560586014
             DEVELOPMENT QF  A TANK TEST METHOD  FOR  A  NATIONAL
                    SURVEY OF UNDERGROUND  STORAGE TANKS
                                    by

                             H.  Kendall  Wilcox
                           Jairus  D.  Flora,  Jr.
                             Clarence L.  Haile
                            Marilyn  J. Gabriel
                        Midwest  Research Institute
                          Joseph W.  Maresca, Jr.
                           Vista Research,  Inc.
                          WORK ASSIGNMENT  NO.  25

                           REVISED  FINAL REPORT

                          Contract  No.  68-02-3938
                        MRI Project No. 8501-A(25)
                            /t/(4y/
                                    For

                   U.S.  Environmental  Protection Agency
                        Office  of  Toxic  Substances
                       Field  Studies Branch, TS 798
                             401 M Street,  SW
                          Washington, DC   20460

                Attn:   Dr. Joseph  J. Breen, Project Officer
                       Michael  R.  Kalinoski, Work Assignment Manager
                       Tom Murray,  Work  Assignment Manager
liP 560/5
86-014

-------
                                 DISCLAIMER
          This document has been reviewed and approved for publication by
the Office of Toxic Substances, Office of Pesticides and Toxic Substances,
U.S.  Environmental Protection Agency.   The use of trade names or commercial
products does not constitute Agency endorsement or recommendation for use.

-------
DEVELOPMENT OF A TANK TEST METHOD FOR A NATIONAL
       SURVEY OF UNDERGROUND STORAGE TANKS
                       by

                H. Kendall Wilcox
              Jairus 0.  Flora, Jr.
                Clarence L. Haile
               Marilyn J. Gabriel
           Midwest Research Institute
             Joseph W.  Maresca, Jr.
              Vista Research, Inc.
             WORK ASSIGNMENT NO.  25

              REVISED FINAL REPORT

             Contract No.  68-02-3938
           MRI Project No.  8501-A(25)
                       For

      U.S.  Environmental Protection Agency
           Office of Toxic Substances
          Field Studies Branch, TS 798
                401 M Street, SW
              Washington, DC  20460

   Attn:   Dr. Joseph J. Breen, Project Officer
          Michael R. Kalinoski, Work Assignment Manager
          Tom Murray, Work Assignment Manager

-------
                                 DISCLAIMER
          This document has been reviewed and approved for publication by
the Office of Toxic Substances,  Office of Pesticides and Toxic Substances,
U.S.  Environmental  Protection Agency.   The use of trade names or commercial
products does not constitute Agency endorsement or recommendation for use.

-------
                                   PREFACE
          This revised draft final report was prepared for the U.S. Environ-
mental Protection Agency under EPA Contract No. 68-02-3938, Work Assignment
No. 25.  The report describes an evaluation of methods for determining leaks
in underground tanks used to store motor fuels and a pilot study conducted in
preparation for a national survey of underground storage tanks.  The work as-
signment was directed by Dr. Ken Wilcox under Dr. Clarence L. Haile, Deputy
Program Manager.   Technical support at MRI was provided by Dr. Jairus D. Flora,
Ms. Marilyn Gabriel, Ms. Sharon Perkins, Mr.  Brad Combs, and Mr.  Ed Olson.

          Two key MRI consultants contributed to the success of the program.
Dr. Joseph W.  Maresca, Jr., Vista Research, Inc. provided additional data
analysis and evaluation methods.  Mr.  William Purpora, Protanic,  Inc., pro-
vided guidance in field testing procedures and data evaluation.

          Several U.S. Environmental Protection Agency personnel  were instru-
mental in the success of the project.   Mr. Michael R.  Kalinoski and Mr. Tom
Murray were Work Assignment Managers for EPA, and Dr.  Joseph Breen, Project
Officer for the program, provided key input to the project and were involved.
in many helpful discussions of the testing procedures and evaluation of the
data.   Additional input from the EPA was presented by Dr.  Frederick W. Kutz,
Mr. Martin Halper, Mr. Joseph Carra, and Mr.  Robert Heath.

          Excel lent-cooperation was received from those responsible for pro-
viding test facilities.   In particular Mr. Andres Talts, Defense Environmental
Leadership Project, and Major Patrick Fink, U.S. Air Force, were extremely
helpful in obtaining cooperation at the various military facilities where
testing was conducted.  Mr. Glen Pugmire, Amoco, provided a site and fuel for
the preliminary testing in Kansas City.   Mr.  Fred Sanders,  Amoco, was most
helpful in providing support on a daily basis during that time.  .Mr. Dave
Carpenter, Texaco, also provided a site and fuel for the temperature study
which was conducted in Kansas City.

          Excellent cooperation was also received from the various test com-
panies who participated in the program.   Helpful comments and information
were provided by Ms.  Joyce Rizzo, Hunter Environmental Services,  Inc.  (Leak-
Lokator), Mr.  Tom Collins and Mr. Frank Senese, ARCO Petroleum Products
Company (ARCO Tester), Mr. John Copus, 0. H.  Materials Company (Petro-Tite),
Mr. William Burkhart, Smith and Denison (Varian Helium Leak Detector),
Mr. John Nedbed,  Fuel Recovery Environmental  Services, Inc. (Certi-Tec),
Mr. Jack Stillwagon, Heath Consultants (Petro-Tite), Mr. Eugene Collins,
National Institute of Petroleum and Energy Research, and Dr.  Jon Young of
Ohio State University.

-------

-------
          This report was prepared by Dr. Wilcox, Dr. Flora, Dr. Haile,
Ms. Gabriel, and Dr.  Maresca.

                                        MIDWEST RESEARCH INSTITUTE
Approved:
James L. Spigarelli
Associate Director
Kansas City Operations
                                        Paul C. Constant
                                        Program Manager

-------

-------
                              TABLE OF CONTENTS

                                                                      Page

Preface	    ii
List of Tables	     v
List of Figures	   vii
Executive Summary	    ix

I.         Introduction 	     1

II.        Preliminary Investigation	     3

               A.  Overview	     3
               B.  General Selection Guidelines	     4
               C.  Identification and Classification of Test
                     Methods 	     6
               0.  Field Observation Visits	    11

III.      Preliminary Testing	    13

               A.  Experimental Procedures 	    14
               B.  Results	•	    17
               C.  Conclusions and Recommendations 	    30

IV.        Development Study Testing. ...•...'	  31

               A.  Overview	    31
               B.  Method Descriptions 	    32
               C.  Experimental Procedures	•	    41
               D.  Results	    44
               E.  Discussion.	    93
               F.  Recommendations	    98

V.         Pilot Study	    99

               A.  Objectives	    99
               B.  Overview	   100
               C.  Data Collection	   100
               D.  Theoretical Precision  	   101
               E.  Data Analysis	   107

VI.        Recommendations for National Survey	   Ill

VII.      References	,	   113

Appendix A - Definition of Terms	   A-l
Appendix B - Glossary of Fuels	   B-l
Appendix C - Sources of Error in Volumetric Test Methods 	   C-l
Appendix D - Temperature Behavior Studies	   D-l
Appendix E - Site Visits to Observe Tests	   E-l
Appendix F - Soil Analysis	   F-l
Appendix G - Facility Descriptions, Tank Characteristics, and Test
               Conditions	   G-l
Appendix H - Development Study Test Data	   H-l
Appendix I - Pilot Study Data	   1-1

                                     iv

-------

-------
                               LIST OF TABLES
Number                                                                Page
   1      Leak Detection Methods Reviewed	     7
   2      Summary of Statistical Analysis of Quantitative Methods
            Tested at Kansas City Service Station Site 	    18
   3      Results Summary from ARCO	    20
   4      Results Summary from Certi-Tec 	    22
   5      Results Summary from Leak Lokator	    25
   6      Results Summary from Petro-Tite	    27
   7      Development Study Test Facilities	    33
   8      Tanks Tested and Their Characteristics 	    34
   9      Matrix of Tests Conducted	    35
  10      Petro-Tite Tank Test Temperature Conditions	    47
  11      Leak Lokator Tank Test Temperature Conditions. .'...'..    48
  12      Summary of Baseline Results and Tank Tests Attempted-. .   .    50
  13      Results o'f Leak Simulation Tests Using ARCO Method ....    54
  14      Sample of Simulated Leak Data for ARCO Test at Damneck .-  .    55
  15      Results of Leak Simulation Tests Using Leak Lokator
            Method	    57
  16      Sample of Simulated Data for Leak Lokator Test at Pitstop
            North	    58
  17      Results of Leak Simulation Tests Using Petro-Tite Method  .    66
  18      Sample of Simulated Leak Data for Petro-Tite Test at Scott
            AFB, Tank 17	    67
  19      Summary of the Leak Lokator Analysis	    80
  20      Summary of the Petro-Tite Analysis	    87
  21      Estimates of the Precision of the Leak Lokator Method. .   .    94

-------
                       LIST OF TABLES (continued)
                                                                                 , *"**
Number
22
23
-
Estimates of the Precision of the Petro-Tite Method. . . .
Summary of the Leak Lokator and Petro-Tite Precision
Results 	
Page
95
96
It*      Mean and Standard Deviation of the Petro-Tite Data
          Analyzed Separately for Up Displacements and for Down
          Displacements	    97
25      Example Spreadsheet Data File	   102
26      Precision of Temperature and Volume Measurements 	   105
27      Analysis of Variance for Interobserver Variability ....   106
28      Summary of Pilot Study Results 	   110
                                    VI

-------
                               LIST OF FIGURES



Number                                                                Page
1

2
3
4
5
6
7
8
9
10

11

12
13

14

15

16

17

18
Precision Testing as Defined by the National Fire Protec-
tion Association, Inc 	
Leak Simulation System Designed and Fabricated by MRI . . .
ARCO Test - Measured Leak Versus Simulated Leak 	
Certi-Tec Test - Measured Leak Versus Simulated Leak. . . .
Leak Lokator Test - Measured Leak Versus Simulated Leak . .
Petro-Tite Test - Measured Leak Versus Simulated Leak . . .
The ARCO Test Apparatus 	
The Leak Lokator Test Apparatus 	
The Petro-Tite Test Apparatus 	
Temperature Profiles Obtained Before and After Selected
Petro-Tite Tests 	
Temperature Profiles Obtained Before Selected Leak Lokator-
Tests 	
Simulated Leak Data for ARCO Test at Damneck 	
Simulated Leak Rate Data for Leak Lokator Test at Pitstop
North 	
Net Leak Rate and Simulated Leak Rate for Leak Lokator
Test at Pitstop North 	
Net Leak Rate Minus Simulated Leak Rate for Leak Lokator
Test at Pitstop North 	
Cumulative Volume Change for Leak Lokator Test at Pitstop
North 	
Simulated Leak Rate Data for Petro-Tite Test at Scott AFB
Tank 17 	
Measured and Simulated Leak Rates 	

5
15
21
23
26
28
36
38
40

45

46
56

59

62

63

64

68
70
  19      Temperature Corrected Measured Leak Rates 	   71

-------
                         LIST OF FIGURES (continued)
Number                                                                Page
20

21
22

23

24
25

26

27
28

29

30

31

32

33

34
35
36
Cumulative Measured Volume Changes and Temperature Related
Cumulative Volume Changes 	
Measured and Simulated Leak Rates 	
Measured Leak Rates for Petro-Tite Test at Fort Lewis
Tank System 10E10 	
Point-by-Point Temperature Compensated Leak Rates and
Temperature Volume Changes 	
Cumulative Temperature Related Volume Changes 	
Cumulative Volume Changes After Removing Smoothed Tempera-
ture Trend 	
Time Series Analysis of the Leak Lokator Data at Pitstop
Tank No. 2 	
Time Series Analysis of the Leak Lokator Data at Damneck .
Time Series Analysis of the Leak Lokator Data at Fort
Lewis Tank No. 2 	
Time Series Analysis of the Leak Lokator Data at Fort
Lewis Tank No. 3 	
Time Series Analysis of the Petro-Tite Data at Damneck
Tank No. 1 	 	
Time Series Analysis of the Petro-Tite Data at Scott AFB
Tank No. 1 	
Time Series Analysis of the Petro-Tite Data at Fort Lewis
Tank No. 4 	
Time Series Analysis of the Petro-Tite Data at Pitstop
Tank No. 2 	
Sample Data Plot 	
Example Data Summary 	
Performance Curve 	

72
73

74

75
77

78

82
83

84

85

88

89

90

91
103
109
112

-------
                              EXECUTIVE SUMMARY


          Leaking underground storage tanks have become a matter of national
concern.   Loss of materials stored in underground storage tanks poses a risk
of contamination of ground water with an attendant potential risk to human
health.   In the event that the material is a flammable liquid such as motor
fuels, leaks from underground storage tanks may pose an additional risk of
fire or explosion.

          The United States Congress recognized this problem and incorporated
a new subtitle in the Hazardous and Solid Waste Amendments of 1984, signed by
President Reagan on November 8, 1984.  As part of the provisions of Subtitle I
of the Resource Conservation and Recovery Act (RCRA), the U.S.  Environmental
Protection Agency (EPA) was mandated to develop regulations for underground
storage tanks (USTs) to reduce the risk of leaks and the resultant risk to
human life and property.   As part of this regulatory process, EPA was man-
dated to conduct a national survey of underground storage tanks used to store
motor fuel.  The objectives of the national survey are:

          1.  To provide estimates of the total number of underground motor
fuel storage tanks.

          2.  To provide estimates of the number of establishments with under-
ground motor fuel storage tanks.

          3.  To provide estimates of the number of such tank systems that
leak-

          4.  To collect information on the characteristics of tanks and tank
establishments.

          A part of the national survey was to include physical tightness test-
ing for a national probability sampling of tanks.  This project provided input
by evaluating and recommending methods of tank testing for use in the national
survey.

          The objectives of this project were:

          1.  To select a method for physical testing of underground storage
tanks used to store motor fuel to determine if they leaked.

          2.  To quantify the performance of the recommended method in terms
of its precision, accuracy, and suitability for use in a national survey.

          3.  To define the standard operating procedure of the recommended
method for use in the national survey, thus ensuring consistent data of high
quality.
                                     ix

-------
          In order to achieve these objectives, several guidelines for judg-
ing methods of testing tanks were developed.   These guidelines related to the
performance of the method and to its suitability for use in the national sur-
vey.  In developing these guidelines, the objectives of the national survey
were explicitly considered.   Methods were judged on their performance in mea-
suring leaks from underground storage tanks.   In addition, they were judged
on their ability to identify and compensate for sources of interference from
temperature changes, tank deformation, water table, etc.   Because of the need
for a large number of tests in a short time for the national survey, avail-
ability of equipment and trained personnel to do the testing was a factor, as
was consideration of the impact and inconvenience to the owner/operator of
the tank.

          In achieving the objectives of this project, a five step approach
was used.  The first phase was a preliminary evaluation of existing methods.
This included both a literature review and field observation of selected methods
during use in testing tanks.

          The second phase consisted of evaluative testing of five selected
methods.  The tests were carried out to measure performance in a controlled
setting.

          The third phase was field evaluation of three potential methods at
five geographically scattered sites.  Performance was checked under a variety
of different environmental and operating conditions in the field.

          The fourth phase, using the data resulting from the previous phases
and comparing performance with the selection guidelines, recommended a method"
for use in the national survey.  The recommendation was then discussed with
EPA.

          The fifth phase of the study was a pilot test of the recommended
method designed to test its performance under the conditions proposed for the
national survey.  This phase checked the field operation for problems and dif-
ficulties.  It also checked the advertised standard operating procedures for
the method to ensure that they were practical and would provide a consistent
set of high quality data on tank test results.  Potential difficulties and
suggested solutions resulting from the pilot study were used to adjust the
standard operating procedures for use in the national survey.

          The principal conclusions of the project are:

          •  After the evaluation of a number of test methods, it is clear
             that none of the standard test methods evaluated in the program
             can consistently and reliably achieve detection of the 0.05 gal/h
             leak rate specified by the NFPA 329, "Recommended Practices for
             Underground Leakage of Flammable and Combustible Liquids, 1983."
             This conclusion  is based entirely on the data collected during
             the method selection phase of the program since supporting data
             which had been requested from the test companies to document
             their performance claims were not received.

-------
               •  While some methods can provide reliable results under some spe-
                  cific conditions, there are many situations which commonly occur
                  in the test environment (such as the presence of a water table)
                  which can invalidate the test results.   Unless the test crew
                  takes specific steps to identify these conditions, the reported
                  results may be either misleading or incorrect.  Most test meth-
                  odologies currently in use fail to take definitive steps to
                  identify one or more of these problem areas.   The test results
                  obtained must therefore be suspect to the degree that these
                  factors have not been recognized.   The effects of not identify-
                  ing these conditions can be substantial and,  in general, cannot
                  be evaluated from the test data after the test crew has left the
                  test site.

               •  Of the methods evaluated during this program, the Petro-Tite
                  method was selected as offering the most promising approaches to
                  identifying potential problems and taking action to correct them.
                  It has stated procedures to identify and correct for tank end
                  deflection, temperature effects, water table, and vapor pockets.
                  It was also judged to provide the most consistently reliable data
                  overall.   On the other hand, the precision of this test method
                  was judged to be inadequate.  The Petro-Tite method, which uses
                  only four 15-min data points would have a probability of only
                  28% of detecting a 0.1 gal/h loss of product if the probability
                  of false alarm is held fixed at 5% (two-sided).  Better preci-
                  sion was required for the national survey.

               •  To improve the precision of the test results and our ability to
                  identify potential interference problems, the Petro-Tite method
                  was modified.  The performance of this modified Petro-Tite method
                  was then characterized as part of a pilot study.   The method was
                  modified by increasing the frequency method and applying specific
                  statistical tests to the test data.  Based on the performance of
                  the modified method in the pilot study, the precision was improved
                  so that an average probability of detecting a leak of 0.1 gal/h
                  was estimated to be 95%.  In other words, the modified Petro-Tite
                  testing method combined with the statistical  analysis of the test
                  data has a 5% change of erroneously classifying a sound tank as
                  unsound.   On the average, it has a probability of 95% of detect-
                  ing a product loss of 0.1 gal/h or larger.

               This modified Petro-Tite method was recommended for use in the na-
     tional survey.
J
                                          XI

-------

-------
I.   INTRODUCTION

          Leaking underground storage tanks have become a matter of national
concern.   Loss of materials stored in underground storage tanks poses a risk
of contamination of ground water with an attendant potential risk to human
health.   In the event that the material  is a flammable liquid such as motor
fuels, leaks from underground storage tanks may pose an additional risk of
fire or explosion.

          The United States Congress recognized this problem and incorporated
a new subtitle in the Hazardous and Solid Waste Amendments of 1984, signed by
President Reagan on November 8, 1984.  As part of the provisions of Subtitle I
of the Resource Conservation and Recovery Act (RCRA), the U.S.  Environmental
Protection Agency (EPA) was mandated to  develop regulations for underground
storage tanks (USTs) to reduce the risk of leaks and the resultant risk to
human life and property.   As part of this regulatory process, EPA was mandated
to conduct a national survey of underground storage tanks used to store motor
fuel.

          The objectives of the national survey are:

          1.  To provide estimates of the total number of underground motor
fuel storage tanks.

          2.  To provide estimates of the number of establishments with under-
ground motor fuel storage tanks.

          3.  To provide estimates of the number of such tank systems that
leak.
          4.  To collect information on  the characteristics of tanks and tank
establishments.

          The national survey was to include physical tightness testing for a
national  probability sampling of tanks.   This project provided input to the
national  survey by evaluating and recommending methods of tank testing.

          The objectives of this project were:

          1.  To select a method for physical testing of underground storage
tanks used to store motor fuel to determine if they leaked.

          2.  To quantify the performance of the recommended method in terms
of its precision, accuracy, and suitability for use in a national survey.

          3.  To define the standard operating procedure of the recommended
method for use in the national survey to ensure consistent data of high qual-
ity.

-------
          In order to achieve these objectives, several guidelines for judg-
ing methods of testing tanks were developed.  These guidelines related to the
performance of the method and to its suitability for use in the national sur-
vey.  In developing these guidelines, the objectives of the national survey
were explicitly considered.

          The specific quideliries that were developed for use in the selec-
tion process are presented in Section II, Preliminary Investigation.  These
guidelines include consideration of method performance, the means that each
method uses to account for sources of interference to the measurement process
(thermal expansion effects,  tank deformation, water table, etc.), the imposi-
tion on the owner/operator caused by the method, and the availability of
equipment and trained personnel to perform the testing needed in the national
survey.

          In this report, the commonly used term "underground storage tanks"
refers to the tank system, which includes the large vessel and all attendant
plumbing, fill tubes, distribution lines, vent pipes, and other appurtenances.
Thus, the tests evaluated measure the integrity of the system and not just
the vessel itself.  For regulatory purposes, the definition of an underground
tank is one which has 10% of its volume (the vessel together with its associ-
ated piping) buried in soil  or backfill.  The usual case is that the tank and
most of its plumbing are buried completely, but the key point is that the sys-
tem includes both the tank and its piping.

     •     There are also cases where several vessels are included in a mani-
fold system of- which there are several types.  One important type is a siphon
manifold in which two or more tanks are connected by pipes forming a siphon
that allows fuel to flow freely from one to the other, thus maintaining the
same level in the connected tanks.  Such systems may have a single fill pipe
and delivery system.  Another type of manifolding is that used when tanks are
not connected by a siphon but are joined to a common delivery line.

          A summary of some of the. terms used in this report is provided in
Appendix A.

          The methods evaluated are for testing tanks used to store motor fuel
only.  A glossary of fuel names and information about products that were in-
cluded in the evaluation are found in Appendix B.

          English units are used throughout this document rather than metric
units because commercial testers measure temperature in degrees Fahrenheit
and volumes in gallons.  In addition, tank owners, station operators, and
others who will use this document operate almost exclusively in English units.

          This project was conducted in five phases.  Subsequent sections of
the report discuss each phase  in detail.  Section II discusses the preliminary
investigations of test methods.  Section  III discusses the design, testing,
and results from preliminary testing of methods in a controlled setting.  Sec-
tion IV presents the experimental approach and results from a field trial of
potential methods used to test at a variety of geographic locations and the
method recommended  for use on  the national  survey.  Section V describes the

-------
results of a pilot test of the method recommended for national testing.  Sec-
tion VI is the final  set of recommendations for the national survey test pro-
cedures.   The appendices contain supporting information either too detailed
for inclusion or that is peripheral to the purpose of the project.

          After the evaluation of a number of test methods (see Appendix C),
it is clear that none of the standard test methods evaluated in the program
can consistently and reliably achieve detection of the 0.05 gal/h leak rate
specified by the NFPA 329, "Recommended Practices for Underground Leakage of
Flammable and Combustible Liquids, 1983."  This conclusion is based entirely
on the data collected during the method selection phase of the program since
supporting data which had been requested from the test companies to document
their performance claims were not received.

          While some methods can provide reliable results under some specific
conditions, there are many situations which commonly occur in the test environ-
ment (such as the presence of a water table) which can invalidate the test
results.   Unless the test crew takes specific steps to identify these condi-
tions, the reported results may be either misleading or incorrect.  Most test
methodologies currently in use fail to take definitive steps to identify one
or more of these problem areas.  The test results obtained must therefore be
suspect to the degree that these factors have not been recognized.  The ef-
fects of not identifying these conditions can be substantial and, in general,
cannot be evaluated from the test data after the test crew has left the test
site.

          One of the major objectives of the activities carried out prior to
the national survey was to identify and characterize a test method suitable
for use on the program.  The Petrp-Tite method was selected as offering the
most promising approaches to identifying potential problems and taking action
to correct them.  Further, it was judged to provide the most consistently re-
liable data for the national survey.  It has stated procedures to'identify
and correct for tank end deflection, temperature effects, water table, and
vapor pockets.  The test time and sample frequency were increased in order to
improve the precision of the test results and to assist in identifying problem
areas when present.  These modifications were then characterized during the
pilot study phase of the project and a test protocol using the modified tech-
nique was developed.   This modified Petro-Tite method was then used on the
national  survey.


II.  PRELIMINARY INVESTIGATION

     A.  Overview

          Preliminary investigation into the status of tank testing tech-
nology revealed at least 20 methods available for the testing of underground
storage tanks.  Since both time and funds precluded a thorough evaluation of
all these methods, it was necessary to develop a screening process to reduce
the number of methods to a manageable level.  This screening process involved
several steps.

-------
          First, a set of general selection guidelines was developed.  Second,
the methods under consideration were classified into their basic leak detection
characteristics and several were selected for further consideration.  Third,
field observation visits were conducted at seven sites to observe a test demon-
stration.   Finally, five methods were selected for field testing under con-
trolled conditions.

     B.  General Selection Guidelines

          A set of general selection guidelines was developed so that methods
which clearly could not provide data which would meet the national survey
requirements could be eliminated from further consideration.  These guide-
lines were intended to be somewhat flexible so that even though a particular
method might not meet all of the criteria, if potential advantages existed
it could still be considered.

          The most basic requirements for selecting a method suitable for
testing underground fuel storage tanks are those related to sensitivity, ac-
curacy, and precision.  However, the practical considerations of the cost of
a national survey and the facilitation of access to private tank systems im-
pose several additional requirements.  The general method selection criteria
used in the preliminary investigation are described below.

          1.  Quantitative Measurement

          Because one of the objectives of the national survey was to provide
an estimate of the amount of fuels leaking from USTs, the method had to pro-
vide reliable quantitative determination of leak rates for each system tested.
This requirement could have been waived, however, if qualitative methods were
found  to be significantly more reliable at detecting very small leak rates.

          2.  Detection Levels

          The National Fire Protection Association (NFPA 1983) has established
the criteria  for precision testing shown in Figure 1.  For purposes of this
program, 0.05 gal/h was taken as the reference point for comparison of various
volumetric methods.   However, as described in subsequent sections of this re-
port,  highly  reliable detection  of 0.05 gal/h leaks using existing test pro-
cedures may not be achievable for many test methods and for some tank systems.
The result of the  evaluations conducted in this study was to provide docu-
mented performance characteristics for each of the methods tested.

          3.   Invasiveness

          The survey  method  should cause a minimum of disruption to the oper-
ation  of the  tank  system.  This  criterion was especially  important to mini-
mize imposition on the owner/operator and to  encourage cooperation with the
survey.
                                                                                   »*.
                                                                                   m

-------
    4-3.10  Precision Test.
    4-3.10.1  Precision Tesc, as used throughout this pamphlet, means
    any tesc that takes into consideration the temperature coefficient of
    expansion of the product being tested as related to any cempcrature
    change during the test, and is capable of detecting a loss of 0.05 gal
    (190 ml) per hour.


    4-3.10.2  A test should be used which  is chosen from  currently
    available technology to reasonably determine whether or not an un-
    derground liquid storage and handling system is leaking. Any testing-
    device used for the Precision Test must be capable of detecting leaks
    as small as  0.05 gal (190 ml) in one hour, adjusted for variables,  a
    limiting criterion widely accepted by most authorities.


    4-3.10.3  The test procedure should measure the amount of liquid
    lost based upon fundamentally sound principles. It should detect  a
    leak anywhere in the complete underground storage and  handling
    equipment. If the net change exceeds 0.05 gal (190 ml) per hour or
    equivalent criterion established for the technology employed, a leak
    is likely to exist, and appropriate corrective action is necessary.


    4-3.10.4  The Precision Test  should account for  all the  variables
    which will affect the determination of the leak rate. An understand-
    ing of what these variables are  and how  they are handled is essential
    to effective performance of the  test. Following is a discussion of some
    of those variables and how  they affect the measurement.
Figure  1.   Precision  testing  as defined  by the National
Fire Protection Association,  Inc.   Source:    NFPA  1983.

-------
          4.   Physical  Reliability

          The test equipment and procedure should be reasonably rugged and
reliable to ensure successful  completion of all testing required in the sur-
vey.

          5.   Versatility

          The test should be applicable to a wide variety of tank system con-
figurations so that a single method could be used to conduct the entire survey.
Since the national survey was intended to include nearly all sizes of service
stations, fleet and other commercial refueling installations, and farms, it
was likely that numerous configurations would be encountered.  The test method
should also be applicable to most test situations (e.g., high water tables).

          6.   Amenability to Performance Monitoring

          The test procedure should allow monitoring and assessment of accu-
racy, precision, and sensitivity.   This method performance information was
required to provide estimates of the confidence of the survey results.

          7.   Cost

          The cost of conducting the test and interpreting the test results
must fit the budget available for the national survey program.

          8.   Availability of Equipment and Manpower
                                             •
          The ability to conduct the nationwide survey on a timely and cost
effective basis was considered to be important.  Methods for which sufficient
equipment and trained manpower were not available to support the national
survey effort were excluded from serious consideration.

          Each of these guidelines was important to the selection of a test
procedure for the national survey.  Since the various methods each operate
somewhat differently, they will exhibit different strengths and weaknesses.
No one method provides optimum performance for all criteria.  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 pro-
cedure with optimum characteristics for all criteria.  Hence, some degree of
compromise was necessary to proceed expeditiously with the survey.

     C.  Identification and Classification of Test Methods

          The initial literature search identified a number of methods, listed
in Table 1, which had been used to test tanks.  The review was not exhaustive
and drew heavily on presentations at the PACE Symposium (PACE 1982) and the
Petroleum Equipment Institute Symposium (PEI 1983) as well as on literature
obtained from various test groups.  A discussion of each of these methods
(and some additional ones) has been compiled in a report on underground tank
leak detection methods which was prepared for EPA by the IT Corporation (EPA
1986).  While these descriptions are based almost entirely on the suppliers'

-------
                  Table 1.   Leak Detection Methods Reviewed
Detection method
Volumetric
ARCO
Certi-Tec
Ethyl tank sentry
Ezy-Chek
Petro-Tite tank and line
testing system
Hydrostatic (standpipe)
testing
Lasar interferometry
Leak Lokator
Mooney tank leak detector
Pald-2 leak detector
Pneumatic testing (air
test method)
Non-Volumetric
Dye method
Vacutect method
Helium leak detection
Liter-
ature
review

X
X
X
X
X

X

X
X
X
X
X


X
X
X
Field
site
visits

X
X


X




X







X
Prelim- Devel-
inary opment Pilot
testing study study

X X
X


XXX




X X







X
  ,  method
  Tracer Research test          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

-------
information rather than on objective,  independent test data, the report pro-
vides a useful overview of the tank testing methodology available at the time.

          The methods were classified according to their major operational
characteristics.   The four categories considered were volumetric (quantita-
tive), nonvolumetric (qualitative), inventory monitoring and reconciliation,
and external  monitoring.   Each of these categories is briefly described be-
low.

          1.   Volumetric Leak Tests

          Volumetric leak tests are based on the principle that the change in
the volume of fluid in the tank due to a leak can be determined by measuring
properties associated with that change in volume.  The usual procedure in a
volumetric leak test is to measure the product level and product temperature
in the tank over a period of time.  The level changes are then corrected for
the expansion or contraction of the product due to temperature changes.  This
gives a temperature compensated volume change over the period of the test.
The leak rate can then be calculated.   The product level is usually measured
directly, but other sensors such as transducers which measure changes in pres-
sure at the tank bottom can also be used.  The major sources of variability
which can affect the test outcome and are common to nearly all the volumetric
methods are summarized below.

          •  Distortions in the tank may occur as the level of product in the
             tank is changed.  Time for stabilization must be provided each
             time the level is changed significantly.  These distortions may.
          '  be affected by the type of backfill material used around the
             tank and the presence of a water table.  These factors are un-
             controllable and will vary widely from tank to tank.

             Temperature changes will cause thermal expansion, or contraction
             of the tank as well as of its contents.  Only small temperature
             changes (on the order of 0.001°F/h  for large tanks) are needed
             to offset or amplify the indicated  leak rate.  These effects can
             be corrected for i f careful temperature measurements are ob-
             tained during the test.

           •  Poekets of air or vapor trapped in  the tank or piping can cause
             misleading results since the volumetric expansion of gases is
             much greater than that for  liquids.  This problem is particularly
             acute for those test methods which  require the tank to be over-
             filled.

           •  Vibrations from external sources, such as traffic, can have  ad-
             verse effects.

           •  Evaporation and condensation may occur in the tank particularly
             on hot, windy days and in partially filled tanks.  These  factors
             can be controlled or minimized  by proper design of the equipment
             and test procedures.
                                    8
                                                                                   *
                                                                                   ^

-------
          Some or all  of these factors may be operating in any given test.
Their effects and the ability of a given test method to compensate for them
are often unknown because essential data may be unavailable.   Failure to
recognize and compensate for these factors can lead to either false positive
or false negative results.   A discussion of these sources of error is pro-
vided in Appendix C.

          Since temperature variation and compensation are the most important
variables, additional  data were collected.  A discussion of these data is pro-
vided in Section IV.D.I, and temperature behavior studies are included in Ap-
pendix D.

          Other factors which can affect the sensitivity of a method include
the size of the tank,  the time available to monitor changes,  the time elapsed
since the last addition of product to the tank, and the presence of an under-
ground water table in contact with the tank.
                                     »
          2.   Nonvolumetric Leak Tests

          Several nonquantitative'methods are available for qualitatively de-
termining if leaks are present in underground storage tanks.   These techniques
may be useful for identifying tanks that are leaking, after which the more
thorough volumetric tests can be applied.

          These techniques rely on measurement of a parameter which is not
directly associated with product loss, but rather indicate the presence of a
hole.  These can include the detection of sound patterns associated with bub-
bles which are formed when a slight vacuum is placed on the tank or with the
presence of helium in the immediate environment around the tank.  In some
cases, the size of the hole and the leak rate can be estimated.

          3.   Inventory Monitoring and Reconciliation

               a.  Manual Record Keeping and Reconciliation

               Careful product monitoring can be used as a first step in de-
tecting the existence of a leak.  Early detection of leaks can sometimes be
achieved by proper product inventory, regular inspection of the visible parts
of the product handling system, and prompt recognition of the conditions in-
dicating that leaks may be occurring in the underground tank or piping.

               This method of leak detection requires good bookkeeping,  is an
indirect measure of leaks, and when properly applied, can provide other in-
formation useful to the tank owner.  Some of the more obvious situations which
can be identified are:

               •  Loss of product in a tank system.   For example, loss of
                  product when the product is not being dispensed usually in-
                  dicates a leaking tank, but it may also indicate faulty ac-
                  counting or metering of the product, extreme temperature
                  change, or theft.

-------
               •   An apparent increase in product level.   Increases in the
                  product level  in an underground tank may be caused by water
                  leaking into the tank if the ground surrounding it is sat-
                  urated with water.   The increase may also be caused by a
                  leaking gauge or fill cap, and these should be examined and
                  made watertight, if necessary, before concluding that the
                  tank is faulty.

               •   Increasing differences between the amount of product received
                  and the amount dispensed.   This may indicate a meter cali-
                  bration problem, theft, or a leak in tanks or piping.

               •   Large differences appearing consistently between the amounts
                  invoiced and the tank gauges after deliveries where dispensers
                  are located remotely from the tanks.  This may indicate a
                  leak in the remote dispenser line.  In such an event, the
                  line should be tested.

               A number of factors can make the inventory method uncertain.
For example, changes in product temperature can lead to expansion, contrac-
tion, evaporation, and condensation of the stored product.  Errors can also
occur through faulty readings of dipstick measurements.  Mathematical errors
may also occur which can lead to incorrect conclusions.

               b.   Automated Methods

               There are a number of automated continuous monitors which
record the level  and temperature in the tank at regular intervals.  These are
usually microprocessor controlled and can be programmed to perform all of the
normal inventory record keeping functions.  The measurement methods are varied.
While these methods have improved, their general level of sensitivity is
usually around 0.2 gal/h.

          4.  External Monitoring

          External monitoring is keyed to the effects produced by hydrocarbon
products which have escaped and can give an indirect indication of the pre-
sence of a leak.   Background concentrations may also result from spills, over-
filling, and rem'ote releases from other tanks.  In many cases, it is difficult
to determine the exact source of the release.  These methods do not provide
any estimate of the leak rate.

          Methods from each category were considered in light of the guide-
lines, and those felt to represent the best potential for use on the national
survey program were identified.  There  is little doubt that some very good
methods were eliminated from further consideration at this stage.
                                    10

-------
     0.   Field Observation Visits

          After all of the methods listed in Table 1 had been reviewed, the
following methods and companies were selected by MRI and EPA personnel for
observation in the field.
                 Method

          ARCO HTC

          Certi-Tec


          Leak Lokator


          Petro-Tite

          Helium leak detector

          Core drilling


          Underground radar
            Company

Atlantic Richfield, Harvey, IL

Fuel Recovery Company, Inc.,
  St. Paul, MN

Hunter Environmental, Inc.,
  Malvern, PA

O.H. Materials, -Find!ay, OH

Smith and Dennison, Hayward, CA

National Institute for Petroleum and
  Energy Research, Bartlesville, OK

Ohio State University, Columbus, OH
Accordingly, visits were arranged for MRI and EPA personnel to observe the
methods in actual use.   The site visits are described in detail in Appendix E.

          Since one of the major objectives was to obtain quantitative re-
sults from the testing, four of-the methods selected were volumetric.  Most
volumetric methods rely on measurement of product temperature and level to
detect leaks.   The four volumetric methods selected have somewhat different
hardware configurations and approaches to temperature and level measurements.

          The methods observed are described briefly below.  They were eval-
uated for suitability for use on the national survey using the general selec-
tion criteria.

          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.

          The Certi-Tec method uses pressure transducers which are located
just below the surface of the liquid to measure level changes.  Seven ther-
mistors 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.
                                    11

-------
          Leak Lokator uses a buoyancy probe to monitor product 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.

          The Petro-Tite' method monitors product level  visually in an extended
fill pipe.  The product level is returned to the reference level at 15-min
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.

          The Varian helium leak detection method, a nonvolumetric method, is
based on the detection of helium outside a tank which has been slightly pres-
surized 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 tested at the same time.

          Core drilling and underground radar are two external monitoring ap-
proaches which were considered.  Core drilling involves drilling a number of
wells in the area surrounding the tank.  The core samples obtained are an-
alyzed for the presence of hydrocarbon using a gas chromatograph or other hy-
drocarbon sensor.  A number of wells may need to be drilled depending on the
hydrogeological conditions in the area.  No estimate of the rate is possible.

          Underground radar depends on th'e detection of a radar signal which
is reflected from differing objects and conditions under the ground.  The
method has been used to detect leaks in water lines, but has not been used as
a hydrocarbon leak detector.  The main advantages are that it is nondisruptive,
little or no site preparation is needed, and the facility can continue to op-
erate during testing.

          The field observation visits were somewhat cursory in nature.  No
data were obtained for analysis; and  in several cases only laboratory type
demonstrations were provided, which were inadequate for reliable evaluation
of the method in actual field use.  Nonetheless, none of the volumetric
methods appeared to provide any clear advantages or disadvantages for use on
the national survey.  Since no data were obtained under controlled conditions,
the actual performance characteristics could not be determined.  It was de-
cided to  evaluate all four of the volumetric methods under controlled field
conditions.

          The helium  leak detector was judged to have potential for  use on
the national survey partially because the personnel at Smith and Oennison
felt that it might be possible to obtain volumetric estimates by using pres-
sure sensing equipment they  had developed in conjunction with the helium leak
detector.  The method demonstrated some potential for providing quantitative
results,  although there were some problems with the pressure measurement part
of  the test.  The method  was observed to be extremely sensitive and  did lo-
cate what appeared  to be  leaks  in both the lines and the tank,  so it was de-
cided to  evaluate this test  method further.

                                    12

-------
          Although core drilling and underground radar both provide useful
information about leaking tanks, both were eliminated due to their limita-
tions.   Two of the major limitations of core drilling were considered to be
its inability to provide leak rates and the ambiguity of the results in
identifying the specific source of the hydrocarbons.   Since one of the prim-
ary objectives was to estimate the number of leaking tanks, the latter limi-
tation was considered particularly serious.  The state of development of
underground radar for use as a hydrocarbon leak detector precluded its use in
the near term.  The interpretation of the test results is relatively complex
and the availability of test equipment is very limited.

          Since inventory reconciliation was to be evaluated in another part
of the national survey, these methods were not considered for use as leak de-
tectors on this project.  One automated unit manufactured by Veeder-Root was
observed during the site visits, however.


III.   PRELIMINARY TESTING

          The primary objectives of the preliminary testing were as follows:

          1.   To develop the evaluation procedure to be used on later phases
of the program.  The leak simulation procedures were of particular importance.

          2.   To evaluate the ability of each test method to track a series
of simulated leaks using normal test procedures.  . (accuracy)

          3.   To evaluate the ability of each method to obtain repeatable re-
sults when given the same leak rate,  (precision)

          4.   To screen out any methods which would not be suitable for use
on the national survey.  The general selection guidelines were used for this
purpose.

          Preliminary testing was performed at a Kansas City, MO, service
station that was no longer in operation but still had underground storage
tanks.   The subject tank was constructed of steel and had a nominal capacity
of 7,400 gal.   The tank was filled with regular gasoline for the testing.
The product remained generally undisturbed in the tank for several weeks dur-
ing the testing.  Some product was added or removed from the tank to meet the
requirements of the particular test method.  Testing was done during July and
August, making it more likely that the ambient temperatures would produce
thermal stratification in the tank for those tests that did not stir the
product.  External interferences with the level in the tank were generally
minimal, however, probably less than would usually be encountered at a sta-
tion in normal operation.  Because all the methods tested found that the tank
did not leak, it was concluded that the test unit was a tight tank.
                                    13

-------
     A.   Experimental  Procedure

          !•   Evaluation Design

          Each method was evaluated by testing over a 2- to 3-day period.
During the evaluations, each test method was operated using normal procedures.
For the quantitative methods, leaks were simulated using a leak simulation
system designed and fabricated by MRI.  Four different leak rates were simu-
lated in random order for each test method.   The simulated leak rates were
not disclosed to the test operators until all tests were completed to avoid
operator bias.  Each simulated leak rate was run for the normal length of a
standard test.  At the conclusion of the test, when the test operator indi-
cated sufficient data had been collected, a new leak rate was established.
In a few cases for tests where water leaks into the tank would be possible
under conditions of high water table (ARCO and Leak Lokator), fuel was pumped
into the tank to simulate a leak-in condition.

          2.   Leak Simulation Procedure

               a.  Description

               The leak simulation system, shown in Figure 2, consists of a
peristaltic pump, flow meter, and a collection bottle.  Product was withdrawn
from the tank system at a controlled rate with the peristaltic pump.  The flow
meter was used to adjust the pump to the selected nominal simulated leak rate.
The actual rate was determined at the completion of the test by measuring the
product collected in the bottle.  During the preliminary method testing con-
ducted at the service station site, the withdrawn product was measured volu-
metrically using a graduated cylinder.  However, for the subsequent develop-
ment study testing, the withdrawn product was measured gravimetrically.  The
collection bottles were tared before the test and were reweighed at the con-
clusion of the test.  The volume, corrected to 60°F, was calculated from the
total mass and the fuel density.

               b.  Precision

               The precision of the simulated leak rates can be calculated
from the precision of the measurements used to calculate the simulated leak
rates.  In the preliminary testing, the leak simulations were measured by
collecting the product and measuring  its volume in a graduated cylinder.  The
precision of  this measurement was ±0.5 mL.  The time period used varied, but
was the same  as  the testing method used.  In no case was the time less than 5
min.  Generally, the minimum volume measured was 10 mL.

               One can assume that volume is measured to ± 0.5 mL and that
any error is  a rounding error, with the uniform distribution.  The variance
of a uniform  distribution with range  R is R2/12.  The 0.5 mL is converted to
gal to express it in the same units as the product is_usually measured in.
Using the formula gives a variance of v = 5.8156 x 10 9 gal squared, or  a
standard deviation of 7.626 x 10"5 gal.
                                    14

-------
-Q


T5
 s-
JD
 c
 to
 0)
 C
 01

 1/1
 II
TJ
 (ft

 >>
 l/l
 C
 o
ITS


E
0)
Ol

3
O3

-------
               For the simulations conducted in the development study phase,
mass instead of volume was used to measure the amount of product removed from
the tank during leak simulations..   The mass determinations were made to ± 0.1
g, which converts to 0.14 mL of gasoline.   With this more precise method of
determining the amount of product removed during a test period, the variance
is reduced to 4.556 x 10 10 gal squared, and the corresponding standard devia-
tion is 2.135 x 10"5 gal.

               The data from each of the tests were analyzed statistically to
determine the precision and accuracy of each method.  An estimate of the ac-
curacy (or bias) was determined by obtaining the average difference between
the leak rates reported and the actual simulated leak rates.  A paired t-test
was used to test the hypothesis that the method was unbiased; that is, that
the mean signed difference was zero.

