to
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Syateai
Inventory
Day
Date
Meter Recordings in Gallons
#1
12
#3
#4
#3
*6
#7
IB
Totals
Sheet
21
A. Openinq Stick (qals) (Yesterday's Line E) |
1
1
Col 2
B. Deliveries (qals) I
\
1
Col 3
C. Total of fuel in Tank (A ~ B) |
1
1 1 1
1 Col 4
D. Closinq Stick (inches) |
1
1 1
\ Col 5
E. Closinq Stick (qals) |
1 1
1
1
Col 6
F. Fuel Gone from Tank (qals) (C - E) I
1
I
I Col 7
22
! A. Openinq Stick (qala) (Yesterday's Line E) I
1
I
Col 2
8. Deliveries (qals)
L_
1
I
i Col 3
1 C. Total of fuel in Tank (A ~ 8) |
1
1 I
t
1 Col 4
D. Closinq Stick (inches) I
L
1
1
i i
1
1 Col 5
E. Closinq Stick (qals) I
i 1
I i
1
1 Col 6
F. Fuel Cone from Tank (qala) (C — E) |
1 1
1
I
1 Col 7
23
! A. Openinq Stick (qals) (Yesterday's Line E) |
1
1
1 I
i
1 Col 2
! B. Deliveries (qals) |
1
1
1
i 1
i
I Col 3
| C. Total of Fuel in Tank (A ~ B) |
1 1
I
I
i
1
i
1 Col 4
I D. Closinq Stick (inches)
1
1
1
1
1 1
i
1 Col 5
I E. Cloainq Stick (qals) |
1
1
1
1 1
i
1 Col 6
F. Fuel Gone fro« Tank (qals) (C - E)
L _
i
1
1
I 1
i
1 Col 7
24
4
| A. Openinq Stick (qala) (Yesterday's Line E) t
i
1 1
I
i
1 1
i
1 Col 2
I B. Deliveries (qala) |
i
1 1
I
I
1 I
i
1 Col 3
I C, Total of Fuel in Tank (A + B) |
1 1 1
I
1
1
1
i
1 Col 4
I 0. Closinq Stick (inches) |
1
1 i
I
I
i 1
i
1 Col 5
I E. Cloainq Stick (qals) |
1
1 1
I
1
1
1
i
1 Col 6
! F. Fuel Gone from Tank (qals) (C — E) |
1
1 1
i
1
1 1
i
I Col 7
•Transfer Tank System Totals to the indicated columns on the correct Inventory Sheet for this Tank Syatea.
-------
MANIFOLDED TANK SYSTEM RECORDING SHEET
Day
Date
Meter Recordings in Gallons
Tank
#1
Tank
12
Tank
#3
Tank
14
Tank
#5
Tank
16
Tank
#7
Tank
*8
Tank
Systea
Totals
Transfer
to
Inventory
Sheet
1
A.
Openinq Stick (qals) (Yesterday's Line E) | | | I | I I I |
Col 2
I
B.
Deliveries (qals) 1 1 1 1 1 1 1 1 1
Col 3
25 |
C.
Total of Fuel in Tank (A + B) | | | | I I I I |
Col 4
D.
Closinq Stick (inches) I I I II 1 1 1 1
Col 5
1
f.
Closinq Stick (qals) 1 1 1 1 1 1 1 1 1
Col 6
1
F.
Fuel Gone from Tank (qals) (C - E) | | | | | | I I I
Col 7
1
A.
Openinq Stick (qals) (Yesterday's Line E) | I I I I I I I I
Col 2
1
B.
Deliveries (qals) 1 1 1 1 1 1 1 1 1
Col 3
26 |
C.
Total of Fuel in Tank (A + B) I I I I I I I I I
Col 4
1
D.
Closinq Stick (inches) 1 1 1 1 1 1 1 1 1
Col 5
E.
Closinq Stick (qals) 1 1 1 1 1 1 1 1 1
Col 6
1
F.
Fuel Gone from Tank (qals) (C - E) | | | | I I I I I
Col 7
1
A.
Openinq Stick (qals) (Yesterday's Line E) | | I I I I I I I
Col 2
1
B.
Deliveries (qals) 1 1 1 1 1 1 1 1 1
Col 3
27 |
C.
Total of Fuel in Tank (A + B) | | | I I I I I I
Col 4
1
D.
Closinq Stick (inches) 1 1 1 1 1 1 1 1 1
Col 5
1
E.
Closinq Stick (qals) 1 1 1 1 1 1 1 1 (
Col 6
1
F.
Fuel Gone from Tank (qals) (C - E) | | | | | | I I I
Col 7
1
A.
Openinq Stick (qals) (Yesterday's Line E) | | | I I I I I I
Col 2
1
B.
Deliveries (qals) 1 1 1 1 1 1 1 1 1
Col 3
28 |
c.(
Total of Fuel in Tank (A + B) | | | | I I I I I
Col 4
1
D.
Closinq Stick (inches) 1 1 1 1 1 1 1 1 1
Col 5
1
1
E.
Closinq Stick (qals) 1 1 1 1 1 1 1 1 1
Col 6
1
F.
Fuel Gone fron Tank (qals) (C - E) | | | | | I I I I
Col 7
"Transfer Tank System Totala to the indicated columns on the correct Inventory Sheet for this Tank System.
-------
MANIFOLDED TANK SYSTEM RECORDING SHEET
Transfer
Tank
to
Tank
Tank
Tank
Tank
Tank
Tank
Tank
Tsnk
Syatea
Inventory
Day
Date
Meter Recordinqa in Gallons
#1
#2
#3
#4
§6
#7
#8
Totals
Sheet
1
1
29 |
i
1
1
A. Openinq Stick Coals) (Yesterday's Line E)
Col 2
1 B. Deliveries (qala)
i
1 Col 3
: C. Total of Fuel in Tank (A + B)
I
1
Col 4
[ D. Closinq Stick (inches)
i
Col 5
I E. Closinq Stick (qals)
i
Col 6
F. Fuel Gone fro* Tank (qala) (C - E)
I
Col 7
30
1 A. Openinq Stick (qals) (Yesterday's Line E)
1
Col 2
I B. Deliveries (qals)
I
1 Col 3
C. Total of Fuel in Tank (A + B)
1
i
1
i Col 4
D. Closinq Stick (inches)
1
I ..
1
1
1 Col 5
E. Closinq Stick (aals)
1
1
1
1 Col 6
F. Fuel Gone fro* Tank (qals) (C - E)
1
1
1
! Col 7
•Transfer Tank Systra Totals to the indicated columns on the correct Inventory Sheet for this Tank System.
-------
INSTRUCTIONS FOR COMPLETING
INVENTORY SHEET FOR TANKS WITHOUT METERED DISPENSING PUMPS
The Inventory Sheet for Tanks Without Metered Dispensing
Pumps is the only sheet to be used with tanks having unmetered
dispensing pumps. Without metered dispensing pumps, it is
difficult to use inventory records to monitor for fuel losses,
because the quantities of fuel being pumped from that tank are
unknown. As a result, inventory calculations must be based on
stick readings alone. You will need an accurate dipstick and
the correct inches-to-gallons conversion chart for your unmet-
ered tank.
You will need to make a series of 30 opening and closing
dipstick readings of your unmetered tank. There should be one
or more days between each of the 30 readings. Figure 2, below,
shows two plans for taking the 30 readings.
Figure 2
Inventory Readings Plans for Unmetered Tanks
PLAN A: Immediately before each withdrawal or delivery of
fuel, enter the date and opening stick readings for
the tank on the inventory sheet. Immediately after
the withdrawal or delivery make and record the closing
stick reading on the inventory sheet. Deliveries
should be entered from the delivery receipt you re-
ceive from the fuel truck driver. (All deliveries
will be made when the facility is "open," since the
delivery will be occur between the opening and closing
stick readings.)
PLAN B: At the beginning of each operating day (before any
withdrawals of fuel) record the date and opening stick
reading for the day. At the end of the day (after all
1lne, as long as you ma
closing stick readings.
-------
If your tank is used very infrequently (once a day or less)
you may wish to follow Plan A. Plan A requires that you record
dipstick readings on the tank each time you use the tank. If
the tank is used more than once a day, you should follow Plan B.
Plan B requires that you record dipstick readings at the opening
and closing of each operating (business) day.
The step-by-step instructions for recording inventory on
the "Inventory Sheet for Tanks Without Metered Dispensing Pumps"
are:
• In Column 1, record the date that the inventory reading
will be made (day and month).
• In Column 2, record the opening dipstick reading, in
inches (to the nearest quarter inch).
• In Column 3, record the opening dipstick reading, in
gallons (as calculated from your inches-to-gallons
conversion chart for this tank).
• In Column 4, record the closing dipstick reading, in
inches (to the nearest quarter inch).
• In Column 5, record the closing dipstick reading, in
gallons (as calculated from your inches-to-gallons
conversion chart for this tank).
• In Column 6, record the amount delivered to the tank
since your closing reading on the line above. (The
"Gallons Delivered" should be taken from the receipt
provided by the fuel delivery truck driver.)
• Finally, in Column 7, please indicate whether the fuel
delivery was made before the opening stick reading on
this line (i.e., when the facility was closed) or
during the time between the opening and closing stick
readings (i.e., when the facility was open).
-------
INVENTORY SHEET FOR TANKS XITHOUT METERED DISPENSING PUM»S
(Naae of Facility)
Tank Number: __
Type of Fuel i
Size of Tank:
Year Installed!
(Street Address)
(City/Town)
(State)
(Zip)
Dipstick* Inventory
Coluan 1
Coluan 2
Coluan 3
Coluan 4
Coluan 5
Coluan 6
Coluan 7
Day
Dste
Opening
Dipstick*
Reading
(inches)
Opening
Dipstick*
Reading
(gallons)
Closing
Dipstick*
Inventory
(Inches)
Closing
Dipstick*
Invsntory
(aallons)
Deliveries
(in gallons)
Mas delivery
made while
open or cloaed?
(Circle one)
Open Closed
111 1 1 1 1 112
2 I I I ! I I 112
3 I ! I I I I 112
4 I I I I I I 112
5 I I I I I I 112
6
1 2
7
1 2
8
1 2
9
1 2
10
1 2
11
1 2
12
1 2
13
1 2
14
1 2
15
1 2
16
1 2
17
1 2
18
1 2
19
1 2
20
1 2
21
1 2
22
1
1 2
23
1
1 2
24
1
1 2
25
1
1 2
26
1
1 2
27
1
1 2
28
1
1 2
29
1
1 2
30
1
1
1 2
•If tank has remote gauge, check here
readings.
and use reaota gauge readings instead of stick
-------
REPORTING RESPdNSIBIUTIES
OF TANK OWNERS AND OPERATORS
On November 8, 1984, President Reagan signed the Hazardous
and Solid Waste Amendments of 1984, amending the Resource
Conservation and Recovery Act (RCRA). Section 9005(a) of RCRA,
as amended, states:
"FURNISHING INFORMATION—For the purposes of developing or
assisting in the development of any regulation, conducting
any study, or enforcing the provisions of this subtitle
[Subtitle I of Title VI, 'Regulation of Underground Storage
Tanks], any owner or operator of an underground storage
tank . . . shall upon request of any officer, employee or
representative of the Environmental Protection Agency duly
designated by the Administrator, . . . furnish information
relating to such tanks, their associated equipment, their
contents, conduct monitoring or testing, and permit such
officer at all reasonable times to have access to, and copy
all records relative to such tanks [underline added for
emphasis]. For the purposes of developing or assisting in
the development of any regulation, conducting any study, or
enforcing the provisions of this subtitle, such officers,
employees or representatives are authorized —
"(1) to enter at reasonable times any establishment or
other place where an underground storage tank is located;
"(2) to inspect and obtain samples from any person of
any regulated substance contained in such tank; and
"(3) to conduct monitoring or testing of the tanks,
associated equipment, contents, or surrounding soils, air,
surface water or ground water.
Each such inspection shall be commenced and completed with
reasonable promptness.
Section 9006, "FEDERAL ENFORCEMENT," gives EPA the authority
to issue compliance orders and to commence civil actions for
noncompliance with the requirements of Section 9005. Section
90006(a)(3) authorizes EPA to seek civil penalties for violation
of such an order, not to exceed $25,000 per day of continued
noncompliance.
r'' • -t /
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGtiNC'
WASHINGTON. D.C. tO-U-JU
I'l'itic.r.i.: .»no
Dear Establishment Owner/Operator:
If there are no underground motor fuel storage tanks located at
your establishment, please sign the certification statement below
indicating this and return in the postage paid envelope provided. If
there are abandoned or out of service underground motor fuel storage
tanks at this establishment you should not sign this statement. If
the interviewer calls after you have mailed the signed statement,
inform him/her that you have done so.
Sincerely,
Martin P. Hq&per, Director
Exposure Evaluation Division
Establishment Name:
Establishment Address:
Establishment Telephone:
CERTIFICATION STATEMENT FOR
ESTABLISHMENTS WITHOUT TANKS
THE OWNER OR THE OPERATOR OF THE FACILITY, OR
HIS AUTHORIZED REPRESENTATIVE, SHOULD SIGN
AND DATE THE CERTIFICATION WHERE INDICATED.
THE PRINTED OR TYPED NAME OF THE PERSON SIGN-
THE CERTIFICATION SHOULD ALSO BE INCLUDED
WHERE INDICATED.
CERTITIFICATION:
I hereby certify that there are no underground motor fuel storage
tanks at the establishment at the above address. I am aware that
there are significant penalties for submitting false information,
including the possibility of a fine.
F-48
-------
OMB No.i 2070-0037
Expirees December 31, 1985
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
UNDERGROUND STORAGE TANK SURVEY
ESTABLISHMENT OPERATOR'S QUESTIONNAIRE
Conducted byi
WESTAT
An Employee-Owned Research Corporation
1SSOB'MJ . Roc-.v.ii. MD SOBBO • 301 SSi-iSOO
F-49
-------
A. SCREENING INFORMATION
A1. What type of establishment is this? [CIRCLE ONLY ONE CODE]
L21I
a. FARM OR RANCH 01 /60-61
b. GASOLINE SERVICE STATION.
[PLEASE SELECT ONE OF THE FOLLOWING SUBCATEGORIES]!
/62-63
b1. FULL SERVICE STATION (WHERE MOTOR
VEHICLE REPAIR WORK IS DONE) 02
b2. LARGE, HIGH VOLUME STATION 03
b3. CONVENIENCE STORE 04
b4. SELF SERVICE GASOLINE STATION 05
bS. OTHER [PLEASE DESCRIBE]
06 /64-65
c. MILITARY FACILITY 07
d. FEDERAL AGENCY OR OFFICE 08
e. STATE AGENCY OR OFFICE 09
f. LOCAL GOVERNMENT AGENCY OR OFFICE 10
g. MARINA 11
h. TAXI SERVICE OR COMPANY 12
i. BUS FLEET FACILITY 13
j. TRUCK FLEET FACILITY 14
k. AIRPORT OR AIRFIELD 15
1. RAILROAD OEPOT. . . ., 16
in. OTHER BUSINESS [PLEASE SPECIFY YOUR
ESTABLISHMENT'S PRIMARY PRODUCT OR /66-67
SERVICE]! 17
n. BULK FUEL PLANT OR TERMINAL 18
o. PRIVATE RESIDENCE THAT IS NOT ASSOCIATED
WITH A FARM OR RANCH 19
p. OTHER [SPECIFY]s 20
PLEASE DO NOT COMPLETE THE
REST OF THIS QUESTIONNAIRE!
PLEASE CALL WESTAT AT THE
800-638-8985 (TOLL FREE NUMBER)
AND ASK FOR THE "EPA SPECIALIST."
BOX A1
IF A1 s MILITARY, FEDERAL, STATE OR LOCAL AGENCY (COOES 07, 08, 09 OR 10), CHECK
HERE ~AND SKIP TO A6. OTHERWISE, GO ON TO A2.
A2. Is this establishment owned and/or operated by a major petroleum company? [CIRCLE ONLY
ONE COOE]
768-69
/70
YES 1 /71
NO 2
F-50
-------
A3. What is the name and address of the owner of this establishment?
Owner's Name
Owner1s Address
A4. What is the name and address of the operator of this establishment?
Operator's Name:
Operator's Address:
A5. What is the motor fuel that is stored at this establishment used for: retail sales, tftole-
sale sales, or for use by the establishment itself? [CIRCLE ONE CODE FOR EACH ITEM]
YES H)
a. RETAIL SALES 1 2
b. WHOLESALE SALES 1 2
c. USE BY THIS ESTABLISHMENT 1 2
d. OTHER [SPECIFY]: 1 2
m
/73
/74
/75
/76-77
A6. Does this establishment have any underground storage tanks that are used to store
motor fuel? [CIRCLE ONLY ONE COOE]
/78
YES [GO ON TO A7]. . 1
NO 2
PLEASE 00 NOT COMPLETE THE REST OF
THIS QUESTIONNAIRE! PLEASE CALL
WESTAT AT 800-638-8985 (TOLL FREE
NUMBER) AND ASK FOR THE "EPA
SPECIALIST."
A7. How many underground storage tanks currently in use are used to store motor fuels?
NUMBER OF TANKS: /79-81
AS. Does this establishment hsve any underground storage tanks that are used to
store used or waste oil? [CIRCLE ONLY ONE COOE]
YES [GO ON TO A9] 1 ,
NO [SKIP TO A11] 2 '
F-51
-------
A9. How many underground storage tanks currently in use are uaed to store used or Maste oil?
NUMBER OF USED OR WASTE OIL UNDERGROUND TANKSs
783-85
A10. What (is/are) the capacity/ies of your uaed or waste oil tank(s)? [ENTER CAPACITIES IN
GALLONS]
a. Capacity of used or waste oil tank #1 gallons /86-91
b. Capacity of uaed or waate oil tank #2 gallons /92-97
c. Capacity of uaed or waste oil tank #3 gallons /98-103
T?_ q O
-------
A11. Please fill out one Tank Description Sheet for each underground storage tank that this
facility uses to store motor fuel. There are six (6) Tank Description Sheets bound into
this booklet. If there are more than six underground storage tanks at this establishment,
either photocopy as many additional sheets as are required to describe all the tanks, or
write the answers to the questions for each extra tank on a plain sheet of paper.
TANK DESCRIPTION SHEET INSTRUCTIONS
1. Use the space on the next page to draw a map of the underground tank area at your
establishment. On the map, show the location of each tank, the pumps/dispensers for
each tank, and any buildings and features associated with the tanks (such as a garage,
driveway, or wall). See the example map below showing a gasoline service station with
three tanks and two pump islands.
2. Assign a number to each underground storage tank at this establishment, and write
that number on the tank in your map. (See example below.) Also write the tank number
in the upper lefthand corner of the Tank Description Sheet for that tank.
3. It is only necesssry to fill out Tank Description Sheets for tanks that are on site
at this establishment. Do not fill out Tank Description Sheets for any tanks that
this establishment may use, own or maintain off site.
4. If another establishment uses or maintains an underground storage tank on your
establishment'8 site/property, you should complete a Tank Description Sheet for that
tank and include it on your map.
5. large establishments with more then one tank area may find it easier to draw
individual maps of each tank area, rather than drawing one large map.
v/>
I
O
Example. Map
SW-ion Ixi-ilofirtf)
X<1 ttnd *
, j-
Pi Pi f5
.t
6r
-------
Intentionally Blank Page
-------
L2£l
TASK DCSCRIPT ION SHEET
TANK
NUMBER!
T2.
75.
Tit.
T5.
To.
17.
T8.
T9.
T10.
T11.
T12.
What is trie capacity of this tank'' >That is, what is
the naximuit number of gallons of fuel it can hold?;
[ENTER CAPACITY IS GALLONS)
TANK 0ES1CM CAPACITY: gallons
16-21
What is the average anount of fuel in this tank just
Before a delivery7 (That is, what is the low point of
the product levePi [ENTER QUANTITY IS GALLONS]
AVERAGE CONTENTS BEFORE DELIVERS gallons
22-27
What is the average amount of fuel delivered to this
tank? [ENTER QUANTITY IN GALLONS)
AVERAGE SUE Of DELIVERY: gallons
28-53
What is the maximum maiber of gallons of fuel that ha9
ever Been atorad in this tank' (That is, how full have
vou actually filled it?) [ENTER QUANTITY IN GALLONS]
LARGEST QUANTITY HELD IN TANK: gallons
.'34-59
In what year was this tank installed?
YEAR Of INSTALLATION:
/ 4D—45
Was this tank new or used when it was installed''
[CIRCLE ONLY ONE COOE)
SEW [SKIP TO T8) 1
USED [GO ON TO T7) 2
DON'T KNOW [SKIP TO T3] . . . 8
/'44
How old waa this tank when it was installed7
AGE IN YEARS:
/45- *
7B8-OT
/66
"790-91
-------
!_03|
TANK DESCRIPTION ShEET (Continued)
T25.
126.
In whst year wae the tank laet tested by this/these
methods?
YEAR LAST TESTED:
/16-19
Of what material is this tank constructed? [CIRCLE ONLY
ONE CODE]
FIBERGLASS-REINFORCEO PLASTIC . 01
STEEL 02
OTHER (SPECIFY ]: 03
T27.
T28.
T29.
/20-21
le the inaide of this tank lined? [CIRCLE ONLY ONE COOE]
YES [CO ON TO T28] 1
NO [SKIP TO TJO] 2
DON'T KNOW [SKIP TO T30]. . . 8
.'22
In wrtat year was the lining installed?
YEAR LINEO:
/2J-26
Of xtiat material is the liner constructed? [CIRCLE
ONLY ONE COOE]
EP0XY-8ASE0 RESINS
FIBERGLASS RE INFORCEO PLASTIC . . .
ISOPHTHALIC POLYESTER-BASED RESINS.
P0LYURETHANE-8ASE0 RESINS
OTHER [SPECIFY]:
01
02
03
04
05
T30.
TJ1.
/27-2S
la the outside of this tank coated? [CIRCLE ONLY ONE
COOE]
YES (GO ON TO T31] 1
NO [SKIP TO T32] 2
DON'T KNOW [SKIP TO T32]. . . 8
,'29
Of «nat material is the coating constructed? [CIRCLE
ONLY ONE COOE]
FIBERGLASS/EPOXY 01
ASPHALTIC MATERIAL 02
URE THANE 03
COAL TAR EPOXY 04
OTHER [SPECIFY]: 05
T 32.
/30-31
la there secondary containment for this tank? (CIRCLE
ONLY ONE COOE]
YES [GO ON TO T33] 1
NO [SKIP 10 f34] 2
DON'T KNOW [SKIP TO T34] 8
33. Is this secondary contairmant a:
concrete Deain? 01
plaatic-lined earth basin? 02
clay-lined baain? 03
double-Mall tank? 04
or aomethmg else [SPECIFY]: 05
/32
T34.
,'33-34
la there aecondary containment for any equipment that ia
attached to this tank (euch ea pipes, pumps, valvea,
etc.)? [CIRCLE ONLY ONE COOE]
YES [GO ON TO T35] 1
NO [SKIP TO f36) 2
DON'T KNOW [SKIP TO T36] 8
T35. Ia this secondary containment a:
concrete baain? 01
plaatic-lined earth basin? 02
clay-lined basin? 03
double-wall piping? C4
or something else [SPECIFY]: 05
/ 35
TANK
NUMBER:
T36. What is the name of the company that installed
tank?
INSTALLER:
T37. Is there a pavad surface over the tank?
YES [GO ON !0 138] 1
NO [SKIP 10 T40] 2
T38. Is this pavement:
asphalt? 01
concrete? 02
gravel? 03
other [SPECIFY]: 04
T39. How thick is the pavement?
THICKNESS:
(CIRCLE ONE]:
INCHES 01
FEE I 02
OTHER [SPECIFY]:
03
T40. Nhet is the distance from the surfscs to
the top of the tank?
DISTANCE TO SURFACE:
T43. How was the leak detected and/or verified?
[CIRCLE all THAT APPLY]
INVENTORY RECONCILIATION ... 01
ENVIRONMENTAL MONITORING ... 02
FACILITY inspection 03
TANK TESTING 04
OTHER [SPECIFY]:
05
T44. Have the lines (piping) foe ,thie>tank ever
6aen 'ound to be leaking? [CIRCLE ONLY
ONE CODE]
YES
NO
DON'T KNOW .
the
,36-39
.'40
[CIRCLE ONE]:
INCHES 01
FEET 02
OTHER [SPECIFY]:
03
T41. Does this tsnk have any of the following kinda
of protection against corrosion? [CIRCLE ONLY
ONE COOC FOR EACH ITEM]
/'41-42
,'43-45
/46-47
/48-50
/51-52
Yee No
a. Paaeive cathodic protection
(using sacrificial anodsa)? . .
, 1 2
/53
B. Cathodic protection using
imprsssed current7
1 2
/54
c. Other [SPECIFY]:
1 2
/55
/56-
Hae this tank ever been found to be
1asking?
(CIRCLE ONLY ONE COOE]
YES [GO ON TO T43]
1
NO [SKIP TO T44]
2
DON'T KNOW (SKIP TO T44] . . .
8
/58
,'59-60
/61-62
/63-b4
,65-66
,'67-68
F-56
-------
02!
tank DESCRIPTION SHEET
TANK
NUMBER:
11. What is tne caoacity of this tank7 Tnat is. wnat 19
trie maximum number of aallons of fuel it can nols7;
[ENTER CAPACH" IN GALLONS)
IANK DESIGN CAPACITY: oailons
16-21
12. trial is the average amount of fuel in tnis tank just
Before a delivery0 Inat is, what is the low ooint of
the proauct level"* [ENTER QUAMITt IN GALLONS]
AVERAGE CONTENTS BEFORE DELIVERY: qallons
.22-27
T). What 19 tne average amount of fuel oelivered to this
tank? [ENTER QUANTITY IN GALLONS]
AVERAGE SI2E Of DELIVERY: qallons
28-35
14. What is the maximum number of gallons of fuel that has
ever been stored in this tank1 ;ir>at is, how full have
vou actually filled it?) [ENTER QUANTITY IN GALLON'S)
LARGEST QUANTITY HELD IN TANK: qallons
51-39
15. In
-------
; 3? i
125,
T26.
T 51.
TANK DESCRIPTION SHEET (Continued!
In what year was tie tank last tested by this/these
¦ethods"*
YEAR LAST TESTED:
/16-19
Of unit material is this tank constructed? [CIRCLE ONLY
ONE CDOtl
FI8ERCLASS-REINFQRCED PLASTIC
STEEL
OTHER [SPECIFY}:
01
02
03
W.
128.
/20-21
It the ineide of this tank lined? [CIRCLE ONLY ONE COOE]
YES [CO ON 10 T2B1 1
NO [SKIP TO T301 2
DON'T KNOW [SKIP TO T30). , . 8
nz
In wnat year was the lining installed''
YEAR LINED:
.-'23-26
T29» OF what material is the liner constructed? [CIRCLE
ONLY ONE COOE]
CPOXY-BASEQ RESINS
FIBERGLASS REINFORCED CLASTIC . . .
ISOPMTHALIC P0LYESTER-8ASED RESINS.
POL YURE TMANE-BASED RESINS
OTHER [SPECIFY]:
01
02
0}
Olt
05
/27-2B
T30. Is the outside of this tank coated? (CIRCLE ONLY ONE
CQ0C1
YES [CO ON TO T31 ]. . . .
NO [SKIP TO T321. . . . .
DON'T KNOW [SKIP TO T32].
/29
Of what material is the coating conatructed? [CIRCLE
ONLY ONE C00£)
FI0ERGLASS/EPOXY. .
A5PHA.TIC MATERIAL.
ORE THANE
COAL TAR EPQXY. . .
OTHER [SPECIFY]:
01
02
03
04
05
/30-J1
T32.. la there secondary containment for thia tank? [CIRCLE
ONLY ONE COOE]
YES [CO ON TO T33) 1
NO [SKIP TO T34] 2
CON'T KNOW [SKIP TO US) J
33. Is this secondary containment a:
/ 32
concrete Basin''. .¦
plaatic-i mad earth basin? .
clay-lined oasm?
double-wall tank1
or something else [SPECIFYJ:
01
02
S3
o»
05
/33-34
134. Is there secondary containment for any equipment that la
attached to this tank .such as pipes, pumps, valves,
etc.!? [CIRCLE ONLY ONE CODE]
"ES [CO ON TO 135 J . . .
NO [SKIP TO T36} ... .
DON'T KNOW [SKI? TO T36]
135. Is this secondary containment »:
concrete beam1. ......
plastic-lined earth Basin7 .
clay-lined baain?
double-wall piping?. . . . .
or something else (SPECIFY}s
01
02
OJ
Oft
05
/»
tank
NUMSERi
TJ6.
What is the nana of the company that installed the
tank?
INSTALLER:
T37, Is there a paved surfece over the tank?
YES [CO ON TO T38)
NO [SKIP TO UQ) .
138. Is this pavement:
asphalt?
concrete7 . . . .
gravel1
other [SPECIFY]:
01
02
03
04
T39. now thick is toe pavement?
THICKNESS: __
(CIRCLE ONE):
INCHES
FEET
OTHER (SPECIFY):
01
02
03
140. What is the distance from the surface to
the tap of the tank?
DISTANCE TO SURFACE:
[CIRCLE ONE]:
INCHES .....
FEET
OTHER (SPECIFY]:
01
02
03
T42.
T43.
144.
,'38-39
40
/41-«2
/u3-45
/ 46-47
/48-50
/51-52
T41. Does this tank have any of the following kinds
of protection against corrosion? [CIRCLE ONLY
ONE COOE FOR EACH ITEM)
Ye» No
a. Passive eethodic protection
using sacrificial anodes)?
. , 1 2
,-53
6. Cathodic protection using
impressed current?
. . 1 2
• 54
c. Other [SPECIF*]:
1 2
Hi
'56-57
Haa this tank ever Been found to be leaking?
(CIRCLE ONLY ONE CODE]
YES [CO ON TO T43]
NO [SKIP TO 144)
. . 2
DON'I KNOW [SKIP TO T44j . ,
. . 8
,58
how «es the leak detected and/or
verified?
[CIRCLE ALL THAT APPLY]
INVENTORY RECONCILIATION . ,
. . 01
/59-60
ENVIRONMENTAL MONITORING . ,
, . 02
61-62
FACILITY INSPECTION
. . 03
. 63-64
tank testing
. . 04
<65-66
OTHER [SPECIFY]:
05
>~6i7 -iS
Have the lines piping) for this
tank sver
been found to be leaking? [CIRCLE ONLY
ONE COOE]
^ "*
YES
. 1
NO
, 2
69
OON'T KNOW
. 8
,36-37
F-58
-------
: 021
TANK DESCRIPTION SriEi!
TANK
NUMBER:
T3.
T5.
T6.
T7.
T8.
T9.
T10.
111.
T12.
T13.
wnat is the capacity of this tank** That is, what is
the maximum number of gallons of fuel it car. hold0:
[enter capacity in gallons]
TANK DESIGN CAPACITY: gallons
• 16-21
What is the average amount of fuel in this tank just
before a deliverv7 ;That is, what is the iow point of
the product level?, [ENTER QUANTITY IN GALLONS]
AVERAGE CONTENTS BEFORE DELIVERS: gallons
22-27
What is the average amount of fuel Delivered to this
tank9 [ENTER QUANTITY IN GALLONS]
AVERAGE SIZE OF DELIVERY: gallons
'28-33
What is the maximum number of gallons of fuel that has
ever been stored m this tank? ;That is, how full have
vou actually filled it7) [ENTER QUANTITY I\ GALLONS]
LARGEST QUANTITY HELD IN TANK: gallons
,34-39
In what year was this tank installed7
YEAR Of INSTALLATION:
/40-43
Was this tank new or used when it was installed?
[CIRCLE ONLY ONE CODE]
NEW I SKIP TO T8] 1
USED [GO ON TO T7] 2
DON'T KNOW [SKIP TO T8] . . . 8
/44
How old was this tank «rt?en it was installed?
AGE IN YEARS:
Z45-47
Is this tank scheduled for replacement or repair
within the next 12 months? [CIRCLE ONLY ONE CODE)
YES 1
NO 2
/40
Has this tank aver been repaired? [CIRCLE ONLY one CODE]
YES [GO ON TO T10] 1
NO [SKIP TO T12] 2
DON'T KNOW [SKIP TO T12). . . 8
/49
In what year was this tank last repaired?
YEAR LAST REPAIRED:
750-53
What types of repairs were done to this tank7
REPAIRS:
754-55
756-57
Which of the following fuel types were stored in this
tank during the past 12 months? [CIRCLE ONE CODE TOR
EACH FUEL TYPE]
a. Leaded gasoline *
b. Unleaded gasoline
c. Diesel fuel I . .
d. Aviation fuel
e. Gasohol
f. Other [SPECIFY]:
Does this tank have a pump?
YES [GO ON TO T14). ..... 1
NO [SKIP TO T?8] 2
YES
1
1
1
1
1
1
NO
2
2
2
2
2
2
/5B-6*
T14. how many pumps are connected to this tank"*
NUMBER OF PUMPS
/6?-69
T15>. Does tnis tank have a "suction" or a "submerged" spres-
sure: pump delivers system7 [CIRCLE. GNU ONE CODE]
SUCTION 0T
SUBMERGED 02
Of HER [SPECIFY]: 03
70-71
T16. How many dispensers \nozzles'' are connected to this tank7
NUMBER OF NOZZLES:
772-74
T17. Oo the product dispensers \nozzles) for this tank have
meters to measure the total Quantity of product that has
been pumped from the tank? [CIRCLE ONLY ONE CODE]
YES 1
NO 2
!V>
T18. Is this tank attached to another tank by pipes or lines?
[CIRCLE ONLY ONE CODE]
YES [PLEASE SPECIFY THE.
TANK NlWBERiS; OF
THE CONNECTED IANK{S)] /76
1
NO 2
Z77-78
Tt9. How is this tank situated in relation to the water
table7 Is it: ICIRCLE ONLY ONE CODE]
Completely above the water table. ... 01
Partially above and partially
below the water table 02
Or, is the top surface of the tank
completely below the water table ... 03
Other [SPECIFY]: 04
/79-80
T20. Does this tank have a manway or other means of being
entered for internal inspection? [CIRCLE ONLY ONE CODE]
YES [CO ON TO T21] 1
NO [SKIP TO T23] 2
DON'T KNOW [SKIP 10 T23). . . 8
/81
T21. Has the interior of the tank ever been inspected?
[CIRCLE ONLY ONE CODE]
YES [GO ON TO T221 1
NO [SKIP TO T23] 2
DON'T KNOW (SKIP 10 T23). . . 8
/82
T22. When waa the most recent internal inspection of this
tank?
MOST RECENT INSPECTION:
/83-B6
T23. Has the tank ever been tested for leaka after it waa
placed in service7 [CIRCLE ONLY ONE COOT]
YES [CO ON TO T24] 1
NO [SKIP TO T26] 2
DON'T KNOW [SKIP TO T26]. . . 8
/87
T24. What test method was uaed to teat the tsnk? (Please
give the brand name of the teat, if kno**\, and deacribe
the teat procedure. If more than one method waa uaed,
describe all methods used.)
METHOD(S):
/88-B9
/66
"790-91
-------
ODD
TANK DESCRIPTION SHEET (Continued; j TANK
NUMBER t
T25. In "Hat yeer wee the tank laat tested by this/these T36. What is the name of the company that inatalltd tha
aethode? tank?
YEAR last TESTED: INSTALLERt
/16-19 / 38-39
126. Of rfiet matarial ia this tank constructed' [CIRCLE ONLY 137. la thaca a pavad surfscs ovar tha tank?
ONE COOE]
YES [GO ON TO 138] 1
FIBERGLm i-RElNFORCEO PLASTIC . 01 NO [SKIP 10 T40] 2
STEEL 02 AO
OTHER [SPECIFY]: 03 T38. Ia thia psvemant:
T27.
T2B.
/20-21
la tha inaida of thia tank lined? [CIRCLE ONLY ONE CODE]
YES [GO ON TO T28] 1
NO [SKIP TO 130] 2
OON'T KNOW [SKIP TO T30]. . . 8
/22
In whet yaar wes tha lining installed?
YEAR LINED:
/23-26
129. Of what Mteriel ia tha linar conatructed? [CIRCLE
ONLY ONE COOE]
EP0XY-8ASE0 RCSINS 01
FIBERGLASS REINfORCEO PLASTIC ... 02
ISOPHTHALIC POLYESTER-BASED RESINS. 03
POLYURETHANE-BASED RESINS 04
OTHER [SPECIFY]: 05
eaphalt? 01
concrete? 02
gravel? 03
other [SPECIFY]: 0*
T39. How thick ie the paveawnt?
THICKNESS:
[CIRCLE ONE]:
INCHES 01
FEET 02
OTHER [SPECIFY]:
03
T40. Whet la the diatenca froai the eurfaca to
the top of the tank?
,41-42
.'43-45
,44-4?
DISTANCE TO SURFACE:
/27-2B
T30. la the outaide of thia tank coated? [CIRCLE ONLY ONE
COOE]
YES [GO ON TO T31] 1
NO [SKIP TO T32] 2
DON'T KNOW [SKIP TO 152]. . . 8
/29
T31. Of rfiet aetenel la tha coating conatructed? [CIRCLE
ONLY ONE COOE]
FIBERGLASS/EPOXY 01
T34.
T35.
(CIRCLE ONE]:
INCHES 01
FEET 02
OTHER [SPECIFY]:
03
Ul. Doee thia tank nave any of tha fallowing kinda
of protection ageinet corroaion? [CIRCLE ONLY
ONE COOE FOR EACH ITEM]
T32.
133.
,30-31
[CIRCLE
,'32
Is there aecondary containment for this tank?
ONLY ONE COOE]
YES [GO ON TO T33] 1
NO [SKIP TO T34] 2
OON'I KNOW [SKIP to T34] a
Ia thia secondary containment a:
concrete besin? 01
plastic-lined aerth beam? 02
clay-lined baain? 03
double-well tank' 04
or something elaa [SPECIFY]: 05
,33-34
Is there secondery conteinment for any equipment thet ia
attached to thia tank (such ss pipes, pumps, valves,
ate.)? [CIRCLE ONLY ONE COOE]
YES [GO ON TO T35] 1
MO [SKIP TO T36] 2
OON'T KNOW [SKIP TO T36] 9
, 35
Is this secondary containment s:
concrete baain? 01
plaatic-lined earth beain? 02
clay-lined oaein? 03
double-wall piping? 34
or something else [SPECIFY]: 35
142. Hss this tsnk aver been found to be leaking?
[CIBCLE ONLY ONE COOE]
YES [GO ON TO 143] 1
NO [SKIP TO T44) 2
DON'T KNOW [SKIP TO T44] ... 8
743. How wea the leek detected end/or verified?
[CIRCLE ALL THAT APPLY]
INVENTORY RECONCILIATION ... 01
ENVIRONMENTAL MONITORING ... 02
FACILITY INSPECTION 0)
TANK TESTING 04
OTHER [SPECIFY):
05
T44. Heve the lines 'piping; for this tank aver
been found to be leaking? [CIRCLE ONLY
ONE COOE ] ->
YES 1
NO 2
DON'T KNOW 8
,48-50
/SI-52
asphaltic material. . .
. . . 02
Yea
NO
URETHANE
a.
Paaaive cethodic orotection
COAL TAR EPOXY
. . . 04
iuaing sacrificial anode*)? .
. 1
2
'53
OTHER [SPECIFY]:
05
b.
Cathodic protection ueing
impreeeed current?. .....
2
/54
c.
Other [SPECIFY]:
1
2
55
.56-57
50
/59-60
61 -62
, 63-64
65-66
,67-68
49
Z36-37
-------
021
T1.
T2.
TJ.
H.
T5.
T6.
17.
T8.
T10.
Tit.
f 12.
tank description sheet
unit 1* the capacity or this tank? (That la, whet la
the maxiaun mabar of gallona of fuel It can hold?;
[ENTER CAPACITY IN GALLONS]
TAN- DESIGN CAPACITY! gallona
/14_21
What la tha average aiaount of fuel In thia tank Juet
before a delivery? (That la, what la tha loo point of
tha product laval?) (ENTER QUANTITY IN GALLONS]
AVERAGE CONTENTS 3EF0RE DELIVERY I
gallona
-'22-27
Whet la tha everage aaiount of fual delivered to thia
tank? [ENTER QUANTITY IN CALLONS]
AVERAGE SIZE OF OCLlVERY: Gallona
/28-J3
Whet la tha Mxiaua nuaber of gallona of fual that haa
aval baan atotad In thia tank? (That la, how Full ha*a
vou actually filled It?) [ENTER QUANTITY IN GALLONS]
LARGEST QUANTITY HELD IN TANKi
In «hat yaar «*a thia tank lnetelled?
YEAR OF INSTALLATION!
gallona
/J4-39
Haa thia tank new or uaed when it *«a lnatalled?
[CIRCLE ONLY ONE CODE]
NEW [SKIP TO T8] 1
USED [GO ON TO T7] 2
DON'T KNOW [SKIP TO T8] . . . 8
Hon old ««a thia tank whan it wee lnatalled?
AGE IN YEARS I
/40-43
/44
/ft 5-47
la thia tank scheduled for replacenent or repair
within tha next 12 montha? [CIRCLE ONLY ONE COOE]
YES
\0.
/48
In what yaar «aa thia tank laat repaired?
YEAR LAST REPAIRED!
/49
Mhat typaa of rapaira ware dona to this tank?
REPAIRSi
/50-53
T54-55
75J-57
WUch of tha following fuel typaa «era stored in thia
tank during tha paat 12 nontr.a? [CIRCLE ONE COOE FOR
EACH FUEu TYPE]
a.
5.
c.
c.
a.
f.
-?aded gaeoline .
.nleaoed qisoline
Dieael fual . ¦ •
Aviation fual . .
Gaaohol
Othar CSPECIFYj:
YES
I
1
1
1
1
T13. Ooea tma :erw t\ava a pump?
YES [20 ON *u T14] 1
SO [SKIP TQ T18) 2
NO
2
Z
2
2
2
2
,58-65
66
TANK
NUMBERt
T14. How many puapa ara connected to thia tank?
NUMBER OF PUMPS
/67-69
T15. Doee thia tank have a "auction" or a "autoaierged" (pree-
aure) punp delivery ayatm? [CIRCLE ONLY ONE COOE]
SUCTION 01
SUBMERGED 02
OTrtER [SPECIFY]! 03
/70-71
T16. Hon many dlapenaera Cnoizlea) ara connected to thie tank?