               A linear regression of the observed leak rate on the simulated
leak rate was calculated.  The scatter of the data around the regression line
was used to estimate the precision of the test.  A test which is operating
properly is expected to have a slope of one; that is, there is a one-to-one
correspondence between the observed leak rate and the simulated rate.  A slope
differing significantly from one indicates the presence of some interfering
factor which is operating during the testing.  Plots of the data for each
method have been provided in subsection B, Results, which follows.

               The bias and precision were then combined to obtain an estimate
of the root mean squared (RMS) error, an estimate of the overall performance
of the test method.                                                    ,

               c.  Comparison with Volumetric Method Precision

               To evaluate the leak simulation results, it is helpful to as-
sess the relative accuracies of the leak simulations and the tank.testing
methods.  A general calculation of the variance associated with a typical tank
test has been provided for this purpose.  The comparison is based on worst
case conditions.

               The volumetric tank tests are most precise when used on small
tanks.  The smallest tank of interest in the national survey is 250 gal.  This
volume is used to calculate the precision associated with a tank test for com-
parison with the precision of the leak simulation derived above.

               There are two measurements made for a volumetric test.  One of
these is temperature, the other is volume change.  Both are relative to a
given time period, which is taken to be the  same as that used for the leak
simulation measurement of volume, i-e., no  less than 5 min.

               Temperature measurements are  not usually more precise than
± 0.001°F.  A temperature change  (or rounding  error) is converted to an equiv-
alent volume by  multiplying by the  coefficient of thermal expansion and then
by the volume of the tank.  The product considered was gasoline with a thermal
                                    16

-------
expansion coefficient of approximately 0.00062/gal and /°F.  Thus, 0.001°F
converts to volume as

                    Vr = 0.001 x 0.00062 x 250 = 1.55 x 10"4 gal
                     (3
for gasoline.   Again, a uniform distribution is assumed for rounding errors
in measuring temperature to calculate the variances in these determinations
as

                    v2G = 8.008 x 10"9 gal squared

for gasoline.   The corresponding standard deviation is 2.685 x 10 4 gal for
gasoline.

               The level is measured in a graduated cylinder and is reported
to the nearest 0.001 gal.  Again, assuming the uniform distribution for the
rounding error, the variance for the volume determination is calculated as

                     o    = (2 x 0.001)2
                    v vol        12

                        = 3.337 x 10"7 gal squared

This corresponds to a standard deviation of 5.77 x 10 4 gal.  The total vari-
ance for a determination of a temperature corrected volume for a single period
(5 min) is the sum of the variance of the volume and temperature measurements.
This becomes

                    v = 3.413 x 10"7 gal squared

The corresponding standard deviation is 5.84 x 10 4 gal

               Comparison of these values with the standard deviation for the
leak simulation volume determination, namely 2.135 x 10 s gal, shows that the
error in the leak simulations is an order of magnitude smaller than the cal-
culated error in the testing measure.  In addition, comparison of the errors
calculated for the leak simulation with the empirical RMS values for each
method indicates that under actual use conditions the simulations are three
orders of magnitude smaller.  Thus, the error in the leak simulation volume
rates can be ignored.

     B.  Results

          A summary of the statistical analysis for the quantitative methods
as a group is presented in Table 2.  Following that table is a brief discus-
sion of the statistical results of each method, accompanied by its respective
data table and plot of the data.
                                    17

-------
      Table 2.   Summary of Statistical  Analysis of Quantitative Methods
                 Tested at Kansas City  Service Station Site
Method
ARCO
Certi-Tec
Leak
Lokator
Petro-Tite
na
15
12
22
18
Bias
(gal/h)
-0.01
+0.25
+0.01
+0.05
Intercept
(gal/h)
0.005
-0.30
-0.01
0.06
Slope
0.95
0.71
0.94
1.05
Standard error
(gal/h)
0.049
0.166
0.020
0.101
RMS
(gal/h)
0.050
0.302
0.021
0.113
R*
94.3
38.9
98.9
75.9
Number of simulated leaks.
                                    18

-------
                1.  ARCO Method

                     The ARCO method was tested using 15 simulated leak rates, in-
      cluding one zero rate.  An average difference of 0.01 gal/h was observed be-  .
      tween the rates reported by ARCO and those calculated by MRI.  This estimated
      bias in the results was not significantly different from zero (t = 0.21, 14
      degrees of freedom).  The estimated regression of the observed leak rates on
      the simulated leak rates was

                     Y - 0.005 + 0.943 X,

      where Y is the leak rate reported by ARCO and X is the leak rate simulated by
      MRI.  The ideal regression equation (in a tight tank) would be

                     Y = 0.0 + 1.0 X.

      With the ARCO method, the "intercept did not differ significantly from zero
      and the slope did not differ significantly from one.   The data are displayed
      in Table 3 and plotted in Figure 3.  The R2 for the regression was 94.3%, in-
      dicating that most of the variability of the data was explained by the regres-
      sion.  The RMS error estimated for the method under the conditions of the
      Kansas City test was 0.05 gal/h.  This is the estimated variability associated
      with a single test.  In this case, each test averaged just under an hour (55.7
      min) in duration.   In order to reduce the variability estimated with the method,
 ,    either repeated determinations or a longer test time would be needed.   Repeated
      determinations of a leak rate under the same conditions would reduce the var-
 }     iance by a factor of 1/n, where n is the number -of determinations.  It is not
"     clear how much variability would be reduced if the test time were increased
      for a single reported leak rate.

                2.  Certi-Tec Method

                The Certi-Tec method was tested using 12 simulated leak rates, of
      which two were set at zero and so represented the normal condition of a tank
      test.  The results of the Certi-Tec test evaluation are presented in Table 4
      and displayed graphically in Figure 4.  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.  That is,  the reported leak rates
      were estimated to be too large (too much product leaking) by about a quarter
      a gallon per hour.  This bias was quite large and was significantly different
      from zero (t = -5.23, 11 degrees of freedom).   The estimated regression of
      the leak rates reported by the method on the simulated leak rates was

                     Y = -0.30 + 0.71 X,

      where Y is the leak rate reported by Certi-Tec and X is the leak rate simu-
      lated by MRI.  This differs substantially from ideal.  In fact, the intercept
      differs from zero at the 5% significance level and the slope differs from one
      at the 5% significance level as well.  The standard error of the regression
.      was 0.167 gal/h.  The R2 of the regression was only 38.9%, indicating that
 ;     slightly less than 40% of the variability in the reported leak rates was ex-
      plained by the simulated leak rates used in the test.

                                          19

-------
                     Table 3.   Results Summary from ARCO
ARCO
leak, rate
(gal/h)
-0.140
-0.080
-0.460
-0.070
-0.210
-0.100
0.040
-0.050
-0.130
-0.250
0.210
-0.060
-0.490
0.240
-0.030

dTime
(rain)
85.0
90.0
60.0
77.0
45.0
69.0
37.0
61.0
54.0
60 .,0
29.0
60.0
32.0
47.0
30.0
MRI test
d Volume
(ml)
-660.0
-485.0
-1,895.0
-635.0
-625.0
-390.0
0.0
-215.0
-115.0
-785.0
370.0
-150.0
-1,150.0
430.0
-41.0

Leak rate
(gal/h)
-0.123
-0.085
-0.501
-0.131
-0.220
-0.090
0.000
-0.056
-0.034
-0.207
0.202
-0.040
-0.570
0.145
-0.022
Bias • -0.010 gal/h
RMS   '0.050 gal/h

Regression   Y = -0.005

R-Squared = 0-943

SE(B0) s O-O^8
  Test (Bias - 0)


0.952 (Sim)
                                                     t = -.21 Non-significant
                                          (SE) = °'°49
                           SE(Bt) • 0-0647
95% Confidence interval for B0
95% Prediction interval at 0
                                 (-0.028, 0.036)
                                 (-0.107, 0.115)
                                   20

-------




*


n
n

B__








i i i i 1
T ^t rO CN t- c
5 6666






•



n
a n
a
On
c n •


°D




i i i i
o* 6 6 6 c
m *-*
0 £
(O
0)

-------
                   Table 4.   Results  Summary from Certi-Tec
Certi-Tec
leak rate
(gal/h)
-0.446
-0.687
-0.506
-0.255
-0.428
-0.350
-0.883
-0.468
-0.188
-0.222
-0.318
-0.276

dTime
(min)
41.0
82.0
90.0
85.0
91.0
90.0
60.0
53 .,0
60,0
60.0
30.0
30.0
MRI test
dVol ume
(ml)
-1,345.0
-710.0
0.0
-309.0
-1,180.0
-240.0
-1,930.0
-760.0
0.0
-195.0
-270.0
-210.0

Leak rate
(gal/h)
-0.520
-0.137
0.000
-0.058
-0.206
-0.042
-0.510
-0.227
0.000
-0.052
-0.143
-0.111
Bias  -0.251 gal/h
RMS    0.301 gal/h
Test (Bias - 0)
t = -5.23
Regression   v<- -0.300 + 0.710 (S1m)

R-Squared = 0.389          Standard error (SE) = 0.166

SE(B0)= 0.067    '         SE(Bl)- 0.282

95% Confidence interval for B0   £0.450. -0.10)
95% Prediction interval at 0     ( 0.70U,
                                    22

-------
                                                                                               in

                                                                                               rO
                                                                                               
-------
          Thus, this method, as implemented during this test, appears to have
substantial bias and relatively low precision.   The precision estimated at
0.167 gal/h for a single leak rata determination is quite high relative to
the desired 0.05 gal/h criterion.   Even though the random error could be re-
duced by taking several repeated determinations of the leak rates and averag-
ing them, the bias would remain a problem.

          3.  Leak Lokator Method

          The Leak Lokator method was tested using 22 simulated 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 duration
was 10.8 min.  The data from these determinations are presented in Table 5
and are displayed graphically in Figure 5.

          The bias in the determinations was estimated to be -0.005 gal/h.
That is, the reported leak rates (out of the tank) were very slightly higher
than the simulated leak rates.  This bias was not significantly different
from zero (t « -0.23, 21 degrees of freedom).  The estimated regression of
the reported leak rates on the simulated leak rate was

               Y = -0.012 + 0.935 X

where Y  is  the leak rate reported by Leak Lokator and X is. the leak rate sim-
ulated by MRI.  The R2 for the regression was 98.9% and the standard error
was 0.020 gal/h.  The estimated slope and intercept agree closely with the
ideal values, but both differ from the ideal significantly at the 5% level,
although not at the 2% level.

          These data'showed a very small  scatter about the regression line,
resulting  in small estimated values for the  standard error of the slope,  in-
tercept, 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  significances  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  the 18 determinations were based  on an
hour's  data.   The  remaining leak  rates reported were each based  on  a 30-min
test.   The average test  time  was  thus 38.3 min, but, as noted, either a  30-
min  or  a 1-h test was  used.   The  data from the tests are presented  in Table 6
and  are plotted in Figure  6.
                                     24-

-------
                 Table 5.   Results Summary from Leak Lokator
Leak Lokator
leak rate
(gal/h)
-0.430
-0.069
0.159
-0.182
-0.081
-0.072
-0.037
-0.008
-0.551
-0.141
-0.238
0.054
-0.031
-0.170
0.005
-0.467
-0.084
-0.154
-0.034
-0.005
0.201

dTime
(nrin)
12.0
18.0
20.0
14.0
8.0
10.0
16.0
12.0
13.0
12.3
9.5
6.0
8.5
10.0
' 10.0
7.0
8.0
6.5
9.0
10.0
7.0
MRI test
dVolume
(ml)
-340.0
-84.0
258.0
-166.0
-39.0
-46.0
-40.0
0.0
-445.0
-98.0
-124.0
31.0
-19.5
-79.0
0.0
-235.0 •
-34.0
-61.0
-21.0
0.0
95.0

Leak rate
(gal/h)
-0.449
-0.074
0.205
-0.188
-0.077
-0.073
-0.040
0.000
-0.543
-0.127
-0.207
0.082
-0.036
-0.125
0.000
-0.532
-0.067
-0.149
-0.037
0.000
0.215
Bias   0.005 gal/h
RMS    0.023 gal/h
                            Test (Bias = 0)
t = 0.23 Non-significant
Regression   Y = -0.012 + 0.935 (Sim)

R-Squared = 0.989          Standard error (SE) = 0.020
SE(B0) = 0.005
                                  = 0.022
95% Confidence interval  for B0   (-0.023, -0.002)
95% Prediction interval  at 0     (-0.055, 0.031)
                                   25

-------
                                                                                           •:>
o      o      o
nn
 a
                                                                             CM
                                                                               •
                                                                             o
                                                                           I-  o
     o
      1
o
 I
^-
 •
o
                                                                                        s
                                                                                        
                                                                                        4J
                                                                                        in
                                                                                        03
                                 X

                                 X.
                                  O
                                  
                                                   S



                                                   -M
                                                   U1
                                                   
                                                   
-------
                  Table 6.   Results Summary from Petro-Tite
Petro-Tite
dTime
(min)
30.0
30.0
60.0
60.0
60.0
60.0
60.0
30.0
30.0
30.0
30.0
30.0
30.0
30.0
30.0
30.0
30.0
dVolume
(gal)
-0.342
-0.030
0.046
-0.220
-0.119
0.060
-0.080
-0.040
-0.080
0.025
0.020
0.000
-0.225
0.010
0.015
-0.030
0.020
Leak rate
(gal/h)
-0.684
-0.060
0.046
-0.220
-0.119
0.060
-0.080
-0.080
-0.160
0.050
0.040
0.000
-0.450
0.020
0.030
-0.060
0.040
dTime
(min)
30.0
33.5
59.5
60.0
60.0
60.0
57.0
23.0
27.0
28.0
29.0
32.0
31.0
30.0
30.0
26.0
28.0
MRI test
dVolume
(ml)
-1,060.0
-295.0
-123.0
0.0
-228.0
0.0
-718.0
-205.0
-360.0
-78.0
-105.0
-305.0
-1,031.0
-128.0
0.0
-350.0
-140.0
Leak rate
(gal/h)
-0.560
-0.140
-0.034
0.000
-0.060
0.000
-0.200
-0.141
-0.211
-0.044
-0.057
-0.151
-0.527
-0.068
0.000
-0.213
-0.079
Bias    0.051 gal/h
RMS     0.108 gal/h
Test (Bias = 0)
t = 2.18 Non-significant
Regression   Y = 0.057 + 1.045 (Sim)

R-Squared = 0.759          Standard error (SE)  = 0.101

SE(B0) = 0.033             SE(Bi) = 0.152
    (-0.02, 0.120)
    (-0.176, 0.276)
95% Confidence interval  for BO
95% Prediction interval  at 0
                                   27

-------
                                                   crt



S
c












e__

a
> oa
aa n
«
^




a
c


Ota
o>
4->
0)
tn

(U
-M
-_ i^
^* i^~
e
4J
O)
Q_
^. «
i i i r i i i Y £
1 -^
^o^-cMn-tmcor^' o»
• • • *^ • • * « «^
D OOOOOOO u-
1 i 1 1 1 ! I
                                                      •. ,J
q/|D6 '
              28

-------
          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 almost significantly different from zero at the
5% level.   If attention is restricted to the 1-h tests, the bias is clearly
not significantly different from zero.  The estimated regression of the leak
rates reported by the method on the simulated leak rates was

               Y = 0.057 + 1.045 X,

where Y is the leak rate reported by Petro-Tite and X is the leak rate simu-
lated by MRI.  The intercept is not significantly different from zero, sug-
gesting that the bias is not statistically significant.  The slope does not
differ significantly from the ideal or theoretical value of one.  Both of
these tests were done at the 5% significance level.   The R2 for the regres-
sion was 75.9% and the standard error of the regression was 0.101 gal/h.  This
standard error is interpreted as the precision of a single leak rate determ-
ination.  It should be noted that the normal test with four consecutive 15-min
rate determinations would 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.

          5.  Helium Detection Method

          Two tests were conducted using the helium detection method.   In the
first, the tank was tested in its original state.   Large leaks were discovered
at the remote pump, a 4-in. capped nipple, and the fittings where the test
apparatus had been installed in the tank.   These leaks were repaired and the
results of the next day's test revealed substantial  reduction in helium loss
at these points.

          While some helium was detected around the tank, the amounts were
generally very small and could have come from pipe fittings or bungs.   Low
levels were encountered at the southeast corner of the tank.   The concrete
was removed for inspection purposes to see if a line was located in that
area.  None was found.  It was concluded that the tank did not have any sig-
nificant leaks.

          An attempt was made to estimate the leak rate by observing pressure
loss from the tank using a differential  pressure monitor.  While this approach
may have some merit, the results of the pressure monitoring were erratic due
to the temperature sensitivity of the transducer used.  Without additional
development, this approach would not be satisfactory for the purposes of the
national survey.

          The primary disadvantage of the helium detection method is its ex-
tremely high sensitivity.  Helium can escape in measurable quantities even
through threaded connections which have not been completely coated with sealer.
Gasoline will not normally pass through such connections under operating con-
ditions.  Hence,  there is a high probability of false alarms with this method.
In addition, the results are not quantitative.   Since the preliminary assess-
ment of the quantitative methods indicated they had considerable promise for
achieving reliable leak rate quantification, the helium method was eliminated
from further study.

                                    29

-------
     C.   Conclusions and Recommendations

          A comparison of the data from the four volumetric methods (shown
earlier in Table 2) indicates that the Certi-Tec method performed relatively
poorly.   Poor performance may have been due to the fact that the method was
still somewhat in its development stage, but this method was eliminated from
further consideration.

          Of the three remaining methods, Petro-Tite had the poorest perfor-
mance characteristics, probably due to the fact that the test crew conducted
a number of 30-min tests as opposed to the standard 1-h tests.  The method
was retained for further evaluation, in part because of its extremely wide
use and long history.

          Both the ARCO and Leak Lokator methods performed well under the con-
trolled conditions of the test.  The conditions were somewhat favorable in
that the product had a long stabilization time prior to the testing.

          In evaluating each of these four volumetric methods against the
selection guidelines, none of them appeared to present a clear advantage.
Very few problems were encountered.  Since the tank system was a simple one
with all of the distribution lines capped off, most of the problems associ-   •
ated with site preparation were eliminated.

          In brief, comparative evaluations of the volumetric methods against-
the selection criteria revealed the following:

          1.  Quantitative Measurement

          All of the methods tested except the helium method provided quanti-
tative estimates of the leak rate.

          2.  Detection Levels   .

          Some questions were  raised as to how reliably the leaks of 0.05
gal/h could be detected, particularly for the Certi-Tec method.  It was
dropped from further consideration primarily for this reason.

          3.  Tnvasiveness

          This parameter was riot evaluated at the preliminary testing site
except that the three crews demonstrated the capability to set up, conduct
the tests, and restore the site in less than a full day.  The Leak Lokator
method required the  least test time and ARCO the longest.

          4.  Physical Reliability

          All four methods were rugged enough to be acceptable.  No mechan-
ical problems were encountered.
                                     30

-------
          5.   Versatility

          This parameter could not be evaluated based on only one test site.

          6.   Amenability to Performance Monitoring

          Performance of all four methods could be monitored satisfactorily.

          7.   Cost

          None of the approaches presented any significant cost advantages at
this early stage of evaluation.

          8.   Availability of Equipment and Manpower

          All three companies were national and had sufficient resources to
conduct a national survey if selected.

          Based on the observations discussed in this section and summarized
above, the ARCO, Leak Lokator, and Petro-Tite methods were selected for fur-
ther evaluation.


IV.  DEVELOPMENT STUDY TESTING

     A.  Overview

          The primary objective of this phase of the study was to collect and
analyze sufficient data from the three methods selected from the preliminary
testing to characterize and compare them.  The data would also be used to in-
dicate where modifications might be made in the method selected for use on
the national  survey.

          To accomplish this objective, tank test experiments were undertaken
at five sites across the United States using the three test methods.   Two
types-of tests were conducted.  The first required the tank test crew to use
the standard protocol- of that test method.   After this test was complete, the
crew was requested to estimate the magnitude of different leaks produced by a
pump that removed the product from the tank at a known, constant rate.  These
data were analyzed to estimate the simulated leak rate and to estimate the •
magnitude of the ambient volume changes over longer test times.  The ambient
time series was generated by subtracting the simulated leak from the volume
measurements.  The instrument performance,  test protocol, test time,  data
analysis and compensation procedures, detection criterion, and detection per-
formance were evaluated using a case study approach, because the data on leak-
ing and nonleaking tanks were limited to less than 10 tank tests per method.
An extensive analysis of the Petro-Tite data provided by 0.  H.  Materials and
of the Leak Lokator data provided by Hunter Environmental Services was per-
formed by Vista Research and MRI.   A detailed analysis of the ARCO test method
data was not undertaken because ARCO (for proprietary reasons) did not disclose
their test procedure and analysis algorithms, or provide the raw data necessary
for analysis.


                                    31

-------
          Testing was conducted at the five facilities listed in Table 7.  A
summary of the characteristics of each tank tested is given in Table 8, and a
matrix of the tests conducted is given in Table 9.  The facility diagrams,
tank characteristics, and test conditions are provided in Appendix G.

     B.  Method Descriptions

          The three methods tested during the development study phase of the
project are described in this subsection.  The descriptions are based primar-
ily on MRI's observations rather than on manufacturer literature.  An effort
was made to represent the equipment in light of its performance during the
study.

          1.  ARCO

          The ARCC leak test method uses a specially fabricated float mech-
anism and spectrophotometer to detect volume changes in underground storage
tanks and their distribution lines.  A schematic of the apparatus is shown in
Figure 7.  To assure reliable t«st results, product levels must be between 66
and 75% of the tank depth.  The float apparatus, consisting of a cup of dye
solution and a photocell, is inserted into the tank at the beginning of the
test.  The fill port must be at least 3 in. in diameter.   The float is placed
at a prespecified level in the tank, where temperature changes will not affect
the float level.  A 1-h waiting period is recommended to allow the temperature
of the equipment to stabilize.

          With changing product levels, the float rises or drops in the tank
and changes the depth of a dye solution between the photocell and a light
source mounted on the probe.  The change in light transmittance is detected
by the photocell according to Beer's Law.  Hence, the change in detector re-
sponse is a function of the change in the product level in the tank.  The re-
sults are recorded on a strip chart.

          The instrument response is calibrated by adding known volumes of
product to the tank.  The attendant response on the strip chart is used to
determine the sensitivity of the device in terms of gallons per division.

          The unique feature of the ARCO leak test is that it is self-
compensating for temperature changes if the float is positioned properly.
(The method may still be sensitive to unstable temperature conditions, but a
separate temperature correction is not needed.)  The procedure takes advantage
of the relationship between liquid density (a function of temperature), liquid
buoyancy (a function of density), and liquid volume to select a point where
the effect of changes in level due to temperature are exactly offset by changes
in buoyancy.

          A disadvantage is that the test must be conducted with the tank only
66 to 75% full; therefore,  it is not possible to detect leaks in the upper
25% of the tank.
                                    32

-------
       Table 7.   Development Study Test Facilities
     Facility
    Location
Oamneck Naval  Combat
  Training Center
Pitstop
Scott Air Force Base
Fort Lewis
Langley Air Force Base
Virginia Beach, VA

(confidential)
Belleville, IL
Tacoma, WA
Hampton, VA
                          33

-------
Table 8.  Tanks Tested and Their Characteristics
Facility
Damneck
Pitstop
Pits top
Scott
Scott
Fort Lewis
Fort Lewis
Fort Lewis
Fort Lewis
Langl ey
Lang ley
Lang ley
Langl ey
Tank
1
1 (south)
2 (north)
1 (17)
2 (18)
1 (8C25 north)
2 (3C25 south)
3 (4194)
4 (10E10)
1 (hydrant
system
tank 3)
2 (hydrant
system
tank 5)
3 (MoGas)
4 (golf course)
Nominal
capacity
(gal)
5,000
12,000
8,000
5,000
5,000
12,000
12,000
12,000
12,000
25,000
25,000
10,000
1,000
Diameter
(in.)
96
108
96
96
96
96
96
96
96
124
124
90
49.5
Approximate
overburden
(in.)
24
38
53
44.5
44.5
10
5
22
4
12
15
20
12
Petroleum
product
Regular
gasoline
Regular
gasoline
Regular
gasoline
Regular
gasoline
No. 2 Diesel
fuel
Unleaded
gasoline
Unleaded
gasoline
No. 2 Diesel
fuel
Unleaded
gasoline
JP-4 jet
fuel
JP-4 jet
fuel
Regular
gasoline
Regular
gasoline
                     34

-------
                    Table 9.   Matrix of Tests Conducted
Facility
Damneck
Pits top
South
North
Scott
17
18
Fort Lewis
8C25 north
8C25 south
4194
10E10
Langley
HS tank 3
HS tank 5
Mogas
Golf course
Tank ARCO Leak Lokator
1 Ta T

T T: Noisy (vibration),
possible vapor
T T

1 Out of time T
2 T T: Manifold (?)

IT T: Leak at about
gasket; couldn't
test
2 T T
3 — T: Poor sensitivity
(-0.171)
4 — T: Results
questionable
•
1 — b T: Multiple leaks,
plumbing problem
2 Tried, but T: Plumbing problems
couldn't test
3 T
4
Petro-Tite
T

T
T

T
T

""•
T
Added 400 gal ;
couldn't fill up
(bad leak?)
T

Would have had
similar problems
if testing had
been attempted
Would have had
similar problems
if testing had
been attempted
— «P
T
bLetter T indicates the tank or tank system was tested.
 Dash (--) indicates testing was not conducted.
                                     35

-------
o
u
                                 a
                                 u

                                 a
                                         ^N^«-^^^«

                                         >ywx^Nj
 CO
I
                 a.
                 0.
                 
-------
          Evaporation effects are minimized by utilizing a spray device to
saturate the atmosphere above the surface of the liquid.  The method is some-
what sensitive to vibration caused by nearby traffic or wind which may cause
a surging effect on the fluid.   There is no attempt to compensate for water
table effects since it is likely that under actual use conditions this con-
dition would result in the finding of water in the tank.  Vapor pockets and
end deflection effects are not a problem because the tank is not overfilled.

          The entire test can be conducted in 3 to 4 h.  Generally, there are
few site preparation problems since the test operates under normal operating
conditions.   Calibration checks are performed at hourly intervals, and test-
ing continues until consistent results are obtained.

          2.  Leak Lokator Method

          The standard operating protocol for Leak Lokator was proprietary.
The description provided here is based on observation of the testing done on
this program.  MRI discussed points of the method with Leak Lokator personnel.
Current practice on standard operating procedure may differ from this descrip-
tion.

          The Leak Lokator method is based on the buoyancy principle.  A sche-
matic of the test equipment is shown in Figure 8.  The sensor (a float) is
suspended in the fluid from an analytical balance.  Any changes in the level
of the fluid will produce a change in the buoyancy of the sensor which is de-
tected by the analytical balance.  The balance is equipped with an electronic
sender which transmits the signal- to a strip chart recorder where it is re-
corded continuously.  To quantify the instrument response, a calibration bar
of known volume displacement is immersed in the fluid and the effect is noted
on the strip chart.  This information (chart deflection per volume) is used
to convert the level changes produced by the sensor into volumes.   The reso-
lution of the device is of the order of 0.001 gal.

          The temperature of the fluid is monitored using a thermistor lo-
cated at the center of the tank.  The temperature is monitored continuously
on a strip chart and on a digital meter with a resolution of 0.001°F.  The
readings from the meter are recorded manually on the strip chart at frequent
intervals.  The meter readings are also used to calculate the temperature cor-
rection for the leak detection.

          The method can be used to test either the tank or the complete sys-
tem by adjusting the level of product to above or below the piping.   It is
assumed that the pressure head of the fluid is sufficient to overcome the
masking effect of any water table that may be present.   Evaporation effects
are compensated for by filling the sensor with the same fluid as is in the
tank.   Evaporation of fluid from the surface of the liquid in the tank is
offset by an equivalent loss from the sensor.

          A minimum of 2 h is required to run the test, including setup time,
if there are no site preparation problems.   In addition, a stabilization
period of 8 h after filling is recommended to reduce temperature fluctuations.
                                    37

-------
                                                             i-
                                                             rtJ
                                                             O.
                                                             a.
                                                             to
                                                             
                                                              
-------
          The testing is conducted by monitoring the level change for a period
of around 10 min.   The temperature is monitored for at least an hour before
and after the test.   Additional 10 min tests can be conducted if necessary to
verify the test results or if the temperature data suggest that it is desir-
able to do so.

          3.  Petro-Tite Method

          This test is essentially a specialized fluid-static (standpipe) test
that provides correction for temperature and water table effects.   A schematic
of the test apparatus is shown in Figure 9.

          The test is conducted by monitoring at regular intervals the fluid
level in the standpipe and the temperature of the product in the tank.  The
fluid in the standpipe is releveled, using a graduated cylinder, to the same
reference point following each reading, and the volume added or subtracted
from the standpipe is recorded.  The observed volume change is then corrected
for temperature effects using the coefficient of expansion of the fluid and
the temperature change.  The corrected volume and time are then used to de-
termine the leak rate.

          There are several key features of the Petro-Tite method which should
be noted.  First,  the fluid is stirred prior to and during the test to produce
uniform temperature behavior in the tank.   A circulating pump draws fluid
thrpugh a suction  tube from at least 6 in.  below the tank top.   The product
is then discharged from a nozzle located near the bottom of the tank but above
any water which may be present in the tank.   The nozzle is directed upward at
an angle of 45 degrees from the center line of the long axis of the tank to
produce a swirling motion of the fluid.  The temperature is monitored with a
thermistor which is located near the inlet to the suction line.   If mixing is
incomplete, the temperature behavior can be erratic.

          Second,  the static pressure on the tank is adjusted to account for
the effects of any external water table that may be present.  The standpipe
is always raised to the level calculated to produce a net positive pressure
at the bottom of the tank that is 4 Ib above the back pressure of the water
table.  If there are no monitoring wells at the location, a small  bore hole
is drilled to determine the position of the water table.  This process elim-
inates any masking effects which might be produced by the water table.  The
higher head pressure also enhances the flow of fluid through any holes which
might be present,  making leaks easier to detect.

          Third, at the beginning of the test, the net pressure in the tank
is established at 5 Ib to accelerate the outward deflection of the ends of
the tank.  It is usually possible to determine when the outward deflection is
complete by inspection of the "high level"  data.  After a period of time, de-
pending on the diameter of the tank, the head pressure is lowered to 4 Ib.
At this point the  ends of the tank should stabilize relatively quickly and
the remainder of the data is then collected at the "low level."

          The temperature resolution of the unit is 0.003°F.  The volumes of
fluid used to relevel the standpipe are recorded to the nearest 0.005 gal.


                                    39

-------
                                                              a.
                                                              a.
                                                              in
                                                              
-------
          The Retro-lite method tests the entire system.  If a tank-only test
is required, the tank must be isolated by disconnecting all the lines leading
to the tank.  Independent tests of the lines are usually included as a part
of the Petro-Tite test if any leaks are detected in the system.  The time to
conduct the test, including set up, is of the order of 4 h if no site prepara-
tion problems are present.

          Vapor pockets are said to be detected by the behavior of the fluid
level in the standpipe.  If any vapor pockets are present, the tank must be
excavated and the vapor removed by installing a bleed valve before the test
can be completed.

          Since the entire system is tested, it is possible that some leaks
which are not environmentally significant will  be detected.  For example, the
Petro-Tite method will indicate a large leak for a system with a broken vent
pipe line.  Since this line does not normally contain product, the leak is
not environmentally significant.

          The normal test procedure involves the collection of four sets of
data at 15-min intervals.-  The observed volumes are each corrected for tem-
perature, and the net values are summed to obtain the product loss for 1 h.
The test can be extended if the results suggest that additional data are needed.

     C.   Experimental Procedures

          1.  Field Testing

          The design for the development study was as follows.  A total of
13 tanks at five facilities were tested.   It was originally planned that each
tank was to be tested by all three methods.   Difficulties in scheduling and
plumbing problems at some sites, however, precluded a complete round of test-
ing.  The tanks to be tested at each site were chosen on the basis of avail-
ability.  In addition, facilities in different geographic areas were selected
to represent a variety of soil types and other environmental conditions.

          The evaluation program included baseline tests and tests under sim-
ulated leaks.  All tests were conducted separately; i.e., tanks were tested
by only one crew at a time.  Also, the leak rates simulated were selected in
random order and were blind to the test crew.   Temperature profiles were
usually taken before and after each test by lowering a thermistor into the
tank at 6- or 12-in. intervals.  These profiles are presented in Appendix D.

               a.  Standard Baseline Tank Tests

               Baseline tests were standard leak detection tests performed by
each method without leak simulation.  Each organization was asked to perform
the test exactly as if it were a routine test to be done on the national  sur-
vey and inform MRI when it was completed.  MRI placed no time or other con-
straints on this part of the evaluation.
                                    41

-------
               b.   Leak Simulation Tests

               After the base'line tests were completed by each company, MRI
asked that additional testing be done by each firm's test crew under the same
conditions but with a simulated leak rate blind to the test crew.  The sim-
ulated leak was induced using the leak simulation system described in Section
III.   The crews were again asked to inform MRI when their test was completed
under these conditions.  In all cases the simulated leak rate was held con-
stant for a period of 1 h or more.  After each crew had performed a test to
its satisfaction under the simulated leak rate, a second simulated leak at a
new rate was induced.  A total of four simulated leak rates were used, in
random order, one of which was zero, and should have duplicated the condi-
tions of the baseline test.  The crews were told neither the simulated leak
rates nor exactly when the new rates started.  They were told, however, when
they should begin testing and were asked to tell MRI when their testing under
that condition was complete*  The exact simulated leak rates were computed by
collecting and weighing the product pumped out of the tank and correcting the
mass to the average tank temperature before converting to volume.  The average
tank temperature was independently measured by MRI by placing a thermistor
halfway into the fuel and taking readings every hour on a digital readout.

               c.   Soil Coring and Analysis

               An ancillary activity conducted during the development study
evaluations was the collection and analysis of soil cores from one of the fa-
cilities.  The objective of this effort was to gain some experience and in-
sight into this type of external monitoring.  The procedures used and results
obtained are described in Appendix F.

          2.  Data Reduction and Analysis

          The data resulting from the testing at the development sites con-
sisted of three data sets:  baseline test data, simulated leak data, and time
series analysis of the ambient volume fluctuations after the simulated leaks
were removed.  In some cases, a zero rate was also simulated.  In these cases
the simulation system's peristaltic pump motor was operated to ensure that
the selected rate was blind.

               a.   Baseline

               The baseline data  for each method were 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 re-
solve the conflict.

               The character!2:ation of a tank as tight or leaking was a sub-
jective judgment by MRI based on  analysis of all of the test data and by visual
observation around the test site.  After the testing was completed, the Pitstop
South tank was excavated and found to be leaking around the fill pipe flange.
Since none of the other tanks were excavated, the judgments were inferred from
the extended test data analysis.


                                    42

-------
               b.  Leak Simulation Regression

               The first analytical approach was to fit a linear regression
to the data from each tank and method, regressing reported leak rate on the
simulated leak rate using only the data collected during the simulations.
Ideally, the slope of this regression should be one.  The intercept of this
regression represents an estimate of the leak rate of the tank system when
there is no simulated leak.  The difference between the intercept of the re-
gression 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 re-
gression line provides an estimate of the precision of the test.  Combining
these two measures yields an estimate of the mean square associated with the
testing method, reported in this document's data tables as RMS (or root mean
square error).

               In general, the three methods performed similarly.  Each had
difficulty with some tanks and performed well with others.

               c.  Description of Ambient Noise Analysis

               The second analytical approach was to remove the simulated
leaks from the data to produce volume, temperature, and temperature compen-
sated 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 induced vol-
ume changes adequately accounted for the total volume changes.   The data re-
ceived from ARCO were inadequate for this type of analysis, so ambient noise
analysis was performed only on the Petro-Tite and Leak Lokator data.

                    (1)  Petro-Tite Method

                    Continuous time series data of the change in volume and
the change in temperature (converted to volume using the product volume and
the coefficient of thermal expansion) were collected for an entire day of
Petro-Tite testing.  These data were generated from the data collected every
15 min by subtracting the simulated leak volume from the measured volume.
The time series included the baseline test data and were 3 to 6 h in duration.
The magnitude of the simulated leak was changed in the field during a 15-min
sample interval after each 1-h Petro-Tite measurement period.   For the con-
tinuous time series, 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 tempera-
ture compensated volume were then generated for analysis.   The temperature
compensated time series were generated by subtracting the temperature (ex-
pressed in volume) from the measured volume on a point-by-point basis.   (This
is the same method that Petro-Tite uses.)  A least squares line was then fit
to the entire cumulative temperature compensation time series to estimate the
temperature compensated volume rate for comparison with the baseline test re-
sults.  The standard Petro-Tite 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).

                                    43

-------
                    (2)  Leak Lokator Method

                    Time series of the cumulative volume and cumulative tem-
perature (converted to volume using the product volume and the coefficient of
thermal expansion) 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 tnin
and included four to nine of the standard Leak Lokator volume rate measurement
periods.  The individual Leak Lokator tests were typically 10 min but ranged
from 6 to 18 min.   Long time series were not generated because of a gap of 3
to 4 min in the collection of Leak Lokator data between simulated leak se-
quences.  The simulated leak rate was subtracted from the uncompensated vol-
ume rate measurements made by Leak Lokator and was 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 sim-
ulated leak sequence was estimated from the reported Leak Lokator volume rate
data.   The mean and standard deviation were weighted to produce values equiva-
lent to a 10-min test period.  Because most test periods were close to 10 min,
similar values of the mean and standard deviation would be obtained without
weighting the data.

                    A continuous time series of temperature was generated 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 volume data for each simulated
leak sequence were used to compensate for temperature.  Except for the three
simulated test sequences which were longer than 1 h, the temperature time
series were generally 1 h.  The temperature data were converted to a volume
time ser-ies, 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.

                    Leak Lokator regarded its standard operating procedures
as proprietary.  In a standard test of the Leak Lokator, as performed by MRI,
the rate of change of temperature was estimated from only two points sepa-
rated by an hour.  A mean temperature compensated volume rate was then com-
puted 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 re-
sults from the baseline test and the other simulated leak test sequences.

     D.  Results

          1.  Tank Temperature Conditions

          Tank temperature conditions could not be controlled during the tests.
The Petro-Tite and Leak Lokator test methods were used to test tanks under
different conditions.  Whenever possible, MRI made temperature profiles with
a single thermistor accurate to 0.01°F in 6-in. depth increments before and
after each day of  tank testing.  In general, a wide variety of different tem-
perature conditions were encountered for each method.  Figures 10 and 11 show
single  temperature profiles made before and/or after some of the Petro-Tite
and Leak Lokator tests.  The approximate location of the Petro-Tite and Leak
Lokator thermistors are indicated.  The product temperature conditions are
summarized in Tables 10 and 11.

                                    44

-------
      CD
        /
*
        o

        i
        ?
        V
11
                                                             1   1   1   1    1   1
                                                                             m •


                                                                             *3
                                                                                 o

                                                                             *  I
                                                                             •O  C
       ••did iiu 10 doi uiaaj
••dy
                   doi 11104
••dij u ij jo do| may
                                                 o
                                               Q.Z
                                                                                                   '»di^ ||'J jo doj.
                                                               40
  I  -M

 2*.
 4->  O
 O!
 O-  £

 "O +•>
 O) -I-J
 +->  o
 o /^

 i—    v>
                                                                                                                                                 ai       OJ
                                                                                               • — •  tt)  4->
                                                                                                   JS
                                                                                                O) +->  01
                                                                                                s-     ^
                                                                                                O ^*-  4J
                                                                                               4-  O
                                                                                                OJ    14-
                                                                                               -Q  O.  O
                                                                                                    O
                                                                                               •a -u  c
                                                                                                ai     o
                                                                                                c  a) -t-
                                                                                               •i-  -C  4J
                                                                                               rC  4->  (13
                                                                                               -!->      U
                                                                                               J2   =  O
                                                                                               O  O r—

                                                                                               yi  M-  0)
                                                                                                +j
                                                                                               Q. on  03
                                                                                                  •>-  O
                                                                                               OJ -D -r-
                                                                                               S-    T3
                                                                                               3  E  C
                                                                                                                                            S-  •<-  2
                                                                                                                                            OJ  X  O
                                                                                                                                            CL to  i.
                                                                                                                                            S  E  s.
                                                                                                                                               4=  01
                                                                                                                                               t- -C
                                                                                                                                            S-  1/1  C
                                                                                                                                            3  a;  ra
                                                                  45

-------
        1
        t   5
        Is
Q.

*  10
                Damncck
                Tank No. 1
                                          08:09
                                         9-21-64

                                          i 16:12
                                                         0.
                                                         I  5
                                                         <*•
                                                         O
                                                         I
                                                         o
                                                                 a.
                                                                Q 10
                                                                Ft. Lewis
                                                                Tank No. 2
                                                                    63
                                                                          64
        3.
        £
        =   5
        uZ
        «••
        o
        §•
                 Plhtop
                 Tank No. 2
                                                         1
                                                         §•
                                                                 g- '0
                                                                a
                                                                 Ft. Lewis
                                                                 Tank No. 4
            62
                  63
                              64
          61
 62
S   5
        u.
        !*•
        O

        J"  10

        o
        V
        a
                Scott Air Fore* SOM
                Tank No. 1
                                               11.02
                   64
                        65
                      T«fnp«fattjra, *F
                                     66
67
                                                                 s.
                                                                 a:
                                                                Ft. Lawis
                                                                Tank No. 3
10
63.5     64
64.5
Figure 11.   Temperature profiles obtained  before  selected  Leak Lokator  tests.
The maximum  distance  from the top of the fill  pipe  is the  approximate bottom
of the tank.   The arrow indicates the position of the temperature  sensor.
                                         46

-------




























in
C
o

.'Jj
•o
c
0
0
CU
3
i %
+»
(O
J-
S
C.
0}
r—

0)
-P
•r—
1
O
s.
+}
CD
a.


•
o
rH

CU
p-«
J2
<2

































4->
lr^
r—
•PB
^t
(0
•P
oo

it- in
0 S-
01

o a
Z i—


H-

u_
xo
ro -**>
^*




.c
•p /— s
N O.-P
~O ->
•r™
2



•P
C C
V O

u in u.
— CO
<*- rtj rH
*t» CX^x
fc
+J
•r~ ^^
U r—
fO fQ
ex cn
rO SB* •*
CJ






(0
u
0
^J




0)
r—
a
fO
4>J
00




^1






<^O
,
o








o
f






to
CM
•
rH
f^






in
CM
^^
CM
tO
o
o
o
.
0



CM
f^
o
M
m






^«
CO
^
x>^
o
r^














rH






^c
u
o>
c
E
(O
o



0)

.a
TJ

00




rH






^»

d








0
^






CM
rH
»
CM
<^






LfJ
f^
00
00
m
o
a
o

C)



o

CO
M
rH
rH





m
CO
cn

o
rH



/^N
s_
fO
0)
^

m
CM
C_3
00

CM



U)
•^»
3
ai
^j

^j
s-
o
u_
0)
^Z
4-
o
s-
a.

o
z




1







•1









1









1








CO
CO
o
cn
m
0
o
o
.
o



*f
^4
cn
M
rH
r-l




^*
CO

J5
^x.
o
rH







^•N
0
rH
UJ
O
rH

^j"



(ft
•r~
3
0)
—1

^>
i»
O
U.



0)

3
(TJ
+J
00




fO







CO

A






CO
0
o'






m
cn
*
00







rH
to
in
m
to
o
o
o
.
o



cn
cn
CO
M
rH
rH





^»
CO
ro
rH

cn







^^
^
4^
3
O
in

rH






Q.
O

in
^j
*^»
CL.



0)

r>
(TJ
+j
00




CO







m

A






cn

o






00
o

cn
to






rH
to
in
m
to
o
o
o

0



^^
O
en
M
fs».






^t*
^^
^x
rH









^^
-C

i-
0
c
v""/
CM






a.
o

in
-P
• pM
a.



cu
^*
_^3
fl
^j
OO




ro







00

ro






^"
00
c»






^-
CM
i
r^
to






p*^
f^.
C\J
ro
to
o
o
o
.
o



to
00
o
M
Lf5






^J-
CO
cn
rH

cn










^\
p^
rH
^~*
rH




CO
U_
^
XJ to
GJ
•a •— •

-P "+- *j
(TJ O ra
s t- -a
•r- Q. __
to 41 +J
0) 5- -r-
3 3
in 4->
ra ra >»
3 S- —
0) r—
CQ.ro
O E U
•i- ^_ ^
•i— s
r— >
ra 'i— i/i
S T3 0)
i. C u)
0) •!- ra
J= 0)
•P £ i-
o o
4— s- ai
O <+- T3

•P T3 CU
C  in
O ra 0)
•z. s cn
47

-------

























I/I
c
0

•+J
•^*
TJ
c
0
0
cu
i-

to
s.
cu
CL
s
cu
^J
u>
cu
r—
.^^
C

TJ U.
1 0

z


e
a
•t— •••• >
in u.
CO
J
^J
•^* ^^
U —
tO 10
cu o»
<8 -«x
u


"

cu
^J








td

.
^n
id




CM

in
cQ
o
o
a

o



^^
rO
o
A
CM
rH





00

co
^^
o
rH







^^
m
CM
u
00

CM



Ul
•f—
^
cu
^J

^^
S-
o
u_

cu
a
to
5J






— <






tn^
.
O








o

^>







r™^
rH
.
*^
id




CM
fO
r**»
*y
o
o
0

C3



^f
CO
0

CM
rH





CO
^^
^t"
^^
CD







^^
^f*
cn
rH
^

co



11
•r-
:E
cu
_j

^j
:.
a
u.

CU
ja
(O
55






I-H






rH
.
O

V






o

^?







CM
CO
.
rH
id




CO
m
UD
vd
o
o
0

d



^»
ro
o
A
CM
rH





eS
*^^
U^
^^
a
rH






^,
O
rH
LU
O
rH

^f



Ul
• ^B
3

^•*
— TJ D
S_ r—

+J
ra ^
^ c
O HI
— 1 4->
-* CU
*TJ ^^
cu *->
cu
>> i-
J3 O
4—
T3 CU
CU J3

IB Ul
S cu

Ul *- (B
CU O -(->
&• fB
ui ova
(B
3 ,
C CU r-

CL S IB
X CU U
0) +J -r-
c
r- r- O
tO IB +•>
E3 0
•a c
CU -r- O
-C > S
4J -r-
•O ui
<*- C CU
O -f- Ul
•P S cu
e o &-
cu s- u
•r- <*- 
O -U 
•<-> 13 tO
0 C .C
Z IB *J
48

-------
          Several observations can be made.   First, the profile made after
the Petro-Tite tests indicates that the circulation produces an isothermal
profile to within ± 0.01°F (i.e., the accuracy of the thermistor used to mea-
sure the profile) regardless of the initial  profile.  Whether the temperature
is also uniform at other points in the tank is not directly known, but if the
temperature were not uniform throughout the tank, large errors in the tempera-
ture compensation would be observed.   Since large errors in temperature com-
pensation are not observed, we infer that the circulation.approach succeeded
in producing a uniform temperature throughout the tank.

          Second, the Leak Lokator one-thermistor compensation approach should
work best for temperature profiles with uniform temperature changes as a func-
tion of depth (e.g., Fort Lewis Tank Nos. 2 to 4 and Pitstop Tank No. 2) but
would not work well for nonuniform profile changes (e.g., Damneck, Scott AFB
Tank No. 1).

          Third, the Petro-Tite thermistor,  located approximately 45-in. be-
low the top of the fill-tube, may not always extend into the tank sufficiently
to avoid temperature fluctuations in the fill-tube and near the top of the
tank which are not representative of the tank as a whole.  More investigation
of the effect of the inlet location is needed.

          A summary of the baseline test results for all three test methods
is given in Table 12.   Data from all  development study tests are contained in
Appendix H.

          It should be pointed out that, because the tanks  were tested at dif-
ferent times by different methods, the conclusions reached  by each method
could be correct only for the times at which the tanks were tested.   This could
occur if either the environmental conditions changed significantly or if the
tank developed a leak after the initial test.   It should be noted that the
MRI conclusions are based on the results of all the methods used'to test each
tank system, together with any other evidence, such as product observed in
soil or water in the area.   In addition, each test company  expresses its test
results as follows:  certified as having passed the test, which means that
the measured leak rate is less than the NFPA criterion; or  noncertified as
having failed the test, which means that the measured leak  rate is greater
than the NFPA criterion or that excessive noise or product  loss results in
an inconclusive result.  MRI expressed its conclusions about these test re-
sults as either tight (certified), leaking (noncertified),  or inconclusive
(noncertified).

          The final conclusion, nonleaking or leaking, was  determined by MRI
after data analysis.  To confirm the results,  excavation would have been use-
ful, but this was not possible because all tanks tested were in use.
                                    49

-------
        Table  12.   Summary  of  Baseline Results  and Tank Tests Attempted
Facility and ARCO
tank (gal/h)
Damneck +0.02 Ca
Pits top
1 (south) +0.02C C
2 (north) 0.0 C
Scott
1 (17) ' Out of time
2 (18) . +0.02C
Fort Lewis
1 (8C25 -0.04 C
north)
2 (8C25 0.0a C
south)
3. (4194)
4 (10E10)
Lang ley
1 (HS tank 3) --
2 (HS tank 5) • Physical problem
with tank
3 (MoGas) -0.03 C
4 (Golf
course)
Leak Lokator
(gal/h)
-0.077 Nb
-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 N
-3.0d N
—
™*
Petro-Ti
(gal/h)
+0.003
-2.892
-0.05
+0.004
-0.812
—
-0.342
-3.0
-0.024
—
—
-2.540
te
C
N
C
C
N

N
N
C


N
MRI
conclusion
Tight
Leak
Tight
Tight
Leak
Tight
Leak
Leak
Leak
Leak
Leak
Tight
Leak
^Certifiable according to NFPA 329.
°Noncertifiable by NFPA standards.
 .Test OK, but leak (possibly in upper part or piping) not found.
 Test appeared OK, but data are inconsistent.
Negative values indicate leak out of tank.
— indicates testing was not conducted at that tank by the test company
     indicated.
                                     50

-------
          2.   Baseline Tests

               a.   ARCO Method

               Several interesting observations can be made from these data.
First, none of the tanks were found by ARCO to be leaking.  The ARCO result
disagreed with the MRI conclusion for three tanks.  It must be noted that the
ARCO system tested tanks approximately 75% full and under no additional head
pressure.  Thus, the ARCO system provides a test more representative of the
usual operating conditions of the tank.   However, the ARCO system will not
detect a leak in or near the top or in the fill pipe, or if there is a leak
in the lines.   While it may be unlikely that all of the leaks encountered dur-
ing the study are in the top of the tank, it is a possible explanation.  In
fact, one of the tanks (Pitstop South) was excavated and found to have a leak
in the fill pipe flange at the top of the tank.  In a second case (Scott No.
18), the results of all the tests were in question due to anomalous behavior
observed by Leak Lokator.

               It should also be noted that ARCO's testing in general has de-
termined the percentage of leakers to be approximately 5% in their own sys-
tems (Collins 1985).  This estimate is considerably lower than was determined
in the national survey but is consistent with the differences between the ARCO
method and the Petro-Tite method.  The ARCO estimate of leakers is also based
on tanks that are tested periodically on a regular basis and for which regular
maintenance is conducted.  It may well be correct for this population of tanks.

           —  b.   Leak Lokator Method

               The results reported by Leak Lokator disagree with the con-
clusions reached by MRI on 4 out of 10 tests.   The MRI conclusions are based
on an extended analysis of all the data, including the data collected by Leak
Lokatcr.  In three cases the tank was reported to be leaking when the MRI con-
clusion was that it was tight; and in one case the tank was reported to be
tight when the MRI conclusion was that a leak was present.

               Several factors should be noted in making these comparisons.
First,, extended analysis of the Leak Lokator data (discussed in subsection 4
below, Ambient Noise Analysis) indicates that the primary reason for the dif-
ferences is probahly in the temperature compensation approach used by Leak
Lokator.

               Second, MRI's conclusions are based in part on the Petro-Tite
data as well  as the Leak Lokator results.  There is no reason to believe that
the condition of the tank was different from test-to-test, but this remains a
possibility.

               Third, since the Leak Lokator tests are generally conducted
with the product level below grade, leaks which are in the piping above the
product level  will not be detected.  This could have been the situation at
the Fort Lewis Tank No. 2.  If this situation existed, the results reported
by Leak Lokator would be correct for the tank, but incorrect for the complete
system.


                                    51

-------
               A large leak in the Scott Tank No.  1 was reported by Leak Lokator
whereas the tank was found by Petro-Tite to be tight.  The conflicting con-
clusion is difficult to reconcile because interactive effects between Tanks 1
and 2 were observed during the testing of Tank No. 2 by Leak Lokator.  This
could be explained if the tanks were manifolded together.   However, if this
were the case, it is expected that a much larger leak would have been observed
during the Petro-Tite test.  In addition, since the tanks are parallel to each
other, interactive end effects which are sometimes observed when tanks are
placed end-to-end seem unlikely.   The reasons for the inconsistencies at Scott
are not apparent from either observations of the crews or the data.

               The reported results at Damneck indicated a small system leak.
An additional test at product level below the piping indicated the tank to be
tight.  Extended analysis of the data indicates that temperature effects, not
accounted for by the Leak Lokator temperature compensation procedure, could
account for these differences.
                         •

               c.  Petro-Tite Method

               The Petro-Tite test results agreed with the MRI conclusion in
all nine systems tested.   In one instance, Pitstop Tank No. 2, extended analy-
sis indicated volume increases were actually occurring which would have inval-
idated this test (due to a volume increase greater than could be accounted
for by temperature effects, by more than +0.05 gal/h) if baseline data had
been taken 1 to 2 h later (discussed in subsection 4 below, Ambient Noise
Analysis).

   ~"-• "        Since this method applies a significantly higher head pressure,
always tests the entire system, and always accounts for water table effects,
it would be expected to report larger leaks with a higher frequency than other
methods.  Where testing of leaking tanks overlaps with other methods, the pre-
diction of higher leak rates seems to be the case with Petro-Tite (Pitstop
No. 1, Scott No. 2, and Fort Lewis Nos. 2 and 3).   However, the frequency of
reporting leaks is lower than that for Leak Lokator.   Of eight overlapping
tests with Leak Lokator, Petro-Tite reports four leaks while Leak Lokator re-
ports six.  Of five overlapping tests with ARCO, Petro-Tite reports three
leaks while ARCO reports none;.

          3.  'Leak Simulation Results

               a.  ARCO Method

               The ARCO method was used to test seven tanks.  Of these, on
one tank, only the standard baseline test was run; no simulated leak tests
were run because the standard test required all of the time available.  An-
other test produced only two data points, the baseline test and one simulated
leak rate.  The other five tests had the baseline leak rate and several simu-
lated leak rate tests.  However, in most cases, only five or six data points
were available.
                                    52

-------
               The results from the leak simulation tests using the ARCO
method are summarized in Table 13.   Sample data are shown in Table 14 and
Figure 12.  This method reported none of the tanks to be leaking.   However,
other methods gave different results for some tanks.   The conclusions about
the condition of the tanks are summarized in Table 12.   The ARCO result dis-
agreed with the conclusion for three tanks,  namely, Pitstop South, Scott Tank
18, and Fort Lewis Tank 8C25 South.  However, it should be noted that the ARCO
system tested tanks at approximately 75% full.   Thus, the ARCO system may pro-
vide results more representative of the usual operating conditions of the tank.
However, the ARCO method would not detect a hole in or near the top, in the
fill pipe, or a leak in the lines.   Hence, the reporting of no leaks by ARCO
is not necessarily inconsistent with the fact that other test methods reported
leaks in the same systems.  In fact, the Pitstop South tank was excavated and
found to have a leak at the fill pipe flange.

               The data in Table 12 indicate that the ARCO test method per-
formed without difficulty at the Damneck, Pitstop North, and Fort Lewis 8C25
South test facilities.   If an apparent outlier is removed from the data, the
method also performed well at the Langley facility.  One of the sites (Scott
Tank 18) showed no apparent trend between the reported leak rates and the
simulated leak rates.  One other test, at Fort Lewis, gave a slope substan-
tially different from one.

               These results indicate that the ARCO method can give a precise
determination of a leak rate under some operating conditions.   It is limited
to detecting leaks in the lower 75% of the tank.   It can detect inflow or out-
flow, but would be defeated if the water table were at a level that approxi-
mately 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 dis-
advantage of not being able to detect potential leaks in the upper quarter of
the tank.   The test relies on a very sensitive level  monitor to detect changes
in the level of the product, so it can be disturbed by vibration of or dis-
turbances in the liquid level in the tank.  It is self-compensating for tem-
perature.

               b.   Leak Lokator Method

               The Leak Lokator method was used to test 10 tanks.   Of these,
two tanks had only baseline tests and no simulated leak tests conducted.  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 did not certify three tanks
that were determined to be tight and did certify one tank that was determined
to be leaking.

               A summary of the results from the leak simulation tests using
the Leak Lokator method is presented in Table 15.  Sample results are shown
in Table 16 and Figure 13.  The leak rates were reported by Leak Lokator over
approximately a 10-min period.  Several leak rates were determined during each
simulated leak period.   These rates were regressed on the simulated leak rates.
Table 12 lists each tank and its baseline leak rate reported by Leak Lokator
before any simulated leak tests were run.


                                    53

-------
        Table 13.   Results  of  Leak  Simulation  Tests  Using  ARCO Method'
Tank
Oamneck
Pitstop south
north
Scott 18
Fort Lewis
8C25 southb
8C25 north
Langley
MoGas
Baseline
rate
(gal/h)
0.02
0.02
0.0
0.02

O.Q
-0.04

-0.03
-0.03c
Intercept
(gal/h)
-0.023
-
-0.092
-0.145

-0.005
-0.094 '

-0.336
-0.027
Bias
(gal/h)
-0.003
-
-0.092
-0.165

-0.005
-0.054

-0.306
0.003
Slope
1.049
-
0.809
-0.044

1.140
0.493

0.419
1.167
SE
(gal/h)
0.022
-
0.041
0.099

0.047

0.367
0.118
RMS
(gal/h)
0.023
-
0.101
0.192

0.072

0.478
0.118
.Negative - leak out;  positive
 Two points only.
 Outlier removed.
= leak in; bias = intercept - base.
                                      54

-------
      Table 14.   Sample of Simulated Leak Data for ARCO Test at Oamneck
      Test
 Rate
(gal/h)
SIM rate
(gal/h)
Base rate
 (gal/h)
ARCODN.SAS; 1
Damneck
m = 1.12, y1  = 0.01
-0.110
-0.370
-0.030
+0.020
-0.170
 •0.075               -0.035
 •0.319               -0.051
  0.000               -0.030
  0.000               +0.020
 •0.166               -0.005

              Mean =  -0.020
           Std Dev =   0.025
                                    55

-------
-0.32   -«
.0.24    -0.2   -0.16   -0.12   -C.<
     Simulated Laak Rote (gal/h)
                                                       -°'04
      Figure 12.  S1.ul.Ud lea, data  for ARCQ test at DamnecK.
                                  56

-------
   Table 15.   Results of Leak Simulation Tests Using Leak Lokator Methodc
Tank
Damneck
(@ 120")
Pitstop south
north
Scott 17
18
Fort Lewis
8C25 south
4194 ,
10E10 NTCD
TCC
Lang ley
HS 3
HS 5
Baseline rate
(gal/h)
+0.0775 @ 125
(+0.008 (9 118)
-0.524
-0.012
-0.299
-0.178

-0.027
-0.171
-0.191
-0.191

-0.448
-3 or more
Intercept
(gal/h)
-0.0825
(-0.005)
-
-0.026
-0.366
-

-0.010
-0.159
-0.596
0.069

-0.641
0.126
Bias
(gal/h)
-0.005
(-0.13)
-
-0.014
-0.067
-

0.017
0.013
0.405
0.260

-0.193
0.126
Slope
0.786
-
0.879
0.839
-

0.734
0.749
0.541
0.835

-1.78
2.43
SE RMS
(gal/h) (gal/h)
0.025
0.209
0.015
0.048
0.047

0.097
0.026
0.165
0.098

0.048
0.304
0.0255
(0.028)
-
0.021
0.082
-

' 0.099
0.029
0.437
0.278

0.199
0.329
 Negative = leak out;  positive = leak in;  bias = intercept of their (adjusted
^for base) regression; intercept = bias plus base.
CNTC - not temperature corrected.
 TC - temperature corrected by Leak Lokator.
                                    57

-------
Table 16.   Sample of Simulated Data for Leak Lokator Test at Pitstop North
                                                                              •Q
Test
LOKLEAK.SAS; 4
Pitstop north
m = 0.882, y1 = 0.012


















Rate
(gal/h)
-0.289
-0.283-
-0.285
-0.297
-0.294
-0.258
-0.289
-0.068
-0.072
-0.080
-0.077
-0.209
-0.188
-0.198
-0.209
-0.201
-•0.197
+0.028
^•0.027
+0.022
•fO.018
SIM rate
(gal/h)
-0.351
-0.351
-0.351
-0.351
-0.351
-0.351
-0.351
-0.098
-0.098
-0.098
-0.098
-0.224
-0.224
-0.224
-0.224
-0.224
-0.224
0.000
0.000
0.000
0.000
Base rate
(gal/h)
+0.062
+0.068
+0.066
+0.054
+0.057
+0.093
+0.062
+0.030
+0.026
+0.018
+0.021
+0.015
+0.036
+0.026
+0.015
+0.023
+0.027
-0.028
-0.027
-0.022
-0.018 .
                                                           Mean =  -0.038
                                                        Std Oev =   0.022
                                    58

-------
      0.05 -i
  "5
  o>
  in
  O
-0.05 -





 -0.1 -





-0.15 -





 -0.2 -





-0.25 -





 -0.3 -
    -0.35
         -0.4
                      •0.3            -0.2             -0.1


                           Simulated Leak Rate (gal/h)
Figure 13.   Simulated leak  rate  data for Leak  Lokator test at  Pitstop North.
                                       59

-------
               The RMS errors for the Leak Lokator simulated leak results
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 end of the range are from a test that MRI's analy-
sis suggests was not valid.   No indication of questionable results was re-
ceived from Leak Lokator.  If that data point is excluded, the upper ends of
the ranges become 0.048 and 0.082.  These error estimates refer to the leak
rates as reported by Leak Lokator.  With the ability of Leak Lokator to ob-
tain 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.  Provided that the
estimates are consistent, this should reduce the error and improve the pre-
cision of the test method.

               The Leak Lokator method records a level measurement determined
by a float and an analytical balance on a strip chart.  The chart is cali-
brated by rapidly inserting and removing a calibration bar of known volume
and observing the displacement of the line on the chart.  The leak rate is
determined by fitting a straight line visually to the strip chart tracing,
reading its slope in divisions of the chart per minute, and converting this
to a volume change by the calibration.  Temperature compensations can be made
on the basis of a single thermistor located at the center of the tank.  (The
Leak Lokator manufacturer has stated that it uses three thermistors .in large
tanks, but its test crew did not do so in any of these tests.  In most of
these tests they did not use any temperature compensation in estimating the
leak rates.)

               Fitting the straight lines visually and reading the slopes from
them introduces an observer factor into the leak rate estimation.  In the ab-
sence of other data on inter-observer variability, MRI had an investigator
with a degree in mathematics independently draw the slope lines and estimate
the slopes.  Only those data selected by Leak Lokator for leak rate estima-
tion were used in this comparison.  The values obtained by MRI and the values
reported by Leak Lokator were compared.  The estimate of the inter-observer
variability was 0.04 gal/h.  While this represents only limited information
on inter-observer variability, and while that variability might be smaller
for trained personnel experienced in the method, it is large enough to cause
concern relative to the desired criterion of 0.05 gal/h.  A substantial im-
provement in the method would seem to be to convert the strip chart data to
digital form so that a slope could be estimated mathematically—say, by least
squares—to remove this source of possible variability.

               Additional analyses were performed on the raw Leak Lokator data.
While the data do not provide a continuous or exactly periodic measurement of
volume changes, they nearly do so when there are no problems.  Gaps occur when
the strip chart scale is  reset and when the balance used to monitor the buoyancy
of the level sensor is adjusted back to scale.  This would typically be about
a minute on each end of  a 10- to 12-min level measurement.  Thus, time  series
analyses can be used to  shed additional light on the performance of the method.
                                    60

-------
               A comparison of the leak rates measured by Leak lokator and
the simulated leak rates for the test at Pitstop North is displayed in Fig-
ure 14.   The differences between the simulated rates and those reported by
Leak Lokator are plotted over time in Figure 15.  The fact that the data are
not equally spaced reflects that the determinations of slopes were based on
slightly different time intervals.  Also, gaps of somewhat different lengths
occurred as the scale was reset.

               The leak rates reported were weighted by the times, interpolat-
ing linearly through the gaps, and summed to give a cumulative volume change.
This is plotted in Figure 16.  Since this is based on the leak rates reported
by Leak Lokator, it should represent that method's temperature compensated
leak rate determination.  Leak Lokator1s temperature compensation appeared to
be based on two points an hour apart.  Also plotted in the figure is a cumu-
lative volume change that one would expect from thermal expansion (contrac-
tion) from the temperature changes as noted on the strip chart.  It should be
noted that substantial differences could occur depending on which two points
are used for temperature.

               Use of a smoothing technique for the temperature data, followed
by use of the smoothed temperature•data for temperature compensation might
result in smaller variability of the Leak Lokator method.  This would neces-
sitate use of a fairly large number of individual leak rate estimates over a
period of 2 h or more.  It would also be desirable to keep the data gaps as
small as possible, as large gaps would complicate and possibly invalidate this
type of analysis.  Simple use of several leak-rate determinations averaged
would reduce the standard error.  Alternative methods of temperature compen-
sation might improve the precision and accuracy of the method as well.

               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 sys-
tem to obtain leak rate determinations in about 10 min once the test is run-
ning would allow multiple determinations and averaging to reduce this varia-
bility.   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 done below the level of such leaks.  If
such testing is done inside the tank rather than in the fill pipe, sensitivity
is greatly reduced because much smaller level changes would be produced by a
leak of a given size.

               In addition, hydrostatic pressure from the water table could
pose a problem for this test.  No check on water table was made.  While the
ability of the method to test at different levels is an advantage, it can
also present difficulties.  Testing did not appear to be standardized to any
specific 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.
                                    61

-------
                                                          s_
                                                          o
                                                          0.
                                                          o

                                                          (f>
                                                          *J
                                                          •r*•
                                                          a.


                                                          to
                                                          I/I
                                                          .

a
                                                   o
                                                          O)
                                                          i.
                                                          o
                                                          u
                                                          (O
                                                          0)
                                                         T3
                                                          OJ
                                                          3
                                                          a
                                                         •a
                                                          c
                                                          01
                                                          (O
                                                          i-
                                                         01
                                                         
-------
                                                            o
                                                            <0
                                                                           i-
                                                                           o
                                                                           a.
                                                                           o
                                                                           3)
                                                                          •M
                                                                          o
                                                                          +J
                                                                          fO
                                                                          jx:
                                                                          o
                                                                          fO
                                                                          O)
                                                                          5-
                                                                          O
                                                                          (T3
                                                                          i.
                                                                          ra
                                                                          0)
                                                                          (V
                                                                          in
                                                                          3
                                                                          C
                                                                          0)
                                                                         (13
                                                                         n:
                                                                         a>
                                                                         O)
                                                                         O)
                                                                         S-
                                                                         3
                                                                         cn
63

-------
                                                                       s_
                                                                       o
                                                                       Q.
                                                                       O
                                                                      4J
                                                                       in
                                                                      a.

                                                                      it?
                                                                      4J
                                                                      ai
                                                                      4J

                                                                      O
                                                                      ITJ
                                                                      O
                                                                      _J

                                                                      01

                                                                      o
                                                                      4_
                                                                      (U
                                                                      c
                                                                      TJ
                                                                      U
                                                                      01
                                                                     o
                                                                     OJ
                                                                     's
                                                                     O)
                                                                     3
                                                                     O5
SUO||D£)
       64

-------
               c.   Petro-Tite Method

               The Petro-Tite method was used to test nine tanks during the
development study.  Four of the systems tested had leak rates so large that
simulation of additional leak rates on the order of 0.2 gal/h was not feas-
ible.   Simulated leak rate testing was performed on five tank systems.  On
these systems the analysis of the simulated leak rate data was done, regress-
ing the measured leak rates on the simulated leak rates.   Additional data
analysis, utilizing cumulative volume changes, after subtracting simulated
leak rates, was also done on these data sets.

               The standard point-to-point temperature compensation consists
of measuring the temperature change, converting it to a volume expansion or
contraction by multiplying by the thermal coefficient of expansion and the
volume of the tank, and then for each point subtracting this change from the
measured volume change.

               A summary of the results from the leak simulation tests using
the Petro-Tite method is presented in Table 17.  Sample data are shown in Table
18 and Figure 17.

               The Petro-Tite test results agreed with the MRI conclusions
about the presence of a leak for all tank systems tested except one.  On this
test field observation documented a leak in a manway gasket that began after
Petro-Tite had finished the standard test.  The method was also able to per-
form a test of all tank systems for which testing was attempted.  However, in
some cases, substantial  line leaks were found, indicating a system leak that
would have to "be repaired or isolated before a separate test on the tank could
be done.

               For two tank test sites with large estimated leaks (Fort Lewis
8C25 South and Pitstop South) it was possible to make a separate estimate of
the standard error of the leak rate even though no simulated leak testing was
done.   This was possible because the Petro-Tite method made several independent
measurements of the leak rate over 15-min intervals.   The variation among these
IS-^min rates yields an estimated standard error.

               The root mean square errors for the Petro-Tite results ranged
from 0.036 to 0.193 for tanks judged to be tight.  The 0.193 is rather large.
That tank posed special  problems.  Its analysis was discussed in more detail,
leading to the conclusion that the 0.193 is not representative.   Error esti-
mates 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.  It should be noted that these error measurements
represent an estimated error for a single determination of the volume change
rate.   The Petro-Tite test uses an average of several determinations, so the
effective error is the standard error rather than the standard deviation.
The standard error is calculated by dividing the standard deviation by the
square root of the number of rates averaged.
                                    65

-------
   Table 17.   Results  of Leak Simulation  Tests  Using Petro-Tite Method'
Tank
Oamneck
Pitstop south
north
Scott 17
18
Fort Lewis
8C25 south
4194
10E10
Lang ley
golf course
Baseline
rate
(gal/h)
+0.003
-2.89
+0.050
+0.004
-0.812

-0. 342
-3.0
-0.024
-2.54
Intercept Bias Slope SE
(gal/h) (gal/h) (gal/h)
-0.009 -0.012
-
+0.069 +0.019
+0.002 -0.002
-0.774 0.038

(Could not
-•0.038 -0.014
(Could not
1.01
-
1.26
1.075
0.608

fill tank)
1.50
keep filled)
0.052
0.240
0.078
0.036
0.109

0.107
0.193
-
RMS
(gal/h)
0.054
-
0.075
0.036
0.115

0.193
-
Negative = leak out;  positive = leak in;  bias = intercept - base.
                                   66

-------
Table 18.   Sample of Simulated Leak Data for Petro-Tite  Test
                    at Scott AFB,  Tank. 17

          Rate                          SIM rate
        (gal/h)                         (gal/h)
          0.004                           0.000
          0.044                           0.000
          0.004                           0.000
         -0.036                           0.000
         -0.076                          -0.078
         -0.076                          -0.078
         -0.076                          -0.078
         -0.060                          -0.078
         -0.180                          -0.192
         -0.160                          -0.192
         -0.220                          -0.192
         -0.076                           0.000
          0.080                     .      0.000
         -0.020                           0.000
         -0.040                           0.000
         -0.360                          -0.321
         -0.320                          -0.321
         -0.340                          -0.321
         -0.380                          TO.321
         -0.360                          -0.321
                           67

-------
        0.1 -

       0.05 -
   •g  -0.05 -
   >—»
   »   -0.1 -
   o
   (X
   x  -0.15 -|
   o
   -3
   •o   -0.2 -
   H
   '8  -0.25 M
   o
   2
       -0.3 -

      -0.35 -
-0.4
   -0
                                                                 Q

                                                                 a

                                                                 a
34   -0.3   -
                                r   r   i    i    i    \   Y   i    r   i    t    \
                           0.26   -0.22  -0.13  -0.14  -0.1    -0.06  -0.02
                                 Simulated  Leak Rate (gal/h)
                                                                                        -r\.
                                                                                        " - i
Figure 17.   Simulated leak  rate  data for Petro-Tite test1 at  Scott AFB Tank  17.
                                     68

-------
               Although the estimated errors for the Petro-Tite method are
not too large, the nature of the data suggests that they might be reduced by
an alternative sampling and an-alysis.  Clearly the error associated with a
leak rate estimate can be reduced by taking additional data over a longer
period of time.  In addition, the procedure for temperature compensation used
in the standard method is to subtract a volume change due to temperature for
each period from the observed volume change.  Since the temperature related
volume change is subject to sampling error in the determination of the tempera-
ture change, the temperature compensated volume changes have variances that
are the sum of two terms.  One is from the error associated with the measure-
ment of the volume change, and the other is from the measurement of the tempera-
ture change.  Use of more frequent volume determinations, use of a larger number
of volume determinations, or smoothing the temperature data before calculating
the temperature correction all could reduce the error estimates.

               Figure 18 is a plot of the observed volume changes each 15 min
compared with the simulated leak rates from the Petro-Tite test at Damneck.
Some variability of the measured volume changes about the simulated rates is
apparent, but there is close agreement.   Leaks out of the tank are denoted by
negative values.

               The temperature corrected leak rates with the simulated leak
rates subtracted are plotted in Figure 19.   In this case, the standard point-
to-point temperature compensation used by Petro-Tite in its standard test
method was used.  Again, while there is some variability, the mean is close
to zero and the variability is relatively small.

               The cumulative volume changes and the temperature induced vol-
ume changes (also cumulative) are presented in Figure 20.  This figure shows
that in this tank test there was a relatively smooth temperature change and
it was uniform throughout the tank.   Clearly, the temperature is accounting
for the volume changes except for some random error or noise.  Although the
point-to-point temperature compensation used by Petro-Tite worked satisfactor-
ily in this case/ the variability illustrated in Figure 19 can be reduced by
a better temperature compensation algorithm.

               Data from the Petro-Tite test on Fort Lewis Tank 10E10 are pre-
sented in Figures 21 through 25.  Negative values denote leaks out of the tank.

               This tank system test represents a case where the standard
analysis was clearly insufficient.  The reasons for this conclusion and a more
detailed analysis are illustrated using the figures.  The measured volume
changes and the simulated volume changes are plotted in Figure 21.   The data
show considerable variability around the actual leak rates.  The volume changes
with the simulated leak rates subtracted out are presented in Figure 22.   This
figure illustrates the variability in the volume changes at 15-min intervals.
The result of the temperature compensation is shown in Figure 23.  In this
figure both the temperature compensated volume changes (or leak rates) and
the temperature compensation are shown.   Two observations are apparent from
the figure.  First, there is a negative serial correlation between the tempera-
ture compensation and the "corrected" series.  Second, the use of this method
has added variability to the data.  Neither of these features is desirable.
Temperature adjustment should remove dependence of the data on temperature,
not add it, and one wants to smooth the variability in the data, not increase
it.
                                                       \
                                    69

-------
     0.4p-
    0.28
    0.16
o
I   0.04
   -0.04
   -0.16
   -0.28
                                      Time (Hours)
           Observed Volume Changes
           Induced Volume Changes
     Figure 18.  Measured and simulated leak rates (Petro-Tite test at Oamneck),
                                       70

-------
     0.2i-
    0.12
    0.04
o
X
 o
O
   -0.12
    -0.2
                         W        V
                                         Time (Hours)
           Point to Point Net Volume Change
         Figure 19.   Temperature corrected measured leak rates (simulated
         leaks subtracted out) for Petro-Tite test at Damneck.
                                       71

-------
                                                                0
                    Time (Hours)
Cumulative Volume Changes
   Related Volume Change
                        72

-------
2.8 r-
               1
      Simulated Leak Rates
      Measured Leak Rates
2               3
     Time (Hours)
      Figure 21.  Measured and  simulated  leak rates (Petro-Tite
      test at Fort Lewis Tank System  10E10).
                                 73

-------
                                                                       *
                            Time (Hour?)
Volume Changes
                             74

-------
o
_o
~o
O
     10r-
      6-
 8.   -2-
        01                2                3
                                        Time (Hours)
        • Point to Point
        • Temperature - Related Volume Changes
   Figure 23.   Point-by-point temperature compensated  leak  rates  and  temperature-
   volume changes (Petro-Tite test at Fort Lewis Tank  System  10E10).
                                       75

-------
               The temperature induced volume changes are cumulated and
plotted in Figure 24-.   This illustrates an increasing temperature trend.  A         •"*>
smoothing technique is used on the cumulative temperature-volume effect.  This     •--•''
may be linear regression or a moving mean of some order.  In this example, a
linear regression was used.  After the temperature trend has been estimated,
it is used to correct the cumulative volume change.  The result is plotted in
Figure 25.  In Figure 25 the temperature compensated cumulative volume changes
are essentially zero,  with little random error for the first period of about
2 h.  After that point, there is a noticeable volume change, which corresponds
to a leak in this case of about 0.13 gal/h.

               Comparison with the original data revealed that the tank had a
manway.  The gasket on the manway was observed to be leaking (noted on the
data sheet) at 2 h, at which time the manway bolts were tightened.  After that,
the gasket around'the manway continued to leak product throughout the remainder
of the test.  This appears to be the source of the leak.  Statistical tests
on the first 2 h of temperature compensated data show no significant differ-
ence from zero.  The leak rate reported after the manway gasket began leaking
was significantly different from zero.

               As a result of the more detailed analysis of Petro-Tite data,
several suggestions for reducing errors involved in the method were developed.
One approach is to extend the time of the test, collecting more data to reduce
the error associated with the average leak rate measurement.  A second approach
is to increase the amount of data by making the volume change and temperature
change measurements more frequently.  There is a practical limit to this im-        ~;>,
provement; as the frequency of measurement is increased, some reduction in         .'_•-']
error size will occur until the error involved in releveling or measuring the  .
volume begins to dominate.  Attempts to obtain data more frequently would then
result in increased error.  A third approach is to smooth the volume and tem-
perature data before incorporating the corrections.  This would be a more
sophisticated algorithm for data analysis and would not require other proce-
dural changes, although maximum error reduction would be achieved through a
combination of these changes.  Different smoothing methods could give slightly
different error modifications.  The approach se-lected by MRI for standard
analysis of the data on the national survey is described next.

               The observed volume changes are accumulated.  Temperature
changes as measured by the thermistor are converted  to equivalent volumes
and these temperature-related volumes are also accumulated.  By the nature of
this process, both of these cumulative curves must pass through the origin.
A straight  line is fit to  the temperature-related volume data through the
origin by least squares.   The predicted values from this line (smoothed tem-
perature  effects) are used to correct the observed volume for temperature.
This is accomplished by subtracting the smoothed temperature effect from the
observed  volume.  The resulting temperature corrected volumes are converted
to gallons  per hour by dividing by the time interval.  The arithmetic mean
and standard error of the  mean are calculated.  The arithmetic mean is  the
estimate  of the leak rate.
                                    76

-------
      2r-
    0.4
 c*
 o

=5 -0.4

O
   -1.2
     -2
                    1
          Cumulative Temperature Effect
                                        Time (Hours)
  Figure 24.   Cumulative temperature related volume changes  (Petro-Tite  test at
  Fort Lewis  System 10E10).
                                      77

-------
                             Time (Hours)
Temperature Compensated
Cumulative Volume Changes
                                         10-1) after removing smoothed
                                  78

-------
               A set of diagnostic procedures was implemented to ensure that
the line fit the temperature data adequately.  If curvature or evidence of
lack of fit was found, special analysis was done for that tank.  These tech-
niques might involve using a quadratic curve to fit the data or using moving
mean smoothing before the temperature correction was done.

               The rationale for smoothing only the .temperature data follows.
The volume changes must reflect the sum of any thermal expansion or contrac-
tion plus the losses resulting from any leaks.   Thus, the volume measurements
are based on an integrated average temperature behavior.   However, the tem-
perature measurements are made by a single thermistor at a single point in
the tank.  The circulation of the product brings different product by this
fixed point.   Consequently, any differences in temperature on a microscale
will be noted by the thermistor to its degree of response (about 0.003°F).
Thus, the temperature-related volume data to be used for temperature compen-
sation is based on a local effect rather than the whole tank average.   As a
result, it would be expected to be more variable than the volume data.   Con-
sequently, smoothing the temperature data to make it comparable to the inte-
grated average of volumes over the tank is appropriate.

               The Petro-Tite method seems capable of identifying and success-
fully 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 valid tests were always obtained.  However, difficulties were en-
countered that increased the error associated with the estimated leak rates
beyond that which is desirable.  In difficult cases, large error rates rendered
measurements on the order of 0.05 gal/h unreliable.   In addition, large error
rates were often associated with large leaks.  Most of the situations with
large error estimates were cases where a substantial leak was present.   The
leak was sufficiently large that the loss in precision did not interfere with
the detection of the leak.

          4.   Ambient Noise Analysis Results

          Since insufficient data from the ARCO tests were received to be able
to perform ambient noise analysis, only the Leak Lokator and Petro-Tite results
were analyzed using this technique.

               a.   Leak Lokator

               A summary of the mean and 95% confidence-intervals on the mean
volume rate,  temperature rate, and temperature compensated volume is presented
in Table 19.   The site, tank number, duration of the test sequence, the number
of Leak Lokator volume rate measurements in the test sequence, and the test
results based on Leak Lokator's 0.05 gal/h criterion are also given.   For com-
parison, the baseline test results are shown.  Several observations about the
data presented in Table 19 are noteworthy.  First, the test sequences 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 se-
quence was used.  The results of the other tank test (Fort Lewis Tank No.  3)
indicate that the tank is leaking, but it could not be determined whether the
flow is into or out of the tank.  Second, temperature, volume, and temperature


                                    79

-------




















in
in

ITJ
c
<£.

i.
O

in

o


ta

1 X


t i— t
Ol Ol CJ
3 nj <-* **
*•> t- .c un
TO ^ en
S. 01 r-
Ol S tl
Q. 3 0!
S r~ *^^
o o
r— >
IX




Ol CJ
•^^
TO <••* a<
s_ j= un
-v. en
01 •—

3 S

O
^>

1 ^^


z


TO Ol C
4^ S *r-
Ov- g
h- -M x_/
01
U
*J C
in 01
01 3
t— cr
01
in
c
r-

C
O
*f™
^J
TO
U
O
—I

3 C
O ••-
O) Ol
c c
.2 .c .c J* ic
TO o) 01 TO 01
Ol -t- -i- 01 -r-
QO co r*>- 10
O r-t CM r-4
0000

o' o* d d



e3 ^^ ^^ un
^ co 00 CD
o o o o
d o' d o'
i i


r*. r-t un r-4
O O O O

d d d d



^H ^^ ^^ CT
^H l^ ^^ C^
O O O C3

d o* d ca
1 1 1 1


^9* ^^ ^^ C3
O O CM i-4
O O O C3

d d d d



r*>» r« 4 un co oo
f*« un rH 
01
M
(O
03
~



^^
U
Ol
c
S

o
3 3
O O
Ol Ol
c c
JT .H 12 -C
Ol TO TO Ol
•r- Ol CO -r-
rH O CM
IO O CO
0 r-40

odd



00 O 00

r-4 r-4 O
d o* o*
i i


en 10 <~o
o o o
• • •
o o o



un VO r-4
r-t en o
CM O O

o'o* d



•«r oo o
rH CM CM
O O 0
* * *
000



r*% i^1* ^^ en
CM rH ^ rH
O O O O

o o o o
1 1

«3-co «J-


«* un o
^" CM <^


Ol
c

r— r-4 CM CO
Ol
in
TO
CO
CM
in

^
01


.^J
S.
o
u.
3 3 3 C
a o o -i-
Ol Ol Ol Ol
c c c c
'<^ 3 '^ '^
fO ^3 ^9 fO
(U Ol Ol Ol
CO CM IO
r— 1 CO ^
000

odd



CO 00 CM
^^ 00 ^O
rH O O
o' d d
1 1


«3- O rH
O O O
• • •
o o o



CO rH ^T
^^^ G3 ^3
O O rH
• • »
O O O
1 1

•
CM rH O
rH CO CM
O O 0

d d d



CM o en CM
P** CO CQ ^f
rH rH O O

O O O O
lilt

un un r»>


r^ un co
co co ^f


0)
c.
•f»
r- rH CM CO
01

10
03
CO
u»
tfff
3t
01


^J
S.
o
u.
3 C C 333333
O-r--f- OOOOOO
Ol O1O1 O1O1O1O1O1O1
c cc cccccc
_!^_r;_ij_^ ^^j— _C-ii ^- v/ ^ j^ Y
TO Ol TO TO Ol Ol Ol Ol TO TO TO TO TO TO
g) •*- aj (Jj -i— -i— -i— -r- D 0) Ol Ol O) Ol
uoo^o uncncnun OI-H^-O
{Q t**» CO CO f"**» P*» ^J" O3 rH CM ^"
ooo oooo oooo

odd dddd do' do



00 00 rH f*» CO CO 00 rH CO CO ^"
^* 00 UO rH C3 O UD ^O en CO rH
OrHrH OOOrH OrHCMCM
odd odd d d o' o' d
tit i t t i


rH^o coooooun unoo ^ ^*
ooo oooo oooo
• •• **** *•»•
ooo oooo oooo



rHrHCO CniOP^-rH CMCMOOOO
uncoo r*.oun«M uocooo
CMOO OOOCM CMOOO
• •• . t . . ••••
ooo oooo oooo
III II


CMoof cnunr*»un cncounr^
unifiCO OOOO ^MOOCO
ooo oooo oooo

dod o'odo dooo



rH CO P*"» 00 CM VO CO ^f CO Cn CO un rH (^O
cnounun i—ienununun cncMcsj^o
rHCMrHi-H OOOOO CSJCOCMCMCM

OOOO OOOOO OOOOO
r i i i i i i i

LO **o un r^ ^^ ^o un uo uo un* un


IJOP-I^H oo^-rxuo un^-iauo
un^^ ^^"Tun un^unco



c c c

r-i— 1 CM CO i— rH CM CO ^T r-t— ICMCO^S"
Ol Ol Ol
in in in
TO TO TO
CO CO CD
«± CM rH

•r— CO
y LJ-
0)0. <
—I 0

4-> (/> 4->
i- +J O
O -r- U
u. a. oo
80

-------
compensated volume rate data exhibit a large range of variability compared to
0.05 gal/h.   The high variability in the temperature compensated volume rate
suggests that the test time is too short and a single thermistor is not ade-
quate 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 (a) an average of four to nine Leak Lokator volume rate
measurements were used instead of one and (b) the average rate of change of
temperature over 1 h was determined by fitting a least squares line to 5 to
10 temperature values over the hour instead of the two end points.   The un-
certainty' in the Leak Lokator temperature compensated volume rate results pre-
sented in Table 19 is about a factor of 5 smaller than the uncertainty of a
single 10 min volume rate measurement and a two-point temperature rate mea-
surement.

               The time series plots of temperature (converted to volume) and
uncompensated volume were generated for each of the 21 sequences of Leak
Lokator data.  Cumulative time series plots illustrate the reasons for the
inconsistent test results and the high variability.   The shortcomings of using
a single thermistor to compensate for temperature are illustrated in Figure
26 for Pitstop Tank No. 2.  Since the tank is not leaking, the volume and
temperature changes should be approximately equal.   Figure 26 shows the re-
sults of the first sequence of measurements on Pitstop Tank No.  2;  seven
volume rate measurements were combined over 48 min.   The temperature compen-
sated volume rate is 0.017 gal/h.  The temperature and volume data are nearly
equal, suggesting that temperature accounts, as it should, for the total vol-
ume change.   The test results would not change significantly for these data
if only a two-point estimate of tempera'ture were used.  Figure 26 also shows
the results of a 56-min test.sequence on Pitstop Tank No.  2, Sequence Mo. 4,
starting approximately 1.5 h later.  The predicted temperature compensated
volume computed for these data is -0.168 gal/h, significantly different from
0.017 gal/h.  Clearly, the temperature changes are not correlated with the
volume changes.  Because the other two test sequences on this tank are rea-
sonably consistent with the first sequence and the volume changes are nearly
uniform, it appears that the single thermistor is not adequately estimating
the rate of change of temperature.

               Similar results were observed in other tank tests.  For example,
the data from the third and fourth test sequence of the Damneck test (a tight
tank although incorrectly declared leaking by Leak Lokator) presented in Fig-
ure 27 and from the first and third test sequence of the Fort Lewis Tank No.
2 tests (a leaking tank incorrectly declared tight by Leak Lokator) presented
in Figure 28 illustrate the same point.  The difficulty in using a two-point
analysis approach is clearly obvious from Figure 29 for Sequence No. 2.   De-
pending on which two points on the temperature line (solid) are taken to de-
termine the slope, large differences in slopes would be estimated.   Hence,
the temperature compensation may be highly unreliable.
                                    81

-------
                                                                    e\j



                                                                     O*
                                                                     Q.
                                                                     O
                                                                     (O
                                                                     1.
                                                                     o
                                                                     O)
 o

ZZ
•a  .
£  2

 -  I
 2  ?
0
•
o


00
o
o

|o6 
-------
.*   o

~»  z

 i   *
a   §
                         c-
                         a
3
o
icS 'auinioA
s
o
aAMoinwn-) -
\
o
tM
CN
O

3
o
1

                                                                                                                u
                                                                                                                
                                                                                                                s.
                                                                                                                3
                                                                                                                en
                                             83

-------
CM
 •
 o
II

o
o

3.
o

o 	 .°. —
J 	 _
0*
'_ 	 ' —
	 s
o

o
o
c

-------
                                                                                                                                    c
                                                                                                                                    
-------
               b.   Petro-Tite

               A summary of the mean and 95% confidence intervals on the mean
volume rate, temperature rate, and temperature compensated volume rate esti-
mated from the long Petro-Tite time series is presented in Table 20.  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-Tite1s 0.05 gal/h detection criterion are also
given.  For comparison, the baseline test result is added to the table.  Agree-
ment between the baseline test results and the long time series results is
good.  The one discrepancy, Pitstop Tank No. 2, will be discussed below.  The
time series from the Fort Lewis Tank No.  4 test was divided into two segments
after a leak was identified that had begun several  hours after the test was
started.