NUMBER OF NOZZLESi
m_u
T17. Do the product diapanaara inoiilee) for thia tank have
metara to Maaure the total quantity of product that haa
been pumped froai the tank? [CIRCLE ONLY ONE COOE]
YES
NO.
T18.
/75
la thia tank attached to another tank by pipaa or linea?
[CIRCLE ONLY ONE COOE]
YES [PLEASE SPECIFY THE
TANK Nl*IBER(S) OF
THE CONNECTED TANK(S)]
/76
NO.
T19.
/77-7S
'no» la thia tank aituated In relation to tha water
table? la it! [CIRCLE ONLY ONE COOE]
Completely above the weter table. .
Partially above and partially
below the water taola
Or, la the top surface of the tank
completely below the water table .
Other [SPECIFY]:
01
02
03
04
T9. Hee thia tank evar baan repaired? [CIRCLE ONLY ONE COOE] T20.
YES [GO ON TO T10] 1
NO [SKIP TO T12] 2
DON'T XNOW [SKIP TO T12). . . 8
/79-80
Doea thia tank have a marmay or othar aaana of being
entered for Interne! inapaction? [CIRCLE ONLY ONE COOE]
YES [GO ON TO T21] 1
NO [SKIP TO F23] 2
OON'T KNOW [SKIP TO T23J. . . 8
T21. Haa the interior of trie tank ever been inapected?
[CIRCLE ONLY ONE COOE]
YES [GO ON TO T22] 1
NO [SKIP TO T23] 2
DON'T KNOW [SKIP TO T23]. . . 8
/81
T22.
/a 2
Xhen «aa the moat recent internal inapaction of thia
tank?
MOST RECENT INSPECTION!
/33-86
T23. Haa the tank aver been teatad for leaks after it waa
placed in service? [CHCLC ONLY ONE COCT]
YES [GO ON TO T24] 1
NO [SKIP ro T2S] 2
DON'T KNOW [SKIP TO T26). . . 8
.•'87
t24. ttiat teat aethod «as uaed to teat the tank"" ..Please
give the brand nana cf the test, if known, and describe
the teat procedure. If sore than one methoa »aa useo.
deacriba all nathoda uaed.;
HETHODCS;i
,88-89
•90-91
¦o 1
-------
TANK DESCRIPTION SHEET 'Continued)
tank
NUMBER:
T25.
T26.
In »nat year was the tank last tested by this/these
method*?
YEAR LAST TESTED:
T36. What is the name of the company that metalled
tank?
INSTALLER:
/16-19
Of Nftat material is this tank constructed17 [CIRCLE ONLY
ONE CODE]
FIBERGLASS-RE INFORCED PLASI1C . 01
STEEL 02
OTHER [SPECIFY ]: 05
127.
T28-
T29.
,20-21
la the inside of thi3 tank lined? (CIRCLE ONLY ONE COOC}
YES [GO ON TO 128] 1
NO [SKIP TO T30] 2
DON'T KNOW [SKIP TO T30]. . . 8
/22
In what year was the lining installed?
YEAR LINED: __
/2J-26
Of what material is the liner constructed? [CIRCLE
ONLY ONE COOE]
EPOXY-BASED RESINS 01
FIBERGLASS REINFORCED PLASTIC ... 02
ISOPHTHALIC POLYESTER-BASED RESINS. 03
POLYURETHANE-BASED RESINS 04
OTHER [SPECIFY]: 05
T37. Is there a paved surface over the tank?
YES [GO ON F0 T38] 1
NO [SKIP TO 140] 2
T38. Is this pavement:
asphalt? 01
concrete? . 02
gravel? 03
other [SPECIFY]: 04
T39. How thick is the pavement?
THICKNESS:
ICIRCLE ONE]:
INCHES
FEET
OTHER [SPECIFY]:
01
02
03
T40. v*iat is the distsnce from the surface to
the top of the tank?
DISTANCE TO SURFACE:
T30.
Z27-28
Is the outside of this tank coated? [CIRCLE ONLY ONE
CODE]
YES [GO ON TO T31 ] 1
NO [SKIP TO T32] 2
DON'T KNOW I SKIP TO T32). . . 8
[CIRCLE ONE]:
INCHES
FEET
OTHER [SPECIFY]:
T31. Of «nat material is tne coating constructed?
ONLY ONE COOE]
/ 29
[CIRCLE
01
02
03
T41.
FIBERGLASS/EPOXY.
01
.30-31
132. Is there 3econdarv containment for this tank? [CIRCLE
ONLY ONE CODE)
YES [CO ON 10 T33J 1
NO [SKIP 10 T34] 2
8
DON'T KNOW [SKIP TO T34]
33. Is this 3econoarv containment a:
concrete basin''
plastic-lined earth basin"' .
clay-lined basin''
uOuole-wall tank?
or something else [SPECIFY]:
, 32
01
02
03
04
05
/33-34
T34. Is there secondary containment for any equipment that is
attached to this tank \such as pipea, pumps, valves,
etc./? ICIRCLE ONLY ONE CCOE]
YES [CO ON TO T35] :
NO [SKIP TO T36] 2
CON'T KNOW [SKIP TO T36] 8
T35. Is this secondary containment a:
concrete basm*7 01
olastic-lined earth basin? 02
clay-lined basin? 03
dcuOle-wall piping? 04
or something else [SPECIFY]: 05
'35
T42. Has this tank ever been found to De leaking'
[CIRCLE ONLY ONE COOE]
YES [GC ON TO T43] 1
NO [SKIP 10 Ta-] 2
DON'T KNOW [SKIP TO T44] ... 8
Tu3. How was the leak detected and/or verified?
[CIRCLE ALL That apply)
INVENTORY RECONCILIATION ... 01
ENVIROWfENTAL MONITORING ... 02
FACILITY INSPECTION 03
TANK TESTING 04
OTHER [SPECIFY):
05
T44. Have the lines ^piping) for this tank ever
been found to be leaking? [CIRCLE ONLY
ONE COOE]
YES 1
NO 2
DON' r KNOW 8
/ 3 8—3 9
JO
,'41-42
.43-45
Z46-47
/48-50
/ 51-52
Does this tank have any of the following kinds
of protection against corrosion? [CIRCLE ONLY
ONE COOC FOR EACH ITEM]
ASPHALTIC MATERIAL. . .
. . . G2
Yes No
URETHANE
. . . 03
a.
Passive cathodic protection
COAL IAR EPOXY
. . . 04
'using sacrificial anodes)? .
. 1 2
55
OfHEK [SPECIFY ]:
05
D.
Cathodic protection using
impressed current'*
. 1 2
'5*
c.
Other [SPECIFY]:
1 2
. 55
56-57
59-60
,61-62
6 3-64
65-66
.'67-63
F-62
-------
TANK DESCRIPTION SHEET
IANK
NUMBER:
Tt.
T2.
13.
F».
T5.
16.
17.
18.
19.
no.
in.
:iz.
Whet is the capacity of this tank' .That is, "hat is
the *a*wum number of gallona of fuel it can hold"'!
[ENTER CAPACITY IN GALLONS]
TANK DESIGN CAPACITY: gallons
16-21
what is the average amount of fuel in this tank just
before a delivery? (That Is, aflat is the low point of
the product level?) [ENTER QUANTITY IN GALLONS)
AVERAGE COMENTS BEFORE DELIVERY: gallons
22-27
ttiat is the average mount of fuel delivered to this
tank? [ENTER QUANTITY IN GALLONS]
AVERAGE SUE Of DELIVERY: gallons
26-33
what is the maximum number of gallons of fuel that has
ever been storad in this tank? iThat is, haw full nave
vou actually filled it?) [ENTER QUANTIFY IN GALLONS]
LARGEST QUANTITY HELD IN TANK: gallons
.34-39
In what year was this tank Installed?
lEAR Of INSTALLATION:
uO-A}
Was this tank new or used when it was installed?
[CIRCLE ONLY ONE COOt]
NEW [SKIP tO T8]. ...... 1
USED [GO ON '0 '7] 2
OON'T KNOX [SKIP TO T8] . . . 8
/»»
How old was this tank when it was installed?
1GE IN YEARS:
,45-47
Is this tank scheduled for replacement or repair
within the next '.2 months? [CIRCLE ONLY ONE COOE]
YES 1
NO 2
'48
Has tnis tank ever Been repaired? [CIRCLE ONLY ONE COOE]
YES [GO QN TO 110] 1
NO [SKIP TO T12) 2
DON'T KNOW [SKIP 10 T12). . . 8
49
In whst year was this tank last repaired?
YEAR LAST REPAIRED:
, 50-53
What types of repairs were done to this tank?
REPAIRS:
.54-55
36-57
Which of the following fuel tvpes were stored in '.his
tank during the past 12 months? [CIRCLE ONE CCOE fOR
EACH ruEL TYPE].
YES
F13.
a. Leaded gasoline
b. Unleaded gasoline
c. Oiesel f'jel
d. Aviation fuel
9. Gasohol
f. Otner [SPECIFY]:
Ooes this tank have a pump?
yES [CO ON TO 114) 1
so [skip 'o rial. 2
NO
2
2
2
2
58-65
Tla. How many pumps are connected to this tank?
NUMBER Of PUMPS
.67-69
T15. Does this tank have a "suction11 or a "submerged* Cpres-
sure) pump delivery system'' [CIRCLE ONLY ONE COOE]
SUCTION 01
SUBMERGED 02
OTHER [SPECIFY]: 03
.'70-71
T16. How many dispensers inozzles) sre connected to this tank?
NUMBER Of NOZZLES:
*T2-7»
117. Do the product dispensers .nozzles) for this tank have
netera to measure the total quantity of product that has
been p>»ped from the tank? [CIRCLE ONLY ONE COOE]
YES 1
NO 2
'75
T18. Is this tank attached to another tank by pipes or lines?
[CIRCLE ONLY ONE COOE]
YES [PLEASE SPECIFY THE
[ANK NUMUER(S) Of
[HE CONNECTED TANK(S)) /76
1
NO 2
/77-78
T19. How is this tank situated in relation to the water
table' Is it: [CIRCLE ONLY ONE COOE]
Completely above the water table. ... 01
Partially above and partially
below the water table 02
Or, is the top surface of the tank
completely below the water table ... 03
Other [SPECIFY]: 0«
79-80
T20. Does this tank nave a mannaV or other means of being
entered for internal inspection' [CIHCLE ONLY ONE COOC ]
YES [GO ON TO F21] 1
NO [SKIP tU T23] 2
OON'T KNOW [SKIP TU 125). . . 8
;ai
T21. Has the interior of the tsnk ever been inspected?
[CIRCLE ONLY ONE L'JDE]
YES [GO ON TO F22] 1
NO [SKIP TO T2J] 2
OON'T KNOW [SKIP TO T23J. . . 8
82
T22. When waa the most recent internal inspection of this
tank?
MOST HECENT INSPECTION:
33-86
T23. Has the tank ever oeen tested for leaka a*tar it was
placed in service'' [CIRCLE ONLY ONE CCOfj
YES [GO ON TO T24) 1
NO [SKIP 10 T26) 2
DON'T
-------
TANK DESCRIPTION SHEET ^Continued;
T25. In what y«»f "as the c,nk l»at tested by this/these
methods?
YEAR LAST T E SI EDi
1)5.
.'16-19
126. Of wnat materiel :s this tank constructed? [CIRCLE ONLY
ONE CCOt)
FIBERCLASS-REINFORCED PLASTIC . 01
STEEL 02
OTHER [SPECIFY]: 0)
127.
T28.
T29.
/20-21
Is the inside of this tank lined? [CIRCLE ONLY ONE CODE]
YES (CO OH TO T28] 1
NO [SKIP TO T30] 2
DON'T KNOW [SKIP TO T30]. . . 8
.'22
In wHat year was the lining installed?
YEAR LINED:
23-26
Of what material is the liner constructed? [CIRCLE
ONLY ONE COOE]
EPOXY-BASED RESINS
FIBERGLASS REINFORCED PLASTIC . . .
ISOPHTHALIC POLYESTER-BASED RESINS.
POLYUREThANE-BASED RESINS
OTHER [SPECIFY]:
01
02
0)
04
05
,27-28
T30. Is the outside of this tank coated? [CIRCLE ONLY ONE
COOE]
YES [CO ON TO T31] 1
NO [SKIP 10 1321 2
DON'T KNOW [SKIP TO T32]. . . 9
.'29
T31. Of "hat «aterial is the coating constructed? [CIRCLE
ONLY ONE COOE]
FIBERGLASS.'EPOXY 01
ASPHALT IC MATERIAL 02
URE THANE 03
COAL TAR EPOXY 00
JTHEK [SPECIFY]: -35
,30-31
T32. Is tnere secondary containment fcr this tank? [CIRCLE
ONLY ONE COOE]
YES [CO ON TO T33] 1
NO [SKIP TO T34) 2
DON'T KNOW [SKIP TO T34) 8
.32
33. Is this secondary containment a:
concrete basin? 01
plastic-lined earth basin? 02
clay-lmed baam1 G3
double-wall tank? 0
-------
L°il
TANK DESCRIPTION SHEET
TANK
NUMBER:
YES
a. Leaded gasoline
b. Unleaded gasoline
c. Oiesel fuel
d. Aviation fuel
e. Gasohol
f. Other [SPECIFY]!
T13. Does this tank lave a pump?
YES [GO CN '3 Tit] 1
NO [SKIP TO 5181 1
NO
2
2
2
2
2
58-65
Tl. What is the capacity of this tank? (That is, "hat is
the insxiauffl nunber of gallons of fuel it can hold?)
[ENTER CAPACITY IN GALLONS]
TANK DESIGN CAPACITY: gallons
. 16-21
T2. *hat i> the average amount of fuel in this tank just
Before a deiiverv? IThat is. whet is the low point of
the product level?) [ENTER QUANTITY IN GALLONS]
AVERAGE CONTENTS BEFORE DELIVERY: gallons
22-27
T3. What is the average amount of fuel delivered to this
tank? [ENTER QUANTITY IN CALLONS]
AVERAGE SIZE OF OELIVERY: gallons
28-35
T
-------
: 03;
FANK DESCRIPTION SHEET ;Continued)
T25. In what year the tank last tested by this,-these
•ethoda?
YEAR LAST TESTED:
,'16-19
f26. Or what Mteriel is this tank constructed? [CIRCLE ONLY
ONE COOCl
FIBERGLASS-REINf ORCED PLASTIC . 01
STEEL 02
OTHER [SPECIFY]: 03
T27.
,20-21
Is the inside of this tank lined? [CIRCLE ONLY ONE COOE]
YES [GO ON TO T28) 1
NO [SKIP TO T30] 2
DON'T KNOW [SKIP TO TJO]. . . 8
/22
T28. In what vesr wis the lining installed?
YEAR LlNEDi
/23-26
T29. Of what aaterial is the liner constructed? [CIRCLE
ONLY ONE CODE)
EPOXY-BASEO RESINS 01
FIBERGLASS HElNfORCEO PLASTIC ... 02
ISOPHTHALIC P0LYESTER-8ASE0 RESINS. 03
POLYURETHANE-BASED RESINS 04
OTHER [SPECIFY]: OS
,'27-28
T30. Is the outside of this tank coated? [CIRCLE ONLY ONE
COOE 1
YES [CO ON TO T31) 1
NO [SKIP TO T32) 2
DON'T KNOW [SKIP TO T32]. . . 8
'29
TJ1. Of what aiatenal is the coating constructed? [CIRCLE
ONLY ONE CODE ]
FlbERilLASSCPOXY 01
ASPHALUC MATERIAL 02
UfiE THANE 03
COAL TAR EPOXY 04
UThEH [SPECIFY]: 05
.'30-31
T32. Is there secondary conteirwent for this tank? [CIRCLE
ONLY ONE COOE J
YES [UO ON TO T33] 1
NO [SKIP TO 134] 2
DON'T KNOW [SKIP TO T54] 3
32
33. is tnis secondary containment a:
concrete basin? 01
plastic-iined earth basin'' 02
clay-lined Basin' 03
loutole-wall tank' 34
or soaetmng else [SPECIFY]: _____ 05
33-34
T34. Is there secondary contairwent for any equipment that is
attached to cms tank isuch as pipes, ptaps, valves,
etc.3? [CIRCLE ONLY ONE COOE]
YES [GO ON TO 135] t
NO [SKIP TO -36] 2
DON'T KNOW [SKIP TO T36] 8
T35. Is this secondary containment a:
concrete basin' 01
plastic-lined eartn basin' 02
ciav-li-'ed basin' 03
double-wall piping? 04
ar something else [SPECIFY): 05
35
TANK
NUMBER:
T36. Xhat is the name of the coopany that installed the
tank?
INSTALLER:
T37. Is there a paved surface over the tank?
YES [GO ON TO T38] 1
NO [SKIP TO T40] . 2
T38. la this pavenent:
aspnalt? 01
concrete? 02
gravel? 03
ether [SPECIFY]: 04
139. How thick is the pavenent?
THICKNESS:
[CIRCLE ONE]:
INCHES 01
FEET 02
OTHER [SPECIFY):
03
T40. irfhat is the distance fro«t the surface to
the top of the tank?
OISTANCE TO SURFACE:
f44. Have the lines 'piping! for tnis tank ever
been found to Be leaking? [CIRCLE ONLY
ONE COOE]
yes :
NO 2
DON'T KNGto 8
,38-39
,40
[CIRCLE ONE]:
INCHES 01
FEET 02
OTHER [SPECIFY]:
03
T41. Does this tank have any of the following kinds
of protection against corrosion? [CIRCLE ONLY
ONE COOE FOR EACH ITEM]
¦ 41 -**2
,43-45
,46-47
, 48-50
51-52
faa No
a. Passive cathodic protection
fusing sacrificial anodes?
. . 1 2
53
o. Cathodic protection using
impressed current?
. . 1 2
54
c. Other [SPECIFY!:
1 2
55
•56-57
Has this tank ever Been found to
be leeking?
[CIRCLE ONLY ONE COOE]
YES [GO ON TO T43)
NO [SKIP :u tii4]
. 2
DON'T KNOW [SKIP TO T44J . .
. 8
58
How was the leak letected and,-or
verified'
[CIRCLE ALL THAT APPLY]
INVENTORY RECONCILIATION . .
. 01
59-60
lnvirowental monitoring . .
. 02
61-o2
facility inspection
03
6 3-64
TANK TESTING
. 04
65-t>6
•JTmER [SPECIFY]:
05
.67-68
69
.36-37
F-66
-------
B. OPERATING PRACTICES
YES 1
NO 2
84. Are tha inventory (atick) readings recorded in a log or journal or other permanent
record such aa a daily inventory report? [CIRCLE ONLY ONE CODE]
YES 1
NO 2
/16
B1. Do you (or another eatabliahment employee) inventory the contents of your tank(a) by
¦eaauring the depth of the contenta with a dipstick? [CIRCLE ONLY ONE CODE]
YES [GO ON TO B2] 1
NO [SKIP TO B5] 2
B2. How often do you inventory the tank contenta? [CIRCLE ONLY ONE COO£]
TWICE OAILY 01
OAILY 02
WEEKLY 03
EVERY TWO WEEKS 04 /17-18
MONTHLY 05
OTHER [SPECIFY]: 06
B3. Do you have a chart (or charts) that show how to convert the depth of the product in
the tank(a) to gallons? [CIRCLE ONLY ONE COOE]
/19
/20
/21
B5. Do any of the underground motor fuel storage tanka at thia establishment have remote
gauges (either float or electronic) that ahow the quantity of product in the tank? [CIRCLE
ONLY ONE CODE]
YES [GO ON TO B6] 1
NO [SKIP TO B8] 2
B6. How often do you (or another eatabliahment employee) inventory the contenta of your
tank(a) by reading the remote gauge(a)? [CIRCLE ONLY ONE CODE]
TWICE OAILY . Ot
DAILY 02
WEEKLY 03
EVERY TWO WEEKS 04 /22-23
MONTHLY 05
OTHER [SPECIFY]: 06
F-S7
-------
B7. Art the inventory (gauge) readings recorded in a log or journal or other permanent record
auch aa a daily inventory report? [CIRCLE ONLY (ME CODE]
/24
YES 1
NO 2
B12. Do you (or another eatabliahment employee) check the accuracy of your diapeneer meter#
to make sure the Meters correctly measure the amount pumped? [CIRCLE ONLY ONE COOE]
YES 1
NO 2
/25
/26
YES 1
NO 2
B8. Do the product dispensere for your tank(a) have metera that record the total
quantity of fuel thet haa been pumped from the tank(a)? [CIRCLE ONLY ONE COOE]
YES [GO ON TO 89].. 1
NO [SKIP TO B16] 2
89. Do you (or another eetabliahaent employee) check and record the dispenser meter
readinge for the tank(a)?
YES [GO <* TO B10] 1
NO [SKIP TO 812] 2
810. How often do you check and record the diepeneer meter readings? [CIRCLE ONLY
ONE COOE]
TWICE DAILY 01
DAILY 02
WEEKLY 03
EVERY TWO WEEKS 04 /27-28
MONTHLY 05
OTHER [SPECIFY] i 06
811. Are the dispeneer meter readinge recorded in a log or journsl or othsr permanent
record such ee e daily inventory report? [CIRCLE ONLY ONE CODE]
/29
/30
£— o 8
-------
013. Doaa anyone other than you or another establishment employee (such aa a state or
comty Waighta and Naaauraa official) check the accuracy of your diapanaar netara?