               The time series of volume, temperature, and temperature com-
pensated volume, generated by removing the simulated leaks from the Petro-
Tite volume-time series, are presented in Figures 30 through 33.  The time
series are three to six times longer than the standard 1-h Petro-Tite test.
The first hour of each time series contains the baseline data.  Several ob-
servations about the strengths and weaknesses of the method can be made from
the data.

               First, the time series for Oamneck Tank No. I (Figure 30) and
for Scott AFB Tank No. 1 (Figure 31) illustrates the high correlation between
the trends of the temperature and volume data required for temperature compen-
sation.  This suggests that the method of temperature compensation, i.e., cir-
culation of the product and measurement of the rate of change of temperature
with one temperature sensor, worked adequately.

               Second, negative, serial correlations were observed between
the temperature and temperature compensated volume rate time series for some
of the tests.  This suggests that the method could be overcompensating for
temperature effects.  This is illustrated well 'by the data obtained from the
Fort Lewis Tank No. 4 test data shown in Figure 32.  These temperature fluctua-
tions are probably caused by inadequate resolution of the Petro-Tite tempera-
ture sensor, although inadequate circulation or incomplete mixing could also
be a cause,  the sensor resolves changes of 0.003°F.  A volume change of 0.021
gal would result from an temperature change of 0.003°F in a 12,000 gal tank
with gasoline.  This increase in the fluctuations in the temperature compen-
sated volume data resulting from the point-to-point correction can be a problem
if the test time is too short.  In a short test, these fluctations can adversely
influence the estimate of the trend.  This is clearly observed in the baseline
test conducted on Fort Lewis Tank No. 4 shown as the first hour of the data
in Figure 32.  Similar correlations between the temperature and the tempera-
ture compensated volume time series shown in Figures 30 through 33 were ob-
served for the tests performed on Damneck Tank No.  1 (Figure 30) and Pitstop
Tank No. 2 (Figure 33).
                                    86

-------





























in
•r—
to
TO
C
^£

CU
4^
IfH
1—
1
o
i.
+J
0)
a.

0)
5
4_
O
>k
i.
TO



OO


.
o
CM
TO
I—












































I/)
4J
H—
3
I/)
cu
s-

4-1
U)
CU
*"*
CU
c
in
CU
in
TO
CD




C
CU O
i. T-
3 4->
4J TO
TO in
i- C
4) CU
to.
s
cu o
t— u



1
£!
3
4>J
TO
S-
cu
a.
g
cu
r-















































0) -O
S- CU
3 4-1
-U TO
TO 10
s_ c
cu cu
CL a.
E S
a> o
r- U














cu
4J
TO ^^
i- .C
"^^
0) i—
§TO
OJ

'o
>





0)
TO /— «.
s_ .c
^x^
CO i—
g TO
3 O!
,« ^^
O
>



CU
+J
CO ^.
J- JZ
^Sm,
cu f—
S TO
3 O!
^«MI \^^
O
^>



























+J
•(-> I—
IO 3
CU V)
h- CU
i.
OJ

TO ^^
1. S.
^.

e "TO
3 O)

*O
>


+J
;•••
3
O)
s_

( '
IO
cu
r-

i— t
U

^p
Lf)
cn






IX



_
0
^O
Lrt
G\






IX


HH

^^
LO
cn






IX




•z.


££
c
uZ




C
o
•r-
4-1
TO
U
O
—I

3 33
O O O
O) O5 OJ
C C C
^> *p* ^J ^J +J '^ *r~
.!=-*.= J= ^3 -^ 0^
rjj TO C^ OJ C3) TO TO
•p» CU 'r" *i~ **•• cU CU
r- — 1 r- r- r- — 1 — 1


CM
CO ^- «f ^f CM CM
O CO CM UfJ O Cn rH
OOO O O 00 03

O O O O O CM O
+ 11 1 II




4_> (J 4-> 4-> 4-J
3 C 3 C 33
O •>- O -r- O O
O) OJ O) OJ OJ OJ
c c c c c c
4-1 -I— 4-1 -r- -1— -f— 4-) •!— -I—
JH v/ r* \x ^^ NX r* vx ^^
OJ TO OJ TO TO TO OJ TO TO
•r« CU *r* CU CO CU ''^ CO CO
1— -J r- — J — 1 —1 r- —I — 1


co o r*»r*.tn i— i m cn ao
O Lf> CMrHCO rH O OO CM
O O OrH^- O O O O

o' o' o' o o o o' o o



rH rH CM^J-00 CO IB CO. CM
CM r^* ^^ r^» cn co C3 r*^> CM
O CO O O O rH O P** r^
• » *•• • < . *
O O OOO O O CM O
1 1 1 1 III


00 00 !D CO O Cn ^" rH LO
O CO CMCO^- O rH UO O
0 0 0 rH 0 0 O 0 O

o o o' o' o' o o' o o'



*^ ^" P*1* rH {Q CO C5 Cn Cn
1C 00 i-HiTJO rH Cn r-. O
O O OOO O rH CM O

O O OOO O O O O'
1 1
oo vo CMLDP-. o r*». cn ^j-
O CO CMCOCM rH rH CO CM
O O OOO O O rH O

o" o' o o o o o o o



CO P^ in CO **f" rH ^* CO *!^
<3- 00 CMCMO LT> 00 Cn rH
O CM OOrH rH rH ^- r*.
« • ••• . . . .
O O OOO O O CM O
III II

o r*N CM cn co cn ^ oo UD
CM CM rH rH CM rH


TO J3
rH CM ^••=a-'3- CM rH rH CM




IO 10
•<—•!— en cn
22 u_ u_
-^ oi oi a. < a. <:
O — J — 1 O O
CU 4-> 4-1 4-1 4-J
C 4-> +J IO -t-> IO 4J
£ i. S- +J O +J O
TO O O *r- ,
-U l/l .—
co c
CO •*-> O
5 cu 2
-c o

o ^
(1^
^ O (+—
o
m ^
CM C
• m o
CM . -r-
CO 4->
•4-> U
IO -M CU
i. in i.
•i- TO •<-
U 	 1 Q
TO .a o
87

-------
             0.35r
             0.30


             0.25
        -3   0-20
         O)
         |   0.15
         _3

         >   0.10



             0.05


                0


            -0.05
_ Cumulative Volume
                 -Temperature Volume
                       r- Temperature Corrected Volume
                           60
                    120        180

                         t Minutes
240
300
Figure 30.   Time series analysis of the Petro-Tite data at  Damneck  Tank No.  1.

-------
    1.100
    1.000
    0.900
    0.800
    0.700
1  0.600
|   0.500
    0.400
    0.300
    0.200
    0..100
       0
  -0.025
       _
        o
                     Cumulative Volume
                                                 Temperature Volume
                                    \
Temperature Corrected Volume
                        60       120      180      240
                                      Time, Minutes
                       300
                                                              360
Figure 31.   Time series analysis of the Petro-Tite data at Scott AFB Tank  No.  1.
                                    89

-------
0
CO
o
 0.40



 0.30


 0.20



 0.10



    0



-0.10



-0.20


-0.30
                                          Temperature Volume
                           •Cumulative Volume
            \J
            V
                 •Temperature Corrected Volume
                 60
                       120       180

                        Time, Minutes
                                             240
300
   Figure 32.  Time series analysis of the Petro-Tite  data
   at  Fort Lewis Tank No.  4.
                           90

-------
o
0.70r

0.60


0.50


0.40


0.30


0.20


0.10
       0
   -0.05
                                       Cumulative Volume—\^.---"
                 Temperature Volume
                              Temperature Corrected Volume
                   60
                           120        180

                           Time, Minutes
240
285
    Figure 33.  Time series analysis  of  the  Petro-Tite data
    at Pitstop Tank No. 2.
                              91

-------
               Third, inspection of the temperature compensated volume rate
time series for each test suggests that a 1-h test with data taken at 15-min
intervals is too short to reliably detect small leaks.  Within a test, the
15-min data show fluctations with apparent periods of 30 to 90 min.  These
can be sufficiently different from the trend exhibited by the entire time
series to mislead the conclusions based on a 1-h test.

               There is a discrepancy between the baseline and long time
series test results for Pitstop Tank No.  2.   The baseline test declared the
tank to be tight (-0.05 gal/h) while the long time series exhibited a large
inflow (0.138 gal/h) into the tank.  The baseline result almost suggests that
the tank may have been losing product.  Since it is impossible to have water
flowing into the tank during a Petro-Tite test, the test result is suspect.
The 4.75-h time series shown in Figure 33 clearly indicates an inflow.  The
reason for the inflow is not known, but probable causes are thermal expansion
of a trapped vapor pocket, inadequate measurement of temperature, or struc-
tural deformation.

               The long time series obtained for Fort Lewis Tank No. 4 (Figure
32) suggests that the tank started leaking 2.25 h into the test.  This is con-
sistent with the observation in the field notes that a small leak was observed
around the gasket of the manway 2 h after the test began.  For this reason,
the time series was divided into two segments for analysis.  Although neither
rate is statistically significantly different from zero, the first 2.25 h in-
dicates a .flow into the tank while the second 3.25 h indicates a flow out of
the-tank.

          5.  Precision of Measurements by Leak Lokator and Petro-Tite

          The precision of measuring volume changes in a 4-in. fill tube with
the Leak Lokator method and in a 4-in. riser with the Petro-Tite test method
is estimated experimentally from the up and down displacements of the product
level produced by inserting and removing an object of known volume.  The pre-
cision, defined as the uncertainty in measuring a known volume change, is
estimated from the standard deviation of the absolute value of the resulting
changes in volume computed for both the up and down displacements.  These
types of data are collected by Leak Lokator as part of their standard tank
tests for calibration of their volume measurements.  Their calibration pro-
cedure requires inserting and removing a 25 cc or 50 cc cylindrical bar a
total of three times before each test.  The raw calibration data collected
during the development study field tests are used to estimate the precision
of Leak Lokator.  The Leak Lokator volume data are recorded in arbitrary di-
visions to the nearest hundredth on a strip chart and converted to gallons
using the displacement volume and the mean change on the strip chart to de-
velop a conversion factor.  Since the Petro-Tite volume measurements are made
with a graduated cylinder, volume calibration measurements are not routinely
made as part of their test procedure.  Special calibration tests, using a
0.0555 gal cylinder, were conducted during three of the Petro-Tite Pilot Study
tank tests to obtain the necessary data to estimate precision.  The Petro-Tite
data, sampled once per minute, were recorded in inches to the nearest hundredth
and converted to gallons using the displacement volume and the mean height
change to develop a conversion factor.


                                    92

-------
          The results of the Leak Lokator and Petro-Tite analyses are presented
in Tables 21 and 22.  Each table gives the test location, tank number, displace-
ment volume, number of displacements, and the standard deviation.  The mean
of the standard deviation for all the tests was obtained by taking the square
root of the average variance of each test after weighting by the number of
displacements.   The mean and median of the standard deviations are presented
in Table 23.  Both methods (Leak Lokator and Petro-Tite) have more than ade-
quate resolution for detection of 0.05 gal/h leak rates.

          A difference in the magnitude of the up and down height changes was
observed for two of the three Petro-Tite tests, so the mean and standard de-
viation of the up and down height changes were analyzed separately and are
presented in Table 24.  The data were converted from inches to gallons using
the mean conversion factor.  The data show a larger change when the product
level was raised by inserting the calibration cylinder.  The reason for this
difference is unknown, but the difference in the means is the reason that the
standard deviation computed from the combination of the up and down data is
larger than the standard deviation computed for either one.  This behavior
was not observed in the Leak Lokator data.

     E.   Discussion

          The important results of the development study testing are summarized
below:

          Each test method was applied to a wide variety of temperature pro-
files (see Appendix D).  It is expected that all of the methods would perform
best in tanks where uniform horizontal and vertical temperature changes, occur.
However, inadequate data exist to assess the frequency with which problems
occur or to recognize the pattern produced when unstable temperature condi-
tions are present.   None of the methods that were evaluated measure tempera-
ture profiles.   When erratic changes at the single thermistor are noted, addi-
tional  stirring (Petro-Tite) or stabilization time (Leak Lokator) is provided.
It is not known what effect erratic temperature behavior would have on the
ARCO tester.

          All three of the methods tested (Leak Lokator, ARCO, and Petro-Tite)
measure product level and temperature with sufficient sensitivity and resolu-
tion to detect 0.05 gal/h leak rates.  This can be shown by theoretical cal-
culation and was demonstrated during the leak simulations.   However, there
are many factors affecting the test data which are not measured during the
test (e.g., vapor pockets, unstable temperature behavior, etc.).   These lower
the precision and accuracy of the test.  For these reasons the noise levels
present during actual testing may preclude achieving consistent detection of
a 0.05 gal/h leak rate.  The Petro-Tite system provided the best capability
for identifying problems with the test.
                                    93

-------
      Table 21.   Estimates of the Precision of the Leak Lokator Method
 Location
    Tank
Displacement
   volume
   (gal)
  Number of
displacements
Standard
deviation
  (gal)
Oamneck



Fort Lewis



Pitstop

Scott AFB

Scott AFB

Total
1
1
1

2 (8C25 rear)
3 (4194)
4 (10E10)

2 (rear)

1 (17)

2 (18)
   0.0132
   0.0132
   0.0132

   0.0132
   0.0132
   0.0132

   0.0132

   0.0066

   0.0066
      6
      6
      6

      6
      6
      5

      6

      6

      6
 0.0016
 0.0009
 0.0003

 0.0008
 0.0009
 0.0017

 0.0006

 0.0002

 0.0002

 0.0010
                                    94

-------
       Table 22.   Estimates of the Precision of the Petro-Tite Method

                          DisplacementStandard
                             volume   '         Number of         deviation
Location       Tank          (gal)           displacements         (gal)


N30078D1        1            0.0555               17              0.0052

N2900601        I            0.0555               19        •      0.0027

N29010D1        1            0.0555               18              0.0055

Mean                                                              0.0046
                                    95

-------
   Table 23.   Summary of the Leak lokator and Petro-Tite Precision Results
                                               Standard deviation
                                      Mean                          Median
Test method                           (gal)                         (gal)

Leak Lokator                         0.0010   -                      0.0008
Petro-Tite                           0.0046                         0.0052
                                    96

-------







>^
^-™
»
(TJ
C in
^C *^
C
(TJ Ol
-u s
(TJ 0)
a u
TJ
0) r—
-u a.
•r- in
U« • ••
p~ *i™
I O
0
s_ c
•p 3
(U O
a. o

a> s-
JC O
-P M-

<*- T3
O C

C
O in
•^ 4J
•P C
(0 0)
> (U
gj U
£3 (TJ
^•w
•a a.
S_ U)
0} -r-
•a a
c
(0 a.
4-J ^i
CO
J_
•a o
c •+-
(TJ

c
(TJ
0)
^^

•
CM
(TJ







C






r~
QJ
>
0)
r™»

^^
(J
3
•a
o
s.
a

c
-r~
(U
in
(TJ
OJ
i.
u
a)
a





•a o
i. -r- /->
(0 +J >—
"O (TJ (TJ
C ••- O)
(TJ > v-<
•P  T3





/••*\
C -—
(TJ (O
(U D>
^^ ^_^





to
4-*
<+- c
O OJ
£
S- OJ
(U (j
-Q (TJ
S r-
3 a.
21 t/)
•5











P>*
a)

 o







in
•P
t- C
o 
-------
          Each method was used to test three to four leaking tanks and two to
three nonleaking tanks, based on MRI's determination.  The location, number,
and size of the actual holes and the true leak rates are unknown.  All three
methods erroneously declared at least one (not the same one) of the apparently
leaking tanks to be nonleaking although final analysis of the one case where
this happened for Petro-Tite showed that a leak started after Petro-Tite had
concluded the standard test.  Leak Lokator erroneously declared three of the
apparently tight tanks to be leaking.   ARCO found none of the tanks to be
leaking.

          The test period of each test method is too short to reliably detect
0.05 gal/h leaks with acceptable probabilities of detection and false alarm.

          The temperature compensation analysis algorithms are generally in-
sufficient.  However, the algorithms did provide sufficient compensation for
some test conditions.

          Sources of largest errors observed were test times that were too
short and temperature compensation that was insufficient.

          Reanalysis of the raw data over longer-than-standard test times and
with different analysis techniques and compensation algorithms improved the
detection performance over that obtained with standard techniques.

          A single thermistor is adequate for temperature compensation only
when the temperature profile changes uniformly over the entire- tank over time.

          The ARCO method compensates for temperature through indirectly se-
lecting the product level.  The effectiveness of the evaporation-condensation
and temperature compensation method used by ARCO could not be independently
analyzed from the test data.

          The detection performance of each method using its standard pro-
cessing algorithm varied considerably.  This conclusion was based on the am-
bient noise and the simulated leak data.

          No test operator demonstrated a consistent ability to limit on-site
pretest setup time to less than several hours.  Time requirements were similar
for all methods.

     F.  Recommendations

          The findings of the development study have resulted in  several rec-
ommendations concerning the method of tank testing to be used in  the national
survey program.  These recommendations were further refined, developed, and
tested  in the pilot study (discussed  in Section V).  The recommendations are
summarized below.

           •  The tank testing method  should include putting a head of pressure
             on the tank.  There are  two reasons for this.  First, proper com-
             pensation for water table effects is necessary if the proper con-
             clusion  is to be reached under high water table conditions.


                                    98

-------
             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.

          •   The tank testing method should provide frequent temperature mea-
             surements with a precise thermistor and adequate temperature com-
             pensation.   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 ac-
             curate temperature measurements at frequent intervals.   The con-
             clusion that the product should be circulated to achieve tempera-
             ture uniformity resulted from the better performance observed
             for the Petro-Tite method over the single thermistor approach
             used by Leak Lokator.

          •   Data on temperature and level changes must be collected frequently.
             This facilitates the use of smoothing algorithms to more accurately
             determine trends and also facilitates detection and mitigation
             of artifacts.

             Data collection must continue longer than the standard test time
             so that sufficient data for a precise analysis can be provided.
             A longer test period with frequent temperature and level read-
             ings provides more data for trend determination.  However, the
             practical considerations of cost and disruption to a facility
             will also influence the test duration.

          •   The test method must incorporate an adequate statistical analy-
             sis of the data to draw supportable conclusions about the leak
             rate.  While none of the techniques was found to collect either
             sufficient test data or to provide adequate analysis algorithms,
             the Petro-Tite method was amenable to more sophisticated al-
             gorithms.


V.   PILOT STUDY

     A.   Objectives

          The results from the preliminary tests and tests at the five U.S.
sites led to the recommendation that modified Petro-Tite equipment and pro-
cedures be adopted for the national survey.   The major objective of the pilot
study was to modify and evaluate the performance of the modified Petro-Tite
method for use on the national survey.  This process included testing to:

             Determine the best sampling interval for collecting the data;
             that is, the time interval at which product in the standpipe
             should be releveled and data readings made.

             Determine the best length of the test;  that is, the number of
             data points and total time of data collection.
                                    99

-------
          •   Develop and test the analysis algorithm.

          •   Implement the procedures operationally in the field to identify
             operating difficulties and correct them.

          •   Provide data sufficient to be used on the national survey.

          •   Field test the entire survey data collection effort, including
             scheduling, data collection, and analysis.

          •   Finalize the test protocols.

          •   Estimate the detection performance of the method.

          0.  H.  Materials under subcontract to MRI collected the data to
achieve these objectives.  Data analysis was performed by MRI and by Vista
Research, Inc.,  also under subcontract to MRI (Vista 1985).

     B.  Overview

          A sample of 25 tanks from the national survey sampling frame was
selected from two primary sampling units (PSU) 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 pre-
sented difficulties.  Recommendations for alleviating 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 at three sites.  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 pro-
cedures for the national survey were developed.  These procedures are deline-
ated in the Test and Analysis Plan (Haile 1985) prepared for the national
survey program.   Data and test review procedures were developed to check each
tank test for Validity and to ensure that the standard analysis is 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 en-
tered  in a spreadsheet data file utilizing a portable computer.  Following
this,  test data were entered as each data point became available.  The spread-
sheet  calculations were developed to provide rapid determination of an esti-
mated  leak rate and to plot the data for visual display after a minimum number
of data points had been entered.  This provided a preliminary analysis and


                                    100

-------
estimated volume change rate that could be obtained on the scene.  The raw
data from these tests and the plots and field analysis results are included
in Appendix I.   An example of the spreadsheet data file is shown as Table 25,
and Figure 34 is the associated plot.

     D.  Theoretical Precision

          The Petro-Tite test (discussed in Section IV.B) is based on two pri-
mary measurements.  The first is the volume change in the standpipe, and the
second is the temperature as measured by the thermistor arrangement.  The
volume is measured by bringing the product level to a reference mark by addi-
tion or subtraction of product in a cylinder graduated in hundredths of a gal-
lon.  Testing personnel try to interpolate to the nearest 0.005 gal of product,
which is one source of error.  The thermistor is a nonlinear device whose elec-
trical resistance is dependent on the effects of temperature.  The usual situa-
tion is that there are about 325 digits 1°F; however, this can vary to as low
as 310, and a lower bound of 300 is reasonable.   Thus, temperature can be mea-
sured only to about the nearest 1/300°F, a second source of error.  A third
source of error is the precision with which a tester can relevel the product
in the standpipe.  Experimental data having individual testing personnel re-
peatedly relevel a standpipe showed that releveling is quite reproducible.
Under static conditions, the error in releveling was estimated to be about
0.00068 gal, while under actual test conditions, the error in releveling was
estimated to be about 0.00270 gal.

          The temperature corrected volume changes upon which the test is
based are the differences of the measured volume- changes and the temperature
changes over the measured time intervals.   As noted above, two sources of er-
ror are associated with the volume changes.  Thus, the theoretical precision
of the measurement can be expressed as the sum of three terms in terms of the
variance of a single measurement.  One variance is for the releveling preci-
sion, one for the precision with which the volume can be read, and'one for
the precision of the temperature measurement.  The time measurements are suf-
ficiently precise that their variability was not included.

          Assuming that the error in reading the volume is a simple rounding
error, if volume is rounded to the nearest 0.005 gal, the error would be that
of a uniform random variable distributed on the range (-0.0025, +0.0025) gal-
lons.  The variance of a uniform random variable is.given by the formula (range
squared)/12.  Thus the variance would be 2.083 x I0"g gal squared.  Actually,
it is probably better to assume that the measurement is to the nearest 0.01
gal, as that is the finest graduation on the cylinder.  This would lead to a
variance estimate of 8.33 x 10"6 gal squared.

          The experimental data suggest that the error involved in releve]ing
the standpipe is also a rounding error with a variance of about 7.27 x 10 6
gal squared, or a standard deviation of 0.0027 gal.
                                    101

-------
Table 25.  Example Spreadsheet Data File
01-





Ti
Hr
1 1
i L
L i
I ~
j, ~ *•
12
12
13
•t "**
13
13
14
14
14
14
15
15
Apr —





me
Min
15
3'!'
45
i.)
15
30
45
0
15
30
45
("i
1 <3
1 vJ
30
45
0
15
65
Sits Code
Test Firm
Tsst Crew
MR I Crew

Level
< d i v )
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12

L2*52Sn
o.H.n.
BS,C'F , J5
BC,MG

V Before
igal )
N/A
0 . 500
0 575
i'. 64
<».7l
0.76
0.82
0.21
0 . 26
0.315
0 . 365
0.41
0 . 455
0.505
0.55
0. 585
0.615
Pags
Fuel Type
Tank Vol
API Dens
Exp Coe-f

V A-fter
(gal)
N/A
0 . 575
0.64
0 . 7 i
0.76
0.82
0.875
0 . 26
0 .315
0. 365
0.41
0.455
0.505
0.55
0.585
0.615
0.65
1
UNLEADED
1 03o
60. 1
0.00061431

Fuel Temp
cdigi cs)
11 784
11813
11343
11873
1 1900
11928
11957
11982
1 20 1 1
12034
12Ofcl
12085
12107
12128
12148
'12166
12135

Date
T digits
T digits/F
Leak Fate

Tcorr dV
•.gal .'
! •! / H
<•;> .017
0 . t.'OS
0 . 0 1 0
-0.0<.'4
0 . Ou4
-0. 003
. 000
-0 . 003
0 . 004
-0 . 009
-0 . 00-3
0 . 006
0 . OO3
-0 . 005
-O . (JO 6
-O . 003

MAP .13, 85
11622
3 1 7
-0. 002
0. OO3
Leak Kate
(gal/h)
M / H
0. 067
0 . 0 1 9
0. 039
-0 . 01 7
0.015
~'j .013
-0 . 00 1
-0 . < -i 1 3
0 . 01 5
-0. 037
-0 . 0 1 3
O.u23
0 .011
-O.021
-0.025
-0.013
                   102

-------






I
0)
<^ w
co *
lO *
03 8,
1 OU
i— i
a




c
c
Li' " ~
a-

0+
Q+
'
Of O
Of 0
Q ~h ^
DH* O
n+ o

n+ o

D 4- ^
Q+ 0
a + o
<>

' »






•





n+ « 1
CH- 0

1 1 I 1 1 1 I ] T
• *•••**«•
3OOGQQOOQ C

a>
0
o
- " o

2
II
_cu £•* g
>d a*
0 CU
« M
^
H


--

01
_t*
u.x i
0
3 Uncorrect
                                                            4-

                                                             O
                                                             (O
                                                            T3
                                                             (0
                                                            00
                                                            oo
                                                            CU
                                                            i-
                                                            OJ
103

-------
          Thus, the variance associated with the volume measurement is the
sum of the releveling variance and the variance in reading the graduated
cylinder, or approximately 1.56 x 10 5 gal squared.  The square root of this
is 0.00395 gal, which is the standard deviation of the volume measurement.

          Similarly, the error in reading the thermistor is assumed to result
from rounding to the nearest digit in balancing the resistance.  If 300
digits/°F is assumed as a lower bound, the rounding error would be uniformly
distributed on the range -17(500° to +1/600°F.  This leads to a variance of
9.259 x 10"7 degrees Fahrenheit squared.

          The temperature is converted to a volume change caused by thermal
expansion or contraction.  The formula for this conversion is:

                              dV = dT x Ce x V0

where dV is the change in volume, dT is the change in temperature, Ce is the
coefficient of expansion, and V0 is the total volume of product.  If the Ce
and V0 are taken as constants for a given product and tank, the variance as-
sociated with the volume change due to temperature is the variance of the
temperature measurement times the square of Ce and times the square of V0.
Thus, this source of error depends directly on the size of the tank.  It also
depends on the coefficient of expansion.  The major difference in coefficient
of expansion occurs between fuels, e.g., gasoline at about 0.0006 and diesel
fuel at about 0.00045.  For the Petro-Tite method, the temperature resolution
is 0.003°F which corresponds to a volume of 0.018 gal and 0.014- gal for gaso-
line and diesel, respectively, in a 10,000 gal tank.

          Table 26 gives approximate theoretical precision values for a single
measurement for gasoline and diesel fuel, for several tank sizes.

          As discussed, the Petro-Tite testing system requires the operator
to relevel a standpipe periodically.  In addition, the operator must read the
amount (volume) of product in a graduated cylinder.  Two experiments were con-
ducted at site N2900601 to determine how much interobserver variability was
encountered in making these level measurements, which require the operator to
read the meniscus in a glass tube.  The first experiment consisted of a" static
releveling test.  A calibration bar of known volume (0.056 gal) was repeatedly
inserted into'or removed from a standpipe of the type used in the Petro-Tite
test.  A sample of four individuals who were to do some of the testing was
taken.  Each person releveled the standpipe five times after insertion or re-
moval of the calibration bar and measured the product recovered or added.  An
analysis of variance was done to test for interobserver variability.  Results
are presented  in Table 27.

          The  analyis showed the inter-observer variability was not signifi-
cantly (at the 10% level) larger than within-observer variability.  The
variability was estimated to be 0.00265 gal.
                                    104

-------
Table 26.   Precision of Temperature and Volume Measurements
Volume (gal)
250
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
5,500
6,000
6,500
7,000
7,500
8,000
8,500
9,000
9,500
10,000
10,500
11,000
11,500
12,000
12,500
13,000
13,500
14,000
14,500
15, 000 ~
Standard deviation
Diesel
3.951167 x 10"3
3.955613 x 10 3
3.973349 x 10"3
4.002733 x 10"3
4.043514 x 10"3
4.095348 x 10 3
4.157824 x 10"3
4.23047 x 10 3
4.312772 x 10"3
4.404188 x 10 "3
4.504165 x 10"3
4.612144 x 10"3
4.727579 x 10~3
4.849936 x 10 4
4.978705 x 10"3
5.113402 x 10" 3
5.25357 x 10"3
5.398785 x 10"3
5.548649 x 10 3
5.702796 x 10"3
' -5.860888 x 10"3
6.022614 x 10"3
6.187689 x 10"3
6.355854 x 10"3
6.526868 x 10"3
6.700513 x 10 3
6.876591 x 10"3
7.054919 x 10 3
7.23533 x 10 3
7.417673 x 10"3
7.601809 x 10"3
(gal)
Gasoline
3.95232 x 10"3
3.960219 x 10~3
3.991658 x 1Q~3
4.043514 x 10 3
4.115014 x 10 3
4.205156 x 10 3
4.312772 x 10"3
4.43659 x 10"3
4.575297 x 10"3
4.727579 x 10 3
4.892171 x 10 3
5.067873 x 10"3
5.25357 x 10"3
5.448242 x 10 3
5.650959 x 10"3
5.860888 x 10 3
6.077281 x 10 3
6.077281 x 10 3
6.299471 x 10"3
6.758945 x 10"3
6.995237 x 10 3-
7.235331 x 10 3
7.47886 x 10"3
7.725499 x 10"3
7.974961 x 10 3
8.226989 x 10"3
8.48135 x 10 3
8.737849 x 10 3
8.996296 x 10 3
9.256529 x 10 3
9.518404 x 10 3
                           105

-------
        Table 27.   Analysis of Variance for Ihterobserver Variability
 Source'Sum of squaresdf             Mean square

Observers           2.190471 x lo"s            3           7.301569 x 10"7
Error               4.813075 x 10"6           12           4.010896 x 10"7
Total               7.003546 x 10"6
F-test ratio 1.820 - not significant at 10%.

Note:  S = 2.65 x 10"3 = 0.0026 gal.
                                    106

-------
          Another source of variability is present in the actual testing in
that the level changes as temperature or other variables in the tank change,
which may introduce additional variability into the volume measurements.  Ten
releveling calibrations with the calibration bar were done on each of two
tanks during testing to determine the size of the variation.   The pooled esti-
mate of variance from these dynamic trials was 7.270 x 10 6,  somewhat larger
than the estimate from the static releveling tests.  Expressed on the same
scale as the measurements (by taking the sguare roots), the variability esti-
mated from the static tests was_6.833 x 10 4 gal, while that from the dynamic
releveling tests was 2.696 x 10~3 gal.   This latter number is probably the
best estimate of total variation in the volume measurement process because it
includes interobserver variation, replication, and some variation due to dif-
ferences in the two tanks under test.  However, there is probably more varia-
tion due to differences in the environment and among tanks than is represented
in the variability between these two tanks.  Thus, this variability should be
regarded as not including tank-to-tank variability.

     E.   Data Analysis

          The data from the pilot study tank tests were analyzed with several
purposes in mind.  One was to determine the best sampling interval.  Data col-
lection intervals of 1, 5, 10, and 15 min were considered.  Data collection
at 1-min intervals was found to be impractical for the large scale survey.
For the 1 min data, it appeared that the additional reduction in standard
error that would result from the more frequent data was lost because of the
additional difficulties in reading accurately in the short time available.
In addition, analysis performed by Vistc! Research, Inc. (Vista 1985) on the
1-min data suggested that the time to independence was about 4 min.  That is,
measurements needed to be about 4 min apart in order that they not be serially
correlated.  Consequently, use of time intervals less than about 5 min might
result in dependence.

          Since the standard error of the estimated leak rate is proportional
to the square root of the number of observations, both the 5-min and 10-min
intervals would provide improvements to the precision of the test data.   The
tests on the pilot study with 15-min intervals gave an average standard error
of 0.0223 gal/h, while the tests with 5-min intervals gave an average standard
error of 0.014 gal/h.  Both of these standard errors are computed for a total
test time of 2 h.  Theoretically, the 5-min times should reduce the standard
error by a factor of 1/(1.732) or should multiply the standard error by 0.577.
The actual reduction obtained with the 5-min data was slightly less than theo-
retical.  That observed was 0.014 gal/h, while the theoretical value would be
0.0129 gal/h.  The difference may be due to chance, or it may suggest that
the 5-min interval is about at the limit of the observers.  Spectral analysis
by Vista Research, Inc., also indicated that the 5-min interval is an optimum
sample interval.

          The 10-min interval was not used in the pilot study.  It would theo-
retically reduce the standard error of the leak rate by a factor of 0.82 com-
pared to that of the 15 min interval for the same total test time.  However,
it is inconvenient for the test crews to use because its intervals do not co-
incide with the 15-min test intervals that are standard.   For this reason and
                                    107

-------
the improved precision provided by the 5-min interval, it was decided to not
use the 10-min interval.

          The pilot study results demonstrated that data collection at 5-min
intervals was practical in the field.   Further, the analysis of the 1-min
data indicated that a 5-min interval  should generally result in independent
observations.  The expected improvement in precision was observed on the
pilot study.  As a result of these considerations, the 5-min interval was
recommended as the standard for the national survey.

          Determination of the best total time of the test was not so clear-
cut.  Precision and accuracy can be improved by longer test times, but this
is achieved at the expense of more inconvenience for the owner/operator and
longer work days for the test crew.  Long test times would necessitate dupli-
cate crews or would result in loss of data because of increased errors caused
by fatigue.  A compromise of 2 h of data at the low level was selected as pro-
viding sufficient data while still proving to be practical for the field data
collection.

          Seventeen of the 25 tanks selected were tested in this pilot study.
The remaining 8 tanks could not be tested due to scheduling problems or other
limitations.  The data from these tank tests were analyzed to estimate whether
a tank was leaking and to provide an estimated volume change rate together
with its estimated standard error.  The results of the final analysis for each
tank were reported on a site data summary form, an example of which is shown
in Figure 35.

          A summary of .the test results is presented in Table 28.  To be con-
sistent with national survey results, the standard errors are all reported
for 2-h test times.  Four tanks had data collected at 15-min intervals and so
had larger standard errors.  Two tank tests are not included because of data
problems that indicate testing difficulties that may invalidate the results.
These tanks were at sites N3000000078 and N290000010 (tank 2 of 3).

          The data reported in Table 28 express the leak rates in gallons per
hour.  The standard errors reported are in the same units.  In addition to
the quantitative leak rate estimate, a qualitative (yes or no) determination
is also presented.  The conclusion about leaking or not is coded as "C" for
certifiable, that  is, not leaking by the 0.05 gal/h NFPA standard.  A code of
"N" is used for noncertifiable.  If the "N" is followed by (I), the tank
could not be certified by the  test, but there were problems with the test
that made the results invalid.

          Three of the systems had leaks.  Two were fairly large (-1.381 gal/h
and -0.263 gal/h), while one was smaller (-0.107 gal/h).  One additional sys-
tem had a system leak rate (-0.056 gal/h) even though the tank was certified
by  the Petro-Tite  test crew, who estimated a leak rate of -0.029 gal/h (later
found to have been taken when  the tank was isolated from the line).  Detailed
analysis of the data from this system revealed that the leak rate of -0.056
gal/h for the system was due to a  leaking delivery  line.  The line test gave
a line leak rate of -0.04 gal/h.  When the tank was isolated from the line,
the tank rate was  -0.018 gal/h..  It was concluded that the system leak con-
sisted of the line leak and that the tank was tight.

                                    108

-------
                             SITE DATA STM1ARY
Survey ID No.  L290Q00528	          Tegt pate  «*r  13. 1985

     Tank No.  1  of 2	          Product   ^leaded

     Tank Size    1Q36
                               TEST RESULTS


                 System       Line 1       Line  2        Line 3        Line

Certified

Rate

S.I.
     Figure  35.   Example  data  summary.   Note:   The  standard error reported  has
     been adjusted to a 2-h  test period  for consistency with future  tests on
     the national  survey.  A negative  sign  indicates  leak out of tank.
                                      109

-------
              c
              o
              vn
              »  (O  O)
       to  dj ^^
           0)
           IQ
           c   .
           HI  O
               o


               0>
                       uu
                                     ou
                                                     uuu
                                                                         ou
                       oounmoofMOr-ioooooooo

                       O O r-4 O O O O* O O* O* O O* O* O* O O* O
                       +  i   i   i  •»••»•*     i   i   i  * •*•  -h   i  •»•  I
                        ooooooooooooooooo

                        o o o* o* o* o o' d o' d d  d o d o' d o*
OOr-IOOOOOOOOOOOOOO
    III      I          I      I             III
                                           z       z z a: a. z     z a:
                                               i^i— tr-IOOOOt'OOO
                                               I~^H-
                       O
                       o
                       o
                       o
                       
   oooooooooooooooo
   oooooooooooooooo
   oooooooooooooooo
   oooooooooooooooo
                                               C3C3O
                                                                      O
                                                                                     •a
                                                                                      a>
                                                                                     •a
                                                                                      a
                                                                                      
N«^ a) ro
   T3 -a
 . » ro
 
t^. <^  qj
•i- _l S_

 S-  --JZ
 O)  S- *->
 U  ro •—

+->  3
 O  05-M
 C  Qj in
    i. O)

 11 ^
Z  

•(J _1 Q.
 ui
    • -•a
  CL-M
 S-     ra
 OJ  II  E
 «J     -I-
                                                                                      (_] CUO
                                                         no

-------
          Of the 17 tanks tested,  one resulted in a clearly invalid test.
One test was problematical, but the system was probably tight.   Three tanks
appear to have significant leaks,  and the remainder appear to be tight.   It
should be noted that the leak rates reported here are the leak rates or vol-
ume change rates observed under the testing situation,  which includes a posi-
tive pressure of 4 psi at the bottom of the tank.   Since this is higher than
the pressure found in the tank during normal operating conditions,  the abil-
ity to detect leaks is enhanced.   Thus the volumes of product escaping into
the environment would be smaller in actual tank operation.   Depending on the
location of the hole where the tank is leaking, the leak rates could actually
be as low as zero in operation; for example, if the hole is in the  top of the
tank and the system never has product at that level.   These rates,  then, are
overestimates of product loss or leakage in operation.

          A preliminary estimate of the performance of the test can be made
from the pilot data.   The estimate is preliminary for three reasons.   First,
only a limited number of tanks were tested.   Second,  since these were in a
restricted geographical area, the  results may not be  representative of the
nation as a whole.   Third, in fine-tuning the sampling frequency to optimize
test performance, not all tank data were collected at the sampling  frequency
finally determined for the modified method.   Consequently, the estimate of
the performance is preliminary and the actual performance of the test method
on the national survey may differ.

          The performance of the test method is primarily determined by the
standard error of the leak rate estimate.  With a given threshold for declar-
ing a leak, or for a specified allowable probability  of type I error, the
standard error can be used to calculate a performance curve.   In this con-
text, a type I error is a false alarm, that is, declaring a tight tank to be
leaking.  The power is the probability that the test  would detect a leak of a
given rate.  The standard errors of the estimated leak rates on the pilot
study ranged from 0.01 gal/h to 0.035 gal/h for tanks that had valid tests
and that were tested with the 5 min protocol.  Consequently, a value of 0.03
gal/h for the standard error is a  reasonable estimate for the performance of
the test.

          Using this and the NFPA  Standard 329 criterion of 0.05 gal/h as the
threshold for declaring a leak yields the performance curve plotted as Figure
36.  In Figure 36, the probability of detection or power is plotted as a func-
tion of the true leak rate.  The significance level is the point where the
curve crosses 0, denoted by the vertical line.   This  is the probability of a
false alarm.  This curve is an estimate of the performance of the test to be
used on the national  survey.  At the threshold value  of 0.05 gal/h  the proba-
bility of detecting a leak of exactly 0.05 gal/h is 50%.   At this threshold
(0.05 gal/h) a leak rate of 0.10 gal/h would be detected with a probability
of approximately 95%.  The probability of detecting a leak depends  on the size
of the leak, the threshold, and the variability of the data from a  particular
tank.
                                    Ill

-------
                                                                              Ol


                                                                              3
                                                                              U


                                                                              u
                                                                              01
                                                                              a.
                                                                              (U
CO
  •
o
CM
  •

O
oo
                            jo
                         112

-------
VI.  RECOMMENDATIONS FOR NATIONAL SURVEY

          The modified Petro-Tite test method, as refined based on the pilot
study, should be used for the national survey.  The key modifications include:

          •   Temperature and volume measurements should be made at 5-min in-
             tervals.

          •   The test period should provide data for not less than 2 h.

          •   The volume and temperature data should be plotted to indicate
             trends and provide visual diagnostic checks.

          •   The results should be analyzed using smoothing techniques for
             the temperature data.

          •   The data analysis should provide standard error estimates as well
             as leak rates and be flexible enough to accommodate unusual trends
             such as curvilinearity and non-monotonicity.

          These recommendations and selection of a modified Petro-Tite method
for the national survey do not constitute an endorsement of a tank test method
by the U.S.  Environmental Protection Agency.


VII.   REFERENCES

Collins, T.   1985.   ARCO.  Personal communication.

Haile CL et al.  1985.  Midwest Research Institute.  Test and analysis,plan
for the tank testing program of the national survey of underground storage
tanks.  Draft plan.  Washington, DC:  Office of Toxic Substances, U.S. En-
vironmental  Protection Agency.  Contract 68-02-3938.

NFPA.  1983.  National Fire Protection Association, Quincy, MA.  NFPA 329.
Recommended Practices for Underground Leakage of Flammable and Combustible
Liquids.

PACE.  1982.  Petroleum Association for Conservation of the Canadian Environ-
ment.  Leak seeking in underground storage tanks, and tank testing:   a way to
limit liability and loss.  Presented by Heath Consultants, Inc.  Proceedings
of the Underground Tank Testing Symposium, May 1982.

PEL   1983.   Symposium on Tightness-testing Systems for Underground Tanks,
Las Vegas, Nevada,  September 1983.  Sponsor:  Petroleum Equipment Institute.

USEPA.  September 1984.  U.S. Environmental Protection Agency.  Leaking under-
ground storage tanks containing motor fuels:  a chemical advisory.  Washington,
DC:  USEPA,  Office of Toxic Substances.

USEPA.  October 1984.  U.S. Environmental Protection Agency.  More about leak-
ing underground storage tanks:  a background booklet for the chemical advisory.
Washington,  DC:  USEPA, Office of Toxic Substances.

                                    113

-------
USEPA.  January 1986.  U.S. Environmental Protection Agency.  Underground tank
leak detection methods:  a state-of-the-art review.  Cincinnati, OH:  USEPA.
EPA/600/2-86/001.

Vista.  1985.  Vista Research, Inc., for Midwest Research Institute.  Analysis
of the pilot study tank test data.  Subcontract 147-8200-7.
                                     114

-------
    APPENDIX A
DEFINITION OF TERMS
     A-l

-------
          The following terms are defined as used in this  report.

          Accuracy - Degree to which a measurement process duplicates  a  know£
true value.

          Backfi11 - Material used to fill in around tanks.  Usually consists
of pea gravel or sand.

          Bias - A systematic error common to all measurements made with a
particular device.  The average difference between a series of measurements
and the true value.  A measure of accuracy.

          Bung - Plug used to close extra openings in top  of tank.

          Dispenser - Unit containing meter and valves for filling vehicles
with fuel.

          Distribution line - Piping which delivers fuel to other  locations
such as dispensers or other tanks.

          Drop tube - Tube that extends from fill pipe to  near bottom  of tank.
Permanent drop tubes cannot be removed without modifications to  the tank.
Removable drop tubes can be easily removed for testing purposes.

          Fill pipe - Opening into tank through which fuel is added to the
tank.  Remote fill pipes do not lead directly into the tank and  are usually
located some distance from the tank.       ,    .                              '*

          Foot valve - A check valve used to prevent fuel  from flowing back-
wards through a  line.

          Fuel - Also called "product".  See Appendix B  for descriptions of
fuel types.

          Manifold system - Two or more tanks connected  together such  that^
fuel flows between them when added or removed from either  tank.   The  term is
also used in inventory analysis to indicate a system where two or more tanks
have a common dispenser, regardless of whether fuel flows  between  them.

          Manway - Access port into tank  large enough to allow a person  to
enter the tank.

          NFPA  recommended standard - The National Fire  Protection Associa-
tion (NFPA)  recommends in its Bulletin 329 that  "precision testers"  be capa-
ble of-detecting leaks of 0.05 gal/h.

          Precision  - Degree to which repeated measurements made by  the  same
device under identical conditions  vary about a central value.

          Pressure pump - Also referred to as a  remote or  turbine system.
The pump  supplying the fuel  to the dispenser is  located  in the tank  below the
product  level.   Fuel  is supplied  to the dispenser under  pressure.


                                   A-2

-------
          Root mean squared error - A statistical measure of the variability
of an estimate that incorporates both precision and accuracy.

          Sensitivity - Degree to which a measurement device is affected by
or responds to changes.

          Standard deviation - A statistical measure of the precision of a
measurement.  It is estimated by the following formula:


                    s =  I (xi - *)a
                    x  = _  Zx.  is the arithmetic mean
                         n

          Standard error - A* statistical measure of the precision of an esti-
mate.  The standard error of the mean is I/ n times the standard deviation,
where n is the number of terms.

          Standpipe - An extension added to an access port into the tank for
the purpose of bringing the fuel level above grade.

          Suction pump - The pump supplying fuel to the dispenser.   Located
outside the tank.  Fuel is drawn into the line under reduced pressure before
dispensing.

          Tank system - The tank and all ancillary connections such as vent
lines, delivery lines, manifolds, and other tanks connected by manifold.  The
entire system is tested as a complete unit.

          Vent pipe - Piping extending from the top of the tank to the atmos-
phere for the purpose of preventing pressure buildup in the tank.
                                   A-3

-------

-------
   APPENDIX B
GLOSSARY OF FUELS
      B-l

-------
I.   AVIATION GASOLINE (FINISHED)

          All special grades of gasoline for use in aviation reciprocating   :.^
engines, as given in ASTM Specification D 910"and Military Specification  MIL-'
G-5572.  Excludes components used in blending or compounding into  finished
aviation gasoline.   Three grades of aviation gasoline:  Grade 80-red, Grade
100-green, and Grade 100Ll.-blue.  Grades 100 and 100LL represent two aviation
gasolines that are identical in anti-knock quality but that differ in maximum
lead content and color.   The color identifies the difference for those  en-
gines which have a low tolerance to lead.


II.  DISTILLATE FUEL OIL

          A general classification for one of the petroleum fractions produced
in conventional distillation operations.  Used primarily for space heating,
on- and off-highway diesel engines (including railroad engines and agricultura
machinery), and electric power generation.  Included are products  known as
No. 1, No. 2, and No. 4 fuel oils and No. 1 and No. 2 diesel fuels.

     A.  No. 1 Distillate

          A petroleum distillate which meets the specifications for No. 1
heating or fuel oil as defined in ASTM D 396 and/or the specifications  for
No. 1 diesel fuel as defined in ASTM 0 975.

          1.  No. 1 Diesel Fuel                                            . |

          A volatile distillate fuel oil having a boiling range between 300°
and 375°F and used in high-speed diesel engines generally operated under  wide
variations in speed and load.  Includes type C-8 diesel fuel used  for city
buses and similar operations.  Properties are defined in ASTM-0 975.

          2.  No. 1 Fuel Oil

          A  light distillate fuel oil used in vaporizing pot-type  burners.-
ASTM 0 396 specifies for this grade maximum distillation temperatures of  400°F
at the 10% point and 550°F at the 90% point, and kinematic viscosities  between
1.4 and 2.2  centistokes at 100°F.

     B.  No. 2 Distillate

          A  petroleum distillate fuel oil which meets the specifications  for
No. 2  heating or fuel oil as defined in ASTM D 396 and/or the specifications
for No. 2 diesel fuel as defined in ASTM D 975.

          1.  No. 2 Diesel Fuel

          A  gas-oil distillate of lower volatility than that of No. 1.  diesel
oil, with distillation temperatures between 540° and  640°F at the  90% point
and used  in  high-speed diesel engines generally operated under uniform  speed
                                     B-2

-------
and load conditions.   Includes type R-R diesel fuel used for railroad  locomo-
tive engines, and type T-T diesel fuel used for diesel-engine trucks.  No.  2
diesel fuel is also used in agricultural machinery and in marine vehicles.
Properties are defined in ASTM Specification D 975.

          2.   No. 2 Fuel Oil

          A distillate fuel oil used in atomizing-type burners for domestic
heating or for moderate capacity commercial-industrial burner units.   ASTM
D 396 specifies for this grade distillation temperatures between 540°  and
640°F at the 90% point and kinematic viscosities between 2.0 and 3,6 centi-
stokes at 100°F.

     C.  No.  4 Distillate

          A fuel oil  used for commercial burner installations that are not
equipped with preheating facilities.  It is used extensively in industrial
plants.  This grade is a blend of distillate fuel oil and residual fuel oil
stocks that conforms  to ASTM Specification D 396 or.Federal Specification
VV-F-815C; its kinematic viscosity is between 5.8 and 26.4 centistokes at
100°F.
III.  KEROSENE

          A petroleum distillate that'boils at a temperature between 300° and
500°F, that has a flashpoint higher ..than 100°F by ASTM Method 0 56, that has
a gravity range from 40° to 46° API,  and that has a burning point in the range
of 150° to 175°F.   Included are the two classifications recognized by ASTM
Specification D 3699 (No.  1-K and No.  2-K) and all grades of kerosene called
range or stove oil that have properties similar to No. 1 fuel oil, but with a
gravity of about 43° API and a maximum end point of 625°F.   Kerosene is used
in space heaters,  cookstoves, and water heaters and is suitable for use as an
illuminant when burned in wick lamps.


IV.   KEROSENE-TYPE JET FUEL

          A quality kerosene product with an average gravity of 40.7° API,
and a 10% distillation temperature of 400°F.  It is covered by ASTM Specifi-
cation D 1655 and Military Specification MIL-T-5624L (Grades JP-5 and JP-8).
Three types of aviation turbine fuel  are Jet A, Jet A-l, and Jet B.  Jet A
and Jet A-l are relatively high flashpoint distillates of the kerosene type
but with differing freezing points.   Jet B is a relatively wide boiling range
volatile distillate.   A relatively low freezing point distillate of the kero-
sene type, it is used primarily for commercial turbojet and turboprop aircraft
engines.
                                    B-3

-------
V.  LIQUIFIED PETROLEUM GASES

          Ethane, ethylene, propane, propylene, normal butane, butylene, anc;
isobutane produced at refineries or natural gas processing plants,  including
plants that fractionate raw natural gas plant liquids.


VI.  MOTOR GASOLINE (FINISHED)

          A complex mixture of relatively volatile hydrocarbons, with or with
out small quantities of additives, that have been blended to form a fuel suit
able for use in spark-ignition internal combustion engines used in automotive
marine, and stationary equipment.  Specifications for motor gasoline, as give
in ASTM Specification 0 439 or Federal Specification VV-G-1690C, include boil
ing temperatures of 122° to 158°F at the 10% point and 365° to 374°F at the
90% point, and a Reid vapor pressure range from 9 to 15 psi.  Motor gasoline
or "Mogas," includes finished leaded gasoline, finished unleaded gasoline,
and gasohol.  Blendstock is excluded until blending has been completed.
(Alcohol that is to be used in the blending of gasohol is also excluded.)

   .  A.  Regular Gasoline

        .  As defined in ASTM 0 439 and Environmental Protection Agency (EPA) '
specifications, gasoline anti-knock designation equal to or less than 3.  In-
cludes both leaded and unleaded regular gasoline.  Excludes any blendstock
until blending has been completed and the blendstock is incorporated in the
finished leaded gasoline and no longer separately identified.         •      .4

          1.  Leaded Regular Gasoline

          As defined in ASTM D 439 and EPA specifications, gasoline anti-knoc
designation 3 produced with the use of any lead additives or which contain
more than 0.05 g of lead per gallon or more than 0.005 g of phosphorus.

          2.  Unleaded Regular Gasoline

          As defined in ASTM 0 439 and EPA specifications, gasoline anti-knoc
designation 2 containing not more than 0.05 g of lead per gallon and not more
than 0.005'g of phosphorus.

     B.  Premium Gasoline

          As defined in ASTM 0 439 and EPA specifications, gasoline anti-knoc,
designation 3 4.  Includes both leaded and unleaded premium gasoline.  Also
includes the entire volume of gasohol, both the gasoline and the alcohol.

          1.  Leaded Premium Gasoline

          As defined in ASTM 0 439 and EPA specifications, gasoline anti-knoc
designation 5 produced with the use of any lead additives or which contains
more than 0.05 g of lead per gallon or more than 0.005 g of phosphorus.
                                    8-4

-------
          2.  Unleaded Premium Gasoline

          As defined in ASTM D 439 and EPA specifications, gasoline  anti-knock
designation 4 containing not more than 0.05 g of lead per gallon and not more
than 0.005 g of phosphorus.

          3.  Gasohol

          A blend of finished motor gasoline (leaded or unleaded) and alcohol
(generally ethanol but sometimes methanol) in which 10% or more of the product
is alcohol.
VII.  NAPHTHA-TYPE JET FUEL

          A fuel in the heavy naphtha boiling range with an average gravity
of 52.8° API and 20 to 90% distillation temperatures of 290° to 470°F-, meet-
ing Military Specification MIL-T-56241 (Grade JP-4).  JP-4 is used for turbo-
jet and turboprop aircraft engines, primarily by the military.  This category
excludes ram-jet and petroleum rocket fuels, which are included in the "Mis-
cellaneous Products" category.


VIII.   PROPANE. CONSUMER GRADE

          A normally gaseous parafinic compound (C3H8), which includes all
products covered by Natural Gas Policy Act (NGPA) specifications for commer-
cial use and HD-5 propane and ASTM Specification D 1835.   Excludes feedstock
propanes, which are propanes not classified as consumer grade propanes, in-
cluding the propane portion of any natural gas liquids mixes, i.e., butane-
propane mix.


IX.  RESIDUAL FUEL OILS

          The topped crude of refinery operations, which includes No.  5 and
No. 6 fuel oils as defined in ASTM Specification 0 396, Navy Special fuel oil
(NSFO) as defined in Military Specification MIL-F-859E including Amendment 2
(NATO symbol F-77), and Bunker C fuel oil.  Residual fuel  oil is used for the
production of electric power, space heating, vessel  bunkering, and various
industrial purposes.
                                    B-5

-------
              Table B-l.   Petroleum Products Included in National
                     Survey of Underground Storage Tanks
Type
Uses
Occurrence
Aviation gasoline
(finished)

Distillate fuel oil,
diesel fuel
  No.  1

  No.  2
Kerosene-type jet fuel
Motor gasoline
(finished)
Naphtha-type jet fuel
Reciprocating aircraft
engine
City buses and similar
  vehicles
Diesel cars and trucks,
  marine crafts, railroad
  locomotives, farm ma-
  chinery, military tanks

Commercial turbojet and
turboprop aircraft engine

Automobiles, light trucks,
motorcycles, various ma-
chinery and equipment

Military turbojet and
turboprop aircraft engine
Small airports
Bus stations

Truckstops, railyards,
  mi 1i tary
Small  airports,  large
airports

Service stations,  con-
venience  stores,  farms,
fleets
0
                       t
Military  facilities
                                    B-6

-------
            Table B-2.   Petroleum Products Not Included in National
                     Survey of Underground Storage Tanks
Type
Uses
Occurrence
Distillate fuel oils
No.  1, 2, and 4

Kerosene
Liquified petroleum
gases
Propane, consumer
grade

Residual fuel oils
Heati ng
Space heaters, water
heaters, cookstoves

Domestic, commercial, and
industry fuels, and re-
finery, and petrochemical
raw materials

Heating
Industry, homes
Homes
Homes, industry,
commercial
Homes, industry
Industrial purposes, space  Industry
heating
                                    B-7

-------
   Table B-3.   Volumetric Coefficients of Expansion and
    Average Densities of Fuels Selected for National
           Survey of Underground Storage Tanks
Fuel
  Leaded regular
  Unleaded regular
  Premium
  Gasohol
    Ethyl alcohol
    Methyl alcohol

Naphtha-type jet, fuel
 VCE
Density (g/on3)
Aviation gasoline
Diesel fuel
Kerosene-type jet fuel
Motor gasoline
. 00056
.0004
NA
. 00063
NAa
NA
.82
.66-.



69
,00062
,00072

 NA
     791
    ,810

    ,665
aNA = not available.
                          B-8

-------
                APPENDIX C





SOURCES OF ERROR IN VOLUMETRIC TEST METHODS
                  C-l

-------
          This appendix presents the various interfering factors that can in-
fluence the results of volumetric tank tests.   Although these factors are dis-
cussed primarily within the context of the methods evaluated during this stud|
much of the discussion is also applicable to other volumetric methods.  The
major sources of influence—factors other than leaks that affect determination
of leaks in the tank—discussed in this appendix are temperature, vapor pocket
tank deformations, vibrations, water table, and evaporation and condensation.

          In general, volumetric measurements involve monitoring the liquid
levels and liquid temperature in a tank over a specific period of time.
Changes in the liquid level are corrected for expansion and contraction of
the liquid as the result of temperature changes which may have occurred dur-
ing the testing.   The net change in level is converted to equivalent volume,
and the leak rate is calculated in terms of volume loss per hour.  Differences
in the quality of the results are due largely to the ability of the test op-
erator to identify and correct for these sources of error.   The success of
the method may be more dependent upon operator skill than on the test ap-
paratus in many cases, although in some situations the test method may be se-
lected to eliminate an error source.  For example, methods which do not re-
quire overfilling the tank eliminate vapor pocket effects in nearly all cases.

     A.  Temperature
              ~—-"™~
          Temperature is the single most important source of error in tank
testing.  Failure to adequately measure and apply temperature corrections-can
easily lead to incorrect conclusions (either false positives or false nega-
tives).  The magnitude of the temperature effect is illustrated by the fact   %
that an error of only 0.01aF will result in a temperature/volume correction
error of approximately 0.06 gal in a 10,000-gal tank.  The effect is shown
graphically in Figure C-l for three tank sizes.  Figure C-2 depicts three
different temperature changes which will produce a volume change of 0.05 gal,
0.10 gal, and 0.20 gal in a range of tank sizes.

          Nearly all volumetric methods require determination of the average
temperature of the product in the tank.  The approach varies widely, but the
assumption that the test equipment accurately represents or compensates for
the temperature behavior of the tank is the key feature.

          One method uses product circulation prior to and during the testing
in an attempt to produce a uniform temperature.  A single thermistor located
near the pump inlet monitors the temperature during the test.  Although cir-
culation seems to work well, limited data are available to document perfor-
mance.  If circulation is not complete, significant errors may result.  Some
data are presented in Appendix 0.

          Several methods which do not use product circulation use either a
single thermistor or multiple thermistors located at various levels below the
fill pipe.  It is usually desirable to wait until stabilization has occurred
after filling before conducting the test.  The basic assumption of the single
                                   C-2

-------
     o
     CO
           0   0.2  0.4  0.6  0.8  1.0  1.2  1.4  1.6  1.8  2.0
                                                           lO.OOOgai
                                                            5000 gal
                                                            1000 gal
Figure C-l.   Change in volume of gasolene as a function of temperature.
                                C-3

-------
   0.09


   0.08

   0.07


   0.06

  • .0.05

i* 0.04

   0.03

   0.02

   0.01
5
                                              Changes in Volume
                                              	0.20gai
                                              	O.IOgal
                                              	0.05 gal
                  2000
                               4000        6000
                                Tank Si» 
-------
thermistor approach is that if temperature stratification is present, all
layers will change at the same rate.   If multiple thermistors are used, the
volume of the tank represented by each thermistor is used to calculate the
temperature behavior of the tank contents.  Again, only limited data.are
available to evaluate the procedure.

          A third approach, used only by Arco, is to locate the level sensor
float at the point in the tank where changes in level due to temperature are
directly compensated by the change in fuel density, or buoyancy.  The net re-
sult is temperature self-compensation.  Placement of the float is dependent
on the tank geometry and may not be obvious.  Even though this method is self-
compensating, it still assumes that there is uniform temperature behavior in
the tank.

          Unstable temperature behavior can often be detected by properly
trained test crews.  Corrections can then be applied by providing additional
stabilization time, additional mixing, or by adding more temperature sensors.
Failure to detect problems with temperature measurements generally results in
erroneous conclusions.

          There are two additional sources of temperature variability which
may contribute to uncertainty in the test results.  First, many of the test
methods fill up the tank the night before but add several gallons (topping
off) immediately before starting the test the next day to raise the level
into the fill tube or above grade.  If the test is begun immediately after
topping the tank with product that is at a different temperature, as it
usually is, then the thermistor array will locally measure the changes in
temperature of the added product as it mixes in the vicinity of the fill tube
and thermistor array.  This effect may last several hours and the temperature
changes will not be indicative of the average temperature change in the tank.
Second, periodic temperature fluctuations occur in the tank which are caused
by interval waves.  The period and amplitude of these waves depend on the tank
geometry and density (temperature) profile.   The waves typically form at the
boundary of two different fluids of different densities.  This is a problem
if only one or two thermistors are being used to measure temperature fluctua-
tions or if the data are undersampled.

     8.  Vapor Pockets

          Vapor pockets are a result of air or vapor trapped in the tank when
overfilling is complete.  These may be due to a tilted tank, manways, air in
plumbing associated with the tank, or other tank configurations.  They can be
a large source of error for two reasons.   First, since the expansion coeffi-
cient for gases or vapors is much larger than for liquids, changes in temper-
atures will produce proportionally larger changes in volume.  If the pocket
remains trapped, the change in vapor volume will be measured as a change in
liquid volume.  This is shown graphically in Figure C-3.

          Second, changes in pressure can affect volume substantially.   This
would be a problem only if a severe change in atmospheric conditions occurred
during the test time.  The magnitude of this effect is shown in Figure C-4.
As can be seen, unless the vapor pocket is large (100 gal), the effect of
small pressure changes is small; but it could in some instances be significant.

                                   C-5

-------
    3.Jr
    3.0
    2.J
    2.13
  J
  ">
    t.i
    i.o
    •3.5
Vapof Pocket Volume                 /

	100 gal
	10 gal  ;

	1 gal



                    O.J              1*            I.J
Figure C-3.  Changes  in measured product volume as a function of
changes in temperature at 59°F for three vapor pocket volumes.
                              C-6

-------
  +0.-30-
  +0.20-
 +0,10 -


 +0.0*-

5    0
 -0.04-
 -0,10
  -0.20
  -0.30
  -O.JO
                          4V - Vf - Vf « VT
                                       •
'(*•
                                              1    rrem   U
               \
                       «
                                   • *.  v
                                           Vapor Pocket Volume
                                           	lOOgai
                                           	lOgai
                                           	   Igai
                                                   i
              -O.I
                         -0.25         0         -KL03
                                < in Hgj  ?; - 30 in rtg
Figure C-4.  Changes  in  product volume as a function of  pressure
cnanges for three vapor  pocket volumes.
                             C-7

-------
          Vapor pockets are not generally a problem for test methods which c
not involve overfilling the tank.   The only exception is vapor trapped  in  .-•
lines connected to the tank.   Vapor expansion due to heating or pressure    :
changes during the test can push fuel back into the tank if the line has a
leaking check valve.

          There are two common ways to eliminate vapor pockets in the tank.
The first is to lower the product level below the top of the tank until the
vapor pocket is exposed and continue the test in a partially filled tank.
Although this approach eliminates the vapor pocket problem, it also increase
the surface area of the product, reduces the head pressure (which reduces th
leak rate), and leads to a significant loss in sensitivity approximately pro
portional to the-ratio of the original fuel surface area to the new surface
area.  The process is, however, simple and straightforward if the test  metho
is able to operate on a partially filled tank.

          The second approach, which must be used if the test method is not
adaptable to a partially filled tank, is to uncover the top of the tank and
bleed the air from the system.  After removal of the air, the tank is retesti
in the usual manner.   While uncovering the tank is expensive and invasive, ij
does facilitate a more reliable result.  In cases where digging is impractic
or expensive, lowering the level provides a reasonable alternative which can
at least detect the larger leaks with reliability.

     C.  Tank Deformation     ' .

          Tank deformation occurs when pressure is applied to or removed frc
the tank.  On an 3-ft diameter tank an end deflection of only 0.0048 in. wil
result in a volume change of 0.05 gal.  Many factors will affect this proces
Backfill material, water table, and tank construction material are all  vari-
ables which cannot be controlled.

          An important source of structural deformation may occur when  the
tank is topped off.  There is a false sense of safety when the tank is  fille
the-night before because the few gallons required to raise the full level
into the fill tube or above grade can easily be an increase in the pressure
head of 2 to 6 ft.  This will cause significant deformation.

          One leak test approach uses a standpipe to force the deflections t<
occur quickly by placing a 5-psig head on the tank just prior to testing unt-
the major deflections have occurred, and then to reduce the pressure to 4 ps'
At this point the ends should stabilize relatively quickly with little  or no
further movement in or out.  Another commonly used approach is to fill  the
tank to the test level (or nearly so) a few hours prior to the test.

          Tank deformation can sometimes be recognized from a decreasing ob-
served leak rate over a period of time as the rate asymptotically approaches
a stable configuration.  Currently there is no simple, reliable way to  deter-
mine if end deflection has stabilized except to wait for a period of time
based on experience and test method.  If outward end deflection is occurring
during the test, the result will be to enhance an existing leak or to produc
                                   C-8

-------
an apparent leak in a tight tank.  Some data collected during the pilot  study
which indicated that the relaxation times ranged from 8 to 40 min.

     D.   Vibration

          Large vibrations produce waves on the surface of the product which
can make accurate reading of the level difficult.  The effect is most severe
for methods that rely on measuring very small level changes.  Vibrations can
occur as a result of nearby traffic or windy conditions.  It is usually  possi-
ble to extend test times, delay testing, or remove the vibration source  to
obtain- valid test results.

     E.   Water Table

          The presence of a water table will decrease the observed leak  rate
by reducing the differential pressure in the tank and by increasing the  re-
sistance to liquid flow.  If the fuel head is balanced exactly by the water
table head, no flow will be observed no matter how large the hole.  If the
hydrostatic head is higher than the fuel head, water will flow into the  tank
producing an increase in product level during the testing.

          Some methods, such as that of Petro-Tite, attempt to compensate for
water table effects by raising the fuel head so that the differential pres-
sure on the tank bottom remains constant for all tests.   This ensures that
fuel will flow out of the tank rather than water flowing in during testing.
It may also lead to misleading conclusions since holes above the water table
will be 'affected differently from holes below.  In addition, a bore hole must
bfe installed at each site to accurately determine the water table level.  This
can be costly and/or inconvenient.

          The exact effects of a water table cannot be calculated readily un-
less the position and size of holes in the tank are known.   The means to
achieve accurate evaluation were beyond the scope of this project; however,
generalizations can be made based on Bernoulli's principle.   This is repre-
sented graphically in Figure C-5 for two sizes of holes.  The behavior is
represented mathematically by the equation
                                   r2


where rx and r2 are the rates at a given hole for two different head pressures
ht and h2.   Tables C-l and C-2 illustrate the differential pressure for vari-
ous water table levels under testing and operating conditions.

     F.   Evaporation and Condensation

          Product loss due to evaporation is a potential problem, particularly
during hot or windy conditions where the tank is partially filled.  If the
test system is tight, no loss will occur.  Evaporation of product from the
standpipe could occur if the cover were left open during testing.  Condensa-
tion is generally a problem only if water condenses on the walls of the fill


                                   C-9

-------
  0.20




  0.18




  0.16




  O'.U




  0.12




» 0.10




  0.08




  0.06




  0.04




  0.02




     0
09
 o
ec
                                            I
                         3456789


                            Product  Height  Above  Hole (Ft)
                                                                  10    11
          Figure C-5.  Change  in  observed  leak rate as a function

          of head pressure  (feet  of  fuel)  above the hole.
                                   C-10

-------
Table C-l.   Differential Pressure (psig) Where Fuel Head is Adjusted
          to Product 4.00 Ib of Pressure at Tank Bottom for
              Various Water Table Levels During Testing
Distance below grade
0 ft (surface)
3 ft (top)
7 ft (mid level)
11 ft (bottom)
Table C-2. Differential
Water Table
Fraction filled
Surface (11 ft)
Full (8 ft)
3/4 (6 ft)
1/2 (4 ft)
1/4 (2 ft)

Below
tank
0.56
1.50
2.75
4.00
Water tab!
Mid
level
2.30
3.23
4.48
4.00
e level
Tank
top
4.02
4.96
4.48
4.00
Pressure (psig) of Gasoline Head
Levels Under Operating Conditions

Below
tank
3.43
2.50
1.87
1.25
0.64
Water tabl
Mid •
1 evel
1.7
0.77
0.14
-0.48
-1.09
e level
Tank
top
-0.03
-0.96
-1.59
-2.21
-2.82

Surface
5.30
4.96
4.48
4.00
for Various

Surface
-1.32
'• -2.25
-2.88
-3.50
-4.11
                               C-ll

-------
pipe and drains into the tank.  One buoyancy method compensates  for this ef-
fect by having product placed in a cup which is part of  the  sensor mechanism..
Evaporation losses from the fill pipe are compensated automatically by evap-tl
oration losses from the cup.
                                    C-12

-------
         APPENDIX D
TEMPERATURE BEHAVIOR'STUDIES
           D-l

-------
I.   INTRODUCTION AND BACKGROUND
                                                                             /"
          The precision arid accuracy of leak detection methods are  largely   '^
dependent upon their ability to recognize significant sources of variability
and to compensate for them accordingly.  The primary physical properties  of
gasoline, related hydrocarbons, and fuels warrant considering temperature as
one of the key parameters.  The approach taken for temperature compensation
for most methods assumes that the temperature changes throughout the tank are
uniform.  Although some data can be found to support these assumptions, other
data suggest that there are situations where they may not be valid.

          Product level changes encountered during testing may be partially
attributed to the contraction and/or expansion of the product being tested
due to temperature changes of the product.  Because these temperature  effects
can be pronounced, they must be accurately measured and appropriate correc-
tions applied to the data.
                                                   •
          The basic approach taken fs to measure the temperature of the produc
using a thermistor (or thermistors), located in the tank.  The observed tempera-
ture change is then used to calculate a volume change, which is then added to
(in the case of decreasing temperature) or subtracted from (in the case of
increasing temperature) the volume changes observed during the test to obtain
a corrected leak rate.

          This appendix summarizes the observations of several types .of tern-.
perature studies conducted by MR! from various sources during the development
and pilot programs.  The information was used to characterize the performance;'.;
of the methods evaluated during the method development study as well as to
assess the impact of temperature uncertainties on the test data.

          During the development study, three types of temperature data were
obtained.

          1.  Vertical temperature profiles of each tank tested during the
development study were measured at the beginning and end of each test.

          2.  A study that monitored temperature changes over an extended time
period was designed and conducted by MRI and Vista Research, Inc.

          3.  Temperature data from two Petro-Tite tests were obtained to
assess completeness of mixing.

          These data were used to supplement the characterization and  evalu-
ations of the temperature compensation procedures which were evaluated.   Each
type of data is discussed separately below.
                                   0-2

-------
II.  TEMPERATURE PROFILES

     A.  Experimental Approach

          Vertical temperature profiles were obtained at each tank by  lower-
ing a single thermistor into the tank in 6-in. intervals.  The thermistor was
allowed to stabilize at each interval until the digital readout showed  random
fluctuations only in the hundredths range (usually this was 30 s to 1 min at
each level).  The thermistor was then moved to the next level and the process
repeated.   Temperatures were recorded manually on data forms prepared for that
purpose.

          When the thermistor reached the bottom of the tank, it was raised
to the top in 6-in.  increments and the temperatures were again recorded.  The
cycle (down/up) was  typically repeated several times before the testing began
and again at the completion of the test.

     B.  Instrumentation

          The instrument used to collect the profile data was a single  thermistor
attached to an approximately 20-ft cable.  The thermistor was connected to a
single channel digital readout which displayed temperatures to 0.01°F.

     C.  Data Analysis

          All of the readings obtained for a given series of measurements were
averaged for each level.  An estimate of the standard deviation was also ob-
tained for series where more than one cycle of data was available.   The aver-
aged values were plotted against depth in the tank and a smoothed curve was
drawn through the data points.

     D.  Results

          Several examples of profiles obtained on the development study are
shown in Figures D-l through 0-4.  Data for each plot are also provided in
Tables 0-1 through 0-4.  The data were chosen to be representative of differ-
ent types of behavior and are not intended to imply a frequency of occurrence.

          Figure D-l is an example of a tank with a well behaved temperature
profile.   Because there is essentially no temperature gradient in the tank,
placement of the thermistor is not critical as long as it is not too near the
top or bottom of the tank.  Calculated temperature corrections should be very
representative of the total tank behavior.

          Figure 0-2 illustrates a less well-behaved system in that a verti-
cal temperature gradient does exist.  The temperature shifts along the gradi-
ent are not uniform, although neither the shifts nor the differences are large.
Considering that the two profiles were conducted approximately 8 h apart, the
differences are not surprising.
                                   0-3

-------
VJ
2


4


- 6
41
41
U.
„
.£
O.
41
a
8







10






12



14

0 Before Test
A After Test
-


A
— A
£
1
A
1
I
— A.
I
A
1
^
1
A
1
— A
1
A
1
A
1
A
— • A
1
A
V
^
I 1




O
1
1
O
o
1
0
1
o
1
o
1
0
1
o
1
0
1
o
1
o
1
o
1
0
1
o
1
0
SL
1 1
    60                 •            61                             62

                             Temperature, °F
Figure 0-1.   Temperature changes in unstirred tank—example of
uniform temperature shift - Tank A.
                            D-4

-------
 Table 0-1.   Temperature Comparison in an
          Unstirred Tank, Tank A

Depth            Mean Temperature (°F)
(ft)         Before testAfter test
 5             61.47              62.16
 5.5           61.49              61.35
 6             61.49              61.33
 6.5           61.50              61.33
 7             61.50              61.33
 7.5           61.50              61.34
 8             61.51              61.34
 8.5           61.51              61.35
 9             61.51        •      61.35
 9.5           61.51              61.35
10             61.52              61.35
10.5           61.52              61.35
11             61.51              61.36
11.5           61.50              61.36
12             61.49              61.36
12.5           61.49              61.36
13             61.47              61.50
                 D-5

-------
                                3000gal Regular

-------
  Table D-2.   Temperature Comparison in an
           Unstirred Tank, Tank B

              Mean temperature (°F) and
                  standard deviation	
              Before testAfter test
Depth     Mean      Standard         Mean
(ft)      temp.      deviation        temp.
  3       65.10       0.146          65.63
  4       65.24       0.0931         65.50
  5       65.31       0.0194         65.38
  6       64.62       0.06           64.89
  7       64.25       0.0497         64.46
  8       63.98       0.0194         64.18
  9       63.78       0.0232         63.90
 10       63.60       0.0343'        63.70
 11       63.34       0.0446         63.27
 12       62.47       0.0527         62.37
                   D-7

-------
                                    5
                41 
-------
 Table D-3.  Temperature Comparison in an
          Unstirred Tank, Tank C

Depth            Mean temperature (°F)
(ft)         Before testAfter test
 3             72.73              74.33
 3.5           73.16              73.38
 4             73.29              73.33
 4.5           73.33              73.33
 5             73.33              73.32
 5.5           73.33              73.32
 6             73.33              73.32
 6.5           73.33              73.32
 7             73.34              73.32
 7.5           73.38              73.32
 8             73.54              73.44
 8.5           73.43              73.44
 9             73.44              73.33
 9.5           73.25              73.05
10             72.88              72.69
10.5           72.54           •   72.31
11             71.37              71.40
                 D-9

-------
    Or
 
-------
Table 0-4,   Temperature Comparison in a Stirred Tank,  Tank 0
Mean temperature (°F)
Depth
(ft)
2
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
7.5 .
8
8.5
9
9.5
10
10.5
11
Before
Mean
temp.







70.50
69.85
68.90
68.66
68.27
67.97.
67.68
67.38
67.03
66.48'
65.89
65.01
test
Standard
deviation







0.843
0.734
0.757
0.265
0.116
0.161
0.169
0.233
0.254
0.245
0.281
0.146
After
Mean
temp.


68.70
69.01
69.17
69.21
69.22
69.23
69.23
69.23
69.24
69.23
69.23
69.24
69.23
69.23
69.23
69.22

and standard deviation
test
Standard
deviation



0.272
0.064
0.0231
0.0208
0.0
0.00577
0.0
0.0115
0.0
0.00577
0.0
0.0
0.01
0.0231
0.00577

Day^fol
Mean
temp.
69.54
69.49
69.42
69.36
69.29
69.23
69.17
69.05
68.80
68.46
68.12
67.52
66.96






1 1 owi ng test
Standard
deviation
0.0287
0.0427
0.0386
0.0419
0.0476
0. 0238
0.0404
0.0337
0.154
0.154
0.226
0.302
0.223




•

                           0-11

-------
          Consider the effect of placing a single thermistor  in the center  of
the tank versus placing three thermistors at the midpoints of one-third  volum
segments of the tank.  The single center thermistor located at the midpoint -
of the tank exhibits a temperature shift of ± 0.11°F over the S-h period.
This leads to a temperature compensation volume of 0.55 gal total or a volume
of 0.07 gal/h.

          For the three-thermistor approach one thermistor is placed at  the
center of each one-third volume segment so that each thermistor represents
one-third of the tank's total volume.  For an 8-ft tank this places the
thermistors at approximately 1.5, 4, and 6.5 ft from the top.  Using the
temperature changes at these locations from Figure D-2 gives a weighted  aver-
age of 0.51 gal total or 0.06 gal/h.  The differences in this case are neglig-
ible.

          Figure D-3 presents a set of data which exhibits unstable stratifi-
cation in the tank.  In this case the placement of the thermistor is importam
The midpoint and top levels of the tank exhibited almost no temperature  shift
while at the bottom a considerable shift was observed.  If the same calcula-
tions are performed for this case as were calculated for Figure 0-2, the resu'
is a temperature compensation volume of -0.063 gal or -0.008 gal/h for a sing'
midpoint thermistor and -0.23 gal or -0.03 gal/h for three thermistors.   In
this case, location of a single thermistor could be critical.  Placement of
the thermistor 12 in. lower would have produced a dramatic difference.   In
this case, the temperature change observed would have been 0.1.°F, correspond-
ing to a value decrease of -0.315 gal or -0.039 gal/h.  The difference betwee
this and the -0.063 gal (-0.008 gal/h) observed at midpoint is substantial.  :

          Figure 0-4 show;; the behavior of a tank which was tested using the
Petro-Tite method.  Profiles were obtained before and immediately after  the
testing.  A third profile was obtained approximately one day  following the
testing.  The data are presented- in Table 0-4.

          As can be seen, the stirring action did provide a nearly uniform
temperature profile by the end of the test period.  The tank  tended to return
to its original profile when left to stabilize again.

     E.  Discussion

          Although the temperature profiles'obtained from this test are  in-
formative, discussion of the data must consider the following.

          1.  The profiles were obtained at gnly one position in the tank,
which was usually close to one end.  The temperature behavior in other loca-
tions is only  inferred to be similar in behavior.

          2.  The test periods generally extended at least two to four times
the  duration  of typical tank tests.  The data collected during a routine test
are  usually on the order of 1 h.
                                   0-12

-------
          3.   All of the testing took place in September and October when
ambient temperature conditions were generally mild.  Extremes of either hot
or cold weather were not encountered.

          4.   All tanks were filled at least 12 h before the testing was con-
ducted.

          Under the test conditions encountered on this program, the tempera-
ture data from unstirred tanks are reasonably well represented by a single
thermistor located at the center of the tank.   The data represented in Fig-
ure 0-1 is typical of that frequently encountered.  However, the limited data
presented in Figures D-2 and D-3 suggest that under some conditions the tem-
perature behavior may be nonuniform.  This raises some concerns as to the fre-
quency and severity of the problem which are not answered by the profile data.
The major questions are:

          1.   What is the minimum stabilization time for an unstirred system?

          2.   Are there significant effects from filling the tank with product
which is much warmer or much cooler than the ground temperature around the
tank?

          3.   What effect is produced by topping off the tank with product
which is either much warmer or much cooler than that already in the tank?

          4.   Does ambient temperature affect the test in any way?

          5.   What is the temperature behavior of tank systems over a long
period of time under a variety of conditions?

          The behavior of the vertical profiles of the stirred tanks as il-
lustrated in Figure 0-4 was typical of all of the Petro-Tite tests-.  However,
it is not known  if horizontal mixing is complete as well.  This is of some
concern because  in most cases the profiles were obtained near the pump loca-
tion, which was  usually at one end of the tank.  The deflection angle of the.
nozzle and the implied mixing pattern seem to avoid this problem.

          In addition,  it is not known when the vertical profiles were brought
to a uniform condition.  These tests were conducted over an extended time
period of 6 t 8  h while the normal Petro-Tite test used data collected after
1 to 2 h of mixing.

          As a result of these observations, additional testing was proposed.
This is described  in Section III.


III.   TEMPERATURE  STABILITY STUDY

     A.  Experimental Approach

          A temperature stability study was conducted  in Kansas City on a
10,000-gal tank  located at a temporarily closed service station.  The testing
involved collecting the following three sets of data:

                                    D-13

-------
          1.   Background temperature behavior of the 4,000 gal of  fuel  in  the
tank before the testing and immediately after the filling of the tank.

          2.   Product temperature over the next few hours wi-thout  circulation

          3.   Product temperature during mixing using a Petro-Tite circulatio
pump.

          The data collected were used to estimate the stabilization  time  re-
quired for testing procedures which do not circulate product and the  circula-
tion time required to achieve complete mixing in the tank.  As is  discussed
in the results section, due to the circumstances of the test (fuel tempera-
tures, ground temperature, and ambient temperature), very little temperature
gradient resulted from the filling operation.  In addition, the electronic
components for one of the thermistor arrays were apparently affected  by am-
bient temperature changes.  Hence, the temperature change data for that array
are suspect.

     8.  Instrumentation

          1.   Thermistor Array

          To accurately study temperature stratification and gradients  for
extended periods of time, a rigid support system was utilized to ensure accu-
rate and systematic thermistor placement.  The. array sytem was designed with
an unfolding mechanical arm extending from the vertical support frame,  thus
enabling the collection.of both horizontal and vertical temperature data as  ;
shown in Figure Q-5.

          The array support systems were constructed of polyvinyl  chloride
(PVC) because of its ability to act as an insulator while remaining relativel
inert to petroleum products.  Each thermistor array was constructed of  2-in.
schedule 80 PVC for the vertical main frame and 3/4-in. schedule 40 PVC for
the extending mechanical arm.  To facilitate placement of the laser interfere
eter height measurement system into the tank, the vertical main frame was
necked down to 3/4-in. schedule 80 PVC.

          Extension and retraction of the mechanical arm were accomplished
through the use of two control cables.  Once positioned, the arm was  secured
to prevent movement during testing and product circulation.  To accommodate
uneven tanks, due to sagging or settling, one array was equipped with a level
adjustment screw.  This allowed for planar placement of thermistors between
arrays.

          Both arrays were equipped with at  least 16 Omega OL-701  general
purpose linear temperature probes (with resolution to 0.01°F), containing  YSI
44018 thermistor composite.
                                    0-14

-------
                      Array 2
                        O
                     Thermistors
                       21-38
  Array 1
   O
Thermistors
   1-20
      38 - Ambient Air
                             Fuel  Level
                 30   29   28   27
             20
             19'
             18
             17
            -16
             15'
             14-
             13 •
             12-
             11 •
                                                   6
                                                   5
                                                   4
                                                   3
                                                   2
                                                   1
                    10
7
Figure D-5.   Placement of thermistors on  arrays  by  thermistor number.
                               0-15

-------
          Prior to installing the thermistors in the arrays, a  calibration
was performed on the sensors and the correction factor obtained for  each  was
used in the final data analysis.  A dewar temperature bath with a  styrofoam :-:
insulation lid that secured two ASTM (0° to 30°C) mercury thermometers  reader
able to 0.1°C was utilized.  After the bath equilibrated, values of  all
thermistors and both thermometers were recorded.  From the two  ASTM  thermom-
eters, a bath temperature was established and correction coefficients were
calculated for each thermistor.

          2.  Laser Interferometer Height Measurement System

          A two-tube laser interferometer was used to monitor the  product
level from the center fill pipe during the tests.  A quartz thermistor  was
attached to the tubes at the midpoint of the tank.  A second quartz  thermis-
tor was installed at the center level of the tank from the end  fill  pipe  as
part of the laser interferometer equipment.

          3.  Circulation Pump*

          To facilitate Phase II testing during product circulation, a  Heath
Petro-Tita circulation pump was used.  The suction inlet is placed at least
6 in. below the top of the tank while product is discharged (~  25  psi)  at the
bottom of the tank, above any water.  The discharge outlet is angled ~  45 de-
grees upward from the bottom of the tank, attempting to create  a spiral-like
motion along the long axis of the tank.  Petro-Tite recommended circulation
time is 5 to 8 min/1,000 gal; during Phase II product was circulated for  16 h
with continuous data collection.

          The Petro-Tite method uses one thermistor in the bottom  of the  suc-
tion tube to monitor the tank temperature.  Temperature changes from this
thermistor, sensitive to 0.003°F, are used to calculate volume  changes  (due
to thermal expansion and/or contraction) and to compensate for  these changes.

          During Phase II testing, attempts were made to operate the circula-
tion pump while Array No. 1 (probes 1-20) was in the same opening.   However,
this was not possible due to physical constraints.  Subsequently,  one array
was removed and temperature" was monitored by the Petro-Tite thermistor.

          4.  Data Logging

          Automatic and efficient recording of the data generated  was accom-
plished through a portable microprocessor-based data logger.  The  Doric
Oigitrend 235, capable of handling up to 100 analog inputs, was equipped  with
remote mounting of two front-end modules RTD (FEW) to allow for maximum flex-
ibility in  instrumentation setup.

               a.  Linear Temperature Network

               The thermistors  used in this study undergo a large  change  in
electrical  resistance from small changes in temperature.  This  probe re-
sistance is converted to a voltage through the use of a linear  temperature
network.  A linear temperature  network was utilized because it  required a


                                   0-16

-------
less complex output circuitry to convert sensor resistance to a useful  tempera-
ture indicator while retaining a high degree of sensitivity.  The original
linear temperature network was redesigned by MRI to reduce the maximum  linear-
ity error from 0.4°F to less than 0.1°F over a range of 40° to 120°F.

               b.  Front-end Module Interface

               The signal input wires from the linear temperature network were
terminated into screw-down clamps in the remote front-end module (FEM)  inter-
face.  Each interface can handle inputs from up to 20 sensors and is  specified
for an operating temperature range of 32° to 158°F.  The remote FEM interface
was then connected by a single cable to the FEM mounted inside the data logger,
where the analog to digital conversion occurs.

               c.  Data Logger

               The data logger used for this study was equipped with  an op-
tional math functions card which enabled on-site mathematical handling  of the
data.  All data channels underwent a continuous 7-s synchronous scan, and a
point average function was utilized to generate 1-min averages for each chan-
nel.  All programs were protected against accidental changes by locking the
data logger in the on (protect) position.

               During testing, the data logger was operated by 110 VAC  with
all memory circuits backed up with battery power.   An on-site hard copy of
all data was obtained by using the thermal alphanumeric dot matrix printer on
the data logger.

               d.  Microcomputer

               A computer-readable record of the data was necessary due  to
the volume of data generated (approximately 2,500 by 40 data points).    Due to
the lack of a portable IBM PC-compatible microcomputer, an indirect route to
the final PC-DOS data files was necessary.  The serial output channel  of the
Doric logger was connected to an Epson HX-20 portable computer.   A short BASIC
program in the Epson waited for each 40-channel scan from the Doric and  stored
output on the built-in microcassette tape.  Each tape side held about 90 min
of data when using 1-min averages.   The Epson also produced a printed log of
each tape file ID as the data were recorded.   Each file entry included  the
current date and time transmitted by the Doric as well as the time-averaged
temperature for each channel.