[CIRCLE ONLY ONE COOE]
YES 1 /31
NO 2
314. How often ia the accuracy of your diapenser meters checked? [CIRCLE ONLY (ME CODE]
IF THE ACCURACY OF YOUR DISPENSER METERS IS NEVER CHECKED, CHECK HERe| | /32
AND SKIP TO B16.
DAILY 01
WEEKLY 02
EVERY TWO WEEKS 03
MONTHLY 04 /JJ-J4
ANNUALLY 05
OTHER [SPECIFY]: 06
B15. About how often ia it neceasary to recalibrate (adjuat the gauge of) your
diapanaar meters? [CIRCLE ONLY ONE CODE]
DAILY 01
WEEKLY 02
EVERY TWO WEEKS 03
MONTHLY 04 /35-36
ANNUALLY 05
OTHER [SPECIFY]: 06
B16. Approximately how often do you receive deliveriea to your tank(a)?
FREQUENCY: /37-39
[CIRCLE ONE]:
PER WEEK 01
PER MONTH 02 /40-41
OTHER [SPECIFY]: 03
B17. Are inventory (atick or remote gauge) readinga of your tank(s) taken inwediately before
receiving a fuel delivery? [CIRCLE ONLY ONE CCDE]
YES 1
NO 2
/42
-------
818. Are inventory (atick or remote gauge) readings of your tank(a) taken Immediately after
receiving a fuel delivery? [CIRCLE ONLY ONE CODE]
YES I A3
NO 2
819. Is the quantity delivered to each tank recorded in a log or journal or other
permanent record such as a daily inventory report? [CIRCLE ONLY ONE COOE]
I /44
NO 2
/45
B20. Do you reconcile your inventory (stick or remote gauge) readinga with your book inventory
(meter readinga and deliveriea)?
YES [GO ON TO 821] 1
NO [SKIP TO 822] 2
821. How often do you reconcile your tank inventory (stick or remote gauge) readings with your
book inventory (meter readinga and deliveries)? [CIRCLE ONLY ONE COOE]
DAILY 01
WEEKLY 02
EVERY TWO WEEKS 03
MONTHLY 04 /4*-47
ANNUALLY 05
OTHER [SPECIFY] 06
B22. Do you ever use water-finding paste to check the water level in the bottom of your tank(a)?
[CIRCLE (NLY ONE C00€]
YES [GO ON TO 823] 1
NO [SKIP TO C1] 2
823. How often do you use water-finding paste to check the water level in the bottom of
your tank(a)? [ENTER FREQUENCY AND CIRCLE UNIT COOE]
FREQUENCY: /49-51
[CIRCLE ONE]:
PER DAY 01
PER WEEK . \ . . ' 02
PER MONTH 03 /52-53
PER YEAR 04
OTHER [SPECIFY]: 05
/ U
-------
C. OPERATING HISTORY
C1. Have any tanks at this establishment ever been replaced? [CIRCLE ONLY (ME CODE]
YES [GO ON TO C2J 1
NO [SKIP TO C4] 2
DON'T KNOW [SKIP TO C4] 8
C2. How many tanks have been replaced?
C3. Pleaae answer the following questions about each tank that haa been replaced, beginning
with the tank that was replaced most recently. [SPACE HAS BEEN PROVIDED TOR UP TO TOUR
TANKS. IF MORE THAN TOUR TANKS HAVE BEEN REPLACED, WRITE THE ANSWERS FOR THE AOOITIONAL
TANKS ON A PLAIN SHEET Of PAPER.]
~H
/16
NUMBER REPLACED* /17-19
Firat Tank
Second Tank
Third Tank
Fourth Tank
C3a. In what year was the
(first/second/third)
tank replaced?
-
(year)
/20-23
(year)
/36-39
(year)
/52-55
(year)
/68-71
C3b. Why was the tank
replaced?
[CIRCLE ALL THAT
APPLY FOR EACH TANK]
a. Because it
was leaking?. . . .
01
01
01
01
b. Becauae other tanks
were being replaced
at that time? . . .
02
02
02
02
c. Because it was no
longer needed/in
use?
03
03
03
03
d. To increase storage
capacity
04
04
04
04
e. Or Tor some other
reason? [SPECIFY]: .
05
05
05
05
(speci fy)
Z24-35
(specify)
/4Q-51
(specify)
/56-67
(specify)
/72-B3
F- / i
-------
1
C4. Have any tanka at this establishment ever been removed without being replaced? [CIRCLE
ONLY ONE COOE]
YES [GO ON TO C5] 1
NO [SKIP TO C7] 2
DON'T KNOW [SKIP TO C7] 8
(Hi
/16
C5. How many tanka have been removed without being replaced?
NUMBER REMOVED: /17-19
C6. Please answer the following questions about each tank that hss been removed without
being replaced. [SPACE HAS BEEN PROVIDED FOR UP TO FOUR TANKS. IF MORE THAN FOUR
TANKS HAVE BEEN REMOVED WITHOUT BEING REPLACED, WRITE THE ANSWERS FOR THE ADOITIONAL
TANKS ON A PLAIN SHEET OF PAPER]
First Tank
Second Tank
Third Tank
Fourth Tank
C6a. In what year was the
(first/second/third)
tank removed?
(year)
/20-23
(year)
/34-37
(year)
/48-51
(year)
/62-65
C6b. Why was the tank
removed?
[CIRCLE ALL THAT
APPLY FOR EACH TANK]
/
a. Because it
was leaking?. . . .
01
01
01
01
b. Because other tanks
were being removed
at that time? . . .
02
02
02
02
c. Because it was no
longer needed/in
use?
03
03
03
03
d. Or for some other
reason [SPECIFY]: .
04
04
04
04
(specify)
/24-33
(apecify)
/38-47
(apecify)
/52-61
(specify)
/66-74
C7. Have any tanks at thia establishment been abandoned in place? ("Abandoned in place" means
that the tank is no longer in use but has not been removed.) [CIRCLE ONLY ONE COOE]
YES [GO ON TO C8] 1
NO [SKIP TO D1] 2
DON'T KNOW [SKIP TO D1] 8
L2II
/16
r — 1Z
-------
C8. How many tanks have been abandoned?
NUMBER ABANDONED:
C9. Please answer the following questions about each tank that haa been abandoned in place.
[SPACE HAS BEEN PROVIDEO FOR UP TO FOUR TANKS. IF MORE THAN FOUR TANKS HAVE BEEN
ABANDONED IN PLACE, WRITE THE ANSWERS FOR THE ADDITIONAL TANKS ON A PLAIN SHEET OF
PAPER]
First Tank
Second Tank
Third Tank
Fourth Tank
C9a. In what year was the
(firat/aecond/third)
tank abandoned?
(year)
/20-23
(year)
/44-47
(year)
/68-71
(year)
/92-95
C9b. Why was the tank
abandoned?
[CIRCLE ALL THAT
APPLY FOR EACH TANK]
a. Because it
was leaking ....
01
01
01
01
b. Because it waa no
longer needed/in
use
02
02
02
02
c. Or for some other
reason [SPECIFY]t .
03
03
03
03
(specify)
/24-31
(specify)
/48-5S
(specify)
/72-79
(specify)
/96-103
C9c. How was the tank
abandoned?
[DESCRIBE PROCEDURE, OR
CIRCLE ALL THAT APPLY]
a. Tank was drained. .
01
01
01
01
b. Tank was washed . .
02
02
02
02
c. Tank was cut open .
03
03
03
03
d. Tank was sand
filled
04
04
04
04
e. Tank was cement
filled
05
05
05
05
f. Other [SPECIFY]:
(specify)
/32-43
(specify)
/56-67
(specify)
/80-91
(specify)
/104-115
-------
D. PERMITS AND LICENSES
CH
D1. Were you required to obtain a special building permit or license in order to have your
tank(s) installed? [CIRCLE ONLY ONE CODE]
YES. . . .
NO ... .
DON'T KNOW
02. Are you required to maintain a special permit or license to store flammable or hazardous
material at your establishment? (Often these permits are called Hazardous Use or Hazardous
.Materials permits, and are issued by the state, county, or local fire marshal.) [CIRCLE
ONLY ONE CODE]
YES. . . .
NO . . . .
DON'T KNOW
1
2
8
/16
1
2
B
/17
r'~ / 4
-------
E. INSTALLATION
E1. What type of fill waa u«ad to backfill around and over the tank(a)? [CIRCLE ONLY ONE CODE]
a. Clean sand (with no large rock)?. ... 01
b. Pearock or pea gravel? 02
c. Soil fraa the excavation? 03 /18-19
d. Or bom other kind of fill [SPECIFY] i . 04
E2. (Ia the tank/are any of the tanks) installed with the bottos reating on or in a concrete
or packed earth pad? [CIRCLE ONE CODE FOR EACH ITEM]
Yea No
a. A concrete pad or cradle? .... 1 2 /20
b. A packed earth pad? 1 2 /21
E3. Are any of the tanka strapped to a concrete pad? [CIRCLE ONLY ONE CODE]
YES 1
NO 2
DON'T KNOW 8
/22
E4. What ia the shortest distance between any of. your tanks and any neighboring uidargroind
tank or other eolid underground atructure (auch as s baaaaent wall, sewer, or utility
vault)? [ENTER DISTANCE AND CIRCLE UNIT CODE]
SHORTEST OISTANCE FROM
UNDERGROUND STRUCTURES /23-28
[CIRCLE ONE]:
INCHES
FEET
OTHER [SPECIFY]t
01
02 /29-30
03
£ f D
-------
F. PROTECTION
F1. Has any type of special equipment or Materials been installed to prevent external
corrosion of the tank(a)? [CIRCLE ONLY ONE CODE]
F2. How often do you inspect your external corrosion protection systsart [ENTER
FREQUENCY AND CIRCLE UNIT CODE]
~EI
YES [SPECIFY AND GO ON TO F2]: /16
1
NO [SKIP TO F3] 2 /17-18
DON'T KNOW [SKIP TO F3]. 8
IF YOU NEVER INSPECT THE EXTERNAL CORROSION PROTECTION SYSTEM, CHECK HEReQ /19
AND SKIP TO F3.
FREQUENCY OF INSPECTION! /20-22
[CIRCLE ONE]i
PER DAY 01
PER WEEK 02
PER MONTH 03 /23-2A
PER YEAR 04
OTHER [SPECIFY]: 05
F3. Since you began using ths tank(a), have you ever had the tank(e) completely drained and
cleaned out? [CIRCLE ONLY ONE CODE]
YES 1
NO 2
/25
F4. Does the tank ayatea have a continuous electronic «onitoring system to detect tank leakage?
[CIRCLE ONLY ONE CODE]
YES [GO ON TO F5] 1
NO [SKIP TO F6] . 2
/26
F-7S
-------
F5. How often ia the electronic monitoring syate® inspected for maintenance? [CIRCLE ONLY (WE
CODE]
a. Annually? 01
b. Twice a year? 02
c. Three or four tinea a year? 03 /27-2S
d. Or at aone other interval? [SPECIFY]:
04
F6. Have preaaure piping (or line) leak detectora been inatalled at thia establiahment to
detect leaks in the piping (lines)? [CIRCLE ONLY ONE CODE]
YES [GO ON TO F7] 1
NO [SKIP TO G1] 2 /29
DON'T KNOW [SKIP TO G1] 8
F7. How frequently are the preaaure piping leak detectora tested to make eure they are operating
correctly?
IF THE PRESSURE PIPING LEAK DETECTORS ARE NEVER TESTED, CHECK HERE | ' |
ANO SKIP TO QUESTION G1.
/30
FREQUENCY: /31-33
[CIRCLE ONE]:
PER DAY 01
PER WEEK 02
PER MONTH 03 /34-35
PER YEAR 04
OTHER [SPECIFY]: 05
F8. Have the pressure piping leak detectora ever given false leak signala? [CIRCLE ONLY
ONE CODE]
YES 1
NO 2 /36
DON'T KNOW 8
F9. Have the pressure piping leak detectors ever dstected actual leaks in the piping syetera?
[CIRCLE ONLY ONE CODE]
YES. . . .
NO ... .
DON'T KNOW
1
2
8
/37
-------
G. INFORMATION NEEDS
G1. Have any of the companies from trfiom you receive your fuel products asked you to keep inven-
tory records (dipstick readings, meter readings and delivery records) for your tank(s)?
[CIRCLE ONLY ONE COOE]
YES.
NO .
Qo)
/16
G2. Has anyone ever given you training or explanatory literature about any of the following
topics? [CIRCLE ONE CODE FOR EACH ITEM. IF YOU HAVE RECEIVED INFORMATION OR TRAINING,
PLEASE IM3ICATE FROM WHOM]
Type of Training
Did you
receive?
If "Yes," from whom?
a. Keeping inventory records.
b. Doing inventory reconciliation
calculations
c. Measuring the quantity of product
in a tank using a dipstick and
conversion chart
d. Checking pump meter accuracy.
e. Line leak detection and testing.
f. Tank or line leak prevention.
g. Tank tightness testing methods.
h. Leak monitoring methods (such as
observation wells)
YES NO
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
•c_7q
-------
GJ. If you found out that (your tank/one of your tanka) waa leaking, Mould you probablyi
[CIRCLE ONLY ONE COOE]
a. Rap lacs it with another tank 01
b. Una it and continue to uaa it 02 /41-42
c. Abandon it in place 03
d. Or aoMthing alaa [SPECIFY]« 0*
G4. Hom Much do you expect it Mould coat you tot
a. Replace a tank? $ /0-48
b. Una a t**?. I /49-34
c. Abandon a tank in placa? .... S /55-A0
G5. Do you have an inaurance policy that covara you againet da«age to paopla or property
cauaed by auddan apille of rntor fuel? [CIRCLE ONLY ONE CODE]
G6. Do you have en inaurance policy that covara you againet daaage to people or property
raeulting fro* non-eudden epille (including leaka) of Motor fuel? [CIRCLE ONLY ONE COOC]
Is" I m
NO 2
F-79
-------
TANK TO DISPEMSER METES FUEL LINE CONNECTIONS
Instructions; Mark (X) in each block for which there
is fuel line (pipe) connection from the tank to the
dispenser meter, (if more tanks than spaces, use
additional sheets.)
Disp.
Meter
Number
t Tank Number ana Product
1 T-l I T-2 1 T-3 1 T-4 | T-5 | T-6 I T-7 ! x-5
1 ! ! 1 !
M-l
M-2
|
M-3
M-4
M-5
|
M-6
j
M-7
M-3
|
|
M-9
|
j
M-l-
! !
M-ll
j
M-l 2
I
M-l 3
M-l 4
M-l 5
M-16
M-17
|
M-l 8
|
i
M-l 9
i
i
M-20
1
I
-
Does the facility have a leak sanitating system (foe tanks oc piping) chat is noe
electronic (such as ocsecvacion weils)?
YZS 1
MO 2
If YES, describe
F-80
-------
Site Observations Recording Sheet
Site Code Label
Date
Tank 1
Tank 2
Tank 3
Tank 4
Tank 5
Tank 6
Size of fill pipe
(I.D.)
Drop Tube
(permanent or
removable)
Site Code Label
Date
Tank 1
Tank 2
Tank 3
Tank 4
Tank 5
Tank 6
Size of fill pipe
(I.D.)
*. *
Drop Tube
(permanent or
removable)
F-81
-------
Intentionally Blank Page
-------
OMB (to. i 2070-OOJ7
Enpireei Oecertter >1, 196}
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
UNDERGROUND STORAGE TANK SURVEY
L.
_l
HAILING ACOBESSi Vilified? ... |_|
LABEL VERIFICATION
LOCATION ADORESSi Verified? ... Q
(ESTABLISHMENT NAME)
(ESTABLISWCNT NAME)
(ACORCSS)
(ADDRESS)
(CITY/STATE/ZIP)
CONTACT WWC AM) PHONE I Verified? ... Q
Contact Neaei
(CITY/STATE/ZIP)
Contact Phonal
A. Queationnaira Statuai
B. Inventory Statue (CIRCLE ONE)
1 > Started
2 * Hot Started
J > Obtained
A • Refuead
5 * Other (SPECIFY)
C. Can Taet (CIRCLE ONE)
1 > No Met ere
2 • Complete
J e Pertiel Coaplete
A e Refueed
J > Other (SPECIFY)
0. Mapping (CIRCLE ONE)
1 > Coaplete
2 « Other (SPECIFT)
E. Debriefing (CIRCLE ONE)
1 * Coaplete
2 • Other (SPECIFY)
F. Confidentiality
1 = Fora Encloeed
2 s Waived
} « Other (SPECIFY)
Conducted byi
WESTAT
An Employee-Owned search Corporation
MO M8BO • 3O1991-10OQ
F-83
-------
[IN YOUR TELEPHONE CALL TO SET UP THE APPOINTMENT, ASK R IF HE/SHE HAS COMPLETED THE QUESTIONNAIRE FORM AND BEGUN THE
INVENTORY RECORDING. IF NOT, ENCOURAGE THE R TO 00 SO.7
INTRODUCTION:
1. Hello, ay nM is (YOUR NAME), from Westst. [SHOW IDENTIFICATION CARD], I'm hers to conduct the interview with
you about your underground etorege tenk(s). The other member of my team ia from
Mldweet Research Institute. (He/She) will be drawing a nap of the tank area(a) and taking some pictures of the
surface «res(s) over the tank(s).
[IF YOU HAVE ANY OBSERVERS ON THIS INTERVIEW, INTRODUCE YOUR OBSERVERS. OTHERWISE, ASK IF THERE IS A PLACE WHERE
YOU AND THE R£SPO»OENT CAN SIT DOWN AND GO THROUGH THE QUESTIONNAIRE. A SIDE-BY-SIDE SEATING ARRANGEMENT IS
PREFERABLE, SINCE THIS ALLOWS YOU TO READ FROM THE RESPOfcSENT'S WORKING COPY Of THE QUESTIONNAIRE. IF THE R DOES
MOT HAVE HIS SURVEY MATERIALS IN SIGHT, SUGGEST THAT HE OBTAIN THEM — THAT THE INTERVIEW WILL BE MORE EFFICIENT,
AND THAT YOU WILL HAVE TO RECORD CERTAIN INFORMATION ON HIS INVENTORY FORMS LATER ON.J
CONFIDENTIALITY:
2. As was mentioned in the letter and in the General Instructions, you can (clsim/sak for) confidentiality for all
or psrt of your responses to the questionnaire. The way you do this is by filling out the form that is in the
General Instructions booklet. Have you decided to cleia confidentiality for any of your answers?
YES 1 [OBTAIN COMPLETED CONFIDENTIALITY FORM FROM
RESPONDENT. PUT ID SUCKER ON TOP OF FORM.]
NO 2
[BEGIN INTERVIEW, READING QUESTIONS, ITEM-BY-ITEM. READ THROUGH THE TANK DESCRIPTION SHEET ONLY FOR THE FIRST TANK.
FOR THE SECOND AND FOLLOWING TANKS, READ THE QUESTION NUWERS AW>/OR ABBREVIATED QUESTIONS.]
INVENTORY RECORDS:
3. Next, we need to review your inventory records. The General Instructions booklet discussed keeping inventory
records for stch of your tanks* Hsve you etsrted to keep these records yet?
YES 1 (GO TO 6)
NO 2 (GO TO 5)
A. (IF YES): That's great! Here la a postage paid envelope in which to send the completed inventories to Westst.
A Westst interviewer will be celling you in s few weeks to check with you on sny problems you night be having
with the Inventories. Mhile I an here, I need to review your inventory sheets and initial them. [IF NECESSARY,
PROBE: If I could take a look at then now, I would appreciate it.] [IF THE IANK(S) DISPENSER(S) ARE METERED]:
2 will alao need to record the results of my Metering can tests of the dispenser Meters on the Inventory sheets
for eech tank. (GO TO 8)
5. (IF NO): Is there any reason why you have not etsrted the inventories?
YES 1 SPECIFY:
(GO TO 6)
2 (GO TO 6)
6. Will you be able to stsrt ths inventories today?
YES 1 (GO TO 7)
NO 2 [PROBE FOR WHEN THEY WILL BEGIN:
] (CO TO 7)
REFUSED.... 7 (TERMINATE)
IMPOSSIBLE. 9 (SPECIFY WHY:
] (TERMINATE)
7. Hers is a postage psid envelope in which to send the completed inventories to Westst.
§, tfcon we receive your inventory(ies) they will be computerized end run through a computer progrsm that checks for
gains end losses thst can't be accounted for, such ss over- snd under-deliverles, theft or pilfering, or leakage.
We will let you know the rseults of thst computer analysis.
P 4
-------
DISPENSER >CTCR ACCURACY CHECKS:
If* have found In the past that ¦ major problem in analyzing inventory records ia that soms dispsnser meter readings
are juat elightly inaccurate. Often theae meter errors rfiow up in the computer analysis ae a»ali leaks. For that
reeeon, we era checking out the accuracy of all dlapenser mstere, ueing a 5-gallon Meter testing can.
Our accuracy checking procedure ia the aa*e procedure that is used by the agenciaa that certify aster accuracy. We
will not ba adjusting your natera if we find that they are miareading. Mhst we will do is record the amount of pro-
duct pumped intgo the can according to the diapenser meter, end record the saouit in the metering can eccording to
the gsuga ot the can. I'll need to record thia on your inventory sheeta as well ss my copy of the questionnsira.
The informstion will be fed into the computer program to correct for metering error in the results.
We will be piping five gsllorts of product into the teat csn from each diapenser that has ita own meter. We will
then be pouring the five gallona of product back into the tank from itiich it was puaped. IF £ HAS BEGUN INVENTORY!
I will need to rscord the returned product as s "delivery" to the tank on your inventory sheet. FOR THE FIRST
MEASUREMENT! First, 1 need to wet the inside of the csn with about s gallon of product, and pour it back into its
tank.
MAKE SURE YOU MILL BE ABLE TQ RETURN THE PROOUCT TO THE TANK BEFORE YOU BEGIN PUMPING. DO ALL THE METERS FOR A
TANK BEFORE MOVING ON TO THE *XT PRODUCT TYPE. AFTER ALL OF THE TANKS ARE DONE, WASH THE CAN OUT WITH DETERGENT
AM) WATER, AND DRY IT AS COMPLETELY AS POSSIBLE.
DEBRIEFINGt
To be completed immediately after leeving the site.
D1. Did R have the gueationneire completed?
YES 1
NO 2
D2. Did R hava the inventory sheets stsrtsd?
YES 1 (GO TO D3)
NO 2 (GO TO D4)
03. Did R hsvs errore or problems in the completed psrts of the inventory?
YES........ 1 (DESCRIBEi )
NO 2
D4. Oid R understand inventory proceas?
YES. 1
NO 2
D5. Did H understand moat/ell of the questions in the questionnsirs?
YES 1
NO 2
06. Was Ri YES NO
s. cooperative? 1 2
b. hostile? 1 2
c. gueesing s lot? 1 2
d. Other (SPECIFY)
1 2
D7. Was it necaaaery to talk to more than one R to obtain ail required information?
YES 1
NO 2
06. Comments
-------
TIME BEGAN: A.M.
P.M.
TINE EHXDi A.M.
P.M.
RECORD OF CALLS
ATTEMPT
NUMBER
DAY
TIME
RESULT
CODE
COMMENTS/PROBLEMS
P«
8 1
P*
PO
S I
p«
p«
p*
i 1
m
tm
p»
11
1 s
m
pa
m
PA
an
P"
?1
an
P"
pa
s s
RESULT CQOC5
PRELIMINARY RESULT COPfS FINAL - MAIL TO SUPERVISOR
1
APPOINTMENT
11
COMPLETE
2
RESPONDENT NOT AVAILABLE
12
PARTIAL COMPLETE
3
RESPONDENT NOT LOCATED
13
ESTABLISHMENT CANNOT BE LOCATED
4
RESPONDENT ILL
14
RESPONDENT UNAVAILABLE
5
REFUSAL/BREAKOrr
15
REFUSAL/BREAKOFF
6
RESPONDENT BROKE APPOINttCNT
16
RESPONDENT AVOIDING INTERVIEW
7
LANGUAGE PROBLEM
17
LANGUAGE PROBLEMt NO INTERPRETER
8
OTHER
18
NO TAMCS
9
0UT-0F-SC0PC
19
OUT OF PSU (MOVED)
2D
OUT OF BUSINESS (CLOSED)
99
OTHER (SPECIFY)
• OO
-------
INSTRUCTIONS
PREPARATIONS FOR TANK TESTING
1. If you are not responsible for making the following testing arrange-
ments, please notify those who are as soon as possible. Please notify
other persons who may be involved, including the tank owner and those
at your firm's or regional offices.
2. Immediately contact your fuel supplier or distributor to make
arrangements for filling your tanks. Explain any tank filling
problems to the test coordinator from Midwest Research Institute (MRI)
when he calls.
3. Fill any business vehicles before the fuel drop off. As necessary,
make arrangements for alternate sources of fuel for those vehicles on
the test day.
4. Fuel delivery must be finished before 8:00 a.m. of the test day.
If the test crew has to wait for fuel drop off, it means that testing
will not be finished until later that evening.
5. i iWMn»l«n f TTn eeiili || Mill until the fuel level comes up into neck of
the fill pipe. Use your tank dipsticks to determine when the tanks
are "full": the fuel depth, as measured by the dipstick, should equal
the tank diameter. (In many tanks, you can see when the fuel reaches
the fill pipe neck. However, for tanks with drop tubes, you must use
the dipstick to know when it is full.) Testing cannot be done if the
tanks are not completely full.
6. Once filled, the tanks cannot be used until testing is complete.
Make arrangements to keep the tanks out of service. Your business
does not need to be closed during this time, but the tanks must remain
inactive.
~
~
~
~
~
~
F-S7
FINAL CHECKLIST
Notify responsible individuals.
~ Owner
~ Main or regional office
~ Others
Contact supplier or distributor
Fill business vehicles before filling tanks
Fill tanks before 8:00 a.m. on test day
Arrange to keep tanks out of service
-------
ENVIRONMENTAL CONDITIONS OATA SHEET
Site Code Label
Date Tank No.
Test Firm
Test Team
Time
Temperature °F
8arometric
Pressure
Surface
Ambient
Subsurface
Comments
1
Climatic conditions.
F-88
-------
TEMPERATURE PROFILE DATA
Site Code Label
Test Team
Test Crew
STarT
time
0 U D
Date __
Tank No.
U D U D
£nd~
time
Figure 3
F-89
-------
Site Diagram and Detail Diagram Sheet
Site Code Label
Map ' Test Firm
Test Team , Date
Sketch Ar*a and Dimensions
F-90
-------
PICTURE DESCRIPTION
Site Code Label
Team
picture No.
1
Date
Description
2
3
4
5
6
7
8
9
10
11
12
13
14
15
F-91
-------
Site Code Label
Critical Features Data Sheet
Teaa
Tank
No.
Type of
Product
Date
Number of
Dispensors
According to
Product
Size Size of Size of Size of Size of Orop Tube
of Fill Gauge Stick Vent. Permanent,
Tank Pipe Pipe Pipe Pipe Removable
Delivery
Systen
Pressure,
Suction
Pimp Pit
If
Present
Depth
of Tank
F row Grade
Surface
Over Tank
Electrical
Power Outlets
Powprlines
Overhead
Waterways
I
u.
tv
T
-------
Site Code Label
Team
Oate
EDIT CHECKLIST
o Site code label on all pages.
o Be sure all maps are numbered sequentially.
~ Photographs of critical parameters.
~ Site code labels on photographs and filed in notebook.
~ Check to see that all data sheets are filled our correctly.
-------
SIMULATED LEAK TEST
VOLUME DATA
Site Code Label
Date
Tank No.
Test Team
Test Firm
Volume, g
T ime
Bottle No.
Final Wt.
Tare Wt.
Total Wt.
Specific Gravity:
Temperature:
L ~ J
-------
SIMULATED LEAK DATA FORM
LEAKING UNDERGROUND STORAGE TANK
Site Code Label
Test Crew Tank No.
Test Team
Rotameter Start End Elapsed Measured
Rotameter Setting Nominal Rate Time Time Time Volume Calculated
No. mm gph Clock Hour gms gph Rate
F-95
-------
H.
II. TANK TO TIST
11. CAPACITY
i* Vtar* m iwfw>iiwt o»«
L"J
U o*.—
17. Ml Uf FOR TEST
He* wmm im—
Oaflona
¦ wp.rr«C« MCOM AMD A/TIJt CACH COMFAATWINT DROP Oft I ACM M(T(RIO OtUVt*V QUANTITY
In Ml hw* tup* « pM
11 SKOAL COMMDONS AND MQCfOURES TO TEST THIS TANK
Cm mm4 mcNm CM Mb«* md pmnIw* i» tof (*>.
~ N» hiiv* Q M* «Mr tt* to tor* iin.-un ~ Un*«J Mi( ft
VAPOR RECOVERY SYSTEM
II. TANK MEASUREMENTS fOR
TSn ASSEMBLY
¦ •(fell* Mid
MorivrL
A44M'tor T t «r •* Mai
M EXTENSION HOSE SETTING
r«*ta»io«nM*
21. TEMPERATURE/VOIUME FACTOR (a) TO TEST TWS TANK
H Mmwf | j CaUwtfj •» PndM«T
-------
Data Chart for Tank System Tightness Test
petro Tite
please print
TANK TESTER
1.
OWNER Property [ j
Tenk(i) ~
N»m«
Address
Representative
Telephone
Nime
Addies*
Representative
Teiephone
2.
OPERATOR
Name
Address
Telephone
3.
REASON FOR
TEST
(Explain Fully)
4.
WHO REQUESTED
TEST ANO WHEN
Nlmt
Title
Company or Affiliation
Date
Address
Telephone
5.
WHO IS PAYING
FOR THIS TEST?
Company. Agency ot Individual
Pmon Authorizing
Title
Telephone
Billing Addiess
City
Steta
I*
Attention ot:
Older No.
Othar Instructions
6.
TANK(S) INVOLVED
Identify by Oirection
Capacity
Brand/Supplier
Grade
Approx. Age
Steel/Fiberglass
7.
INSTALLATION
DATA
Location
Cover
Fills
Vents
Siphones
Pumps
North Inside diivawey.
Rear ol station, ate.
Conciete, Black Top.
Earth. etc
Sue. Titefill make. Oiop
tubas. Remote Fills
Size. Manifolded
Which tanks >
Suction. Remote.
Make if known
8. UNDERGROUND
WATER
Depth to the Water table .
It the water over the tank ?
~ ~ N°
9. FILL UP
ARRANGEMENTS
Tanks to be filled .
. Date Arranged by .
Telephone
Extra product to "top off" and run TSTT. How and who to provide ? Consider NO Lead.
Terminal or other contact
for notice or inquiry
Company
Telephone
10. CONTRACTOR.
MECHANICS.
any other contractor
involved
11. OTHER
INFORMATION
OR REMARKS
Additional information on eny items above. Officials or others to be advised when testing is in progress or completed. Visitors or observers present
during test etc.
12. TEST RESULTS
Teat a war* mad* on tha above lank ayatama In accordance with laal procaduraa prescribed for BftfOTrtf
aa detailed on attached taat chart* with reaulta aa lot!owe: Torwr
Tank Identification
Tight
Leakage Indiceted
Date Tested
-
13. CERTIFICATION
Serial No of Thermal
This It to certify that thaaa tank ayatama wara taatad on tha data(a) shown. Thoaa Indlcattd aa "Tight" moat tt>a eritarla eatabtishad by
tha National Flra Protection Aaodation Pamphfat 329.
Testing Contractor or Company. By: Signature
Technicians
TT-q 7
-------
LIQUID VOLUMETRIC LINE LEAK TEST
DATE OF TEST_
JOB #
1 LQCATtO*:.
i mmt*-.
* OftMTOW:.
« IUKM m n»? _
• T1$T mOUHTtO »T:
« tftcuu. IMTKUCTIOMa:
j cowtwackj* on covrtm mabmo rtn
MtCNAMC(i)
• It A K M TANK Till TQ M
mam wmt itm uni rtrri
D #0
nmMMkTmt m taucs
• MAKI AND Ttft Of
PUMP OR CHfHMIftt
covin
©vi* unit _
11 IOIWHFT
(ACN liM
as tnno
» n*i
(MIUTANV) '
n toe op nw nootsuM*.
AMCtCWT TIMPtftATUftt.
WWTW. fit.
«« CUJWIU
ft VOtVMl
Ml OK tP»
•UMCTTI HCAOINO
NIT
UMM
ATM*
•fPOM
AFTI*
CMANQI
-
CMMIMIlfti
APP*0XI»A7<
_ oirr* _
1« TtSf MSULtS
COftCLUttONl WMU AMP COMDHNTI
©Cof»yii9^t
10"*^ K»rtf MaO»P OVCKV IfiO^l
-------
11 lOiNTIFY
IACH UNI
At IWTCD
11 T1MI
(MfUTART)
U 100 OF TIST RROCIOURfS.
AMBItNT TIMMRATURL
WtAThlR. STC.
14 FRCSSURl
•URITTI
11 VOLUME
ptl OR k*»
RIAOING
NIT
CHANOC
MFORC
AfTtR
IIFORE
AFTER
i> list risulti
CONCLUSIONS. RIRAIftft AND COMMINTS
SCAll: Q 1 RACC-1 FT. HA SOOAM-THrS SHUT - 1«T l 1U'
[~) 2 PACK - • FT. HR SOUARI-ThtS •HlfT • JJT I U»
17 SKCTCH Of 10CAT>0N
SNOW NORTH t. STRUTS. STATION RUILDING. TANKS. ISLANDS. HHNO (»F KNOWN.
OR REST INFO). RUMRS OR OtSRINSIRS (USC NUMRCRS ONLY IF MRMAHINUf MARKED).
Form No. U7S-14S (J.2M1MI}
1601 SOUTH JACKSON STfWT »4AC*80N. MOWN 4MM
TOOL DIVISION T-tac+trm 517/784-6961
F-99
-------
APPENDIX G
NATIONAL UNDERGROUND STORAGE TANK SURVEY
NATIONAL SAMPLE OF FARMS
I. INTRODUCTION AND SUMMARY
The survey of underground motor fuel storage tanks is
designed to provide national estimates of the number of
underground motor fuel storage tanks at the end use point and tt
number and percent of these tanks which leak. The survey desigr
defined three segments of the overall target universe of
establishments with underground motor fuel storage tanks:
o Fuel establishments (gas stations and establishments i
other fuel-related or fuel-using industries) which by
the nature of their business are likely to have such
tanks;
o Large establishments (20 or more employees) which by
virtue of their size may have an underground motor fu€
storage tank; and
o Farms, of which over half have motor fuel storage
capacity, but an unknown proportion store motor fuel
underground.
The sample design for the survey is a two-stage cluster
design. The first stage is survey locations, called Primary
Sampling Units (PSUs) and consisting of counties or groups of
counties. The contiguous United States was divided into six
survey regions, based on rough similarity of soil type and
condition, as defined in Figure G-l. Thirty-four PSUs were
drawn, six from each region, except four PSUs were drawn from
Region 5.
-------
Figure G-l. Six regions for National Survey of Underground
Storage Tanks
Northeast
3
— Midwest
Maine
Wisconsin
New Hampshire
Minnesota
Vermont
Iowa
Connecticut
Missouri
Massachusetts
Illinois
Rhode island
Indiana
New York
, Ohio
New Jersey
Michigan
Pensylvania
Maryland
4
— Central
Delaware
Virginia
North Dakota
West Virginia
South Dakota
Washington, D. C.
Nebraska
Kansas
Southeast
Oklahoma
Texas
Kentucky
Tennessee
5
— Mountain
Arkansas
Louisiana
Montana
Mississippi
Wyoming
Alabama
Idaho
Georgia
Nevada
North Carolina
Utah
South Carolina
Colorado
Florida
Arizona
New Mexico
6 — Pacific
Washington
Oregon
California
-------
Among the three survey segments, fuel establishments and
large establishments are both concentrated in the same areas,
where the population is. Drawing a sample of FSUs which is
optimal for both of these segments is therefore no problem,
because they occur together. Farms, however, tend to be found in
the opposite places, those with sparse population. So optimizing
the design for farms is in direct opposition to optimizing the
design for fuel establishments and large establishments. Since
the fuel establishments are the major focus of the survey,
accounting for about 800 of the approximately 920 expected
establishments with underground motor fuel storage tanks, the
sample of PSUs was optimized for fuel establishments by being
drawn in proportion to the number of fuel establishments in each
PSU. As noted above,' the resulting sample of PSUs is not optimal
for studying farms.
The second stage of sampling is the sample of establishments
within the selected PSUs. Three sample frames (master lists)
were developed for the 34 sampled PSUs — one for fuel
establishments, one for large establishments, and one for farms.
Samples were drawn from each list:
o 1618 fuel establishments;
o 600 large establishments; and
o 600 farms.
These establishments were contacted to determine whether they
were eligible for our survey; that is, whether they had
Nj *
-------
underground motor fuel storage tanks. The eligibility rates were
(approximately):
o 50 percent for fuel establishments;
o 15 percent for large establishment; and
o Less than 5 percent for farms.
This appendix discusses the national farm sample of 600
farms to be screened. Subsection II discusses the target
universe of farms and describes the farm sampling frame on a
national basis. The 1982 Census of Agriculture conducted by the
Census Bureau is taken as the standard count of farms, and a list
developed by the Agricultural Stabilization and Conservation
Service (ASCS) of the U.S. Department of Agriculture (USDA) is
the sample frame used. For the nation, overall, this frame
offers good coverage of the farm universe. Subsection III
reviews the survey design with reference to the farm sample and
compares Census figures with ASCS figures for the selected PSUs.
In this subsection, it is seen that the coverage of farms by the
frame is weak in some parts of the country. Section IV concludes
the appendix with a discussion of the ratio-adjustment weighting
method proposed to minimize total sampling error in the farm
estimates.
II. TARGET UNIVERSE OF FARMS AND SAMPLING FRAME
A. Two Farm Data Sources
Two sources of information on farms were used in designing
and conducting this survey. One is the 1982 Census of
Agriculture (the most recent) conducted by the Bureau of the
G-4
-------
Census. This source is used as the most reliable source of
national statistics about farms. The second is the "1983
Deficiency Master File"•developed by the Agricultural
Stabilization and Conservation Service (ASCS) of the U.S.
Department of Agriculture (USDA), which is used as the list, or
sampling frame, for farms.
The Census of Agriculture is a data collection and
tabulation effort which is as inclusive as possible. The 1982
Census lists 2,240,976 farms in the U.S. A farm is defined by
Census as "any place from which $1000 or more of agricultural
products were sold or normally would have been sold during the
Census year." Tables provide breakdowns of these farms by size
of farm, value of sales, type of crop, etc., both nationally, by
state and by county. Some of these figures are reviewed later in
this section.
What the Census of Agriculture does not provide is a list of
farms or farm operators in specific places. Thus, for an actual
sampling frame we used the USDA/ASCS 1983 Deficiency File. This
is a list of farms developed by the USDA containing about
1,942,000 listings (87 percent as many as the Census total). The
original impetus for the development of the file was to provide a
mechanism for payment distribution for the PIK (Payment-in-Kind)
program for 1983. In 1983, the PIK program was so popular that
USDA believes that almost everyone engaged in growing PIK program
crops (which include various cash grains and upland cotton)
applied for it, and hence is listed on the.Deficiency File.
Because they saw a chance to have a near-Census of farms on a
data file, USDA made a special effort to also include listings of
farms not eligible for the PIK program. The basic data were
gathered by the ASCS county agents.
o "~
-------
The official USDA/ASCS statistics indicate that of 2,018,000
farms known to the ASCS, 1,942,000 (or 96 percent) are listed on
the Deficiency File. The ASCS definition of a farm is all of the
land farmed under one operation.
Only about 57 percent of the farms listed on the Deficiency
File (1,116,000 farms) are farms that are eligible for the PIK
program. The remaining 43 percent of farms on the list are not
eligible for the PIK program. Some portion of the ineligible
farms are ineligible because they were not growing PIK program
crops, others because they did not choose to apply for the PIK
program. Because of the 96 percent coverage of farms known to
them, ASCS believes the Deficiency File is a very complete list
of farms in the U.S.
In exploring the universe of farms and comparing the two
data sources, we take the 1982 Census of Agriculture as the
primary source of information on the nation's farms. Although
the ASCS total is less than the Census total, it is probable that
the ASCS list is not simply a subset of the farms counted by the
Census, but a partially overlapping list. This is due to the
fact that the two lists are constructed by different
organizations for different purposes, are based on different
information, and have different definitions as for including and
counting specific cases. However, we can get a summary of the
nation's farms from the Census and a rough idea of the ASCS
coverage of those farms.
B. Summary of the Target Universe Based on the 1982 Census
of Agriculture
The figures presented here are taken from Vol. 1, Part 51,
U.S. Summary and State Totals of the 1982 Census of Agriculture.
G-o
-------
The first table lists total numbers of farms by size and sales
categories.
It seems likely that farms with small acreage or low sales
volume would be less likely to have underground motor fuel
storage tanks and would also be less likely to be included on the
ASCS file than large farms. Table G-l indicates that a number of
farms are quite small, with 8 percent of farms reported having
one to nine total acres. Also, many farms have quite low sales
figures. Nearly one-quarter of farms reported on had less than
$2,500 in sales in 1982.
The Census also gives figures for storage of various fuels
(although unfortunately for our survey, no question was asked as
to whether the storage was underground). Table G-2 summarizes
the storage capacity data for 1982.
This indicates that roughly half of all farms reported
gasoline or gasohol storage, and about 4 0 percent reported diesel
storage. The overlap of the two groups is not given but is
presumably fairly high. However, the number of farms with
substantial storage capacity is much less — 2 percent reported
2,000 gallons or more diesel storage capacity, and 1 percent
reported that much gas storage capacity. Taking 1,000 gallons or
more as a cutoff, 7 percent of farms reported this much gasoline
storage capacity and 8 percent reported this much diesel storage
capacity.