          5.  Data Transfer

          After re-turning to MRI, the tape data were transferred to disk files.
A'separate BASIC program read the data tapes and transmitted the data in ASCII
over a local phone modem link to a Radio Shack Model I which used the Vidtex
communication program to record the data on diskettes.  A separate file con-
version utility (Michtron TRSDOS/IBM Transfer) produced a PC-DOS 1.0 disk
replica of the original Doric logger output.
                                   D-17

-------
          The final conversion step was a BASIC program which  spliced togethei
the random length data files and rearranged the data  into  1-h  blocks  in se- 4
quential data file format.,  Each data scan consisted  of an alphanumeric datej|
time entry followed by 40 numeric temperature readings.  Each  hourly  file namt
was defined as "ISTDdhh" where d is the final digit of the Julian  day and hh
is the hour (0 to 24 h time).   This form allows the subsequent data handling
programs to find any desired date block without operator assistance.

          The contents of the final data files were transferred to a  printer,
and a manual check of randomly selected data periods  verified  that the various
data transfers were free from errors.  Scattered gaps occurred in  the data
when the data tapes were changed or as a result of unrecoverable cassette data
blocks.  Data recovery was > 95% for the overall test period.

          Data were transferred to the HP110 3-1/2-in. discs and processed.
The HP-110 is equipped with Lotus 1,2,3  and BASIC programming software.   For
field testing applications, the computer, disfe drive, and  printer  are battery-
powered.

     C.  Test Scenario

          Field testing began late afternoon with the tank (8  ft x 27 ft)
filled to approximately 40% nominal capacity (4,000 gal).   The fuel was al-
lowed to remain undisturbed for a period exceeding 48 h prior  to testing.
When testing commenced, each thermistor data point was evaluated every 7 s
and the 1-min averages of these instantaneous values  were  recorded.
                                                                             s-v;

          1."  Preliminary Phase I                                            '"*

          Initially, a single thermistor array was 'placed  into the partially
filled tank and allowed to collect data fjor a period  exceeding 2 h.   Nineteen
thermistors were used for this period, five submerged and  the-  remaining in
the vapor space above the product.  After the above data were  collected,  the
tank was then filled to > 99% nominal capacity with continuous thermistor mon-
itoring.

          2.  Phase I

          The second phase of the testing involved implementing both  thermisto
arrays  (33 thermistors submerged, three thermistors in the fill  pipe  vapor
space), the laser  interferometer height measurement system and the two quartz
thermometer probes.  Data from all acquisition systems were collected for a
period of 19 h, without product circulation.

          3.  Phase II

          At this  point one thermistor array was removed and a Heath-Petro
Tite circulation pump was installed  and allowed to circulate for 16 h, during
which data were collected from the available systems.
                                    0-18

-------
          Ambient weather conditions and temperatures were monitored through-
out testing on site and additional information was obtained through the  local
National Weather Service.  Total testing and data collection spanned approxi-
mately 37 continuous hours.

     D.   Data Analysis Techniques

          Temperature data received by the data logger were corrected using
the calibration factors for each thermistor.  The data were then plotted as a
function of time.

     E.   Results

          Four sets of test results are shown.

          1.   A plot of the ambient temperature data from thermistor No. 38
is shown in Figure 0-6.

          2.   Plots of the volume changes observed from level measurements
are plotted with the calculated temperature induced volume changes from  the
two thermistors associated with the laser system are shown in Figure D-7.
The data are provided in Table D-5.

          3.   Plots of the temperature behavior of four thermistors located
on array No.  2 are shown in Figure D-8.

          4.   A plot of thermistors 12 and' 32 is shown in Figure D-9.

          Two problems encountered in this testing limit the interpretation
of the results.  First, due to the temperatures of the fuel, ambient air
ground,  little or no temperature gradient was observed during the course of
the testing.   Second, the apparent electronics drift in the results of array
No. 2 limit the comparison of the data collected at the end of the tank with
the collected in the center.   Nevertheless, some useful information can be
extracted.

          The laser system had two thermistors located at the vertical mid-
point of the end and center which independently measured temperatures.   These
data indicate that the temperature behavior at these two locations tracks
closely.  These data also indicate that all of the product level changes are
attributable to changes in temperature, as shown in Figure D-7.

          It is evident that all of the thermistors on a given array track
closely  together as is shown in Figure D-8.  This suggests that vertical
temperature changes at a given location are uniform.

          The data presented in Figure D-9 show the difference in behavior of
two equivalently placed thermistors on different arrays.   The differences shown
are not  believed to be real because of close tracking of the two thermistors
associated with the laser system.   Independent laboratory checks on the in-
strumentation did verify that some electronic components on array No.  2 were
temperature sensitive while those on array No.  1 were not.


                                   D-19

-------
ft  —  o
                                            ri  —  o
               if? T  T  T  *r  rr
c jl)   3 H
                0-20

-------

-------
 Q>



 i.
 i.
 Oi
 c
•f^
 i-
 3  41
o  e
   •^«
 U) i—
 01  O
 CD Ul
«j -a
    4)
   4)
              •o
           3  ••-
          U  €
              a.
              o
            -o
 — -a
 4) ._
 a s
           •U  CU
           —  o
OOOOOOOOOOOOr-l









d d d d d d  d d d d d  d i-J









ddddddddddddd










o d a* o' d o'  d d d d d  d d     d










ddddddddddddd     d










d d d d d d  d d d d d  o* d     d










d d d d d o*  d d d d d  o' d










ddddddddddddd










ddddddddddddd








p^i—Ii-Hi—tfHt—It"trHf—It—(i—tr—(CM     S
                                                             D-22

-------
                          TEMPERATURE STUDY
                            During Circulation



_
u_

-------
                                                                                                 »n

                          in

                          °
                           *
                                 0

                                  OJ
^  §
Z  r
2  °   !
u«   '
<  i
as =.
!£  o
2  «•
^  
                                                     D-24

-------
          Although the data were seriously compromised, they do  suggest  that
under the conditions of the test, a. single thermistor located at the  midpoint
would have predicted volume changes quite well.  Since  little gradient existed
in the tank, the results of stirring were uninformative.


IV.  MIXING COMPLETENESS STUDY

          This study presented additional data supporting the conclusion that
the circulation procedures used by the Petro-Tite system produce uniform tem-
perature behavior, both vertically and horizontally.

     A.   Experimental Approach

          Two sets of data were collected.  The first involved the  use of a
modified Petro-Tite system.  Three thermistors were mounted vertically in the
access hole of the tank.  Temperatures were monitored at 15-min  intervals dur-
ing the test.

          The second set of data was collected using three thermistors placed
along the top of the tank.  In this case, three Petro-Tite units  were in the
tanks, but only one end and the center unit were used to circulate  product.
Data were collected at 15-min intervals.

     B.   Instrumentation

          Petro-Tite equipment was used to collect both data sets.  Tempera-
tures were recorded manually from the standard Petro-Tite temperature units.

     C.   Data Analysis

          The temperature changes were first converted  to the equi-valent
volume changes for the tank being tested.  The data were plotted  as cumula-
tive volume change versus time and volume change versus time for  each ther-
mistor.   The plots .were then examined visually to determine trends.  The
volume changes for each thermistor were calculated for  the entire tank and
for the segment represented by each thermistor.

     0.   Results

          The results of the measurements from the vertical array are given
in Table D-5.  Plots of the cumulative volume change are shown in Figure D-10.
A plot of the volume changes for each thermistor for each 15-min  interval  is
shown in Figure D-ll.

          Analysis of variance of the temperature changes in the  vertical pro-
file showed no significant differences among the- thermistor at different depths.
The F statistic was 0.11 with 2 and 24 degrees of freedom.   The  estimated stan-
dard deviation of a 15-min temperature-related volume was 0.0133  gal.  The
mean temperature-related volume change per 15-min period for each thermistor
is shown at the bottom of Table D-5.
                                   D-25

-------
                                                                                  o
\ \
 v \
    •\'\
     '*  \


     \ \
it
                     • v\
                       •v\
                            '\
                                     \-.


                                      S
                                          \
                                                              ,\
                                                                              f- ^-

                                                                              i   r—•,
                                                                              i-
                                                                                 !!(    • UJ
                                       D-26

-------
                                                              ITS


                                                              i.
                                                              i.
                                                              
-------
          The results of the measurements from the horizontal array are  give
in Table D-6.   Plots of the cumulative volume change and volume changes  are
shown in Figures 0-12 arid D-13, respectively.

          Analysis of variance on the volumes corresponding to temperature
changes at three horizontal positions in a tank also showed no significant
differences by position.  The F statistic was 0.72 with 2 and 18 degrees  of
freedom.  The estimated standard deviation for a 15-min temperature-related
volume change was 0.0088 gal.  The mean volume resulting from the temperatur
change per period is shown at the bottom of Table D-6.

     E.  Discussion

          The test data collected during these two field tests support the
assumption that circulation is complete within the mixing and high level
phases of the test.
                                                                       •
          There are, however, some limitations presented by the data.  First
since the absolute temperature accuracy of a Petro-Tite thermistor is ±  3.0°!
absolute temperature differences in the tank were not determined.  A completi
multipoint calibration of each thermistor used would be required for this con
parison.  Second, nothing is known about the horizontal profile of the tank
used in the vertical profile test or the vertical profile of the tank used ir
the horizontal profile tast.  For the vertical profile test, the tank had be
filled immediately prior to the test.  The horizontal profile tank had been
tested three days earlier and had been sitting undisturbed since that time.

          These factors suggest that large temperature gradients were probat:
not present, and the conclusions must be judged in that light.  However,  sin
the pumps add heat to the tank (as evidenced by the increasing temperatures)
irregular temperature patterns would be expected if mixing was incomplete.
That this was not observed lends support to the conclusion that mixing durin>
the testing was at least adequate.

          The analysis of variance showed no significant differences in  tem-
perature-related volume changes for the vertical array thermistors.  The  ob-
served differences were quite small, which supports the conclusion that  the
mixing was adequate to attain uniform temperature changes vertically through-
out the tank.

          The observed differences in temperature-related volume changes  for
the horizontal array of thermistors were small and not statistically signifi-
cantly different.  This supports the conclusion that the mixing was adequate
to attain uniform temperature behavior horizontally throughout the tank.  To-
gether, these data lead to the conclusion that the mixing succeeds in attain-
ing uniform temperature behavior for the product in the tank.
                                   0-28

-------
'. \ \
h&v
   o
   S^
                                                                     rO

                                                                     S_
                 \
                                                     IS
  o
L "
r «
               +J
                C
                O
                IS

                s-
                o
                0)
                01
                                                                    JZ
                                                                    u
                                                                    c
                                                     ho
                O

                

                                                                    3
      i^nTO-^ 7
                            0-29

-------
                                               
-------
          The vertical profile test results shown in Figure 0-12 show some-
what more erratic results than were expected for a small (1000-gal) tank.
Calculation of the volume correction using the top thermistor only versus a
average of all three thermistors gives values of 0.605 and 0.624, respectively,
for the 2-h test.  This difference of 0.019 gal seems well within the range
of variability experienced in the development study and expected for the na-
tional survey test results.

          The horizontal profile results are very consistent.  The cumulative
volume changes using probe No. 3 (the master probe in this test) versus the
average gave essentially identical results for the 2-h test.  In this case
two circulation pumps were used which may have improved the circulation pat-
tern .  A 10,000-gal tank does not normally require two circulation pumps.

          There are some limitations to the data in that not all relevant ini-
tial conditions could be tested.  However, the data obtained support the con-
clusion that the mixing achieves uniform temperature behavior in the tank.
                                   0-32

-------
 ID
o
a
a
   •a
 C  41
•T-  10


 OJ
 C    

a.
    O3

 10 -r-
4-1  C
 C •—
 o  ^
 N 4J
•i-  C
 s-  a
 o «j
 i
o
   01

              £-§
              3  S-
              O  ^
             i— CJ
              o
              >   a*

              £ "0
              3 5-
             CJ &
              a
              £ co

             "o ja
             > o
              01
              £ CM

              r—    a

              <3  Q.
                o
•(-> en

5) at

a  o
    s_
<3  a.
              o  o
                  i.
              <  a.
              O3 O)

              a  o

              
                                     o oo oo
                               «•> o
                                               -33 r- ao  us r-
                           unomoinoinoinoin     c
                           rH  en ^f c3 rH en  *^* e? rH en ^*     to
                           ^^  L^? ^^ iiO n3 ^^3  10 r^* r^ ^^ ^^*     0}
       a.

       3
       a.

       c
       o
        (O


        u

       •1—
        u
    a.  o

    c
•a  o  £
 C •!-  O
 O 4"^  S-
                                                                 c 3 -a
                                                                 

                                                                 3 0)  tfl
                                                                 CL-M  O
                                                                    c  a.
                                                                 O <1>  Q.
                                                                 C U  O

                                                                 I   I   I

                                                                rH CM en
                                                                              O  O O


                                                                              4J  &) 3)

                                                                              o  a o
                                                                              s_  i. s_
                                                                              a. a. a.
                                                          0-31

-------
         APPENDIX E





SITE VISITS TO OBSERVE TESTS
           E-l

-------
     A.   Objectives

          Arrangements were made to visit each of the sitas tentatively  se-
lected for the pilot study.  The primary purpose was to assess the method  ir
terms of its suitability for use on the National Survey.  Methods judged to
present satisfactory potential for the National Survey were selected  for fur
ther testing.   A summary of each site visit follows.

     B.   Leak Lokator Site Visit

          A site visit was conducted on April 18, 1984, to observe the Leak
Lokator system in operation at a site near Philadelphia, PA.  The system was
operated by Hunter Environmental.  Dan Heggem and Mike Kalinoski from EPA  an
Ken Wilcox from MRI observed the test.  The principal contacts from Hunter
Environmental  were Oonna Hymes and Joyce Rizzo.

          1.  Description of Method

          This method is based on measurement of the level of liquid  in  the
tank.  The sensor is a patented buoyancy probe connected to a balance that i<
suspended over the tank fill pipe.  As the liquid level changes occur, the
buoyancy of the sensor changes and the change is measured by the balance.
The changes are monitored electronically and transmitted to a signal  process
where they are continuously recorded on a strip chart.  The temperature  of
the tank is continuously monitored with a single thermistor which is  located
at the mid!eve! of the tank.  The. temperature is also recorded on a strip
chart.  More sensitive temperature readings from a digital voltmeter  are   <
manually recorded on the chart at periodic intervals.  The system is  cali-
brated prior to each use and again at the conclusion of the test by insertin
a metal rod of known volume into the tank.  This allows the effect of the
known volume change to be calculated and removes the necessity to know exact
tank dimensions.

          The precision of the method is stated to be 0.05 gal/h if testing
is conducted on an overfilled tank.  The test can be conducted on a partially
filled tank, but the precision of the method will be lower, particularly wit]
large tanks.  The normal procedure when testing a tank is to test with prod-
uct level raised to around 28 in. in the fill pipe.  This checks the  overall
system tightness including the .piping system which leads to the dispenser.
If a leak is observed, then a second test is conducted at a level just above
the top of the tank to test only the tank.  This information is used  to  dif-
ferentiate between leaks in the tank and leaks in the piping system.  The  pip
ing system can also be separately pressurized if necessary.

          2.  Observations

          This method is well developed and the operator, Hunter Environment-
has considerable experience in conducting tank tests.  Over 12,000 tanks havi
been tested over the past 3 yr.  Hunter currently has 15 units operating in
the field and plans to add several more.  Hunter can generally be at  a site
within one week's notice and can test several tanks at one location in one
day.  The only adverse weather condition which seems to present a problem  is


                                   E-2

-------
high wind.   This condition causes small movements in the balance which are
recorded as noise on the strip chart recording the level changes.

          Hunter also places a high value on the training of its personnel.
This training includes a thorough knowledge of tank plumbing systems as
Hunter considers (correctly) an understanding of plumbing essential for  cor-
rect interpretation of results.

          3.  Recommendati pns

          MRI recommends the Leak Lokator method for further study.  The sys-
tem is well designed, it seems to have the requisite accuracy, and Hunter has
had considerable experience in conducting testing.  In addition, since Hunter
has IS units available, it provides the necessary flexibility in meeting test
demands.

          Since Hunter does not sell or lease its equipment, any testing us-
ing this method will have to be contracted through them.  Some additional data
relative to the sensitivity and accuracy of the method have been requested
from Hunter.

     C.  Certi-Tec Site Visit

          A site visit was conducted on April 20, 1984, at the Fuel Recovery
Company,  Inc., St. Paul, MN,. to observe the Certi-Tec tank testing system in
operation.   Fuel Recovery's main business is the recovery of spilled petroleum
products and'the restoration of ground water supplies which have been contam-
inated.  The test was observed by Mr.  Tom Muir from EPA and Or.  Ken Wilcox
from MRI.   The principal contact from Fuel Recovery was Mr.  Dan Bigalkie.

          1.  Description of the Method

          The Certi-Tek method is based on measurement of the level of liquid
in the tank by measuring pressure changes which occur within the tank as a
result of loss of product.  A-pressure transducer is attached to a rod which
is lowered into the.tank until it rests on the bottom.   The rod is then clamped
to the.top of the tank.  Five thermistors are also attached to the rod at
intervals of approximately one-fifth of the diameter of the tank.  Temperature
and pressure readings are then collected by a data logger which updates a CRT
every 5 s'.   The test values are printed every IS min on a tape.   Although the
data logger did not average the readings during the site visit,  it does have
this capability.  After data are collected for 1 to 2 h, the data from the
tape are entered into a computer manually and the loss of product is calculated.

          The 6-ft diameter steel tank observed during this test was not buried.
The capacity of the tank, which was filled with water,  was 1,000 gal.   The
test was conducted in the shop area of Fuel Recovery Company.   The testing
was conducted by introducing a "leak" by opening a valve which was located at
the bottom of the tank and capturing the water with a bucket for volume mea-
surement at the end of the test.   After approximately 1 h of recording data,
a measured volume of 0.50 gal had leaked from the tank.  The computed leak
value was 0.59 gal.


                                   E-3

-------
          2.  Observations

          This test method seems to have high potential for  use  in  leak de- «
tection of underground storage tanks.  One of the major attractions  of  the  ;
method is the relatively low cost of the equipment.  The method  is  a simple
one and the equipment can be installed at the test site very quickly.   The
system as it is now configured at Fuel Recovery needs some further  refinemenl
and the company states tnat they are continuing to work on this.  There is n(
field test data available so that evaluation of potential problem areas cannc
be concluded at this time.  Refinements which could be made  include  the addi-
tion of a continuously recording strip chart and the direct  input of data to
the computer.  Considerably more precision and accuracy data will need  to be
collected before the method can be used for routine tank testing.
        i
          3.  Recommendations

          MRI recommends the Leak Lokator method for further study.   Fuel
Recovery is in a position to provide this support and will have worked  out
some of the details regarding its use before the pilot study begins.

          If the method is selected for the National Survey, EPA should con-
sider options such as purchase or lease of this equipment for the study.

     0.  Varian Helium Detector Site Visit

          A site visit was conducted on May IS, -1984, at Smith and Oennison,
Haywood, CA, to observe a tank test using the Varian Spy 2000 leak detection .
system.  The test was observed by Ken Wilcox, MRI, and Tom Muir, EPA.   Bill  ?
Burkhardt of Smith and Oennison and Jack Farmer of Varian also observed the
test.

          1.  Description of the Method

          Helium, which diffuses readily through soil and even concrete or
asphalt, is used to pressurize the tank to be tested.  The helium, which will
diffuse rapidly through the leak and rise to the surface within  a matter of
minutes, is detected by an instrument based on mass spcctrometric techniques.
In most cases it is necessary to drill small holes in the concrete or asphalt
covering the tank.  A'matrix of holes is drilled for the initial evaluation.
Additional holes are then drilled in areas whe're helium is initially detected
With experience, the operator can learn to make judgments as to  the  size of
the  leak, as well as its approximate location.

          The method seems straightforward, and trained technicians  should be
able to operate the monitors without difficulty.  The tank must  be drained
completely  in order to test the bottom, which may represent  some risk in area
where high water tables exist.  Quantitative estimates of the leak  rate can
be obtained using a differential pressure system developed by Smith  and
Oennison.
                                   E-4

-------
          2.  Observations

          The tank tested was constructed of fiberglass and was believed  to
be leaking because water had been found in the tank to the extent that  it was
necessary to pump out an estimated 400 gal/wk.  Approximately a year earlier
the tank had been uncovered and the fittings on the top checked for evidence
of leaks.  The tank had been installed approximately 10 yr earlier.  The  water
table in the area was observed to be above the tank while it was uncovered
the previous year, but based on evidence in a sump hole near the tank,  the
water table was estimated to be approximately 4 ft above the bottom of  the
tank during the present test.

          Initially, a grid was laid out above the tank at 5-ft intervals.
Helium was used to pressurize the tank to approximately 5 psig through  the
dispenser which had been disconnected.   The vent pipe for the tank was  also
capped off.

          Holes were drilled in the concrete pad above the tank on the  grid
marks using a 5/8-in. concrete bit.   Two helium detectors were used to  moni-
tor the leaks.  The Varian Model Spy 2000 owned by Varian was operated  by Jack
Farmer of Varian, and a similar instrument owned by Smith and Dennison  was
operated by Bill Burkhardt.

          During the test, each hole was monitored with a leak detector and
the maximum concentration obtained at each .hole was recorded on the grid.  If
the detector went off scale, a shorter sample time was used.   All of the  holes
were then monitored for the shorter sample time.   Additional  holes were drilled
where high concentrations-of helium were detected and the test sequence was
repeated.  The resulting pattern was then used to determine the approximate
leak location and obtain an estimate of its magnitude.  Some operator judgment
is necessary in selecting additional holes and in assessing the data.

          3.  Recommendations

          This test method is extremely sensitive for pass/fail determinations.
Since helium diffuses readily through very small  holes, "leaking" pipe  joints
caused by failure to dope the threads during assembly could lead to a false
positive result even though the leaking of product through such a joint would
be negligible.  Additional data can be collected if the tank fill pipe, dis-
penser connections, and vent pipe can be adequately sealed.   Further work is
proceeding on this technique.  In spite of the extreme sensitivity of the
technique, the helium leak system should be evaluated in the pilot study.  It
is readily portable and can be easily operated by a trained technician.   The
cost for the instrument is relatively high, but it may be possible to make
lease arrangements for the equipment.

     E.   Site Visit, National Institute for Petroleum and Energy Research
           (NIPER)

          A site visit was conducted on May 2, 1984,  to the National  Institute
for Petroleum and Energy Research (NIPER) in Bartlesville, OK.   Soil  coring
                                   E-5

-------
and drilling procedures used by NIPER to locate naturally occurring  oil  poo'
were observed.   It was hoped that these techniques could be adapted  to  idenl
fying leaking fuel tanks.   NIPER has had considerable experience  in  this ar*
and feels that the technique shows strong promise.  The site visit was  con-
ducted by Mike Kalinoski,  EPA and Ken Wilcox, MRI.  The principal contacts c
NIPER were Herbert Carroll, Jr. and Gene Collins.

          1.  Description of the Method

          The presence of specific trace organic compounds can be detected i
both soil core samples and ground water samples taken in the vicinity of sto
age tanks.  Portable gas chromatographs using a variety of detectors can be
used on site to detect low ppb levels of contaminants.  In both cases small
holes or wells are drilled around the perimeter of the storage area.  Soil
cores obtained are transferred to airtight storage containers.  After a  suit
able equilibration time, a'sample of the headspace in the container  is  an-
alyzed by the gas chromatograph.   A sample line may also be lowered  into the
hole to detect the presence of hydrocarbons in the ground if desired.  Groun<
water samples obtained from these wells may be similarly analyzed for specif
species which are volatile and of low solubility in water.  If a series  of
holes is drilled systematically,  it is possible to identify the probable
source of the leak.  If a more precise determination is required, the sample
can be returned to the laboratory and a full extraction with GC/MS analysis
performed.

          The ability of this method to detect hydrocarbons is high.  Howeve
the specific source of the hydrocarbons could be subject to questions.   Port
able instrumentation is readily available and has a high sensitivity if  prop
erly used.  However, where the storage tanks are located beneath concrete
drives, as may be the case for most service stations, the drilling of enough
holes to locate the leak may be impractical.  Useful information regarding
the soil characteristics around the tank could also be obtained from the cor
ing operation.

          2.  Observations

          During the visit to NIPER, a test drilling was observed to a depth
of several feet.  The drill can be mounted on a trailer and. pulled to the te:
location. . A fully equipped van with a power supply and gas chromatograph is
available to perform on-site analysis so that decisions as to the location o
the next test hole can be determined.

          A GC analysis was also performed by NIPER in the laboratory to
demonstrate that capability.  The analysis time was around 10-12 m'in per sam-
pie.  The results were recorded on a strip chart so that the sample  signatur
could be visually compared with known standards.

          3.  Recommendations

          Since the signatures of a leaking fuel tank have not been  reliably
determined, it seems unlikely that the method can be used without further
development.  Nevertheless, it is desirable that at least a few core samples
be collected from sites where fuel is known to have been leaking and analyzec
The number of samples to be collected may depend on the available funds.

                                   E-6

-------
          Since the NIPER personnel seemed to have very positive  feelings
about this technique, they were asked to prepare a short proposal  to  provide
details of what could be done and estimate costs.  This proposal  will  be eval-
uated before final selection for the pilot study is completed.

     F.  Petro-Tite Site Visit

          A site visit to 0. H. Materials was conducted on May 9,  1984, to
observe the Petro-Tite method in operation on a company owned tank.   Dan
Heggem, EPA, and Ken Wilcox, MRI, observed the test.  Petro-Tite  has  some 600
franchised dealers in the United States.  The method, known in the  industry
as the Kent-Moore test, has probably been used to test more tanks  than any
other tank testing method.  The National Fire Prevention Association  (NFPA)
test code is based largely on this test method.

          The principal contacts at 0.  H. Materials was John Copus.   0. H.
Materials was recommended by the Petro-Tite Company as being one  of their
better testing groups.  Jack Stillwagen of the Heath-Petro-Tite Company also
observed the test.

          1.  Description of the Method                       , •

          The Petro-Tite test is a hydrostatic test that compensates  for tem-
perature, pressure, and viscosity variations.  The Petro-Tite test consists
of exerting a pressure head of 5 Ib on the tank.   A pump is used  to circulate
the product throughout the tank in order to promote a uniform temperature.
Using a thermal sensor, the temperature changes are monitored to  account for
the expansion and contraction of the product.  This test is conducted by mea-
suring all product added or removed from the standpipe in order to maintain a
constant head.   By comparing the product added or removed with anticipated
volumetric changes resulting from temperature variations, it is possible to
detect leaks in the tank system.

          For storage systems with submerged pumping, the Petro-Tite  test must
be run separately on the tank and on the piping to give good results.   On suc-
tion delivery tank systems, the test checks the entire system simultaneously.

          The Petro-Tite test requires  several hours (usually one work day)
to complete.  Generally, during the first few hours of the test,  there is a
drop in the standpipe level because of the increase in the internal tank pres-
sure.   This is  compensated for by reducing the tank pressure after 2 h to con-
trol tank expansion.   A relatively difficult test,  it must be performed by
skilled technicians.   Shutdown of the facility is required during the test.

          2.  Observations

          The equipment and installation of the  equipment on the  tank were
observed upon arrival  at the test site.   The process seems straightforward,
except that for conditions of high water table the fuel  head on the tank is
quite high.   During the test observation the water table was within an inch
or two of the surface requiring the fuel reservoir to be located  approximately
18 ft above the ground.   A scaffolding  had been  erected for this  purpose.
This high head  required that the vent pipe be capped to prevent overflow.

                                   E-7

-------
          After installation of the equipment the product was  stirred  for
about 1 h to stabilize the temperature.  The temperature was recorded  manu-/
ally from a digital thermometer.  Attempts to record the volume changes  ass-j
ciated with the initial 5-lb pressure required addition of considerable
amounts of fuel to the reservoir.  This rapid drop continued for  several
hours, but the rate was decreasing steadily, indicating that the  tank  end  de
flections were stabilizing.  Unfortunately, it was not possible for  the  ob-
servers to remain at the site long enough for complete stabilization to  occu
with the reduction to 4- Ib of pressure for the test.

          Working on a high scaffolding such as this in a high wind, as  was
the case during this observation, could lead to difficulty in  accurately rea
ing the fuel level.  Considerable shaking of the platform occurred.

          3.  Recommendations

          Since the Petro-Tite method has been used widely and is  recognized
in the industry, it could be included in the pilot study program  for compar-
ison purposes.  The test method can be applied to a "tank only" test only  if
the plumbing to the tank can be disconnected.  Otherwise both  the  tank and
plumbing are tested at the same time.  Partially filled tanks  cannot be  test

     G.  ARCO Site Visit

          A site visit was conducted at the Atlantic Richfield Research  Cent
in Harvey, IL, for the purpose of observing the ARCQ tank testing  system in ,
operation on one of their tanks.  The test was observed by Mike Kalinoski  fi£
EPA, and Joel Pavelonis and Ken Wilcox from MRI.  The principal contact  at
ARCO was Tom Collins.

          1.  Description of the Method

          The ARCO method is based on detecting level changes  in  a partially
filled tank.  The sensor consists of a photocell contained in  an  "ink  well"
and a light source.  These are attached to a hinged float in such  a way  as t<
change the depth of the ink over the photocell when changes in level occur.
This changes the light attenuation which is detected with a voltmeter.   The
voltage changes are recorded on a strip chart as the test progresses.  The
system is calibrated by adding known volumes of fuel to the tank.

          One of the novel features of the method is that the  float  level  car
be positioned in the tank at a point where it is insensitive to temperature.
changes.  This is due to the fact that changes in level caused by  expansion
or contraction of the product with temperature are exactly compensated by
changes in  the density (and float buoyancy) with temperature.  This  eliminat
the need for temperature measurements and corrections.

          2.  Observation;;

          The test method appears to be extremely sensitive, making  it pos-
sible to test a partially filled tank in a period of a few hours.  The equip-;
ment can be installed fairly quickly.  ARCO has tested approximately 6,000  '


                                   E-8

-------
tanks using the technique and has verified the test results  in several cases
by removing the tanks for inspection following the test.  They are currently
in the process of franchising independent groups to provide  the testing at
sites other than those controlled by ARCO.

          3.  Recommendations

          The ARCO system should be studied further.  It has considerable
sensitivity, it is simple to operate, and the equipment costs do not seem to
be excessive.   Some precision and accuracy data are available from Arco.

     H.   Underground Radar Site Visit

          On May 10, 1984, Ken Wilcox visited the University of Ohio, Columbus,
to discuss the possible application of underground radar to  locating leaking
storage tanks.   The principal contact there was Or.  Jon Young.

          1.  Description of the Method

          The underground radar system sends radar waves into the ground which
are then reflected back to the surface by various objects which may be buried
there.   The signals receive computer enhancement and are printed out showing
differing densities and depths of buried objects.  The waves can reach down
to around 15 ft, but unfortunately resolution decreases as the depth increases.
An experienced operator can learn to recognize specific types of objects and
their composition from the reflected waves.   The success of the method is
largely a function of the skill  of the operator.   A hand-held unit which does
not use computer enhancement is also available.

          2.  Observations

          Underground radar has not been applied to the detection"of liquid
hydrocarbon leaks.   It has been used to detect water and natural  gas leaks.
In both of these applications, leaks are detected by locating areas where soil
characteristics are different from those in the surrounding areas.   It is not
clear that a pattern characteristic of a leaking storage tank can be easily
identified.  The cost to construct a radar unit and collect sufficient data
to develop the method could cost as much as $100,000.   A hand-held unit might
prove useful,  however, in locating piping system, etc.,  around service stations.

          3.  Recommendations

          This method should be considered for inclusion in the pilot study.
Although the state of the art of this method,  its high cost,  and  long develop-
ment time preclude its use at this time, the method shows some promise and
has some attractive features if it can be made to work.
                                   E-9

-------

-------
 APPENDIX F





SOIL ANALYSIS
    F-l

-------
          Soil samples were collected at Fort Lewis, Tacotna, Washington,  dur
ing the pilot study.  The coring was conducted by Norton Corrosion  of Seatt*
Washington.  The soil analysis was conducted by Warren  Rogers  Associates,  It
Newport, Rhode Island, and the National Institute for Petroleum  and Energy
Research (NIPER), Bartlesville, Oklahoma.  Four holes were  drilled  at two  of
the test sites, 8C25 and 4194.  Two soil samples were taken'from each core,
one at midpoint (6 ft) eind one at the bottom (12 ft).   In addition  to these
samples, MRI took four surface samples at each site.  Two of the surface  sara
pies were soil composites taken near the tanks at 1-ft  and  2-ft  depth.  The
other two were taken at 50-ft and 100-ft distances  from the tanks.   All soil
samples were dry, sandy, and well drained.

          The samples were analyzed for percentage  moisture, conductivity,
pH, sulfides, and number of hydrocarbons present (by Warren Rogers  only).
NIPER analyzed the surface samples taken at 50 ft and 100 ft from the tanks.
The results of the analysis are shown in Table F-l.

          Warren Rogers Associates determined the mean  age  and the  probabili
of a leak from the analysis and other system factors.   The  mean  ages to leak
for site 8C25 ranged from 14.9 to 16.5 years, if the surface samples were  ex
eluded.  For site 4194, they range from 15.7 to 17  years, also excluding  the
surface samples.  In both cases, surface samples yielded slightly higher mea
ages (17.7 and 17.3 years) due to slightly less acidic  and  less  conductive
soils in those samples.

          The samples were dry.  Hence, the mean ages were  computed for sat-
urated soil-s of like chemical composition.  This allows for the  possibility
of fluctuating water tables.  These ages range from 10.5 years to 10.9  for
site SC25 and from 10.5 to 11.3 years for site 4194.

          The tanks were reported to be 50 years old.   Therefore, the condi-
tional probability of a 'leak given point corrosion  would have  reached unity
approximately 29 years ago.  In addition, there is  no evidence of hydrocarbo
in the soil.

          This suggests that a condition of uniform corrosion  exists.   It  is
extremely unlikely that leakage could persist for so long a period  without
detection -and without some evidence of ground contamination.   A  single  tank
tightness test could confirm this.  If the tanks tested'tight, uniform  corro-
sion could be assumed with probability one and no further testing would be
warranted.

          Tables F-2 and F-3 show the Warren Rogers Associates findings at
sites 8C25 and 4194, respectively.
                                      F-2

-------


























U)
>^B
(fl
>
03
C
O

.,
1— t
1
LI-
CU
»^
^3
nj
f^

































C
o
-a

M
(J
o
i.
•O
>

,*-s
S
a.
a.
v>
CU

«*••
it-
i-
OO




x
a







>,
4J ^^
•- s
. > u
4J ~U>
u o
3 .C
•0 E
C i
o <-»



CU
3
4J
in
•f—
O
2
3-?






















in
s_
cu
OJ
O
<^


2:
O£
UJ
a.
i— «
z.

in
i.
CU
OJ
O
OS

3:
a:
U4
*;
z.
in
i.
cu
OJ
^
^

LU
^
•z.


in
i.
OJ
OJ
o
OS

as
LU



in
s_
cu
c?
Ci

3;



0
e

cu
^»
a.

ra
LO



^j
+j
•r—
VI





O O i-H O O





o o o o a







o o o o o




CSJ f^l Cn i""* O
LO r*- f^ OT csj
LO Lfl UO ^f Urt


L^l QQ t_p ^^ ^**
un to LO to tn


f*? LO rH <^ LO
LO LO /^ LO ^*J
LO ^^ r^ n f^i






LO CT> r*>» <— 4 CT>
Lp ^j» ^o *-.O OO

L^ ^^ L^ LO ^f*
r-^ r**» CO LO C\l
O^ LT^ O O rH
_H^ MT-J j^pj




LO C^ f*^ LO CO
CP C\l CSJ ^ LO
cn cn cn i— t cn
>-*








fl3 Q
-3 4j -a 4j -a
— o — o —
s ca s ca s
i— 1 r— t CM CM CO



^^
cn
r-^
^





o





0







o




o
1-1
m


f"*
m


v
P^
CO






/v.
^

to

CO




CO
in
r^.










_j
0
CO
ro












o





o







0




m
in



*
m


to
.o-oao
cM co in
«g-«rcMcoin<3-CMCMCM-r-stor>.r-itoin
r™1^ r^



tO CO 00 CM CM r"^ tO CO CD  4J
a. a.
cu cu

4J *J

rH CM •
^/
4J *J C
— — S
a. a. o
r^j t5 *+—
LO t/J
i i
cu cu •+-
u u
(TJ 03 O
• • M- 4- in
CU 01 1. S.
-— -— 3 3 C
Q. C. ui ui CU

(0 03 CU CU 03
01  J-> — J
•f— 
g
O
i.

4->

O
O
r— 1

C

N>
03
*J

CU

'a.
£
03
00

F-3

-------
         Table F-2.  Determining the Probability of  a  Leak,  Site 8C25
EPA Washington State Locations

Station ID:  Building No. 8C25, Fort Lewis, Tacoma, Washington
Age of oldest tank:  50 years
Mean age:   15 (average)
Years after the mean:  35
Probability now leaking::  1.00
Cathodic protection:  No
Probability of leak in 1 year:   1.00
                       2 years:  1.00
                       3 years:  1.00
                       4 years:  1.00
                       S years:  1.00
Hydrocarbons in soil:  Mo
Ground water encountered:  No
Active transit system nearby:  No
High voltage lines nearby:  No
Cathodically protected structures nearby:  No
                          Year                                      Interior
Tank        Size        installed       Type         Product         corrosion


 1         12,000         1934          Steel        Unleaded          Rough

 2         12,000         1934          Steel        Unleaded,        Rough



Additional Data

Underground aquifer neartjy:  No
Basements:  No
Sewers:  Yes
Utility vault or conduit:  No
Potable water source:  No
Navigable water way:  No
Leak detectors:  No
Impact valves:  No
Vapor recovery systems:  Yes
Lines:  Steel
Flexible connectors:  No
                                      F-4

-------
         Table F-3.  Determining the Probability of  a  Leak,  Site  4194
EPA Washington State Locations

Station ID:  Building No. 4194, Fort Lewis, Tacoma, Washington
Age of oldest tank:  50 years
Mean age:   16 (average)
Years after the mean:  34
Probability now leaking:  1.00
Cathodic protection:  No
Probability of leak in 1 year:   1.00
                       2 years:  1.00
                       3 years:  1.00
                       4 years:  1.00
                       5 years:  1.00
Hydrocarbons in soil:  No
Ground water encountered:  No
Active transit system nearby:  No
High voltage lines nearby:  No
Cathodically protected structures nearby:  No
                          Year                                      Interior
Tank        Size        installed       Type        Product       -corrosion


 1         12,000    '     1934          Steel       Diesel            Rough

 2         12,000         1934          Steel       Regular      '     Rough



Additional Data

Underground aquifer nearby:   No
Basements:  No
Sewers:   Yes
Utility vault or conduit:   No
Potable water source:   Yes
Navigable water way":   No
Leak detectors:   No
Impact valves:  No
Vapor recovery systems:   No
Lines:   Steel
Flexible connectors:   No
                                     F-5

-------

-------
                 APPENDIX G
FACILITY DESCRIPTIONS,  TANK CHARACTERISTICS,
             AND TEST CONDITIONS
                   G-l

-------
          This appendix presents the descriptions and tank  characteristics  c
the tank systems tested at the five facilities selected  for the  development
study.  Specific test conditions are also described  for  each method  tested.
Table G-l lists the tanks or tank systems tested and their  characteristics.


I.  FACILITY DESCRIPTIONS

     A.  Damneck Naval Combat Training Center, Virginia  Beach, VA

          1.  Site Description

          Testing was conducted by all three test crews  at  a tank  located at
the Training Center's Public Works Division where the auto  compound  and  heav;
equipment motor pool and service garage were located.  The  access  road to ttr
Public Works Division carried a high volume of traffic,  including  periodic
heavy equipment transportation, throughout the day.  The site diagram  is
shown in Figure G-l.

          Just to the east of the access road was a marshy  area  whose water
level fluctuated with the tide.  Two product dispensers  were on  a  concrete
pad.  The tank tested was to the north of the dispensers.   The tank  was  lo-
cated under a dirt surface.  A pit measuring approximately  1 ft  x  3  ft x 6  f
had been dug around the test tank's fill pipes.

          2.  Tank Description

          The tank tested was a 5,000-gal (nominal; 5,084 actual)  tank mea- ;
suring approximately 96 in. in diameter and 13 ft 6  in.  in  length.   The  age
of the tank was 28 years..   When the crew arrived, the tank  was holding 4,746
gal of regular gasoline.  The tank was covered with about 2 ft of  soil ma-
teria-1.  The tank was made of steel and the type of del i very, system  was  suc-
tion.

     B.  Pitstop

          1.  Site Description

          Testing was conducted on two tanks at this site by all three test
crews.  The site was a convenience store which had a moderate level  of busi-
ness.  The site diagram is given in Figure G-2.  An oil  sheen observed on th>
surface of the stream that ran adjacent to the property  had led  to the deci-
sion to test at this site.

          2.  Tank Description

          Two tanks were tested, both containing regular gasoline.   The  manv
fold connecting the tanks was disconnected prior to testing.  The  south  tank
was an 12,000-gal steel tank measuring 108 in. in diameter.  Subsequent  to
testing, the fill pipe connection into the tank was  found to be  leaking.
                                   G-2

-------










U
'^
in
•i—
i.
0)
+j
U
fl
^
fl
.c
J_
*f»
0)
J^
^•B

•o
c
•o
in
CU

in
^^
C

1— t
1
o
1
i —




















3 *•»
CU U
r- 3
O T3
i- 0
*J S.
CU Q.
a.

CU
fl CU
S > -^
•f O •
X U C
o -f
S- r— ^/
c^ *f
Q. 0

c u •—
S Q. Ol
O fl ^^^
z u















•24
c
t—

+J
^3
•1—
(J
u_
. co o o
W1 Z r^ i— 4 CO CO
^^ -^x ->^ "••x <*^ -^^
i— t i— » CM r-t CM ft CM
in in
^ a a. 1 1
U O O _J _J
C 01 in 4J -u j_) >j
S +j +j O O S- S_
ro •— -r- cj u a o
O a. cu 1/1 — '
cn
M
(U

+j
i^
0
u.





in
t& ^*
cn i
o a.
S -3






«3- CM
i— 1








i>O ^"
cn CM
ft


O '0
o o
o a
CM m
r-t CM


cn
c
(T3
t—
S
CU
m
^^
1/1

4J
^^\ ^
O 13
r-l 5-
uu -a
0 >
r-t S
•*** ^— '
^
—1 0)
+j cn
s_ c
O fO
U. — 1






^^
1
a.
"3






in
i— i








^*
CM



o
o
0
in
CM


m
c
J
"T"
^— '
CM

.^

nj
cn
o







CM
r- 1






m

cn




0
0
o
ft.







CU
10
i.
—
o
CJ

<4—
r^
o
o

«•


CU
'cn
c
— i
G-3

-------
                                 Road
            \
Dirt
Surface
                 \
                     \
                                     Asphalt1 Surface
                    Vents
                  A
    3" Fill Pipes
       Regular
           Vent
                    \.\
          3" Fill Pipes
                                 2300
                                 Gal.
Diesel
                           \
                               \
                                   \
                                 Fence
                                                            1
                          5000
                          Gal.
Unleaded
                    Manhol«
                      All Tanks Are Suction Delivery

                Figure G-l.  Damneck  site  diagram.
                              G-4

-------
                                                   and Shrubs
                                           Street
.£  o
 
-------
The north tank was an 8,000-gal steel tank measuring 96  in.  in  diameter.
Both tanks contained 4-in.  fill pipes.  A third, 12,000-gal  tank  containing
unleaded gasoline was not tested.  The depth of the soil  cover  above  the
tanks was 3 ft.  The delivery system for these tanks is  a pressure  system.

     C.   Scott Air Force Base, Belleville. IL

          1.  Site Description

          Testing was attempted at this site (Tanks 17 and 18)  by all  three
test crews.  A site diagram is given in Figure G-3.  The  water  table  in this
area varied from 4- to 7 ft below the surface, the average being 4.5 to 5  ft.
The tanks received much use during the day, and a nearby  road carried  a mod-
erate level of traffic.

          2.  Tank Description

          Both tanks, which were approximately 35 years  old, had  been  painte
on the outside and epoxy coated on the inside before relocation to  this site
in 1981.  The volume of each tank was 5,000 gal, and each lay under 44.5  in.
of soil.  Tank 17 contained Mogas, while Tank 18 contained No.  2  diesel fuel
Both tanks have a suction delivery system.

          Although none of the engineering drawings showed connections betve
the two tanks, evidence obtained during testing by Leak  Lokator suggested so
type of connection between the two.  Although this was never verified, if tr
it would cast doubt on trie test results for these two tanks.  Neither  of  the
other" two test crews observed any problems, which is difficult  to explain fo
the Petro-Tite test.