In conclusion, based on the 1982 Census of Agriculture,
there were about 2.2 million farms, of which 8 percent were
smaller than 10 acres, one-quarter had less than $2,500 in sales
for the year, and perhaps 10 percent have 1,000 gallons or more
fuel storage capacity. This last assumes a substantial overlap
between storers of gasoline and diesel fuel. If there is little
a-7
-------
Table G-i. Faras by acreage and sales
(1982 Census of Agriculture)
Total U.S. Farms 2,240,976
By acreage
1-9 187,665
10 or more 2,053,311
10 - 49 449,252
50 - 499 1,238,162
500 - 1,999 301,320
2,000 or more 64,577
By sales
Less than $2,500 536,327
$2,500 or more 1,702,973
$2,500 - $9,999 560,010
$10,000 or more 1,142,963
$10,000 - $99,999 840,583
$100,000 - $499,999 274,580
$500,000 or more 27,800
(1,676 abnormal £arms not reported by sales - institutional,
research and experimental farms, and Indian reservations.)
G-8
-------
Table G-2. Fuel storage capacity, 1982*
(1982 Census of Agriculture)
Farms reporting fuel expenses
Gasoline
and Gasohol
Diesel
fuel
Storage capacity reported, farms
1,123,463
924,863
1,0001s gallons
583,853
648,605
Farms with storage capacity of:
1 - 499 gallons
500 - 999 gallons
1,000 - 1,999 gallons
2,000 or more gallons
616,650
352,925
136,455
17,433
471,646
262,902
140,896
49,419
Storage capacity reported as
"no", farms
451,895
150,210
Storage capacity not reported, farms
422,083
245,380
~Includes above-ground tanks and containers, as well as under-
ground tanks.
-------
overlap, as many as 15 percent of farms may have 1,000 gallons or
more motor fuel storage capacity.
C. Comparison of Census and Sample Frame
The sampling frame, the ASCS 1983 Deficiency File, is
primarily a data base of farms rather than a source of
statistics. Hence, we do not have extensive national or state
statistics on this file. Nationally, we can compare the number
of farms from Census (2,240,976) and the ASCS file (1,942,437),
showing that the sample frame file has 87 percent as many farms
as the Census. (Note that these are not necessarily completely a
subset of the Census farms, as mentioned above.)
We also can compare total cropland acreage between the two
data sources. The Census shows 445,3 62,028 acres of total
cropland on 2,010,609 farms with cropland, while ASCS shows
443,850,049 acres of total cropland on its 1,942,437 farms. The
ASCS definition of cropland is "tillable soil" — the land does
not have to have been planted, only to be suitable for planting.
The Census definition includes three categories:
o Harvested cropland;
o Cropland use only for pasture or grazing; and
o Other cropland.
The two definitions appear to be quite similar.
The sample frame thus covers 99.7 percent of the total
cropland reported in the Census and has 96.6 percent as m&nv
farms as those reporting cropland in the Census. It appears that
farms with no cropland is an area of sparse coverage for the ASCS
"•-1.0
-------
list. The major categories of land in farms not included in
total cropland are:
o Pasture and rangeland other than cropland and woodland
pastured (418,264,264 acres);
o Woodland (87,088,255 acres); and
o Land in house lots, ponds, roads, etc. (36,082,032
acres).
So farms with pasture, rangeland or woodland and no cropland are
more likely to be in the Census but not the ASCS list. However,
in the Census 90 percent of farms listed had cropland, so farms
with none are relatively rare.
Other types of farms which may tend to under-represented by
the ASCS list (based on discussions with Tom Meyer of ASCS) would
be growers of fruits and vegetables. Most farms grow more than
one crop, and so many fruit or vegetable farms may also have a
PIK-eligible crop or may be listed as an ineligible farm on the
ASCS file. According to Census data, 69,109 (3.1%) of farms
reported vegetables harvested for sale and 123,663 (5.5%)
reported land in orchards. On a national basis, these farms do
not represent a major portion of the target universe, although on
a regional basis their proportion varies. These figures are
presented as a way of assessing the potential for undercoverage,
but we have no direct way of determining the ASCS coverage of
these types of farms.
Ill- SAMPLE DESIGN FOR UST SURVEY. FARM SEGMENT
In this subsection we again review the survey sample design,
emphasizing the aspects relevant to the farm sample. The design
was a two-stage cluster design. The contiguous U.S. was divided
-------
into six survey regions, as presented in Figure G-l shown
earlier. The first stage of the sample was survey locations,
known as Primary Sampling Units (PSUs). These PSUs consisted of
counties or groups of counties and were chosen by region with
probability proportional to number of fuel establishments. The
second stage was the within-PSU selection of farms. Farms were
selected from a sampling frame based on the ASCS list for the
selected counties with within-PSU probabilities determined so
that the overall probabilities of selection would be equal for
all farms. We give more details in the following sections.
A. First Stage Sample of Survey Sites fPSUs)
The first stage in the two-stage sample design was of PSUs,
which were counties or groups of counties. Within each region,
six PSUs (four in the Mountain Region) were selected with
probability proportional to their number of gas stations and
fuel-related establishments. As discussed in Subsection I, this
is the optimal design for studying fuel establishments — the
main focus of the survey.
Table G-3 shows some statistics on number of farms, by
region, based on the 1982 Census of Agriculture. The first two
columns give the total farms in each region and the corresponding
expected sample size, by region, for an equal probability sample
of 600 farms to be screened for underground motor fuel storage
tanks. Regions 1, 5 and 6 have expected sample sizes of less
than 100, with Regions 5 and 6 less than 50. Next, in column 3,
we have used the inverse of the PSU probability of selection as a
PSU weight and weighted the 1982 Census of Agriculture farm
counts for the selected PSUs up to the regional level. By
comparing these figures with column 1, we see that our sample of
PSUs has considerable variance from the actual totals. As
-------
Table G-3. Farm summary based on 1982 Census of Agriculture,
all farms
Agriculture
Expected
Weighted count,
Expected
Region1
Census count
farm sample2
sampled PSU's
farm sampli
1-Northeast
222,099
60
123,714
36
2-South
548,926
147
283,226
82
3-Midwest
725,699
195
908,358
264
4-Central
464,680
125
494,029
144
5-Mountain
121,777
33
147,071
43
6-Pacific
152,630
41
104,164
30
Continental
U.S. Total
2,235,811
601
2,060,562
599
^¦Regions are defined in Figure G-l-
^These farms are to be screened for the presence of underground ir
fuel storage tanks.
-------
mentioned in Subsection II, this is due to the PSU sample
selection being based on the number of fuel establishments, a
measure inversely correlated with the number of farms.
Finally, column 4 gives the expected sample size based on
the 1982 Agriculture Census counts for our PSUs. Regions 5 and 6
are still very low, and Regions 1 and 2 have a lower sample size
than expected from the regional totals.
B. ASCS List for Selected PSUs
The actual sample was drawn from a sample frame based on the
ASCS 1983 Deficiency File. This file was described in Subsection
II above on a national basis. Here, we compare the ASCS file
counts to the Census counts for our sampled PSUs and present some
relevant Census figures on a regional basis. The actual sample
frame used was a modification of the ASCS file, which we describe
below, leading to the final sample sizes.
In Table G-4, the Census of Agriculture counts are compared
with the ASCS file counts for the sampled PSUs on a region-by-
region basis. The third column shows the percent coverage the
ASCS file had. For the 76 counties in our 34 PSUs as a group,
the ASCS file had 70 percent as many listings as there were farms
counted in the Census of Agriculture. On a region-by-region
basis there is quite a bit of variation in this coverage. The
ASCS list has good to excellent coverage of Regions 2 through 4,
which together contain 7 0 percent of all farms according to the
Census; and fair to poor coverage of the rest of the country.
For Region 3, the Midwestern region, ASCS actually has more
listings — 118 percent as many as the Census. For Region 2
(South) and 4 (Central), the ASCS had fairly good coverage — 90
percent and 79 percent as many listings, respectively, as the
-------
Table G-4. Raw farm count based on sampled PSUs (1982 Census
of Agriculture and 1982 ASCS Defiency File)
Raw counts,
sampled PSU's
1982
1983 ASCS
Percent
Agriculture
Deficiency
Coverage
Region1
Census
File
ASCS File
f-Northeast
3,743
1,573
421
2-South
6,619
5,969
90S
3-Midwest
13,367
15,787
118*
4-Central
11,025
8,706
79S
5-Mountain
4,472
2,305
521
6-Pacific
10,851
504
5X
Continental
U.S. Total
50,077
34,844
70*
^Regions are defined in Figure G~1
"¦-IS
-------
Census. For Regions 5 (Mountain) and 1 (Northeast), the coverage
was only about half — 52 and 42 percent as many listings,
respectively, in ASCS as the Census count. Finally, for Region 6
(Pacific), the coverage was very low — the ASCS list had only 5
percent as many listings as the Census for this region.
Several attempts to understand these discrepancies have met
with limited success. The two data sources rely on different
bases to get their lists of farms and farm operators, employ
different (and to a great extent not thoroughly documented)
definitions of "a farm" and have different basic philosophies of
the importance of complete coverage. We were able to determine
that our ASCS list is a list with one record per farm, as defined
by the County Agent, so that the comparison in Table G-4 is the
relevant one.
We expected that vegetable, fruit or livestock farms would
be at greater risk of under-representation on the ASCS list, so
Table G-5 presents the counts of these types of farms by region,
with the percent of all farms in the region, based on the 1982
Census. A farm may, of course, have crops in more than one
category. For example, a cattle ranch with pastureland would
likely also grow feed grain and be eligible for the PIK program.
Farms with land in vegetables or orchards might also have PIK-
eligible crops, or be on the ASCS File as ineligible. The most
striking statistic in Table G-5 is that, while nationally 5.4
percent of farms have land in orchards, in Region 6 (Pacific),
33.7 percent of farms have land in orchards. It seems quite
probable that this is a contributing factor to the severe
discrepancy between the ASCS frame and the Census in that region.
Region 1 (Northeast) has a higher rage of farms with vegetables
(7% versus 3.1%) than the national average but scarcely enough to
account for listing less than half of all farms in that region.
-------
Table G-5. Regional data from 1982 Census of Agriculture on farms
with land in vegetables, orchards, and pastureland
Farms with land
Farms with land
Farms with
In vegetables
In orchards
pastureland
Region*
Number
Percent
Number
Percent
Number
Percent
1-Northeast
15,458
7,OX
12,740
5.7*
151,287
68X
2-South
19,978
3.6*
28,063
5.1*
355,467
65X
3-Mldwest
17,629
2.4*
11,784
1.6*
413,446
57*
4-Central
4,761
1.0X
12,524
2.7X
353,149
76*
5-Mountaln
2,858
2,3*
5,271
4.3*
82,766
68X
6-Paclflc
7,638
5.0*
51,456**
33.7X
71,679
47X
Continental
U.S. Total
68,322
3.1*
121,838
5.4X
1,427,794
64X .
~Regions are defined in Figure G-l.
** California has 39,801 farms with land In orchards, Including 10,481 with grapes, 7,512 with citrus,
6,119 with avocados, 3,664 with plums and prunes, 2,904 with apples and 2,898 with peaches.
Washington has 6,946 such farms Including 5,406 with apples, 2,235 with pears, 2,066 with cherries
and 1,042 with grapes.
Oregon has 4,709 such farms Including 2,053 with apples, 1,717 with cherries and 1,316 with pears.
-------
The basic pattern in Table G-4 is good coverage to over-
coverage in those parts of the country which contain the majority
of all farms (Regions 2, 3, and 4 contain 1,739,305 farms, or 78
percent of the total, see Table G-3), and fair to poor coverage
in the remainder of the country. This underlying distribution of
farms, combined with the pattern of over- and under-coverage and
the PSU selection probabilities, results in a fairly decent
national estimate of number of farms based on weighted ASCS data,
even though the regional estimates are poor. These weighted
figures are shown in Table G-6, along with the expected sample
size based on weighed ASCS file counts. Regions 1, 5, and 6
continue to lose sample cases due to list undercoverage of those
regions.
D. Sampling Frame and Actual Farm Sample
In order to use the ASCS list as a sampling frame, two
modifications were made. First, the list of farms was collapsed
into a list of farmers by aggregating records with the same name
and address. We would thus be able to increase the number of
farms sampled without increasing the costs by sampling 600
operators and interviewing them regarding "any farm land you own
or operate" in the specific counties they were sampled for. For
those few who reported underground storage tanks, we then
determined which distinct farms have such tanks and how many.
The second frame modification was due to the use of a purchased
list for the large establishment segment of the overall survey.
Any large establishments with agricultural SICs were removed from
the large establishment frame and matched against the ASCS list.
If they did not already appear on it, they were added to the
frame.
-------
Table G-6. Weighted farm counts from ASCS 1983 File, expected and
actual sample sizes
Weighted counts
, sampled PSU s
1982
1983 ASCS
Farm sample
Agriculture
Deficiency
size expected
Region1
Census
File
from ASCS file2
1-Northeast
123,714
52,376
15
2-South
283,226
301,055
86
3-Midwest
908,358
1,105,519
314
4-Central
¦494,029
512,376
146
5-Mountain
147,071
132,621
38
6-Pacific
104,164
5,652
2
Continental
U.S. Total
2,060,562
2,109,599
601
^Regions are defined in Figure G-l
2The->e farms are to be screened for the presence of underground motor
fuel storage tanks.
G-l 9
-------
From the final frame of farm operators thus established, a
sample of 600 cases was drawn with within-PSU probabilities set
so that the entire sample had equal probability. Table G-7
reviews the results of farm operators by region, column 1 shows
the distribution of farm operators by region, column 2 gives the
number of distinct farms this represents, and column 3 shows the
farm estimate based on the unadjusted sample weights. Comparing
these estimates back to the Census totals in Table G-3, we see
that there is quite a bit of region to region variation, although
the grand total is fairly close. This indicates that a ratio
adjustment would improve the sampling error of estimation for
this survey, which we describe in the next subsection.
IV. STATISTICAL ADJUSTMENT OF WEIGHTS TO MINIMIZE SAMPLING
VARIANCE
In the previous subsection, it became apparent that the
actual sample of farms based on the ASCS list does not accurately
reflect the regional distribution of farms as measured by the
1982 Census of Agriculture. Further, in subsection II we found
that the underground tank survey regions are very unequal in
numbers of farms. In order that our final estimates of number
and proportion of farms with underground tanks reflect regional
variation and totals more closely, we propose a system of
adjustments to the sample weights by region. Since some of the
six survey regions have such small sample sizes, we also propose,
for farm estimates only, consolidating the survey regions into
three areas which have about the same number of farms and which
will have over 100 sample cases each. The proposed consolidation
is given in Table G-8, which shows the three consolidated
regions, their Census totals, the unadjusted sample estimates,
and the approximate adjustment factor to apply to the sample
weights so that our final sample estimates (of numbers of farms)
-------
Table 6-7. Results of farm sample draw
Number of
Number of
Weighted
farmers (operators)
farms
number of farms
Region1
sampled2
operated
using sample weight
1-Northeast
11
17
53,395
2-South
88
94
295,242
3-Midwest
324
354
1,111,868
4-Central
142
159
499,398
5-Mountain
33
33
103,649
6-Pacific
2
2
6,282
Continental
U.S. Total
600
659
2,069,834
1Regions are defined in Figure 6-1.
^These farms are to be screened for the presence of underground motor
fuel storage tanks.
n-?i
-------
Table G-8. Consolidated regions for farm estimates and ratio
adjustment factors
Regions
Consolidated
region
1982
Census of
Agriculture
Weighted sample,
selection weights
Ratio
adjustment
factor
(rounded)
- Northeast
and Southeast
East
771,025
348,637
2.21
3-M1dwest
Midwest
725,699
1,111,868
0.65
4,5W>-Central,
Mountain
and Pacific
West of the
Mississippi
739,087
609,329
1.21
-------
will equal the Census totals. The actual adjustment was made
after the field work had been completed, so that the final number
of actual farms contacted was used. After this adjustment, the
ratio of largest to smallest weight was about 3.4 to 1, not an
excessive gap.
In assessing the quality of the final estimates for farms,
for these three consolidated regions and nationally, we have
computed sample variances based on the final weights. There is a
qualitative aspect to the accuracy as well, in which we
acknowledge that coverage of the far West Coast especially is
fairly low, and the estimates for the Western consolidated region
may contain some bias if these three states are strongly
different in terms of underground motor fuel storage from the
rest of the west. However, since the West Coast accounts for
only 20 percent of farms in Survey Regions 4, 5 and 6, it would
have to be extremely different for the survey estimates of this
consolidated region to be significantly affected.
-------
APPENDIX H
ENVIRONMENTAL DATA COVERAGE
I. INTRODUCTION
Environmental data coverage by existing data bases and literature
was explored for geographic locations of the OTS Leaking Underground
TOO
Storage Tank survey. ,£-'J Data sources were located and subsequently
reviewed for their usefulness. From the pertinent literature and data
sources found, environmental data sets were derived for survey areas
and organized within an automated data base. Parameter choices were
directed toward use in leak analyses and fuel migration modeling
studies. The data sets were compiled into a Basic Site Information
File containing locators, descriptors, and cross-reference keys
pointing to additional soil, climate, and groundwater information for
the sites in the survey. Fuel component chemical and physical data
were also compiled and tabulated.^
^¦"Literature Searching for Leaking Underground Storage Tank
Project," General Software Corporation, 1985.
"Environmental Scenario Assemblage for Leaking Underground
Storage Tanks," General Software Corporation, 1985.
^"Environmental Scenarios Supporting Movement of Complex Mixtures
to Groundwater," General Software Corporation, 1986
4"Chemical-Physical Parameters and Processes Effecting Petroleum
Fuel Migration", General Software Corporation, 1985.
H-l
-------
II. DATA SOURCE AVAILABILITY AND COVERAGE
In the search for soil, climate, and groundwater information,
only major readily accessible sources were considered. These sources
include, among others, the County Soil Surveys of the Soil
Conservation Service, USGS publications, and the NAWDEX data base. A
summary of the sources located and descriptions of the information
which they contain are presented in Table H-l.
The County Soil Surveys of the Soil Conservation Service provide
the most complete and comprehensive information on soil
classification. The survey status of the original 76 counties in the
LUST survey is provided in Table H-2. The SCS Soils-5 computerized
data base contains most of the information covered in the published
surveys. There were 914 site locations recorded, and of these, over
450 were covered by modern soil surveys, but approximately 150 of the
latter were designated as urban land or mixed land complexes and were
not fully described.
USGS publications and the NAWDEX Groundwater Site Inventory
provide variable coverage for groundwater and subsurface geologic
information. For areas not covered, regional ranges were recorded
from "Ground-water regions of the United States" by R.C. Heath or from
the ENVIRLOC database as cited in Table H-l. These ranges must be
used with caution, however, since they are broad geographic
approximations only.
-------
To obtain up to date, reliable climatic information/ parameters
were requested directly from the National Oceanic and Atmospheric
Administration (NOAA). Currently, NOAA is compiling parameter
summaries from approximately 3000 UA Weather stations from their
databases for the Exposure Evaluation Division of OTS. Publications
summarizing portions of this data include the Climatic Atlas of the
United States and the Statistical Abstract of the United States. Soil
Surveys frequently contain brief climate summaries as well.
m. BASIC SITE INFORMATION FILE
The Basic Site Information File was designed in support of the
Leaking Underground Storage Tank survey from the work performed in a
preliminary study described in the Task 8 report of EPA Contract
68-02-3970. Data, data ranges, and cross reference keys covering a
variety of locator, climate, soil, and groundwater information were
included in the file to enable the user to have a general
understanding of site location and conditions, and to obtain further
information as necessary.
The file itself contains four sections: site location and
identification, climate, soil, and groundwater/geologic. The
parameters in the file and their corresponding lengths are shown in
Table H-3. Tables H-4 through H-6 are examples from the Basic Site
Information File.
H-3
-------
A. Site Location and Identification
The site location and identification portion includes identifiers
ranging in resolution from general region to specific site. These
locators aid in the determination of the number of sites within a
particular state, county, or region, and in the location of the actual
site on a USGS topographic map.
Hie LUST Regions (Pacific, Mountain, Central, Midwest, Northeast,
and Southeast) are the largest divisions contained in the file,
dividing the United States into six parts for survey purposes. The
PSU, or primary sampling unit, is a further division of the LUST
Region which encompasses one or more counties. There are 34 PSUs
included in the LUST Survey which cover a total of 76 counties.
The state and county FIPS codes, or Federal Information
Processing Standards, are numeric codes for each state and county.
The state and county codes are two and three digits respectively, and
are sometimes combined into a single five digit identifier. Being a
standard identifier, the FIPS Code helps to avoid confusion due to
spelling errors and nonuniform abbreviations.
The USGS Topo Quad information is provided for easy reliable
geographic location. This information includes the name of the
topographic quadrangle on which the site may be found, the map scale
of the quadrangle, and the bottom right coordinates of the map. This
information may be useful in the future for digitization of mapping
and site location.
-------
Survey sites were usually received marked on a USGS topo map.
Sometimes, however, sites were marked on nonstandard or state road
maps, or occasionally not marked at all. If a topo quad could be
determined for a site, this information was included in the file,
otherwise it was omitted.
The Soil Survey Area information provides the name of the Soil
Conservation Service County Soil Survey covering the site, the year
the survey was published, and the survey area code. County soil
surveys cover a county, group of counties, or sections of counties.
Sites located in areas with no current published soil survey are
labelled "Area not surveyed" at this point. Sites not marked, or
marked on large scale maps are labelled "Site not specifically
marked", or with some other pertinant descriptor. The Survey Area
Code is obtained from section one of the Soil Conservation Service Map
Unit Use File (MUUF). Every current county survey has a corresponding
code, which is the county FIPS code for single whole county surveys.