     D.   Fort Lewis Army Base, Tacoma, WA

          1.  Site 8C25 Tanks North and South

               a.  Site Description

               Testing was" conducted at this site by all  three  test crews.
The site diagram is shown in Figure G-4.  These tanks appeared  to be buried
in pea gravel.  The depth of the top of the tanks below  ground  surface was
3 ft.  There were no streams or overhead power lines nearby.  A lightly
traveled road  lay about 50 ft from the nearer of the two  tanks.   Soil  coring
that had been conducted in the area indicated that the water table was below
the bottom of the tanks.

               b.  Tank Description

               Each of the two 12,000-gal steel tanks tested was  96 in. in
diameter and 30 ft in  length.  At  the time of the test the tanks  contained
Mogas and were being used primarily for long term fuel storage.   The  tanks
were 50 years old.  The delivery system was a suction pump housed in  a nearby
shed.
                                   G-6

-------
             5000 Gal. Tanks
                                                               Sandy Surface
Suction
Pump

a
a


Pump Isl

4" Fil
	 V*ot
O ®

and 0. pipe
4" Fill
Vent
O ® ,

                                                                      Tank  17

                                                                      Mo-Gas
                                                                      Tank  18

                                                                       Diesel
                       Figure G-3.  Scott AFB site diagram.
                                       G-7

-------
   o
 4" Fill
Manway
2" Stick


   O
 South
  O
 4" Fill
Manway
2" Stick

   O
 North
                                                                  50
                                                                  O
                                                                  a
                                                                  a.
                            Pumps
                                         a
                                         a
                                              Dispensers
            Figure G-4.  Fort Lewis  site SC2S diagram.
                               G-8

-------
               Openings into each tank consisted of one 4-in. diameter  fill
pipe with a removable aluminum drop tube located on a 24-in. manway,  a  2-in.
diameter stick pipe, a second 2-in. opening which was capped, and  a  2-in.
diameter vent pipe.  The approximate soil cover above the  north  tank was  10
in., and that above the south tank was 5 in.

          2.  Tank 4194

               a.  Site Description

               Testing was conducted by Leak Lokator and attempted by Petro-
Tite at this site.  The tank was located at the backup station to  the main
high volume station across the street.  The site diagram  is shown  in  Figure
G-5.  No streams or overhead power lines were nearby.  A heavily traveled
four-way stop intersection was within 50 ft of the tank.   Soil coring that
had been conducted in the area indicated that the water table was  below the
bottom of the tank.

               b.  Tank Description

               The tank, was a 12,000-gal steel tank 96 in. in diameter  and
30 ft in length.  At the time of the test the tank contained No. 2 d-iesel
fuel.  This tank had a level soil cover with the top of the tank 22  in', below
grade.   The tank was 35 years old.   Its delivery system was a suction.pump  '
housed in a nearby building.  There were two openings into this  tank:   a 4-in.
diameter fill pipe located on a 24-in. manway and a 2-in.   diameter vent pipe.

          3.  Tank 1QE1Q

               a.  Site Description

               Testing was conducted at this site by two of the  test  crews.
The site diagram is shown in Figure G-6.  The tank was buried in pea  gravel.
Built-up earth mounds 24 to 36 in.  above the surrounding ground  surface cov-
ered the tanks.   The depth of the top of the tank below ground surface  was
2 ft.  No streams or overhead power lines were nearby.   A  lightly  traveled
road lay approximately 50 ft west of the tank.  Soil coring that had  been con-
ducted in the area indicated that the water table was below the  bottom  of the
tank.

               b.  Tank Description

               The tank was a 12,000-gal steel tank 96 in. in diameter  and
30 ft in length.  The tank at the time of the test contained Mogas and  was
being used primarily for long term fuel storage.   The tank was 35  to  50 years
old.  Its delivery system was a suction pump housed in a nearby  building.
Openings into this tank consisted of a 2-in. diameter stick hole,  a 4-in.  di-
ameter fill pipe located on a 24-in.  manway, and a 2-in. diameter  vent  pipe.
                                   G-9

-------
"9

OS:
           Manwoy
            Diesel
                                      Dispensers
                               Dispenser
                       Manway
                                                      Dispenser
                                                       "5"
                                                 Pumps
                                         Valve
                                         Pit
4 way
Step
                              Road
   Figure 6-5.  Fort Lewis tank  4194  site diagram.
                       G-10

-------
               Road
Dispensers
           a
              Pumps
                 2" Stick
                 O
                                    Manway
Figure G-6.  Fort Lewis  tank  10E10  site  diagram.
                       G-11

-------
     E.  Lang ley Air Force Base, Hampton, VA
                                                                            v
          Testing was attempted at several locations at  Langley  Air  Force '"•'
Base, Hampton,  Virginia.  Testing was conducted by only  Petro-Tite at the
golf course site, and by only ARCO at the Mogas site.  Both  Leak Lokator am
ARCO attempted to test at the hydrant system site.

          1.  Hydrant System Tanks 3 and 5

               a.  Site Description

               These two tanks were located approximately 1/4 mile from the
main runways of the airfield.  At the hydrant complex, there were eight tank
all with piping to the ramp area.  On adjacent roadways  and  runways,  traffic
appeared light to moderate.   A site diagram is shown in  Figure G-7.   Two of
the test crews attempted tests at this site.

               b.  Tank Description

               The two tanks tested were 25,000-gal steel tanks  124  in.  in
diameter containing JP-4 fuel.  A manway extended down to the top of  the
tank, and a 4-in. diameter riser extended out of the manway  12 in.  The ma-
terial above the tank was gravel.  Some difficulties were encountered in in-
stalling equipment into the tank, which suggested that a ladder  may have bee
present extending from the manway into the tank.

     .         It was impossible to overfill these tanks due to  the turbine::
pump system which presented numerous problems with sealing the system.

          2.  Mogas Tank

               a.  Site Description

               The Mogas tank was located, adjacent to the runway.  Both ve-
hicle and air traffic nearby appeared light to moderate.  The Mogas tank was
buried inside a truck compound.  Although there was no through traffic,  fuel
delivery trucks passed by the tank during the testing.   This traffic  was
routed as far as possible around the test tank since one crew was conducting
testing at the time.  A site diagram is given in Figure  G-8.

               b.  Tank Description

               The tank tested was an approximately 10,000-gal fiberglass
tank, 1 to 2 years old, whose diameter was about 90 in.  The distance from
the top of the fill pipe to the top of the tank was 49-1/4 in.,  and to  the
bottom of the tank was 139-1/2 in.  This tank contained  Mogas fuel, and the
depth of the soil cover was 20 in.  The tank was 1 to 2  years old.
                                   G-12

-------
                                 Lines to Runway Ramp

*0

0 5

O 3

O J







1


1


^
f
Control
Room

Pumps
Pumps
Pumps
Pumps
J


*
8 O

a 0

4 O

2 O
                                                                Manways
Figure G-7.   Langley AFB  hydrant  system  site diagram.
                         G-13

-------
                                 -a
                                  
-------
          3.  Golf Course Tank

               a.  Site Description

               This tank was located at the base's golf  course  in  a  grassy
area isolated from vehicular traffic.  Some 100 ft to the west  of  the  tank
was a small creek.  A site diagram is shown in Figure G-9.  One test crew
tested this tank system.

               b.  Tank Description

               The test tank, which contained Mogas, had approximately 1,000-
gal capacity, was 49-1/2 in. in diameter and 10 ft in length, was  8  years old,
and was constructed of steel.  The tank was buried under a soil  cover  of  about
1 ft.   One dispenser was connected to the tank.  The delivery system was  suc-
tion.
II.  TEST CONDITIONS

          Data specific to each tank tested in the development study are  pro-
vided in this subsection.  A matrix of the tests conducted at each tank is
shown in Table G-2.

     A.   Damneck Naval Combat Training Center

          1.  ARCO

          Tests were attempted over a span of 2 days (October 11-12, 1984).
On the first day of testing, the ARCO equipment did not arrive on time.   The
first day was spent doing temperature profiles and lowering the fuel level  to
3,756 gal (73.9% capacity).   On the second day, the ARCO test crew was unable
to get its equipment to operate properly and could not get a baseline.  Eventu-
ally, testing was called off (after 18 h).  Weather conditions were clear and
breezy.  • Ambient, surface, and tank temperatures were recorded both days  but
could not be used because no testing was conducted.  Wind and vibrations  ap-
peared to cause the most problems for the ARCO test.

          2.  Leak Lokator

          On the first of this 2-day test (September 20-21, 1984), the test
crew arrived at the site at 7:45 a.m.   Testing did not begin until noon,  how-
ever, because approximately 2 h was spent on setting up equipment and the rest
of the time spent was related to fuel  (topping off) delays.  The first day's
test was run from 12:45 to 3:00 p.m. at a sample rate (simulated leak rate)
of 0.09 gal/h.   Samples were taken at 15-min intervals during this period.
During the 12:45-1:00 p.m. "calibration pump priming," thermal shutoff on the
pump shut down.   The pump was restarted in about 15 s.  The test company
(Hunter Environmental) claimed this "messed up" its run.   During testing,  am-
bient temperatures ranged from 81° to 85°F, and surface temperatures ranged
from 69° to 72°F.  The tank temperature (from a thermistor placed at 7-ft
depth) stayed at 73.64° to 73.65°F.


                                   G-15

-------
             Creek
       Suction
       Pump
                        Fill    Vent
                         \    i
                            o   o
N
     Golf Course
Figure G-9.  Langley AFB golf course site diagram.
                         G-16

-------
                    Table G-2.   Matrix of Tests Conducted
Facility
Damneck
Pits top
South
North
Scott
17
18
Fort Lewis
8C25 north

8C25 south
4194
10E10

Lang ley
HS tank 3
HS tank 5
Mogas
Golf course
Tank ARCO Leak Lokator
1 Ta T ,

T T: Noisy (vibration),
possible vapor
T T

1 Out of time T
2 T T: Manifold (?)

IT T: Leak at about
gasket; couldn't
test
2 T T
3 — T: Poor sensitivity
(-0.171)
4 — T: Results
questionable

1 — b T: Multiple leaks,
plumbing problem
2 Tried, but T: Plumbing problems
couldn't test
3 T
4
Petro-Tite
T

T
T

T
T

__

T
Added 400 gal ;
couldn't fill up
(bad leak?)
T


Would have had
similar problems
if testing had
been attempted
Would have had
similar problems
if testing had
been attempted
—
T
bLetter T indicates the tank or tank system was tested.
 Dash (--) indicates testing was not conducted.
                                    G-17

-------
          During the second day of testing, tests were conducted  at  two  dif"
ferent leak rates, each for 1 h 45 min.  Ambient temperatures during this  da;
tests ranged from 76° to 78°F with surface temperatures ranging from 70.64°
to 74.43°F.   A thermistor was again placed into the tank at 7 ft  (from the
top of the fill pipe), and during testing, tank temperatures were 73.32° to
73.33°F.   The weather was clear (no barometric pressure data were collected).
Hunter Environmental had some problems while MRI changed the simulated leak
rates and in achieving a stable baseline due to product constantly being re-
moved from the tank.

          3.  Petro-Tite

          The test was conducted on October 4, 1984, by 0. H. Materials.
Prior to testing, a bore hole was dug by the test crew to determine  the  water
table.  The water table was determined to be 6 ft below grade; therefore,  the)
test crew erected the standpipe up 99 in. to compensate for the water table.
The crew began setup of equipment at 10:00 a.m. with the actual test startin
at 1:45 p.m.  Testing appeared to go smoothly; the initial simulated leak  te
was begun at 3:00 p.m.  Three 1-h simulated leak rates were conducted.   Tem-
perature profiles before and after testing were fairly stable.  Ambient  tem-
peratures during testing ranged from 64° to 76°F, with surface temperatures
from 66.3° to 66.9°F.

     8.  Pitstop

          1.  ARCO

          The Pitatop north tank was tested on October 4, 1984, starting at
10:00 a.m., and the Pitstop south tank was tested on October 5, 1984, start-
ing at 10:00 a.m.  For both tests the fuel level had been adjusted to approx-
imately three-fourths full the day before.  Weather conditions during the
tests were partly cloudy to clear on October 4 and clear on October  5.   Wind
in the morning of the first day caused some noise on the sensor.   No particu-
lar problems were noted during the testing.

          2.  Leak Lokator

        .Testing was conducted on the north tank on October 8, 1984.  Test-
ing started at approximately 11:00 a.m.  Fuel had been delivered  the night
before.  During testing in the morning the weather was still heavily overcast
after considerable rain had fallen during the prior two days.  By the after-
noon of the test the sky had cleared.  The testing was conducted  routinely
and no problems were encountered.

          Testing was conducted on the Pitstop south tank on October 9,  1984,
starting at 9:00 a.m.  Fuel had been delivered to the tank the morning before
so that approximately 24 h of stabilization could take place.  The sky was
overcast during the test, and temperatures were in the low to mid-60s.   Two
problems were encountered during the test which led to uncertainty in the  tes'
results.  First, there appeared to be a vapor pocket in the system which couli
not be readily removed.  Second, traffic through the station created vibratio
which affected the test quality.  It was apparent, however, that  the tank  had
a leak.

                                   G-18

-------
          3.   Petro-Tite

          Petro-Tite conducted testing at Pitstop on September  13  and  14,
1984, for the north and south tanks, respectively.  The north tank was  filled
on September 13 and the south tank was filled on September 14.  The water
table was found to be below the bottom of the tank.  During testing the wea-
ther was mild and sunny.  Petro-Tite encountered no difficulties in testing
either tank.

     C.   Scott Air Force Base

          1.   ARCO

          Of Scott Air Force Base tanks 17 and 18, ARCO tested  only tank 18
(diesel).  On the test days (October 1 and 2, 1984) the weather conditions
were clear and sunny with no wind.  Because of the amount of difficulty in
setting up the test equipment, there was no time to test tank 17.  Some vi-
bration problems were observed when aircraft took off in the area.

          2.   Leak Lokator

          Testing was conducted October 11, 1984, on the Mogas  tank, and
October 12 on the diesel tank.  Weather conditions on both days were heavily
overcast, and some rain fell on the second day.

          During the testing of tank 17 (Mogas), leaking around a  manway gas-
ket was observed.  An attempt was made to stop the leak, but this  was unsuc-
cessful, and the test continued.   During the testing of tank 18 (diesel),
some interactive effects between the two tanks were noted which gave a  strong
indication of manifolding between the two tanks.  The presence  of  a manifold
could not be confirmed, but connections of some type between the two tanks
seem highly probable.  As a result, the test results at Scott are  highly ques-
tionable.

          3.   Petro-Tite

          Testing was conducted on tank 18 on September 18, 1984,  and on tank
17 on September 19.

               a.  Tank 18, Diesel

               The test crew arrived at the site at 7:45 a.m.   The ambient
temperature was 57°F.  At 9:00 a.m. Petro-Tite measured the water  table at
96 in.  The level of fuel in the drop tube was measured to be higher than the
actual fuel level in the tank.  Petro-Tite needed extra time to set up  equip-
ment on the tank,'which required removing the manway cover and disconnecting
many pipe connections from the cover.  A second delay before testing could
begin was the topping off of the diesel tank, with 916 gal. delivered within
a temperature range of 58° to 65°F at 3:15 p.m.   The weather was clear  and
sunny and the temperature at 3:30 p.m.  was 75°F.
                                   G-19

-------
               Only two leak simulation tests were conducted  for  this  day.
The average ambient ana: surface temperatures during these tests were 55°  and :
72°F, respectively.

               b.   Tank 17, Mogas

               The test crew arrived at the site at 7:45 a.m.  Even though
Petro-Tite had modified the manway cover for this tank the previous night,
delays were encountered.  Obtaining fuel to top off the tank  just prior to
testing was necessary.  The skies were clear and at 1:15 p.m. the ambient
temperature was 84°F and the surface temperature was 72°F.  Four  leak  simu-
lation tests, each 85 min in length, were conducted.

     0.  Fort Lewis Army Base

          1.  Tanks 8C25 North and South

               a.   ARCO

               Fuel was delivered to a partially full tank on September 17,
1984.  ARCO tested the north tank on September 13 and 19 and  the  south tank
on September 19 and 21.  Testing began at around 10:00 a.m. with  temperature
profiles being conducted first.  The test crew began setup at approximately
10:00 a.m. and actual testing started at 12:00 p.m.  Testing was  conducted
for approximately 6.5 h.  Baseline testing took 5 h and simulated leaks were
conducted for a total of 8 h for both days.  During the testing the tempera-
tures were mild (70°-80°F) with only light breezes.  Very little  veiiicle
traffic was observed during the test, and there did not appear to be any
vibration problems.  Some problems were encountered in obtaining  smooth
traces during the early phase of the testing.   The cause of the fluctuations
was not clear.  Otherwise, there is no reason to question the validity of the
test results.

               b.   Leak Lokator

               Leak Lokator transferred product from the south tank to the
north tank on October 2, 1984, and the manway gasket began leaking, causing
the test on this tank to be cancelled.  Product was transferred back to the
south tank and again the gasket leaked.  The gasket was removed and replaced
with a new one on the south tank.  The tank was filled on October 3 for test-
ing.  Testing began at around 8:30 a.m. after a temperature profile was con-
ducted.  The test crew began setting up at approximately 8:30 a.m. and the
actual test started at 12:00 p.m.  Testing was conducted for  approximately
5.5 h.  Baseline testing took 1 h while simulated leaks were conducted for
4.5 h.  The temperature was miltf (60°-70°F) with a light breeze.  Very little
traffic was observed during the test, and there did not appear to be any  vi-
bration problem.  The only problems encountered were the leaking  gaskets, and
these were repaired on the tank which was tested.  There is no reason  to  ques-
tion the validity of the test results.
                                   G-20

-------
               c.  Petro-Ti'te

               Petro-Tite did not attempt to test the 8C25  north  tank because
of the gasket problem discovered by Leak Lokator.  Fuel was transferred  into
the south tank on October 9, 1984, in order to conduct the  test.   Testing  be-
gan around 8:30 a.m.  with a temperature profile being conducted first.   The
test crew began setup at approximately 8:15 a.m. and the actual test  started
at 1:00 p.m.  Testing was conducted for approximately 4.5 h.  Baseline test-
ing took 4 h, while simulated leak testing was conducted for 0.5  h.   During
the testing the temperature was approximately 60°F.  Light  rain began to fall
at 1:00 p.m.  Wind was 5 to 10 mph during the day.  Very little traffic  was
observed during the test.  A leaky cap on the manway 4-in.  pipe postponed  the
start time by 2.5 h.   The baseline test was conducted with  no problems.
Shortly after the simulated leak test started, the base's electrical  power
went out and the simulation test had to be terminated.

          2.  Tank 4194                                             	i.__

               a.  Laak Lokator	

               Fuel was delivered on October 3, 1984, to fill the  tank,  and
the tank was topped off on October 4.   Leak Lokator tested  the tank on
October 4.  Testing began at around 12:15 p.m. with a temperature  profile
being conducted first.  The test crew began setup at approximately 11:00 a.m.
and actual testing started at 1:30 p.m.   Testing was conducted for approxi-
mately 6 h.   Baseline testing took 1 h and simulated tests were conducted  for
5 h.   During the testing the temperature was about 60°F.  The sky  was  cloudy
and there was a light breeze.   It rained all morning.  There was heavy traf-
fic all day.  Leak Lokator had some problem achieving good  sensitivity with
its equipment.   The cause of this is unclear.   Otherwise, there is  no  reason
to question the validity of the test results.

               b.  Petro-Tite

               The tank needed to be topped off on October  io, 1984,  due to
delivery October 3.  Petro-Tite arrived at 7:30 a.m.  and conducted a  tempera-
ture profile first.  The test crew began setup at approximtely 8:30 a.m.,  but
the actual test never started.   On the test day the weather was cold  and rainy
(55° to 60°F) and there was a light breeze.   It rained off  and on  all  day.
Over 400 gal of product were added to the tank with no success in  filling  the
tank or in putting a head pressure on it.   Piping was dug up and no visible
leaks were observed.   It is suspected that the tank was slanted and an air
pocket could not escape within the 3-in.  piping that led from the  tank to  the
pump.
                                   G-21

-------
          3.   Tank 1QE1Q

               a.   Leak Lokator

               Fuel was delivered October 3, 1984, to fill the tank.  The
tank was tested on October 5.   Testing began at around 8:30 a.m. with a  tern*
perature profile being conducted first.  The test crew began setup at 8:30
a.m. and the actual test started at 10:15 a.m.  Testing was conducted for sp
proximately 6 h.  Baseline testing took 1 h and simulated leaks were conduct
for 5 h.  During the testing,  the temperature was mild (about 60°F) and  ther
was a light breeze.  Some traffic passed by but not enough to cause vibrstio
problems.  No observable problems were encountered during the testing.   The
data became questionable once the results were received from Leak Lokator.
After analyzing the data, no explanation of what happened could be found.
The data without temperature compensation compared better with the stmulatid
rates than the temperature-compensated data.

               b.   .Petro-Tite                                         -——

               Fuel was delivered on October 3, 1984.  The fuel level was no
lowered after Leak Lokator had tested the tank.  Petro-Tite tested the tank
on October 11.  Testing began at around 8:30 a.m. with temperature profiles
being conducted first.  The test crew began setup at approximately 9:00  a.m.
and the actual test started at 10:15 a.m.  Testing was conducted for approxi
mately 3.5 h.  Baseline testing took 4.5 h and simulated leaks were conducte
for 4 h.  The temperature was mild (about 60°F) and there was a light breeze
during the testing.  Some light traffic was observed but it caused no inter"
ference with the testing.  The test crew experienced a problem with the  sen-
sor box during the test,  They installed a new battery but there was still a
problem every third or  fourth 15-min reading.  Because of this, the reported
results differ more from the simulated rates than at the previous test sites

     E.  Langley Air Force Base

          1.   Hydrant System Tanks 3 and 5. (Leak Lokator)     .       	

          Testing at Tanks 3 and 5 at the hydrant complex spanned 2 days.  0
testing tank 3 on September 22, 1984, leaks were encountered on above-ground
plumbing-and pump seals.  The fuel level was lowered to avoid this, and  test
ing proceeded with few problems.  During testing of tank 5 on September  23,
the test crew had problems throughout the day primarily with apparent air
pockets.  Several attempts were made to bleed the air out of the piping.  On
tank 5, the test crew informed MRI that the tank was losing 3 gal/h at the
start of the test.

          During both days of testing, the weather was clear and sunny,  with
a very slight breeze.  Ambient temperatures remained in the 70s to 80s with
no drastic changes, and ground temperature fluctuated by about 2° to 3°F.
Tank temperature (with  thermistor placed at 12 ft) remained stable both  days,
The tank system was a complex one to test because of the excessive and un-
known piping.  The on-site military personnel were to have "isolated" the
tank from the rest of the system, but on tank 5 that was questionable.


                                   G-22

-------
          2.  Moqas Tank (ARCO)

          Original efforts were to test tanks 3 and 5, but complications  pre-
vented this and the Mogas site was selected instead.  ARCO experienced  problems
achieving a good baseline but testing did begin in the late afternoon.  A total
of four simulated leak rates were completed.  The fill pipes were  recessed
below ground level and water had to be pumped out about every 10 to  15  min
during tasting.  Weather conditions were clear to partly cloudy with some
breeze.   Ambient temperatures during testing ranged from 73° to 84°F while
the tank temperature remained stable (one probe at 9 ft).  The wind  appeared
to cause a lot of problems with the testing.

          3.  Golf Course (Petro-Tite)

          Petro-Tite attempted to test the golf course tank on October  5,  1384.
One day of testing was spent o'n the tank, and eventually the test  had to  be
called off because of a large leak in the tank.  When the first pressure  head
was established, a large pool of fuel appeared on the ground surface.   When
the soil was removed to expose the top of the tank, the suction line to the
dispenser was found to be partially detached (at an elbow joint).  The  line
was disconnected and the tank was capped to isolate it.  The final pressure
head was established and the leak exceeded 2-1/2 gal/h, with no visible leaks,
so the test was called off.   The water table was determined to be  about 5  ft
below the ground surface.   Ambient temperatures and weather conditions  were
not a factor because testing had to be aborted.
                                   G-23

-------

-------
      APPENDIX  H





DEVELOPMENT STUDY TEST DATA
           H-l

-------
LEAK LOKATOR SIMULATION DATA
             H-2

-------
TEST
RAZE
SIM RATE   BASE RATE
LLXSEWMX.SAS;! 0.073
DAMHEOC 0.082
0.066
M-.786, Y'»0.005 0.077
0.068
0.077
0.071
0.075
0.160
0.172
0.180
0.171
0.164
0.165
0.156
0.132
0.171
0.317
0.229
0.272
0..271
0.220
0.198
0.213
0.226
-0.001
-0.008
0.001
0.005
-0.023


0.090
0.090
0.090
0.090
0.090
0.090
0.090
0.090
0.187
0.187
0.187
0.187
0.187
0.187
0.187
0.187
0.187
0.317
0.317
0.317
0.317
0.317
0.317
0.317
0.317
0.000
0.000
0.000
0.000
0.000
Mean *
Std Dev »
-0.017
-0.008
-0.024
-0.013
-0.022
-0.013
-0.019
-0.015
-0.027
-0.015
-0.007
-0.016
-0.023
-0.022
-0.031
0.005
-0.016
0.000
-0.088
-0.045
-0-. 046
-0.097
-0.119
-0.104
-0 . 091.
-0.001
-0.008
0.001
0.005
-0 . 023
-0.030
0.034
                      H-3

-------
.-3
    -0.05
          r
                                LEAK  LOKATOR
                        i    I     I    1    I
                           0.08      0.13     O.iS
                                           Lettk Hate
                                        h-4

-------
TEST
RAZE
SIM RATE    BASE RATE
LOKLEAX.SAS;4
PITSTOP NORTH
M-.882, Y'*-0.012



















0.289
0.283
0.285
0.297
0.294
0.258
0.289
0.068
0.072
0.080
0.077
0.209
0.188
0.198
0.209
0.201
0.197
-0.028
-0.027
-0.022
-0.018


0.351
0.351
0.351
0.351
0.351
0.351
0.351
0.098
0.098
0.098
0.098
0.224
0.224
0.224
0.224
0.224
0.224
0.000
0.000
0.000
0.000
Mean *
Std Dev «
-0.062
-0.068
-0.066
-0.054
-0.057
-0.093
-0.062
-0.030
-0 . 026
-0.013
-0.021
-0.015
-0.036
-0.026
-0.015
-0 . 023
-0.027
-0.028
-0 . 027
-0 . 022
-0.018
-0.038
0.022
                LEAK  LOKATOR
                       Pitatcp
                         H-5

-------
                TEST
                            RATE
               SXH RATS    BASE HATS
LLSCT.SAS;!
SCOTT #17

M«.S39, Y* —0.067



















0.127
0.043
0.078
0.092
0.016
0.052
-0.104
-0.102
-0.113
-0.111
-0.112
0.239
0.190
0.216
0.2S4
0.066
0.087
0.063
0.079
0.087
0.071


0.090
0.090
0.090
0.090
0.090
0.090
0.000
0.000
0.000
0.000
0.000
0.352
0.352
0.352
. 0.352
0.198
0.138
0«198
0.198
. 0.198
0.198
Mean- *
Std D«v * "
0.037
-0 . 047
-0.012
0.002
-0.074
-0.038
-0.104
-0.102
-0.113
-0.111
-0.112
-0 . 113
-0.162
-0.136
-0.098
-0 . 132
-0 . Ill
-0.135
-0.119
-0.111
-0,127
-0.091
0.050
       0.3
'a





3


It
3
•a
4.
>•

•3
 0.2S -



  0.3 -




 o.ia -




  0.1 -



 0.05 -




    0
-0.05 -



 -0.1 -g




-o.ie —
                                  LEAK LOXATOR

                                        Scott APB  :To.  17
0.1
                                           0.3


                                         Leaic Rats  (*
                                                                  0.3

-------
     TEST
              HAZE
                                 SIM RATE   BASE BATE
LOKLZAK.SAS;3
FT. LEWIS 8C25
BACK
M-.734, T-0.01








0.106
0.077
0.103
0.105
0.252
0.301
0.343
0.287
0.031
0.042
0.065
0.078


0.130
0.130
0.130
0.130
0.270
0.270
0.270
0.270
0.270
0.270
0.070
0.070
Mean *
Std Dev =*
-0.024
-0.053
-0 . 027
-0 . 025
-0.018
0.031
0.073
0.017
-0 . 239
-0.228
-0.005
0.008
-0.041
0.092
                        LEAK  LOKATCR
                          Ft. Lewis  JTo.  8C25 Bacic
0.07
0.09
0.11
0.13   0.15    0.17    0.13    0.21    0.23

  Simulated. Leaic Hate <•
0.25
                                 H-7

-------
                    R&ZS
                       SIM RASE   BASE  BASE
LOXLEAJC.SAS;2
FT. LEWIS 4194

m*0.749, 7' -0.013

•

















0.025
-0.006
0.021
-0.003
0.035
0.015
0.023
0.190
0.213
0.176
0.174
0.125
0.140
0.128
0.280
0.271
0.290
0.229
0.289
0,223
0.248


0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.230
0.230
0.230
0.230
0.230
0.230
0.230
0.370
0.370
0.370
Oc370
0.370
0.370
0.370
Mean »
Std Dev *
-0.015
-0.046
-0.019
-0.043
-0.005
-0.025
-0.017
-0 . 040
-0.017
-0.054
-0.056
-0.105
-0,090
-0.102
-0.090
-0.099
-0 . 080
-0.141
-0.081
-0.142
-0.122
-0 . 066
0 . 042

0.04>
                        LEAK LOKATOR
                             Ft.  Lcrria ITc.
0.08
0.13     0.13      0.2      0.3*

       Simulated  Letik Bate
0.53

-------
                           RATE
SIM RATE   BASE RATE
LUCRLS2.SAS;!
FT. LEWIS 10E10
B-0.335, y'— 0.26
















-0.094
-0.235
-0.124
-0.089
-0 . 129
-0.104
-0.062
-0.11S
-0.029
-0.046
-0.382
-0.336
-0.380
-0.430
-0.070
-0.023
-0.086
-0.092
-0.127


0.080
0.080
0.080
0.080
0.080
0.080
0.080
0.140
0.140
0.140
0.000
0.000
0.000
0.000
0.280
0.280
0.280
0.280
0.280
Mean *
Std Dev »
-0.174
-0.315
-0.204
-0.169
-0.209
-0.184
-0.142
-0 . 255
-0.169
-0.186
-0.382
-0.336
-0.380
-0.430
-0.350
-0.303
-0.366
-0.372
-0.407
-0.281
0.094
-0.36
-0.46
       Q
       i
                              LEAK  LOKATOR
                                  Ft. Le^ris  No.  10SJ.O
                0.04-
0.08       0.13       0.1S

   Sira-ulated. Lenic Bate
                    0.3
0.3*
0.
                                      H-9

-------
              TEST
       RAZE
SIM BATE   BASE
UHG71. 3*5*1
LAJJGLEJ JP-3

m— 1.779, y'»0.19









0.034
0.043
0.047
0.050
0.238
0.185
0.197
0.152
0.192
0.187
0.269


0.050
0.050
0.100
0.100
0.000
0.000
0.000
0.000
0.000
0.000
0.000
Mean »»
Std Oev *
-0.016
-0.007
-0.053
-O.OSO
0.238
0.185
0.197
0.152
0.192
0.187
0.269
0.118
0.117
     0-Z8-
                                 LEAK  LOKATOR
                                      Laxxsisy APS  JF—3
,?
a
          Q
     0.3S <-}


     0.3* 4


     0.23 -)
0*03
                                       0.0*           0.03


                                            Le?dc Bate Cs
                          0.03
                                         H-10

-------
           TEST
                      RATE
               SIM HATE    BASE HATE
LLNG72.SAS;!
LANGLEY 'JF-5
a«2.429, y'— 0.12









0.443
0.442
-0.557
-0.109
0.04S
0.94S
0.345
0.633
0.504
-0.029
-0 . 064
-0 . 231


0.080
0.080
0.080
0.080
0.080
0.300
0.300
0.300
0.300
0.000
0.000
0,000
Mean =
Std Dev »
0.363
0.362
-0.637
-0.189
-0.032
0.648
0.045
0.333
0.204
-0.029
-0 . 064
-0.231
0.064
0.328
                             LEAK LOKATOR
                                           APB  J"P—6
 0.9 -
 0.3 -
 0.7 -
 o.e -
 0.5 -
 0.4 -
 0,3 -
 0.3 -
 0.1 -
   0
-0.1
a
              0.04-
0.08
                             0.13       O.ifi       0.2
                                  LctLk Hate  C*
0.2*
0.3S
                                       H-n

-------
PETRO-TITE SIMULATION DATA
           H-12

-------
                        RAZE
SIM RATE   BASE RATE
FEEROTITE

PtZTSCT . 3AS ; 1 4 0 . 236
DAMNECK 0.160
0.236
JB»I.OI, y'-o.oo9 o.ieo
0.200
0.240
0.276
0.200
0.040
0.156
0.080
0.120
-0 . 044
-0.044
0.116
-0.040



01 OJ\
. 190
0.190
0.190
0.190
0.220
0.220
0.220
0.220
0.070
0.070
0.070
0.070
0.000
0.000
0.000
0.000
Mean »
Std Dev =»


U . w*rO
-0.030
0.046
-0.030
-0.020
0.020
0.056
-0.020
-0.030
0.086
0.010
0.050
-0 . 044
-0 . 044
0.116
-0.040
0.011
0.049
0.38
                            PETBO-TITS
0.04
0.08
      0.13
                                                     O.ifl
0.3
                                   H-13

-------
               S5ST
         PCTRO:PS.SAS;14
         P1TSTUP
         m-1.26,
 i-0.069
•0.100
•0.020
•0.080
 0.000
•0.100
•0.060
•0.104
 0.000
 0.040
•0.020
 0.284
 0.100
 0.100
 0.224
 0.405
 0.324
 0.180
 0.200
       0.000
       0.000
       0.000
       0.000
       0.000
       0.000
       0.056
       0.056
       0.056
       0.056
       0.187
       0.187
       0.187
       0.187
       0.278
       0.278
       0.278
       0.278
                            •0.100
                            •0.020
                            •0.080
                             0.000
                            •0.100
                            •0.060
                            •0.160
                            -0.056
                            •0.016
                            •0.076
                             0.097
                            •0.087
                            •0.087
                             0.037
                             0.127
                             0.046
                            •0.098
                            •0.078
                                           Mean *
                                        Std Dev *
                                      •0.040
                                       0.073
     .0.3
                                 PETBO-TITE
•A
          0
O.CH
0.08
0.12       O.iS

   Letik Bate
                         0-3
0.3*
                                        H-14

-------
TEST
 RATE
SIM RATS   BASE RATE

pfitSCT. SAS ; 15
SCOTT *1S

m-0.608, j' -0.774











0.300
0.864
0.788
0.796
1.140
0.940
0.960
0.920
0.760
0.828
0.752
0.692



0.000
0.000
0.000
0.000
0.238
0.238
0.238
0.238
0.155
0.155
0.155
0.155
Mean *
Std Dev a

0.800
0.364
0.788
0.796
0.902
0.702
0.722
0.682
0.605
0.673
0.597
0.537
0.722
0.107
                     Scott .4PB Xo.  IS
    0.04-
 0.08         0.12         O.ifi
Simulated Lcaic Hate (^

           H-15

-------
                TESX
                                RATE
 SIM RAZE   BASE RAZE
PETSCT . SAS ; 1 6
SCOTT *17

m«1.075, y'»-0.00


















-0.004
-0.044
-0.004
0.036
0.076
0.076
0.076
0.060
0.1SO
0.160
0.220
0.076
-0.080
0.020
0.040
0.360
0.320
0.340
0.380
0.360


0.000
0.000
0.000
0.000
0.078
0.078
0,078
0.078
0.132
0.192
0.192
0.000
0.000
0.000
0.000
0.321
0.321
0.321
0.321
0.321
Mean *
Std De* »
-0.004
-0.044
-0.004
0.036
-0.002
-0.002
-0.002
-0 . 013
-0.012
-0.032
0.028
0.076
-0 . 080
0.020
0.040
0.039
-0.001
0.019
0.059
0.039
0.008
0.036
                                 PETBO-TITE
                                    Scott APB No. 17
13

a
a
X
_u
a
a
a
      Q.3 -

     Q.3S -

      0.3 -

     0.10 -
     -0.1
0.04
0.08
0.13
0.10
                                                  0.3
                                                         0.34
0.3S

-------
            TEST
BATE
                SIM HATE    BASE RATE
PETROLS. SAS; 12
FT. LEWIS 10E10

m-1.5, y'»0.038














0.208
0.020
-0.168
0.196
0.048
0.120
0.524
0.000
0.044
O.SOO
0.440
0.760
0.1SO
0.084
0.232
0.340


0.097
0.000
0.000
0.000
0.000
0.098
0.099
0.101
0.270
0.270
0.270
0.270
0.118
0.120
0.120
0.119
. Mean *
Std Dev »
0.111
0.020
-0.168
0.196
0.048
0.022
0.42S
-0.101
-0.226
0.230
0.170
0.490
0.062
-0.036
0.112
0.221
0.098
0.187
0.3
                             PETRO-TITE
                               Ft. Leiria  ITo.  10E10
             0.0*
0.08       0.13       O.iS

   Sim-vilateti Leetk Bate

             H-17
                           0.3
0.2*

-------
ARCO SIMULATION DATA
        H-18

-------
\
"i
t
K
0
0
n
0
C
                                   ARCO
                                    DamnecJc
               OJ34     0.08     0.13    0.18     0.2 .    0.24

                            Simulated Leak  Rate (gal/h)
                       0.28
0.32
                           RATE
SIM RATE    BASE RATE
ARCODN.SAS;!
DAMNECX
ia=1.12, y'=0.01-




0.110
0.370
0.030
-0.020
0.170


0 . 075
0.319
0.000
0.000
0.166
Mean =
Std Dev =
0.035
0.051
0.030
-0.020
0.005
0.020
0.025
                                     H-19

-------
         TEST
                      HATE
SIM  RATE    BASE RATE
ARCO. S AS; 10
PITSTOP NORTH
n»0.78, y'»0.06




0.2SO
0.150
0.100
0.000
0.390


0.252
0.069
0.000
0.000
0.326
Mean *
Std Dev »
-0.002
0.081
0.100
0.000
0.064
0.049
0.042
                                     ARCO

                                   PlUtop Hortlx
\

'a
a

e
a
o



•8
e
0.35 -





 0.3 -





0.35





 0.2 H





0.13





 0.1





0.05 -I

                                                  i	1	1	1	1
         0
           0.04    CUB     042    0.18     0^     0.24    OJ38


                        Simula tad  Leak  Rau (gal/h)

-------
           TEST
                          RATE
            SIM RATE    BASE RATS
ARCOSCT.SAS;!!
SCOTT #18
n— 0.408, y'=. 038







0.140
0.120
0.230
0.270
0.090
0.100
-0.020
0.000


0.218
0.218
0.183
0.183
0.305
0.305
0.000
0.117
Mean *
Std Dev =
-0.078
-0.098
0.047
0.087
-0.215
-0.205
-0.020
-0.117
-0.075
0.102
o
•n
a
S
 038 -
 0.24 -
 0.22 -
  0.2 -
 0.18 -
 0.18 -
 0.14 -
 0.12 -
  0.1 -
 0.03 -
 OJD8 -
 0.04 -
 0.02 -
   0 -
-0.02 -f
                                        AECO
                                    Scott APB  No.  18
 T   T
0.04
I    I
   0.12
0.08      0.12     0,18     0.2     0.24
     Simulated  Leak Rate (galxti)
                                                                0.28
                                                                             0.>3
                                        H-21

-------
TEST
BATE
SIM RAT£   BASE HATE
ARCO.SAS;? 0.150 0.150 0.000
FT. LB*IS 8C25 0.280 0.300 -0.020
NORTH 0.240 0.300 -0.060
m»0.587, y' -0.072 0.160 0.060 0.100
0.040 0.000 0.040
0.140 0.060 0.080
0.100 0,160 -0.060

Mean « 0.011
Std D«v » 0.059
G
ARCO
0.28 -
OJ28 -
'£ . °~* "
\
i 0.22 -
«
0.2 -
e
| 0.18 -
A 0.16 -
a
•J 0.14 -
» 0.12 -
h
• 0.1 -
a
S 0«08 -
OJD6 -
OJH -5
(
Ft. Lawis Ho. 8C25 North

s* *•
,*
^
.^
a ^x^^^
^-''x°
a ^s^
^^^
.s^ o
^^

t I I ) I i \ I i 1 t i i t
) 0.04 0.08 0.12 0.13 0.2 0.24 0.28
                Simulated Laak Rate
                        H-22

-------
          TEST
                      RATE
SIM  RATE    BASE RATE
ARCOLNGY.SAS?!
LANGLEY MOGAS
m=0.74, y'=0.22




0.470
0.450
0.750
0.030
0.000


0.300
0.420
0.060
0.000
0.000
Mean =
Std Dev »
0.170
0.030
0.690
0.030
0.000
0.184
0.260
\

a


v^*



0
•+*

6
a
i>



•tf
ii
t.