For partial county and multi-county surveys, codes are 600 numbers.
These codes are found by searching the MUUF for state and survey area
name, and are used for finding cross reference keys to specific soil
information.
The specific site locators are the site ID, latitude, longitude,
and approximate elevation. The site ID is an alpha-numeric code taken
from the marked topo maps as received. The number includes the PSU.
For sites with multiple tanks, a letter is tacked onto the end of the
-------
ID (i.e. A, B, C, etc.) identifying each tank, so that each tank has
its own unique record in the event that soil conditions may differ.
The site coordinates were determined by measuring those marked on
USGS topo maps with a gridded ruler to the nearest 1/16 inch and then
performing the necessary calculations. The coordinates were presented
in the file in degree:minute:second format. Sites received marked on
maps with insufficient scale or resolution were included with general
information only (i.e. no specific coordinates). The elevation was
taken from the topo map.
The Hydrologic Unit, or HU Code, is a numeric code assigned to a
drainage basin or distinct hydrologic feature by the Office of Water
Data Coordination. Although the HU Code is applied mainly to surface
water, it is sometimes used to organize groundwater studies. An
example of this is D.K. Todd's major water resource divisions in
Ground-Water Resources of the United States. These major divisions
correspond to the first two digits of the HU Code, as shown in
Figure 1. HU Codes are available from ENVIRLOC.
B. Climate
State Climatic Divisions (SCDs) are areas within states which
have similar climates. The National Weather Service has defined 353
divisions in the United States which frequently follow county
boundaries. These divisions, which were retrieved from GEOCOLOGY, for
survey locations, will help determine the closest applicable weather
station from which to take climate data. NOAA will provide rainfall
H-6
-------
statistics to the Exposure Evaluation Division for those stations
recording hourly precipitation as well as mean temperature and
humidity by SCD.
C. Soil
The soil information included in the basic site file provides
some parameters plus soil type keys for obtaining additional data from
Soils-5, the soil data base of the Soil Conservation Service.
The Soil Map Unit is an alpha-numeric which is obtained from the
Soil Conservation Service published soil surveys. The unit is found
by locating the site on one of the soil maps in the county survey,
usually by comparison with the marked topo map. The Soil Map Unit and
the Survey Area Code are then used to extract the Soils-5 Recnumber
from the Map Unit Use File (MUUF) section three. The Soils-5
Recnumber consists of the two character state abbreviation and a four
digit number which together determine the record to access within the
Soils-5 data base. The additional information include such parameters
as permeability, pH, percent clay, etc. A sample of the available
data is shown in Table H-7.
If a county or part of a county did not have a current published
soil survey, a soil type inference was made using surrounding county
soil surveys, making either an individual soil type inference-or a
major association inference as shown in the site file. Soils-5
-------
Recnumbers were then found as before. If an inference could not be
made with reasonable confidence, then no inference was made.
Additional information in the Basic Site File includes seasonal
high water table, availability of C-Horizon (subsoil) parameters, and
relative corrosivity to steel and concrete, all of which could be
useful for the prediction of possible tank leaks. The seasonal high
water table information provides a depth range, water table type, and
the months of common occurrence. Availability of C-Horizon
information is a yes or no indication of whether countysoil survey
data include the mineral subsoil. Risk of Corrosion is a relative
parameter (low, moderate, high) determined primarily by drainage class
and texture, total acidity, resistivity at field capacity, and
conductivity of saturated extract, as described in part 603 of the
National Soils Handbook of the SCS.
D- Groundwater/Geologic
R.C. Heath divided the United States into major groundwater
regions (referred to in the site file as Heath Regions) in his report
"Ground-water regions of the United States". Figures 2 and 3 show the
boundaries of the fifteen regions. Heath established ranges for
transmissivity, hydraulic conductivity, and porosity for these
groundwater regions, which may be used if actual data is not
available. These ranges are very general, however, and should be used
with caution.
H-ti
-------
Space is provided for the NWWA (National Water Well Association)
subregion for future input. The NWWA is currently organizing
hydrogeologic parameter ranges for subsets of the region of R.C.
Heath.
A literature search was performed in the National Water Well
Association bibliographic data base to locate articles and studies
describing aquifers in the areas of interest. Most of the
publications were USGS reports which contain good groundwater and
geologic descriptions. These USGS publications were used to develop
the groundwater file which is cross referenced in the Basic Site File.
Extensive searches were performed in the NAWDEX Ground Water Site
Inventory to obtain water and well information. Site Resolution
(position with respect to aquifer), Well Usuage Description (domestic,
public, industrial, etc.), and Depth to Groundwater were usually
obtained from the GWSI. Well sites within five minutes latitude and
longitude of a survey site were used to determine the parameters at
that site. If no well sites were within this radius of the LUST site,
Depth to Groundwater was taken from ENVIRLOC (this appears as a
range). The other literature sources previously mentioned were
occasionally used when available.
The Basic Site Information File, Soils-5, the Groundwater
Information File, and the future NOAA weather data, will be useful
tools providing reasonable environmental scenarios to the modeller.
-------
Intentionally Blank Page
-------
Table H-l. Information Source Summary (1 of 5)
Source
Literature:
Parameters
Geographic Coverage
and Frequency
County Soil
Surveys USDA Soil
Conservation
Service
soil type, level, slope,
permeability, pH,
available moisture
capacity, temperature,
precipitation, soil
texture, % fragments,
sieve analysis, liquid
limit plasiticity,
index, shrink/swell
potential, erosin
factors
most OS counties
(down to 60 inches
only)
depth to groundwater soil
bulk density, cation
exchange capacity, organic
content, clay content
some counties
USGS Publications
Water Resources
Data
Guidebooks for
Fieldtrips
Water Resources
Bulletins
surface water data
observation well nunber,
location, hydrologic unit,
groundwater level,
well characteristics,
aquifer type, groundwater
quality
thickness and
characterization of
rocks and water
bearing formations
hydrogeology of
principal aquifers,
saturated thickness
ranges, temperature,
water level,
characterization of
core samples, analysis
all US states
most OS states,
site specific
OS, site specific
(usually to bedrock),
info variable by
state
OS, site specific,
info variable by
state
-------
Table H-l. Information Source Sunmary (2 of 5)
Source
Water Resources
Bulletin
Geological
Circulars
Water Resources
Investigations
Parameters
of rock samples,
hydraulic conductivity,
specific gravity,
particle size, porosity,
water quality
hydrogeology of
prinicipal aquifers,
saturated thickness
ranges, temperature,
water level,
characterization of
core samples, analysis
of rock samples,
hydraulic condictivity,
specific gravity,
particle size, porosity,
water quality
soil chemistry,
tranmissivity,
hydraulic conductivity,
thickness, sieve
analysis, soil layers
well data, water
quality, pimping
and drawdown studies
Geographic Coverage
and Frequency
US, site specific
info variable by
state
US, site specific,
info variable by
site
US, site specific,
info variable by
site
Open File
Reports
Resources of the
United States
D.K. Todd, 1983
Premier Press
Water level, aquifer
description
precipitation,
occurrence of
groundwater,
storage coefficient,
evapotranspi ration,
base of fresh water,
potenticroetric contours,
basement slope
US, site specific,
info variable by
site
US major groundwater
regions, info availab
for most regions
-------
Source
Statistical
Abstract of
the United
States, 1984
US Dept. of
Commerce,
Bureau of Census
Climatic Atlas
of the United
States, 1968
U.S. Dept of
Commerce,
Environmental
Science Services
Administration,
Environmental
Data Service
Hourly
Precipitation
Data, NOAA,
US Environmental
Data Service
(monthly
publication by
state)
Topographic Map
Series, USGS,
Reston, VA
Ground-Water
Regions of the
United States,
R.C. Heath,
USGS Geological
Survey Water-
Supply Paper
2242
Table H-l. Information Source Summary (3 of 5)
Geographic Coverage
Parameters and Frequency
mean temperature, selected US cities
precipitation, days
w/precipitation
greater than 0.1 inch,
average snowfall,
average percent sunshine,
average windspeed
temperature, US (maps)
precipitation,
state climatic
divisions, humidity,
evaporation, snowfall
radiation, skycover,
wind speed
hourly precipitation US weather stations
elevation, coordinates US, most areas
groundwater regions,
descriptions, ranges of
transmissivity,
porosity, hydraulic
conductivity, and
recharge
US groundwater regions
-------
Table H-l. Information Source Sanmary (4 of 5)
Source
NOAA
(National
Oceanic and
Atmospheric
Administration)
Parameters
temperature,
wind speed,
precipitation,
state climatic
division, sky cover,
humidity
Geographic Coverage
and Frequency
OS weather stations,
data collected variable
by station
Data Bases:
Geocology,
Oak Ridge
National
Laboratory
(contained in
GEMS)
NAWDEX
(National Water
Data Exchange)
Ground Water
Site Inventory,
USGS, Reston, VA
National Ground
Water Information
Center Data Base,
National Water
Well Association,
Worthington, OH
monthly temperature
monthly evaporation
state climatic divisions
within counties
soil great groups
well description,
groundwater level,
water use,
lithology,
tranamissivity,
hydraulic
conductivity,
storage coefficient,
water quality
bibliographic,
key word search
covers current
literature including
USGS publications
OS state climatic
divisions
eastern OS counties
OS
eastern OS
OS site specific,
data variable
by site
global, major emphasis
in OS, literature
dependent
ii-xi
-------
Table H-l. Information Source Summary <5 of 5)
Source
ENVIRLOC,
Soil/HU Code,
General Software
Corporation
Landover, MD
Soils-5, USDA
Soil Conservation
Service,
Washington, D.C.
Parameters
approximate depth
to groundwater
ranges, soil
parameter ranges,
Hydrologic Unit
Code, Heath
Groundwater region
number
essentially same
information and
coverage as published
surveys
Geographic Coverage
and Frequency _
continental US by
Zip code or coordinate
most US counties, info
for most counties (with
modern published surveys
only)
H-15
-------
Table H-2. Status of County Soil Surveys (1 of 6)
County
Arkansas:
Garland
California:
Alameda
Los Angeles
San Mateo
Colorado:
El Paso
Teller
Connecticut:
Hartford
Tolland
Florida:
Duval
Illinois:
DuPage
Indiana:
Grant
Survey Name
Survey Status
Year
Published
Mapping in
progress
Alameda Area (excludes Complete
western section)
Los Angeles County, West Complete
San Fernando Valley Area
San Mateo Area Complete
(excludes northern
section)
1966
1979
1961
El Paso County Area
(excludes northwestern
section)
Complete
Mapping not
started
1980
Hartford County
Tolland County
Out of print 1962
Complete 1966
City of Jacksonville,
Duval County
Complete
1978
DuPage and Part of
Cook Counties
Complete
1979
Grant County Out of print 1915
Mapping in
progress
H-15
-------
Table H-2. Status of County Soil Surveys {2 of 6}
County
Iowa:
Pottawattamie
Kansas:
Johnson
Waynedotte
Kentucky:
Bullitt
Jefferson
Oldham
Minnesota:
Ramsey
Mississippi:
Issaquena
Warren
Missouri:
Caldwell
Carroll
Chariton
Clinton
DeKalb
Survey Name
Pottawattamie County
Johnson County
Leavenworth and
Waynedotte Counties
Jefferson County
Oldham County
Washington and Ramsey
Counties
Issaquena County
Warren County
Caldwell County
Carroll County
Chariton County
Clinton County
DeKalb County
Status Survey
Out of print
Mapping in
progress
Complete
Complete
Mapping in
progress
Complete
Complete
Complete
Complete
Complete
Complete
Out of print
Mapping in
progress
Out of print
Complete
Complete
Year
Published
1914
1979
1977
1966
1977
1980
1961
1964
1974
1912
1912
1983
1977
H-17
-------
Table H-2. Status of County Soil Surveys (3 of 6)
County Survey Name
Gentry
Montana:
Hill
Liberty
Toole
Nebraska:
Arthur
Arthur and Grant
Counties
Blaine
Blaine County
Custer
Custer County
Grant
Arthur and Grant
Counties
Hooker
Hooker County
Logan
Logan County
Loup
Loup County
McPherson
McPherson County
Thomas
Thomas County
New Hampshire:
Hillsborough Hillsborough County
Rockingham Rockingham County
Year
Survey Status Published
Mapping
complete
Mapping not
Started
Mapping not
started
Mapping not
started
Complete 1979
Out of print 1954
Complete 1982
Complete 1979
Complete 1964
Complete 1974
Out of print 1937
Complete 1969
Complete 1965
Complete 1981
Out of print 1959
H-13
-------
County
New York:
Albany
Essex
Fulton
Hamilton
Queens
Rensselaer
Ohio:
Greene
Miami
Montgomery
Preble
Oregon:
Clackamas
Rhode Island:
Bristol
Kent
Washington
Table H-2. Status of County Soil Surveys (4 of 6)
Survey Name
Albany County
Rensselaer County
Greene County
Miami County
Montgomery County
Preble County
Survey Status
Out of print
Mapping in
progress
Mapping not
started
Mapping not
started
Mapping not
started
Mapping not
started
Out of print
Mapping complete
Complete
Complete
Complete
Complete
Year
Published
1942
1937
1978
1978
1976
1969
Clackamas County Area
Complete
1985
Rhode Island Complete 1981
Rhode Island Complete 1981
Rhode Island Complete 1981
H-1.9
-------
County
South Carolina:
Lexington
Richland
Tennessee:
Chester
Henderson
Madison
Texas:
Brooks
Collin
Harris
Hays
Kenedy
Travis
Willacy
Williamson
Utah:
Salt Lake
Tooele
1
Table H-2. Status of County Soil Surveys {5 of 6)
Survey Name
Lexington County
Richland County
Henderson County
Madison County
Collin County
Harris County
Survey Status
Complete
Complete
Year
Published
1976
1978
Mapping complete
Complete 1960
Complete 1978
Mapping in
progress
Complete 1969
Complete 1976
Canal and Hays Counti«
Travis County
Willacy County
Williamson County
Salt Lake Area
(excluding eastern
section)
Complete
Mapping not
started
Complete
Complete
Complete
Conplete
Mapping in
progress
1984
1974
1982
1983
1974
H-20
-------
Table H-2. Status of County Soil Surveys (6 of 6)
County
Washington:
Cowlitz
ling
Snohomish
Wahkiakum
Wyoming:
Campbell
Survey Haas
Cowlitz Area (eastern
part excluded)
King County Area
(eastern part
excluded)
Snohomish County
Area (eastern part
excluded)
Campbell County
Survey Status
Complete
Complete
Complete
Mapping complete
Out of print
Mapping in
progress
Year
Published
1974
1973
1983
195S
Johnson
Sheridan
Johnson County, Southern
fart
Sheridan County
Complete 197S
Out of print 1939
Mapping in
progress
H-21
-------
Table H-3. Parameters and Record Lengths included
in the Basic Site Information File.
LUST Region 30
PSU 2
State FIPS 2
County FIPS 3
USGS Topo Quad 30
Scale 9
Bottom Rt Latitude 8
Bottom Rt Longitude 9
Soil Survey Area Name 80
Year Published 4
Survey Area Code 3
Site ID 11
Latitude 8
Longitude 9
Elevation (ft) 5
HU Code 10
SCD 3
Weather Station 35
Soil Map Unit 5
Series 53
Soils5 Recnunber 6
Soil Inference 14
Inference From 33
Inference Associations 56
Inference SoilsS Numbers 54
C-Horizon Info 3
High Water Table 43
Corrosivity to Steel 13
Corrosivity to Concrete 13
Heath Region 27
NWHA Subregion 23
GW & Geologic Description 207
Site Resolution 35
Well Usage Description 35
Depth to OW (ft) 7
GW Cross Reference 25
-------
Table H-4. b<»si<: Site information for Arthur County, NE.
Lust Region: Central
PSU: 23
State PIPS: 031
County FIPS: 005
USGS Topo Quad Name: Arthur
Scale: 1:62500
Bottom Ricfit Latitude: 41:30:00
Bottom Right Longitude: 101:30:00
Soil Survey Area Name: Arthur and Grant Counties
Year Published: 1977
Survey Area Code: &oi
Site ID: N230000635 Latitude: 41:34:05 Longitude: 101:41:25 Elevation(ft): 3730
HU Code: 10180014 SCO: 02 Weather Statical:
Soil Map Units VaE Series; Valentine Fine Sand
Soils5 Recnwtoer: NEG091
Soil Inference: Inference Prcnu
Inference Associations:
Inference SoilsS Numbers:
C-florizon Info: no High Water Table: gt 5.0ft
Corrosivity to Steel: To Concrete:
Heath Region: 5 High Plan NWWV Subregion:
Of and Geologic Description: Dune sand aquifers - unconsolidated fine sand and clay with shallow water table
Site Resolution: dune sand
Well Usage Description: irrigation
Depth to Groundwater(ft): 15.00
GW Cross Reference: 52
-------
Table H-5. Basic Site Jnforinacion for Grant County, IN.
Lust Region: Midwest
PSU: 17
State FIPS: 18
County FIPS: 053
USGS Topo Quail Name: Sweetser
Scale: 1:24000
Bottom Right Latitude: 40:30:00
Bottom Right Longitude: 85:45:00
Soil Survey Area Name: area not surveyed
Year Published:
Survey Area Code:
Site ID: N170001264 Latitude: 40:30:29 Longitude: 85:49:34 Elevation(ft)
HU Code: 5120101 SO): 05 Weather Station:
Soil Map Unit: NA Series:
Soils5 Recmmber:
Soil Inference: Ba.Pw Inference Frcm: Miami County, 1979 (103)
Inference Associations: Blount-Pewamo Association
Inference Soils5 Numbers: IL0014.MI0042
C—Horizon Info: yes High Water Table: 0-3.Oft,perched apparent Dec-May
Corrosivity to Steel: high To Concrete: low
Heath Region: 6 Nonglaciated Central NWWA Subregion:
GW and Geologic Description: Unconsolidated sand and graval deposits over water bearing limestone and
dolomite bedrock
Site Resolution: over unconsolidated and bedrock aquifer
Well Usage Description:
Depth to Groundwater (ft) : 3.2-10
QW Cross Reference: 29
-------
Table H-6. Basic Site Int->r;uat ion for Duvai County, FL.
Lust Region: Southeast
PSU: 07
State FTPS: 12
County FIPS: 031
USGS Itopo Quad Name: Jacksonville
Scale: 1:24000
Bottom Right Latitude: 30:15:00
Bottom Right Longitude: 81:37:30
Soil Survey Area Name: City of Jacksonville, Duval County
Year Published: 1978
Survey Area Code: 031
Site ID: D070000154 Latitude: 30:20:52 Longitude: 81:44:44 Elevation(ft)
HU Code: 3080103 SCD: 02 Weather Station:
® Soil Map Unit: 26 Series: Pelham Fine Sand
ui SoilsS Recnunfcer: GA0015
Soil Inference: Inference From:
Inference Associations:
Inference SoilsS Numbers:
C-Horizon Info: no High Water Table: 0-1.Oft, apparent Jun-May
Corrosivity to Steel: high 'I'd Concrete: high
Heath Region: 10 Atlantic & Gulf Coastal NWWA Subregion:
GW and Geologic Description: Layers of clay, sand, shells, and limestone, very shallow water table, springs
and seeps common
site Resolution: over shallow & Florida aquifer
Well Usage Description: irrigation, public, domestic
Depth to Groundwater(ft): 23.00
GW Cross Reference: 23
-------
Table H-7.Example of the type of Information available in Soils5.
SO0I02 50ft INTERPRETATIONS RECORD
t«.RA: 53C• 55C. 63A. 6 38 SULLY SE<»ie5
Rf!V. MWS.LOi. 2-8A
TYRIC USTORTHENTS. COARSE-SILTV. NIXED «CALCAREOUS I . NES1C
the sw.Lv semes consists or oefp. well oraineo soils forced in loess om uplands and terraces. the surface h»!» is
CRAY ISM BROWN SILT LOAM 3 INCHES THICK. TME SUBSTRATUM IS Lir.Mf BROWNISH GRAY CALCAREOUS SILT LOAN. SLOP'S RANGE FRO" 0
TO *0 PERCENT. AREAS ARC USED FOR RANGELAND ANO CROPLAND.
£SIl?lfiIE2_5UlL_B3DeEBIl£S
I
UNIFIED I AASHTO
DEPTH| |
(IN.I| USDA TEXTURF I
J :_] 1
0-J ISIL l«L. rL. CL-NL
0-3 |YFSL l«L. CL-NL
3-601SIL. VFSL INC. CL-NL. CL
I I
I I
DFPTMICL AY TnoTsT SULK? PfRMEA-
«IN.I||PCT»| DENSITY | OIL IT Y
1 l_ifi3 iNi_iu«ti_ar_e4SSitic_siEYC_eiQ*_i li-it iticityi
lIBCIi.l__5 i—10—i—4Q—1_2 Qfl—1 ll*eC5_l
0 I 100 100 95-100 90-1001 25-*0 I 3-15 I
0 I 100 100 90-100 70-95 I 20-35
0 | 100 95-100 90-100 ItS-IOOl 20-AO I 3-15
I
3-10 I
|A-». A-6
I A - A
|a-». a-a
I II III
I II III
> V A IL ABLE I tniL T SALInTty T SHRINK- I EROS IOn7*7kO lOPGANICl CORROSIYITY |
WAT PR CAPACITY|RFArTICNKMMHOS/CN)| SWELL I F&C1QSS • E'OO . INA T Tf P I |
l «eai_l ieQiccfiuu._5_i_i_ifiBQuci_iecii_i_5iEtu__icc!«catiEi
l*.6-7.S |
|e.«-7.s |
I7.A-S.A I
I I
J I
-------
SOUUfS it! P KAINY
—tskrwVw
09l
f
[1st W tN(,LANI>
«nt
PAMHl NORIHWm
s \ .. V
(itfcAl I.Ak*S
04
MISSOURI HASIN
MID AILANIIC
I.Ml AI BASIN
iwi
lifri h|n»i.o
OIORAImJ
OHIO
CAI II Oil NI A,
*C
~ xn ana iion
SKkAMSAK WHIM Itt l>
•c
RIO
SOUTH ATLANTI<-«il'IF
03
V
20 0
Figure 1. Major surface hydrologic units, corresponding also with groundwater
regions as described in Todd, 1983.
-------
2. Alluvial Batin
Jrri/r^.'rS ^JS,^ V
»J-w-w4sk!^^ ¦ \ ^ r5"-*
f-8'?? AI Weslsrr
<^C'i JaTa* .. '/Jf \» ' ', (sA1
r*>W>/'f'4?Qj>:
9. NonhiriSst,
1J Western,
Mountain «
v flaofle* 9*
»
;t
»^|6vW^ (Ifi ^>i-
^\m S* \/ <
5 High\\ A/ , f0 ,
Plaint \ \ " x f'Ori
'«C»
. Npnelaciated
rm region <
r«;o!
/'
'¦ *tf/M
•sftSSfelG
14. ALASKA \ r.
r-V A»y 'i-M
¦XC *;• i >4;
->.»v«^'i»*s/ 1 .• t "iv v r» a. " •
%" 0~ v, .- A *'•« » l •/• .• ./ *#V
r £ <¦•'
?J ?••$¦ .. - - ~l '
x ^ A j *v ' ) „' CMrt*1
# ¦¦¦/ . v/GuW.r>>-
1 / • »¦• <
* /i .vo' V
«%-V
.*o~
' y
-------
ALASKA
HAWAU
Figure 3. Heath region 12
•tV
. a J \M
-------
APPENDIX I
MULTIVARIATE ANALYSIS
I. INTRODUCTION
While the tables presented in Section 9 provide a useful
descriptive look at leaking tanks and conditions under which
leaks occur, they do not take into account the simultaneous
effects of many variables. To respond to this analytical need,
multivariate statistical models have been developed to examine
the relationship between leak status (1 = leak, 0 « no leak) [or
leak rate (gallons per hour)] and various explanatory variables.
The advantage of the multivariate analysis is that it
provides a method of assessing the contribution of individual
explanatory factors, while simultaneously controlling for other
variables. The procedures used also allow a step-wise approach
(i.e., first finding the one variable that best predicts leak
status [or leak rate], then the second best predictor, etc.) and
a test for the statistical significance of coefficients of each
variable in the model. The results of the multivariate analysis
have been summarized in the next subsection so that the reader
may learn the outcome of the multivariate analysis without having
to go through all the mathematical details. The technical
details on mathematical formulation can be found in later
subsections, along with the final equations for the multiple
regression and logistic regression models developed.
T — 1
-------
II. SUMMARY OF MULTIVARIATE ANALYSIS RESULTS
The major results of the modeling efforts are presented
below. The reader should also note the caveates and limitations
at the end of this summary.
a. gprrelatjons
The multiple correlation coefficients (R) from the final
regression models (which retained only variables with significant
regression coefficients — see Subsection C for confidence
levels) were about .30 for leak status and .45 for leak rate,
demonstrating low to moderate predictive ability. This
corresponds to R2 values of aboue .08 and .20, respectively.
Since R2 can be interpreted as the fraction of variance accounted
for by the model, it is clear that the models do not account for
most of the variance in leak status and leak rate.
B. Predictors of Leak Status
Based on the coefficients in the regression and/or logistic
models, the probability that a tank system leak tends to increase
for:
o Older tanks,
o Tanks with no leaded gasoline stored,
o Tanks with passive cathodic protection, and
o Tanks for which no log of deliveries is kept.
T-O
-------
The positive relationship between leak probability and passive
cathodic protection might seem surprising. A possible
explanation is that passive cathodic protection tends to be used
in areas which have a history of corrosion/leak problems.
Another explanation could be that passive cathodic protection is
strongly correlated with the storage of aviation fuel and, thus,
might be a proxy for this fuel type. (The multivariate model
equations for leak status may be found in Section III, which
follows.)
C. Predictors of Leak Rate
Among leaking tank systems, the leak rate tends to be larger
for:
o Fiberglass tanks;
o Tanks not on a concrete pad;
o Tanks both old and steel (i.e., an interaction
effect) ;
o Tanks attached to other tanks; and
o Tanks in establishments with operators trained to check
for line leaks.
The above factors are not indicators of leak likelihood, but of
larger leak rates among leaking tank systems. The last factor
may well be a case of reverse causality — i.e., where tank
systems leak heavily, operators are trained to detect line leaks
(rather than vice versa).
*More precisely, fiberglass tank systems show less increase in
leak rate as they get older.
T —">
-------
D. Limitations and Caveats
In addition to the comments about the limitations of the
scope of the study presented in Section 8, the following
limitations and caveats apply to the multivariate analysis:
o Only business, government and military sectors are
included (no farms).
o Manifolded tanks that could not be separated for
tightness tests are not included.
o Although a long list of 49 potential explanatory
variables were included, there are other possible
variables which were not in our data base and whose
effects are, therefore, not accounted for. In
particular, soil characteristics were not available for
analysis and use in the models. However, backfill
around the tank (e.g., sand/gravel) is included and may
be more relevant.
o The multivariate analysis finds "measures of
association" rather than causality. Naturally, since
the variables used were suspected of affecting leaking,
the discovery of a statistically significant
association tends to affirm a causal linkage. But the
reader is cautioned that a different covariate could be
the real causative factor, as in all statistical
correlation studies. For example, the variable "age of
tank11 could represent the effects of aging, per se, or
age of tank could be a proxy for different installation
techniques which changed over time, or different resins
used in the manufacture of fiberglass tanks in
different production years.
-------
III. MULTIVARIATE MODEL DEVELOPMENT PROCEDURE
A. Overview
Two regression models (one to predict leak status and one to
predict leak rate) were developed using the variables in Table
1-1 as candidate predictor variables. (Table 1-1 also appears as
Table 9-31 in Section 9 of this report.) The regression analysis
followed a number of preliminary steps before arriving at the
final models. This included elimination of variables with too
many missing variables (X13, X16, X18) and variables with nearly
constant values (X8, X9, X21, X23). Stepwise regression runs
were made to obtain a reduced set of variables which best
predicted leak status or leak rate. Finally, individual
regression coefficients were examined to ensure statistical
significance. Sample sizes are shown below for the final model.
Model
Sample
Size
Leak Status Regression
Leak Status Logistic
Leak Rate Regression
380
327
99
-------
Table 1-1. Simple Correlation of Leak Status with Explanatory Variables
Eaplenetory
Variable
Meaning
Definition
Correlation^)
leak atatua
(1 2 Leak; 0 z
with *1,
No Leak)
Correlation'1) »ith yz.
Leak rate (gel/Hr),
¦nong leaking tanks-2)
XI
Gas Station
1 = Yea;
Q i No
-.08
-.06
X2
1 Underground tanks
Number at facility
.12
.10
X3
Tank capacity
Cailone
.14
.34
Xk
Average lo« fill level"'
Aa fraction of tank capacity
-.05
-.07
X52
(Age of tank)^
in ' years i2
.11
-.20
X6
Leaded gaaoline
1 - yes;
~ > HO
-.26
-.11 1
X7
Diesel fuel
1 = Yes;
Q = >io
.26
-.08
xa
Aviation fuel
1 S Yes;
~ 3 So
.13
.07 j
X9
Gaaohol
1 s tes;
0 i No
-.07
j
X1Q
Other
1 * Yes;
Q s No
.08
.29 |
XII
Suction pimp
1 = Yes;
0 > Mo
.003
-.12 I
XI2
Depth buried
Inchee fro® surface
to top of tenk
.10
-.006
XI3
Mater level
inches from surface
to water tabie^
-.15
-.005
xi 5
Tank tested
1 if tasted after pieced
In service; 3 otherwise
.03
.01 j
xu
Years since test
Since most recent test
.0ot
-.21
X17
Tank nateriel
1 = steel: 0 * fiberglass
.02
.
-.09
XI3
Tank lined
1 = Yes;
0 = No
.07
.02
X19
Tank coated
1 s *es;
0 a Mo
-.01
-.25
X20
Passive cathodlc protection
1 s Yes;
0 : No
.10
i
.05
X21
[¦pressed current cath.
protection
1 s Yea I
0 i No
0
j
0
X23 l
t
Other protection
1 * yes;
0 I So
-.08
1
1
0
*24 j
Previous tank leak
1 * Yes;
0 » Nd
-.05
»
1
-.04
*25
Previous line leak
1 = Yes:
0 x No
.05
i
i
*26 !
1
frequency of delivera
Number per year
-.05
i
i
-.003
*27 |
Sand fill . j
1 = Yee;
0 a No
.03
!
-.10
*28
Gravel rill 1
1 = Yes:
0 = No
.006
.16
*29
Concrete pad j
1 s Yes;
0 * No i
.07
1
-.09
*30 j
Packed earth pad {
1 * Yes;
0 > No 1
.03
1
-.09 |
*31 |
Qist. to nearest tank or
structure |
(feet;
1
-.0#
|
-.09
!?earion'3 correlation coefficient: Kendall's Tau-8 alao calculated for til Y1 correletione and found to ba tha same for
nearly every variable.
-I
"Uainq data only frcw individual leaking tanks »ith quantifiable leaks.
'i.e., juat before product la added.
4
At time of ta(t.
-------
Table 1-1. Simple Correlation of Leak Status with Explanatory Variables
(Continued)
Explanatory
Variable
Meaning
Definition
Correlation'1'
leak atecus
(1 > laakt 0 x
xith Y1,
No laek)
Correletion'1> »ith Y2,
Laek rete (gel/Hr),
Mong leaking tanka'2'
X32
Interaction) age 4 aaternl
(X5 ) (1-X17)
-.03
..07
X33
Interaction: geaohol 4
aatari«l
X9 (1-X17)
0
0
X34
Penit to inatail
1 > Yea; 0
« No
.12
X3S
Perait to atora
1 s Yea: 0
• No
.02
.09
*36
Averaqe nigh fill level*'6'
Aa fraction of tank capacity
..06
-.09
XT J
Average fuel dellvary
in gallone
(to ona tank)
.1J
.23
XT4
Max. ever »tor«d
gallons
.11
.29
XT ISA
Attached to other tank
1 > Yes: 0
= No
.22
.24
XT19
Tank proxiaity to oatar
tab la
1 s ifeove:
*ovb: J
2 i ..lrtially
x belo«t A a other
.11
.28
XT20
Marmay «ith tank
1 < Yea; 0
« No
.19
.13
XT36
Hot aelf-inatalled
1 x Yea) 0
< No
.12
.12
m
Reaote gauge
1 x Yaei 0
> No
-.00i
.0}
»19
Log of deliveries
1 x Yeei 0
• No
-OJ
.002
XC7
Any abandoned tank'9'
1 > Yaai 0
x No
-.03
.03
XC8
# Abandoned tonka
(coded aa zero if none)
• .12
-.09
xfia
Corrosion prevention equip./
aat.
1 • Yee: 0
x No
-.02
-.12
XC2D
Trained to check punp
1 * Yeai 0
= No
.14
.24
XC2C
Treined to check line leak a
1 x Yaei 0
x No
.10
:
.IB
XC2F
Trained to check leak
prevent ion
1 * Yee: 0
* No
.10
.1»
XC2G
Treined to check leak
aonitoring
t > Yea: 0
* No
.15
.17
SAt that facility.
'i.e., Juet aftar product ia delivered.
-------
B. Multiple Regression Models
Two models were constructed:
[1] Leak Status Model:
(among all tanks
with tightness test)
[2] Leak Rate Model:
(among leaking
tank systems only)
Dependent Variable, Y1 -
1 if leak
0 otherwise
Dependent Variable, Y2 -
leak rate in gal/hr
Both models were run using the predictor variables in Table
1-1. The general form of the model is:
Y - b0 + b^ + b2X2 + ...
where a few of the variables were interaction terms and the b's
are regression coefficients estimated by a least-squares
procedure. In addition, a non-linear transformation was used for
one of the X variables. Age2 was used rather than Age because
data plots suggested a non-linear increase in the percentage of
tanks that leak as a function of age.
-------
C. Logistic Regression Model
For the leak status model, an alternative logistic
regression model was run. The dependent variable can be
reexpressed as an odds ratio*, in the form:
rial log Probability of Leaking Tank m
Probability of Tight Tank ~
bo + blxl + b2xx +
This alternative formulation of Model [1] should more nearly
satisfy the homogeneity of variance assumption for regression.
The coefficients (b's) for the Logistic Model are estimated
by maximum-likelihood methods rather than least-squares.
IV. FINAL MULTIVARIATE MODELS
Using the procedures defined above, linear and logistic
regression models were developed for leak status. For leak rate,
a separate linear regression model was developed. The final
models appear below.
The assumed underlying model for the logistic regression is
Y - 1/[(1 + exp (-bQ-b^Xi - b2X2 ....)]. From this expression
it can be shown that log [Y/(I - Y)] « bQ + b^X, + b2X2 + ...
In this equation Y is the probability that the tank system leaks
and 1 - Y is the probability that it does not leak.
T-9
-------
Leak Status Models
[1] Regression Model*: ^
Y1 = .22 + .00019 X5 - .25 Xg + .0044 X12*** + .18 X20
[la] Logistic Model****:
log Prpbability pf Leak _ 1#3 _ #63 x _.0 17 x
y Probability no Leak 6 12
- .38 XB19
All coefficients significant at the 94 percent confidence level
or better (except coefficient of X20 at 78 percent confidence
level).
**(Age)2 was used rather than Age because this non-linear
transformation showed a stronger correlation with leak status.
The regression model found a + coefficient, but the logistic
model found a - coefficient. This may be a case of X12's
collinearity with other variables. However, no strong
collinearities were detected with X,,. (See Tables 1-2 and 1-3
in Section V.) Therefore, the relationship with X12' depth tank
is buried, is inconclusive based on this mixed result.
****A11 coefficients significant at the 94 percent confidence
level or better.
-r . l
-------
[2] Leak Rate Model*****:
Y2 - .91 - .67 X17 - .54 X29 - .0068 X32
+ .62 XT18A + .25 XG2E
The reliability of the model was examined in several ways.
For the regression models, the multiple correlation coefficient,
R, provides some overall measure of the predictive ability of the
model. These results are shown below.
Multiple Correlation
Coefficient, R R2
Equation Unadjusted Adjusted Unadjusted Adjusted
[1] .30 .29 .093 .081
[2] .50 .45 .25 .20
*****A11 coefficients significant at the 97 percent confidence
level or better.
JL JU JU JL
This is an interaction term which was included to capture
the more than additive effect of age and material type together.
-------
The "adjusted" values of R and R2 adjust for degrees of
freedom in the model and, therefore, provide a better estimate of
how reliably the model might predict leak status and leak rate
for other tank systems beyond the modeling data set. The R2 term
can be interpreted as the proportion of the variance in Y that
can be explained for by the model. Thus, the model is able to
account for less than 10 percent of the total variance in leak
status and only about 20 percent the variance in leak rate.
The reliability of the coefficients of the X's in equations
[1], [la] and [2] were also examined to ensure that the value is
not likely to be a chance occurrence. The probability that these
coefficients are not chance occurrences is 94 percent or more for
each of 9 of the 10 parameters in these equations. The remaining
coefficient had a 78 percent probability of being a non-chance
occurrence (i.e., there is a very low probability of the observed
coefficient occurring if its true value were zero). It should be
noted that these probabilities of non-chance occurrence applies
one variable at a time — i.e., with many variables tried in the
model, the probability of at least one chance selection of a
variable increases.
V. RELATIONSHIP BETWEEN EXPLANATORY VARIABLES
(COLLINEARITY1
Multicollinearity frequently exists in large data sets.
Pairwise collinearity is one sample form, and is relatively easy
to visualize. In order to test for such "first order"
collinearity in the models, the correlations between all pairs of
independent or predictor variables (i.e., X's) were computed.
The results shown in Table 1-2 indicate low pairwise
collinearity, except for X17 (tank material) and X32 - [(1 - tank
-------
Table 1-2. Collinearity (intercollelation) of X'a in models
A- Leak status regression and logistic models
Pearsons Correlation Coefficient between
explanatory variables
2
X5
X6
X12
X20
XB19
2
X5
1
-.03
I
•
o
-j
00
0
•
1
.10
X6
1
-.06
-.12
.002
X12
1
.07
.09
X20
1
**
0
•
1
XB19
1
B. Leak rate regression model — Pearson's Correlation
Coefficient between explanatory variables
X17
x29
X32
XT18A
XG2E
X17
1
.09
o
00
•
1
. 13
.05
X29
1
-.07
.38
.08
X32
1
l
•
O
-.11
XT18A
1
-------
material) x (Age)2] in the leak rate model (correlation of -.80).
The variable, X32' is an interaction term. The correlation of
X17 with X32 is close to the correlation of Age2 with -Age2.
Therefore, a large intercorrelation would be expected.
Table 1-3 shows correlations between variables in the models
and variables not in the models. (Variables with small
correlations, less than .20, are not included.) Any large
correlations could be considered as proxies (or substitutes) for
the model variable with which they are strongly correlated. For
example, in the leak status model, passive cathodic protection
(X20) is strongly correlated (correlation coefficient - .62) with
aviation fuel (X8). Therefore, the apparent increase in the
likelihood of a leak with passive cathodic protection, might be
due, in large measure, to its relationship with aviation fuel
storage.
-------
Table .1-3. Correlation Between Model X's and X's not in the Model
A. Leak Status Model
Pearson's
Correlation
Coefficients
Model X Non Model X's (> .20)
2 o
X5, (Age of Tanks)
Xg, Leaded gasoline
X-^2. Depth buried
Xy (Diesel fuel)
^20, Passive cathodic X2 (# Underground tanks)
Xg (Aviation fuel)
*B19'v
deliveries
l18
(29
(Tank lined)
(Concrete pool)
X^isa (Attached to other tank)
Xj*20 (Manway with tank)
XG2e (Trained to chek line leaks)
X^2f (Trained in leak protection)
Xq2h (Trained in leak monitoring)
X-^3 (Water level)
*16 (Years since test)
*34 (^ermlt to install)
X35 (Permit to store)
None
-.39
None
.33
.62
.34
.38
.29
.41
.24
.27
.31
.30
.34
.20
.20
-------
B. Leak Rate Model
Pearson's
Correlation
Coefficients
Model X Non Model X's (> .20)
Xiy, Tank material Xi (Gas station) -.21
X7 (Diesel fuel) .22
Xll (Suction pump) .42
Xjj (Water level) -.29
Xjg (Tank tested) -.28
Xi6 (Years since test) -.37
Xlg (Tank lined) -.35
X19 (Tank coated) .66
X32 (Interaction: Age & material) -.80
X2g, Concrete pad X2 (# Underground tanks) .46
X4 (Average low fill level) ,24
X16 (Years since test) -.48
X2q (Passive cathodic protection) .26
X^q (Packed earth pad) -.20
x34 (**ermit to install) .24
*36 (Average high fill level) .28
X^.3 (Average fuel delivery) .20
Xq.^gA (Attached to other tank) .38
X^2q (Manway with tank) .52
Xq2|| (Trained in leak monitoring) . 24
X32. Interaction: Xj^ (Suction pump) -.29
Age & material X|g (Tank capacity) .26
X|6 (Years since test) .49
X|7 (Tank material) -.80
X|g (Tank lined) .53
X|g (Tank coated) -.54
*T18A* Attached to X2 (# underground tanks) .48
other tank X3 (Tank capacity) .22
X7 (Diesel fuel) .23
X13 (Water level) -.30
Xig (Years since test) -.23
X25 (Previ°us line leak) ,28
X29 (Concrete pad) .38
X^q (Packed earth pad) -.29
X34 (Permit to install) \ 29
X35 (Permit to store) ,25
X36 (^erage high fill level) .25
XT3 (Average fuel delivery) ,35
(Maximum ever stored) .33
X^2q (Manway with tank) .40
Xq2d (Trained to check pump) .24
I-15
-------
Leak Rate Model (Continued)
Pearson's
Correlation
Coefficients
Model X Non Model X's (> .20)
*G2E' drained to X2" (# underground tanks) .25
check line X7 (Diesel fuel) -.25
leaks Xg (Aviation fuel) .25
X^g (Other fuel) .41
*16 (Years since test) -.44
X28 (Gravel fill) .21
Xj^g (Tank proximity to water table) .39
X^20 (Manway with tank) .22
Xg2Q (Trained to check pump) .40
-Xg2f (Trained in leak protection) .89
Xg2H (Trained in leak monitoring) .68
1-17
------- |