?
n
a
o

s
3.7 -





0.0 -





OJ5 -





0.4 -





0.3





0.2





0.1 -
                                     ARCO

                                 Lanaiey AfB  Moeas
                       0.1            0.2            0.3


                             Simulated Leak Rate Cgal/bJ
                                                           0.4
                                     H-23

-------

-------
  APPENDIX I





PILOT STUDY DATA
        1-1

-------
Survey L3 So.  L29000Q528

     Tank Ho.  1  of 2
     Tank Size    1Q36 Gal.
                              SITS DATA SEMMAfiY
Test Data  Mar 13, 1985
Product   ffnleaded
                                T2ST RZSTJ1TS

Certified
Rate
S.E.
Sgsten
yaq
-O.G1
0.0098
T.iae 1
M/A


                                            Line 2
           Line 3
Line 4
Notes:
        The standard error reported haa been adjusted to a. two-hour test
   period for consistency with future testa on the national survey.   A
   negative sign Indicates product loss (leak out).                     "
Fora  8A29-Q01SDS
                                     1-2

-------
I -Aor—8S
Pace 1
 LH  ISite Code L230000S23 Fuel Type UNLSA06D   Date       MAR.13,35
aig? !Test firm O.H.rt.     Tank Vol        1335 T digits       !1S22
N  Y  Teat Crew SS.DP.JS   API Dens        30.1 T dagAts/F      317
   »  MR I Crsw  9C.MS      £*a Goer1  3.300614.31 Leak Rate      3.307
Tune Level
Hr ilin' (div)
1 I
1 1
1 1
12
12
12
12
13
13
13
13
?4
(4
14.
14.
15
IS
IS
33
4S
3
12
30
4S
0
IS
30
4-S
0
ts"
30
45
3
IS
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
y SeCore

-------
I-
w
                                                    Liisj<_gA/icn
— -•-

rt H


n /


u.*
U.3


0.-i»


C.J


3.2,


W* 1

Wi-

1 t <
••• • '

1
i
1
-4
i
i
-i
i
^^ ^i

i a *"
i ^
i 2
"! « ^
i —
i ** "
•^ 31
i *
i — •»•
-f u
i *
i i-
^ —
i 5,>?5'><»rt'>^..
i
i

i i I i i i
n i 2
i
L

ff * '
-» •- '
i i
IT j
, * j
i
i
i
i
i
j
j
i
j
j
j
i
i
i
i
A i
•9
i

: I i
J i
                                                       !T*    h
                                              1-4

-------
                              SITS  DATA SUMMARY


Survey HD Ho.  L250000523	          Test Date

               2 of 2                         ,
     Tank tfo.
      :ani Siza   1UJO Gai«
                                TEST RESULTS


                 	        Line  1       £ine 2       Line 3       Line 4

Certified

Rate   __         	

S.E.               Q«013
tfo tes:
        The standard error reported has been adjusted to a two-hour tast
   period for consistency with future tests on the national survey.  A
   negative sign indicates product loss (leak out).
Fora 3429-001SDS
                                    1-5

-------
u
R 5
sis? !
N
»
Y

Tine
Mr
13
13
14
14
14
14
IS
IS
IS
IS
IS
IS
IS
IS
17
17
17
Kin
30
4S
9
IS
30
4S
3
IS
30
4S
3
IS
30
4S
0
IS
30
Site Cods
Teat Firm
Taat Craw
MR I Oat*
Lav«i

M/A
0,330
0.S
a. 33
9.70S
0,74
9.73S
3.3E
''
-------
                                                           V
                                        TANK 2 REGULAR
J. A-Carrecis
^|t>6) P'JUii|o/,p e/\f|C|iiu»i3

u./
0.3
0.5
0.4.
U.J
2.2
Q.1
ij

.
J. c
t 3
a
a
4 3
i
I ' a
1
I y.
J
1 .c
"!«*'•
i *
! *
i
i i i i i i i i
2 1 2 3
Scpsad Time (h)
•*• £sp«nsian •? Carractad — — — Ras
                                          1-7

-------
                             SITE DAIA ST3MMART
Surrey ID No .  S2900000Q6 __          Test Data Mar.  15.  198$


     Tank ftp.  1  of 1 __ ______          Product  Unleaded _ _


     Tank Size   11
-------
    Page
LR i
31
N

a? i
Y
»
rifle
Hr
IS
IS
IS
IS
IS
IS
IS
IS
17
17
17
17
13
13
18
13
flin
0
IS
30
4S
0
IS
30
45
0
IS
30
45
0
IS
30
4S
Site Coda
Test rtrrc
Taat Crew
MR I Cre»
Lavel
(div)
12-
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
N23000000S
O.H.ft.
3S.DP.JS
M6
V 3«/orvs
( gal )
N/A-
3.380
a. 375
3.3F
3.34.
3.32S
3.4-7
3.4=7
3.465
3.48
3.44S
3-.. 435
*. 42
3 ..40S
3T.33
3.375
Fuel Type
Tank Voi
API Dens
£*0 Ccef
U Aftar
(gai)
N/A
3.375
3.0S
3.34,
3.325
3.31S
3.47
3.465
3.46
3.4>45
3.. 435
0.42
3.40S
3.33
3.375
3. 35
UNLEADED
I 1207
S3. 7
3.30061357
Fuai Temp.
( digits >
12237
12237
12235
12235
I22SS
12233:
12233
12232
12231
12230
12238
12238"
1228S
12234
12234.
12232
Data
T digits
T diqita/F
Laak Rate
Tcorr dU
(gal)
N/A
-0'. 305
3.308
3.303
3.33-1
-"3.3-10
3.300
3.313
3.313
3.308
3.3-35
-0 . 3 1 S
3.331
3.331
-0.315
3.331
flftfl.15,85
12237
323
3.333
Laak Rat a
( gal/h )
N/A
-3.323'
3.332
3.312
3. 124
-3.340
3.300
3.372
3.372
3.332
3. 144
-0.060
3.124
3. 124
-0.360
3. 124
1-9

-------
      a.2 —
                                  fc.  N 2 ^ U U U U U U
                                          1  UNL£*C£2
^
 o-
      0.7  -t
   j.
u -if
                               -a   52    -   .,
    -G..3
                                        1-10

-------
                              SITE DATA SUMMARY
Survey ID So.  C-290Q00123

     Tank No.  1 o£ 1
     Tank Size
                             Test Data   Mar 16, 1985
                             Product:   Diesel
                                TEST RZSUITS
Certified

Rate

S.2.
System

 Mb

 -0.056

  0.0067
                               Line 1

                                No
-0.04
 0.005
             Line 2
Line 3
Line 4
Kotes:
        The data report from OHM certifies this tank as tight.  Their astiaated
   leak rate is -0.029, within the NFPA standard-.  The system appears to 'have
   a slight leak, rate astiaated as -0.056 gph.  However, viewing the tank
   data only for the time that the line was iaoi*te
-------
 LR   ISit«  Code  522(3000123  Fuai  Tyoe OIE3SL     Oats
31Q?  !T«at  Firm  O.H.rt.      Tank  Vol        I03S T digits       10383
N  Y  Taat  Osw  3S,flP,JS    API Qena        32.3 T digats/F      308
   *  flRI Osw   fl.S. ,K.U.   £*o Co*?  3.30044635 Lsak Rate    -3.357
Tl
Hr
1 1
1 1
1 I
1 1
1 I
[ I
1 1
1 1
1 1
1 1
1 1
1 1
12
12
12
12
12
12
12
12
12
^
2
2
3
3
3
3
3
3
3
3
13
13
13
13
1 4
14,
14
14
! 4
! 4
! 4
14
1 4
1 4
! 4
1 4
ne Lav«i
ilin 
-------
IS
1 5
IS
IS
13
IS
IS
IS
IS
IS
IS
IS
15
IS
IS
IS
IS
IS
IS
IS
1-S
IS
IS
! S
1 7
3
S
10
IS
29
25
30
35
4.0
4S
SO
ss
a
s
13
IS
23
25
30
35
40
45
S3
S3
3
12
12
12
12
1Z
12
12
12
12
12
II
12
12
12
12
12
12
!2
12
12
!2
12
12
12
12
9.4SS
0.4S
9.4-7
3.43
9.45
3.S
a. si
3.52
9.525
3. S3
3. 54
3. 55
9. So
3 .555
3.S7
9.S75
3.sa
a.sas
a.sas
9.595
3.S
3.S0S
3.S2S
3.335
3. 34
3.45
9.4T
3.4S
3.4-9-
9.S
3-.S1
3. 52
3.S2S
3.23
9. 54.
3.S5
3.SS
3.355
3.S7
3.S75
3.53.
3.S9S
3.535
3.525
3.S
3. 335
3. SI
3.535
3. 54
3.S4S
137S31
13T79
13777
13733
13-731
13739
13835
1381!
13813
13824
13832
13835
13344
I 3853
I38Si
13850
138S7
13874
13880
I3S8S
13893
13899
1 390S
13912
13918
-3 . 304
-3.301
-0.301
3.301
-0.302
-0 . 302
3.301
-<3'. 304
-0-. 30S
3.301
~9 . 302
3 . 304
--3,307
-#.304
-0 . 304
-0 . 30 1
-3.306
-3.31 1
3.301
-
-------
C I i  u   :'"„'' :'-,l :  •;•''';'"!'
          Tim:  1   '

           1-14

-------
                             SITE DATA  SUMMARY
Survey ID ffo.  8290000010

     Tank No.   1  of 3
     Tank Size  6006
                            Test  Data Mag  18.  1985

                            Product    Regular
                               TZST RESULTS
Certified

Rate

S.E.
JTotes:
System

 Tea

 0.036
                              Line  I

                                JKT
3C7
Line 2

 •OKo

 -0.031
                         Line 3
SCV
            Lzaa 4

              Tea
                                        -0.01
         Tw of the 'line? could not be certified because they had bad check
    valves  (SCV).  One line had a noticible lead (-0.031 gph), and the other
    had  an  estimated leak below the line threshold of -0.025 gph.
Form 3A29-001SDS

-------

-------
                                 Page
 Lfl  iSita Coda N230000010 Fuel fyoe RESUUW    Data      ttAR.13  SS
siO? iTeat Firm OHtt        Tank Vol        3006 T digits      ! i <=SS
N  r  Teat Crau RS.OP.JS    API Dens        Si.3 T digita/F     313
   »  HRI Crew  SR ,W8       Exo Coef  3.93052538 Leak Rate     3.374.
Tine- Laval
Hr din < div >
12
12
12
12
12
12
13
13
13
13
13
13
13
13
13
13
13
13
1 4
! 4
1 4
1 4
I 4
4
4
4.
4
4.
4
4-
s
s

s
w
c
5'
**

c
c
IS
IS
IS
IS
IS
1 S
IS
15

30
35
4.0
4S
S3
S3
3
S
10
IS
20
25
30
3S
401
45
S0
SS
0
S
10
IS
20
25
30
35
40
AS
sa
Ss
0
c
13
IS
23
2S
30
33
40
4S
S3
Sa
3
S
10
IS
23
2S

12
12
12
12
12
12
12
12
12
12
13
12
12
12
12
12
rz
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12

V Before
(gal)
N/A
0.34S
3.375
0.4JS
0.4,5
0.43
9.S
a. 34.
0.SSS
0.S05
0.S4
0.345
0.37S
0.41
0-.43S
0.475
0.S05
0,55
0.8
0.S5
0..S3-
0..T3
0.235
0.32
9. .37
9.4
3.43
0.47
3.32S
0.555
9.SS5
0.5.4
•a.sa
a . 725
0.7S
3.23
'3.31
3.35
3. 33
9.4!
9.4S5
0,425
3.S3S
0.S7S
a . s i s
0.SS
3.73S
0.74

U Af-tar-
( aai )
N/ft
3.37S
3.415
0.4S
9.49
9.S
0.S4
0.SS5
3.S0S
0.S4
0.S75
0-.37S
3.4!
3.435
3.475
3.S0S
9.S5
0.S
0.55
0-.S3
9.73
0,775'
a-. 32
3.37
3-. 4
9.43
0.47
0.S2S
9.S53
0.S2S
3.S4
3. S3
9.725
a. 75
0 . 725
9.31
9.35
3.33
9.41
9.4S5
3.4SS
0.S35
3.S7S
0.SIS
3.SS
9.70S
3.74
9.78S
1-16
Fuel TSWQ
(digits)
12065
12268
1 2972
12375
12973
• 12081
12983
12085
12288
12932
12035
120SS
12939
12130
12194
12108
121 1 1
121 14
1211S
12129
12124
12124
12123
12130
12132
1213S
12133
12140
12143
!2US
t 2 1 S-3
12133
I21S4
I21SS
I 21 S3
I21S3
121S1
I21S4
121S3
12179
12171
!2l74
!2!75
12(73
12130
'2132
12133
1 2 133

Tcarr aV Laak Rat a
(gal) (gai/h)
N/A
9.306
-9 . 307
.300
-0.317
-0 . 3C4
3.91S
0.301
9.305
-0.912
.3.00
9.918
.300
9.313
-3 . 307
-3.317
3.319
0.91S
0.325
-#. 307
-0 ..307
3 . 345
-tf.934
0.338
3 . 306
-•3 . 305
-3-. 90S
0.931
-•a . 30s
a . 3 1 s
-3 .314
a . 90s
3.333
3.91 1
3.31 1
3 . 306
9.323
a . 90s
-<3 . 304.
-3 . 902
0.923
9.905
0 . 0 1 S
3.313
3.321
3.321
3 . 323
3.919

N/A
3.97S
~3 . 388
-3.30S
-^. 208
-0 , 044
a . 1 36
9. 913
9 . 954
-0 . 1 43
-<3'. 90S
9.213
-4.906
a. 133
-0 . 983
-9.233
9.114
3. 174
9.31S
-0 . 388
-3.388
9.540
-^, 4"!£
3.4S8
3. 375
-a, ass
0 . 3S"4
9.375 •
-'3 . 365
3. 1 SS
-a . i 70
3.954
9.3S8
3. 1 33
3. l 35
9 . 375
3.333
3 . 3S4
-<3 . 344
-9.923
9.333
3.354
9. I3S
0. I3S
3.2SS
9.2SS
9 . 273
3. 1 14


-------
2S-«ai—35
IS
IS
ts
IS
IS
Iff
17
17
17
17
17
IT
IT
17
IT
17
17
17
13
13
ia
13
13
ia
13
13
18
13
33
35
49
45
50
55
0
5
13
IS
20
2S
33
35
43
45
S3
S5
3
S
13
IS
23
25
30
35
49
45
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12.
12
12
12
12
12
12
12
12
12
12
a-.73£
3.31
®.34S
3-, 33
3.2*
31.275
3.3
3.33
3.355
3.33
3.41
3.43
0.45S
3.475
3.5
3. SIS
3.54-
3.SS
3.57
3.55
9-. si
3 .525
fr.ff*
3-.S3S
3,33
«.71
9.73
a. TS
3',3-t
9.345
3.38
3.3
3.275
3.3
3.37
3.355
3.33
3.4!
9.43
3.455
3.475
3.5
3. 515
3. 54
3. so'
3.57
3.53
3.51
3.S25
3.54
3.555
9.53
3.71
3.73
3.75
3.77
12! 33
12131
12131
1219?
12135
12135
12133
12232
12232
12235
12237
12213
I 2233
12212
12214
12215
1221S
12213
1222!
12222'
12223
12224
I222S
12227
12233
12223
1223!
12232
3.331
.399
3.335
-9.394
a. 31 1
9.313
3.336
-9.322
3.325
-9 . 305
-0 . 304.
-9.313
3.332
-0.313
-0 . 303
3.313
3.308
-9.325
-9.304.
3.308
3 . 303
3.303
9.301
3.313
-9. 315
3.332
-9 . 304
3 . 308
3.3IS
-9 . 308
3.420
-9.344
3. 135
3.153-
3.375
-0.253
3.330
-9.365
-9.344
-9.125
3,332
-9. 125
-9 . 1 04
3.153
3.338
' -9.306
-9.344
3.338
3.333
3.338-
3. 315
3.158
-0. 196
3.382
-9.344
3 . 338

-------
           2.S
           74.4
           1 A -I


           -4
           . .  i
           i.x -t
               i
            1 4
               i
           2.3 -j
                                       N 2 9 0 0 0 0 01 0

                                      TANK  ! REGULAR
                                                                    .-cS
                             -OS**
L/ XCorr-^r-
                                                   hi
                                        1-18

-------
                             SITS DATA  SUti&ABZ


Surrey H3 tfo.  2S90000010	           Test Date  Mar   18»
     Tank No.   3  o£ 3                         Produce  Unleaded
     Tank Siae   6006  Gal.	


                               TEST RESULTS
                              Line  1        Line 2       Line 3       Line
Certified         Yaa   '        to            tea

Rate              0.013         -O.OJ2       -0_-015

S.E.              0-035
Noces:
 FcsEn 3429-001SDS
                                     1-19

-------
                                 Pago- t-
 UJ  lSit« Coda N2S0000010 Fuel Type UNLSWED   Data      Mftfl  Id  35
315? iTaat Firrt OHM         Tank Vol        S00S T digits       12171
N  Y ,'Test Crew RS.DP.JS   API 0«ns        Si.3 T diqita/F     320
   *  ttRI Crew.  SR.US       Sxo Coaf  3.30060863" Laak Rats
Tine
Hr mn
13 4S
14 0
1 4 S
14 10
4- IS
4 20
4 25
t — rt i
4 ja
4 35
4 40
i . ff>
4 45
4. 50
4 S5
S.K
0
15 5
1 C? (4
IS ! 0
t & i ^
15 IS
IS 20
If™ *^r*
5 25
15 33
!^ «••••
S 35
I 3 40
IS 45
IS S3
IS S5
i £ ,-i
I a •
7 40
If . (^
7 45
t T ^ Tl
1 7 30

L«V«i
(div)
12
12
12
12
!Z
12
1Z
12
12
12
12
12:
12
12
12
12
12
12
12
12
1Z
12
1Z
12
12
12
IZ
12
12
12
12
12
IZ
12
II
1*9
2
12
12
1 2
12
12
12
IM
2
12
l 2.
12
12
12

y 3afor»

-------
2S-tlar-35                          Page  2
17
13
19
13
13
13-
13
13
!3
13
13
(3
13
19
13
13
13
13
13
19.
13
13
33
5
19
IS
23
2S
39
35
44
4S
50
S3
3
s
19
IS
29
25
59
35
A3
AS
1Z
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
3.42
9.s
a. S3
3. SS
3,3
9. S3
3.35
9.335"
9.4.2
3.4S
9.475
3. SI
9. 54.
3. SSS
9. S3
3,S2
3 S3
3.S57S
. 9.24S
3. ,23
9.30S
0.3IS
3.44S
3. S3
3.35
a. a
0.53
^.SSs
3-.32S
^.42
0.4S
3.47S
9.S1
3.S4.
3. SSS
3. §9
3.32
4.S5
3.S7S
9.T
3.23
a. 3as
9. 315
3.37
12422
12423
12431
12433
12435
124413
1244!
12440
12441
12447
12443
124S3
124Sa
124S7
124S7
12*53
12464.
1246S
124S7
12473
12471
12474
3.392
-
-------
                                            *                 "
 o
£


I
 o
..2
•s
3

2.3 -f
2.4. -|
1
•> -*
^ I
' -3 ""1
t t
G.Q -} J-
<3.o - rSf
^» * <*^~
Q-z 1 a*? ^oo*^
« i < -^- 	
- - - «»•'-
^-» 1
>•.«. | |
Q
TANK J UNL£AD£D
-^
^nfl"
^tfP" 4-**1"
-rP=^^,r^*
^T-^" *•*•
-^T -^ , -v ,-j-
^ff^ , 1-- 1- '
cC^" ^^'
rff^^
-^3 **
_^~"
^^ -H-**
-a— +*
-a?: **"
OST ^^-»
-ofJV*
-C*.^
rj^^v*
'" *rt**s*c«»o
-------
                              SITS DATA STJMMAKr                                     -••*.


Surrey ID Ho.  8310000109	          Test Data  Mar 20.  1985	

     Tank So.  1 °* 1	          Product  ^leaded	

     Tank Siaa   8QO° Gal*	


                                TEST RESULTS


                                            Line  2        Line 3       Line

Certified

Rate
Motes:
         This  taat had  a line that co\iid not ba cartlTied because of a bad
    ciiack valve.

         The tank taat  data had some problems.   There ua,2 a, 25 ainuta gap In
    tha  data caused by  the  need to break concrete to i^olata a line from the
    tank.   A theratLrtar box MAS givlaaf  erratiscreadings and had ta ba changed.
    When the thdraistar box was changed,  there  uas a discontinuity in the
    tamperatare readings.   Finally,  there was an entry error in the data.
    To analyze,the data,  the erroneous  entry was corrected, then only data
    taken aftar the thersistar box was  replaced ware used.  With these asodifications,
    the  data were adequate  and the analysis  satisfactory.
 Form 3429-001SDS
                                     1-23

-------
04-Juii-«5
  18  iSiti Cade X310000109 fuel  Type UXUAOE3   Date     MR.29,35
 si?? !Test Fin O.O.     Tan*  Vol        7714 T digits      13347
 »  f }T«t Crs» 3S,JP,JS   API  8«ns        37.2 T digits/F      324
    »  ffil Cr«*  *.3.,5.4.  £ip Corf  0.90039372 t«* Satt     O.OtS
TiM
Hr «n
13 40
13 45
13 30
13 33
14 0
14 3
14 10
14 13
14 20
14 23
14 30
14 33
14 40
14 43
14 30
14 S3
17 0
17 3
27 10
17 13
17 20
17 23
17 30
17 33
17 40
17 43
17 SO
17 33
13 9
ia 5
18 10
18 13
18 20
18 23
18 30
18 33
18 40
13 43
13 50
18 53
19 0
19 S
19 10
19 13
1? 20
1? 23
1? 30
Iml
fdiv)
12
12
12
12
12
12
12
12
- 12
12
12
12
12
12
12
12
22
12
12
12
12
12
' '12
12
12
12
22
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
V 3elore V After Fuel Twp
(gji) (gaiJ (digits)
If/A
0.340
9.333
0.34
0.34
0.343
0.373
0.133
0.143
0.13
0.13
0.183
0.1?
0.19
0.193
0,193
0.195
0.2
0.21
0.21
0.21
0.21
0.203
0.203
0.203
0.203
0.203
0.203
0.203
0.203
9.2
9,2
0.195
0.193
0.193
9.19
0.19
0.31
0.31
O.J1
9.3
9.3
0.293
0,293
0.293
0,293
9.293
X/A
0.333
0.34
0.34
0.243
O.J73
0.38
0.143
0.28
0.13
0.133
0.19
0.19
0.193
0.193
0.193
0.2
9.21
0.21
0.21
0.21
0.20S-
0.203
0.203
0.203
0.203
0.203
9.203
0.203
0.2
0,2
9,193
0.193
9.193
9.J9
9.19
9.J9
9.31
9.J1
9.3
0.3
9.293
0.293
9.293
• 9.293
9,293
0.293
13347
13347
13347
13347
13347
13549
1334?
13345
13347
13349
13349
13544
13349
23549
25549
13349
13349
11343
13344
13344
13244
13244
13344
12244
12544
15544
15544
13544
13544
15543
13545
12543
15345
15345
153&
13343
135A3
13543
13243
13343
13343
13343
13543
13343
13343
13543
15543
Tcorr 4V Uik Rate
(qai) (qji/ft)
N/A
9.01300
0.00300
0.00000
0.00500
-0,01811
9.00500
0.03421
-4.01311
-4. 92811
0.00500
0.04714
-0.04214
0.00500
9.00000
0.00000
0.90500
0.02403
0.92311
0.00000
0.00000
-0.00500
0.00000
0.00000
0.00000
0.00000
0.00000
0.90000
9.00000
9.00903
9.00000
-0.00500
9.00000
9.00000
9.02312
0. 90000
0,00000
9,00000
9.00000
HJ. 91000
9.90000
-0. 90500
9.90000
0.90000
0.90000
9.00000
9.90000
X/A
9.180
9.040
9.000
9.040
-0.217
9.040
9.433
-0.137
-4.337
9.040
9.244
-4.304
0.940
9.900
9.000
9.940
9.239
0.337
9.900
9.000
-4.940
0,000
9.000
9.900
0.900
9.900
9.900
9.900
0.199
9.900
-4.940
9.900
9.000
0.277
9.000
0.000
9.000
9.900
-4.120
0.900
-4.940
9.900
9.900
9.900
9.900
9.900
                                          1-24

-------
                                                          01
                                                          O
                                                         IT
(je6)
                1-25
                                                          T3
                                                          <0
                                                          •MB
                                                          u
                                                          (0
                                                          u

                                                         *«
                                                          °
                                                          u
                                                        <0
                                                        a
                                                        _o
                                                        ul
                                                           fl
                                                          £
                                                          O
                                                          CL
                                                          X
                                                          Ui
                                                          13
                                                           (0
3

a

-------
                             SITS DATA  STMHARY
Survey H3 y0.   L310000557
     Tank No.    1 °f 1
     Tank Siza  1036 Gal.
Certified
Rats
S.E.
System
 ?aa
 -0.012
  0.024.
              Test Date   M*r 21.  1985
              Product
                                              ^leaded
                               TEST RESULTS
                              Line  I
3CY
            Line 2
Line 3
Line
Notes:
         The Line  teat found a bad check valve, 30 no rate could be estiaated.
Fora 3429-00ISOS
                                    [-26

-------
32-Aor-3S
                                 Paq«
  .*7  !
                 OHM
Fual T«no
                                                   Tcorr dU Ueak Rats
                                                    (
147S2
147S3
14734
1 4306
14823
1 4838
I43S2
14863
14887
14304
U323
14333
14348
I 4355
I43S8
14,384
1S301
1331 1
IS327
i esiA^
1 3^ *
IS233
I S!!i9
I ^>4» *»14
1S24S
IS2SS
'. S2S*
1 C""^
1 —«B t 
-------
                                         "vt
 O
 ^
 "5
TANK 1 UNLEADED
i
1 1 -i
, t
' ' u
3-* -i _ a*2"*
Q.fi -f no5**
0.7 -^ S?4-^4'
. 1 -a3^^
u.3 -( aw T '
0.4. 4 S°-^
"°~ 1 nZ"*"
1 _s'"" .. *, ./%«,..
!
«' ' i 1 1 1 1 1 t
0 1 2 j

.,n=--a!-
J^ -t*
t
r
1
t
r
TT5 	 ~
i r
4
srrected
__           EJcosed Tirrr« (h)
Sxponsicn            o   Corrected
                                      1-28

-------
                            SITS DATA SUMHABS




              1,310000561
                  las
    LXieo



Eate

                   0.010

S.E.
                                             *„*»**   Unload,
                                            _ .       Liae 1       ^rae ±
                 Svsrea
    Form

      .   •                               I-Z9

-------
 LS  !Sit» Coda- U313000SS1 Fue-1 Tyaa UNLSAOSO    Oats      ttftfl 24,35
sig? ITeat Firn OHM        Tank Vol         1334. T diQi+.a      14032
M  Y ITeat Crsu RS,JS,OF   API Qena         S3.3 T dlgita/F      325
   *  MR I Cr*w  SR.U8      cxo Coef  3.300SH503 Lsak Rata    -0.333
Tit
Hr
0
0
0
1
1
1
1
1
1
1
1
!
I
1
1
12
12
12
12
12
12
12
12
12
12
12
12
13
13
13
13
13
13
f3
13
13
13
13
13
1 4
14
14
14
14
14
14
14
«* Lavei
nin (div)
45
S3
Sa
0
5
10
IS
20
25
30
35
40
45
53
Sa
3
S
10
IS
20
2S
33
35
40
45
S3
55
0
S
10
IS
23 '
25
20
35
40
45
S3
55
3
3
13
IS
23
25
33
35
12
12
12
12
12
12
12
12
12
12
12
T2
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12.
12
12
12
12
12
12
52
V Be for* V After Fuel Tecia
(Sal) 
-------
1-31

-------
                       S [ T E   '	51 0 0 Cy 0 D 6
                              TANK 1 CJNCSAOEB



X V
«
6
>w
1>
£
|
0

15
1.1 -
1 -
n a -
...
C.2. —
i* T
^ • »
G.fi -
u.3 -

IJ.4. -

+ **
+ **> (J ••
i* 3
, -r u
**•**" a3"3
^**^ n^
-* . n«

^•f -n"
j. — ^
^> •+•*'"' CJ4*
> -a"
*"** -3"" i
 0.2 4

   t J    ^  -«

   n -•
     T
•G.I 4
Tv
J
1
i
0'
^orrr — - — — „
- ^vO^oo^o^O
! 1
1
n ft "T n i i i'i "^"^~i - i _
• * « >• ^-v <^v *^^^^r
I i i
' 2
i
2
:aO'?O«*»>O ^^j
I , r
4.
                    Expcnsicn
£3cos«d T?m« (h)
         o   Corrected
                                 1-32

-------
                           SITS DATA
                  Ho
Rate




S.£.
                                           Line
                                                        LLae_3
                               No
      ?ocm 3429-OQ1SBS
                                          1-33

-------
                                  raqe
r»   '•Sit* Ccce i_j!ws!   ue  .y.s* ysi.      Gat*       >'•!.•»*  23.55
"'?  ! TSST ~ij"?? 'jH?f         Tiny,  '/ci         7335 7* dibits       .'7513
    '
:T : : 8
» w».
Tine 1
Hr* ?1i« i
IS 33
IS 35
Iff? A A
5 40
1 5. 45
15 30.
1 5 55
1 5 3
!5 5
IS !0
IS 15
1 5 23
( 5 7C
!S 33
'5 35
'. 5 40
'S 4.5
IS 50
iff 55
! f "0
1 7 5
t 7 10
17 '5
17 70
'7 2S
1 7 30
!7 35
fT 40
17 45
i r 50
1 7 55
1 3 0
i a 5
1 3 '0
13 15
I fl 70
'3 25
'3 30
i -3 35
13 40
i g 45
13 50
'3 55
1 3 0
' 5 5
18 ' 0
'5 13
r -3 7-3
I 3 "7~

s \ ur sw .'
I Crs« 5
-avei
( div )
12
12
12
17
12
17
12
12
!2
12
17
17
12
12
17
12
!2
12
T7
12
!2
!2
!2
12 •
17
12
17
12
1 2
12
1 7
17
17
17
i 7
!2
* !2
t 7
17
12
17
12
1 2
17
12
[2
1 2
1 2

3 •—'!* .."tS ffl
R .MS ' £
(i 3»f<3r»« i
(sai)
M/A
a. =30
3.4S
3.53
3.75
3.555
"3741
3.335
3. 35
3.3
3.34.
3.455
3,35
•3. 335
3.775
3-.5A5
3. 485
3. 3 AC
3.13 •
3.32
3.73
3.52
3. AS
3.325
3. 55
3.33
3.SS
3.52
3.485
3.35
i
3.3-4S
3.7i
3. =55
3 . 405
3 . 255
3. 355
0.375
3 . "25
3.33
3,735
3.35
3.42
i
3 . 87
3.7s
"3 . 5"^
3 . 435

"I Usns
*rr Co*!* 3
; Aftar !
( S2i )
N/ft
3.450
3.23
a.' ?i
3.335
3.4A
3.335
3.13
3.8
0.S4.
3.435
3.35-
3,225
3. TTS
3 . 345
0.4*35
3.345
0. ta
3.335
3.73
3.32
3.4^
3.325
3. !S5
3.33
0 _ 55
3.32
3.4J3S
3.35
3.213
3.545
3.7!
3.555
3 . 405
3 . 253
'3 . 355
0.375
0.77C
3. 33
3 . 43
3. 33
3.42
3.27
3. 37
3.7'
3 . 37
3.435
3 .235
r-34
23. 7
.30045255
ru«i r*."!s
f. diai*s>
17430
17439
1 7430
17433
1 7423
'7*23
!7423
1 741 3
17417
I74!5
f 74 1 2
17407
17407
17403
17403
! 7337
i 7307
17332
i 7337
17338
17354.
i 7334.
ITTS
17373
' 7375
! ?37'
'737?
i?3B2
t 7353
'7353,

17351
i 7345
17342
! "340.
i 7335
1 7^?^^
i 7377
i 7375
1 73! 3
1 73' 3
173? 4
17323
17305
1 7304
17235
I72SS
17252

T dibits/?
Lidk Rate
Tc err* '"".'
< gai )
N/ft
-3 . 1 40
-3 . ! 70
-*. 123
-3.3S7
—3 . '55
-3 . '35
-3 . 3s 3
-3. ! 33
-3 . 1 33
-0. i 1!
-3.335
-vl 1 A<=
«• * «•
-0 , M S
—3 . 1 33
-3 . 353
-3 . ! 40
-3 . '355
-3 . 1 55
-0 . 335
-0 . 113
-0 . 1 40
-3 . 333
-3. 1 23
-0 . i 30
-3 i 75
-3 . ' 4ig
3 .385
-0 'j}<31
-3 . 373
"3 . i 33
-3 . ' 35
-<3 . 38 S
~3 . ! 23
-0.1 ! fi
—3 . ' ' 3
-3 . 38 S
-0 . '383
-•a i 73
• *. •• w
-•3. 383
-0 . i 7.!
-0 . ! 33
—3.217
3 . '333
•0 . 143
~3.3« 1 \^Z I* 7
( 52i/H )
MM
-! .333
-7 . 040
-f 4.4*1
• « ^^"^»
-! .333
-' . 350
-l S70
-1 .353
-' . 555
-i SCT
• • ^ ^ •
-' . 337
-f .355
— I 740
-1 , 3B7
-1 .250
-i,i j 3
-' .533
-i rag
• - *• V
-I .350
-' . ! 42
-' .332
-! .533
-i .353
— i S 1 7
- ' . 350
- i . -07
-i .cg0
fl 7qa
• . J V
— f '/IflT
-<3. 533
-1 C3 1
-• 1 57 (/I
- ! . '353

••"*•—
-' .332
- ' . 337
_.;i 3<3~
- ! i7l
J „ *• '
-3. 333
- 1 4C7
-1 . SS]
-2. 337
3.457
-' . 73S
- 1 . v)0fl
-' .217
-' .252


-------
SC
12      *.29S      ?.!4.
                                                         -1.33?
                         1-35

-------
   V^ITP"  i  ~ i nnnsfi i

-1
f
" "w
~S
—A
f

I -fc
i
I
4
I
!
!
t
-t
t
L
I.
J
(
I
I
I
i
0
TANK 2 DIESEL
" Q ^X,
wo ' ^*—
2=a s •^^ i
"a=s== "^H
— g
2a.
-rr
i i i i i r i T
1234
        Scored Tint's C h}
Exocnsian        o  Corr«ctad
          1-36

-------
Survey 12 No.   L310000561
     Tank Ho.   3 of
     Tank Size    7869 Gal
Certified



Rate



S.E.
                 System
                  -0.263
                             SITE DATA SUMMARY
                                             Test Data  Mar 2T.
                                             Product.   Diasel
 TEST RESULTS





Line 1



Jb—
Line Z
                                                         Line  3
Line
Motes:
        Hetfa tfaft task and liae appear  ta  leak«   the line leak was not
 Fora S429-OQ1SLS
                                   1-37

-------
32-*pp-3S
  LH  ISit* Code L31300531   Fuel Type QISScL     Cats-     MAR 23.35
 atg? ,'Taat rim QH«         Tank Voi        78S5 F digits      ! S382
N Y ITeat Cr*w RS.JS.QP API Qena 35.7 T digita/F 325
*• URI Cr-u SR.U8 £ (digits) (gal) (gal/h)
14
14.
14
IS
IS
IS
IS
IS
IS
IS
IS
IS
IS
15
IS
IS
IS
IS
IS
IS
tS'
IS
IS
IS
IS
IS
IS
17
17
17
(7
17
17
17
17
17
17
17
i •?
13
13
13
13
13
13
13
19
13
AS
S3
55
3
c
10
IS
20
25
30
35
40
4S
50
55
•3
5
10
IS
20
25'
30
35
40
45
50
35
3
c
10
IS
20
JC
30
35
4.9
AS
50
55
•3
5
10
IS
20
25
30
35
-13
12
12
12
f2
12
12
12
12
12
12,
1Z
12
12
12
12
12
12
12
12
12
(2
12
12
12
12
12
12
12
12
12
12
12.
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
N/ftt
0.4.30
3.335
0.375
3.735
0.7S
9.73
0.7
0.S7
'3.34.
0.S0S
0-.S7S
0.S4
0.515
a.Aa
0.4£
0.4.3
0.4.
0.735,
0.77
0.73
0.70S
0.sa
0.5?
0.S2
0.S
3.575°
3.35
9.S2
3.4-5S
3.3
3.375
3.355
3.335
3.32
a. 9
3.73
3.7-iS
0.7?
3.S3S
0.S75
0.S55
3.S3S
3. SIS
3.53
3. 57
3.55
0.525
N/A
3.335
3.375
3.34.
3.7S
0.73
0.7
3. 57
ff.S*
3.005
3-.S75-
3.34.
3. SIS
0.*3
3. iff
3.43
3.4
3.375
0.7T
3.71.
0.705
3.S3
0.S5
3.S2
0.S
3.57S
3.55
3.52
3". 435
3.4S5
3.375
3.353
3 . 335
3.32
3.3
3.73
3.74S
3.72
3 . 535
3 . 575
3.555
3 . 535
3.515
3. S3
3.37
3.33
3 . 225
'3.3
15054.
1 5054.
15053
IS0S3
15053
1 5053
15051
15352
15352
15053
13050
IS0S0
1S3S1
1534.3
r 504.3
1S04.S
13343
15343
15343
1534.7
13343
T534S
1 534?
15348
IS34S-
15044.
15344
15344.
13344
15344
13342
15351
1 334 r
15341
15340
13343
13333
I 3333
13337
13337
!S3o3
15333
IS333
15335
13333
13333
15333
13333
N/A
-0.335
-3 . 305
-<3 . 335
-3 . 335
-3 . 330
-<3 . 308
-3.341
-3.330
-3.313
-3". 330
-3.335
-0.335
-3.313
-3.323
-3.313
-3 . 333
-3.325
-3 . 32S
-3.323
-3.33S
-3 . 303
-3.333
-3.333
-#.323
-3.303
-3 . 325
-3 . 330
-3 . 325
-3 . 330
-0 . 303
-3 . 1 20
3.331
-3.315
-3.309
-3 . 323
-3 . 3 1 J
-3 . 325
-3 .314.
-3 . 323
-3 . 33 1
-3 . 320
-3 . 323
3.308
-3,323
3.302
-3.325
-3 . -325
N/A
-3 . 420
-3 . 1 0S
-3.420
-3 . 420
-3. 350
-3.333
-3.434
—3.330
-3. 153
-3.3S0-
-'3 . 420
-3 . 434
-3 . 1 S3
-3.240
-0.22S.
-3 . 350
-3 .300
-^3 . 309
-0.34S
-3.434
-3 . 333
-3 . 3S3
-3. 33 3
-9.243
-3 . 333
-3 . 300
-3 . 350
--3 . 330
—3 .'35 3
-3 . 333
- ! . 443
! .337
-3 . 1 30
-3 . 1 08
-3.243
-3'. ! 33
-3 . J00
-3 . 1 S3
-3.240
-3.374
-3.240
-3.240
3.131
-3.240
3.327
-3 . 330
-3 . 330
                              1-38

-------
                            —3.353
                            -3.398
1-39

-------
        Q «jg^gai»'»»"
                                 L  L J T (J (J U (J
                                     TANK J C1ESSL
    -n 1  J   as.            ' '' ••"•-MM.-*-.
 5»  —G.4- -I
N-'         I
       •W 1
                              a.  «.
                               "n v
                                  a.
           i       r       i
          a              t
 A  -u.' -t                            ~a_   -o
^         I                              wn
•3  —n s J                                Q-
^         I                                  -QM '
                                                       •n

                                                          "ua-
                                            npn« (n i
                    •K   Sitocnsion            o   Ccrr«ct«d
                                    1-40

-------
                             SITE DATA SUMMARY
Survey ED Ho.
Tank Ho.
Tank Size
Certified
Rate
S.S.
L310000561 test Date ^ 2<*, 1985
1 o£ ± Product Qiasal
10152
TEST RESULTS
Systea Line 1 Line I Line 3 Line 4
Xes Ho —
0.009 "Large* ____ __ 	 ( 	 	
0.0107 —
Hoces:
        Tha line taat report Included  the  note  that  there  -.as  a  "l
   leak in the line*  but no  quantitative number.
Fora  8429-001SDS
                                   1-41

-------
 IH  iSite Code L313000SS1  ruei Type OISSEL     Oat*      ttrtR.24.35
sig? (Teat Firm OHM         Tank" Vol       !01S2 T digits       US3S
N  Y iTest Cr-« JS ,QP ,RS    fliPl Dens        38.1  T digita/F"      32S
   «•  MR I Crsw  SR.U8       Sxp Caef  3\3W4>534S Laak Sate     3.01Z"
Tine Level
Hr Win (div)
JS 35
IS 40
IS 45
IS 53
IS S3
1 S 0
IS S
IS 13
!f* « f*
S 1 S
IS 23
IS 25
IS 33
IS 35
1^ A /%
S 40
IS 45
IS S3
!** *****
S 55
(urn *m
7 3
17 S
I am t /K
7 (3
Itm j f»
7 1 S
. I» ^A
!7 20
. ** *^^>
1 7 2S
!•• dB/%
7 j0
!« ^••»»
7 35
1 3 3
t fS (^
1 3 S
13 !3
1t*\ i ••
3 1 s
1 3 23
! 3 25
1 3 30
13 35
! 3 40
1 3 43
13 5*3
13 S5
! 9 3
1 3 S
13 10
IS ! S
1 3 23
1 3 25
. M ^ *\
1 3 j3
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
1Z
12
12
12
12
!2
12
12
12
12
12
12
12
12
12
12
12
1 1
12
12
12
12
12
12
12
12
12
12
12
12
12
y Before
(gal >
N/A
3.720
0.715
0.71
3.7
3.335
3.535
0.S7S
a.ss-
0.555
3.545
0.535
0.325
3.SZ
0.535
0.535
0.53
3. 575
3 . SaS-
0.555-
3.54
3.525
0.52
• a. 5
0.43
3.39
0.375
3.3S5
0.3S
3.34
3.33
3.315
3.3
3.385
3.37
3.35
3.34
3.325
3.325
3.31S
3.305
3.73
3.735
3.73
3.77
2.75
0.75
3'7* T_,
y After Fuel 7««0
 (di'gita)
N/ft
3.71S
3.71
3.7
3.S3S
3.535
3.575
3. So
3.55=
3.S4S
0.S3S1
3.525
0-.S2
3. 505
3.535
0.53
3.575
3.SS5
31. 35=
3.54-
3.S2S
3.32
3.5
3. 43
3.48
3.375
ff.SSS
3.25
3.34
3.23
'31. 31 S
3.3
3". 385
3,37
3.35
3.34
3 . 325
3.322
3-.3IS
3.30S
3.73
3.73S
3.78
3.77
3.73
3.73
3.74
,,3'72
14623
14623
14613
14613
14618
146IS
14616
14513
14613
14614
14614
14612
14512
14613
14513
I4S0S
U60S
14603
14603
14606
14636
14604
14604
14604
F460r
14601
I4S3S
14S33
I 4S53
14537
1 4S3T
14S3S
1-4533
14S33
14332
14332
I4S30
14330
I4S38
14S3S
14=35
14335
14536
14534
14334
14334
I 4532
US32
Tcorr dV
(gal)
W/ff
-0.30S
3.3Z4
-0.313
-0-.30S
3.313
-0.313
-0.301
-0.30S
3.304
-0.013
3.313
-0.305
3.314
-0.313
-0.301
-0 . 30S
-0.310
-3.313
3,323
-0.31S
3.324
-0 . '322
-0.313
3.33T
-0.315
3.3T3
-0.313
3 . 304
3 . 304
-0.315
-0 . 30 1
-3.315
3 . 322
-a . 306-
-3.213
3.31 4
3 . 300
3.319
3.313
-0 .313
-3 . 30S
-3 . 30S
3.319
-0.313"
-3.310
3.313
-3.323
Leak Rate
fgai/h)
N7A
-0 . 360
3.236
-0.123
-3.360
. 3 . 223
-0 . 1 23
-0.307
-0.360
3.3S3
-0 . 1 23
3.22S
-0 . 350
3. 1 S3
-3. 122
-0 . 307
-0.363
-0 . 1 23
-0. (23
3.340
-0 , 1 30
3.233
-0.240
• -0.122
3. 400
-0. 1 30
3.223
-3.180
3.333
3 . 3S3
-3 . 1 30
-3.307
-3 . 1 30
3.340
-3 . 337
-3 . 1 22
0. 1SS
3 . 300
3 . 223
3.225
-3 . I 33
-3.360
-3 . 353
3.223
-3. 123
-3 . 1 20
3.223
-3.240

-------
1-43

-------
                               /*«•» t ^^ f*^  i   ^* s*  .-•* t^  s^. /—+. r^ s*+ j*

                               OJ  ! L  l_O i  UUUUO'G  i
        .1
 (7

.*!


 o
 /"•

 i
-0.1  -f
»   -a.2  -»
     —U.+
     -<3.5
                       Sxp-ansion
                                                   •>    uorr*cr«c
                                          1-44

-------
                             SITE DATA SUMMARY

Survey ED Ko.    G31QOOO*»64	          Test  Data  Mar 29.  1985
     Tank tfo.     1  of 3                     Product   Diesel
     Tank Siae    10152 Gal.

                               TEST RESULTS

                 System       Line 1       Line  2        Line  3        Line
Cartiiied
Rats              Q.Q26
S.E.              0.011
 Farm 3^29-OOlSDS
                                    1-45

-------
32-Aqr-aS
Fage
  LH  ISiteCode G3130304S4. Fuel Tyce QIESSL     date      MAS 23.1383
 3ig? !Teat Fir* QHft         Tank Uoi       10TSZ T digits      f33S4.
 N  Y '.Test Crsu RS.JS.OP    API Deng        3S.4 T diaita/F      325
    *  MR! Crew  SR ,US       Exa Ca«r~   0-.3004S-48 Leak Rate     3.333
Tine Level
Hr flin ( div >
13
13
13
13
13
13
13
13
14
14
14
14
14
14
14
14
14
!4
14
14
ts
IS
IS
IS
IS
IS
IS
IS
IS
IS
15
IS
IS
IS
IS
IS
IS
13
IS
IS
IS
IS
IS
IS
17
17
17
17
23
25
30
35
40
45
50
55
0
5
10
IS
23
25
30
35
40
45.
S3
55
3
5
10
15
20
25
30
35
40
45
30
S3
3
3
10
IS
20
25
33
35
40
45
S3
»»«•
3
i»
10
IS
12
12
12
IZ
rz
12
12
IZ
12
12
12
12
12
12
IZ
12
12
12
12
12
12
12
12
12
12
IZ
IZ
12
IZ
IZ
12
12
12
t •*
i i.
!2
* n
12
12
12
12
12
12
12
12
12
12
12
12
U Before
( (diaita)
M/A
3.325
J.32S
3.34-
3.34
0.355
3.365
0.37
0.SS
3.33
fr.gr
0.305
3.315
3.32
0.295
0.3
3. 31
3.32
0\3Z
0.33.
3". 335
3-. 345
fr.35
0.355
3.35
3. 353
3.37
3.38
3.32
3.4
3.4
3 . 4 1 S
3'.4!S
3.425
3.42S
3.435
3". 445
3.4=5
3.4S
0.47
3.475
3.43
3.43
3 . 4S
3.3
3.3
•3.31
3.3IS {-46
13833
13890
F33S3
13830
13893
13333
13392
13332
C3834
13834
13834
13834
13334
13894
V3894
13835
13835
13835
13335
13855
133S5
I383S
i332S
I383S
C383S
I332S
T33S7
13837
138 S3
13893
13300
13330
13230
13230
I3S33
13301
13231
13232
f3S3Z
13232
13334
13234
1 3204
13234
13234
1393S
13236
13236
Tcorr dV LsaK Sate
Cgal) (gal/h)
N/A
3.00S
3". 300
0,315
3.300
0.315
-3.313
0.30S
-3.013
0.010
3.310
0.035
0-.010
3.335
3 . <3<3S
-3.313
3.3*13"
"3.3T0
0.300
9.31-0
-3.313
3.310
3.30S
3.305
3 .305
3.305
-3 .313
3.313
-3 . 305
3.310
-3 . 323
0.315
3 . 3.03
3.313
3'. 330
-3 . 305
0.310
-3 . 005
3.335
3.313
-3.324
3 . 305
0 . 303
3.313
3 . 3 f0
-3.323
3.313
3.305
N/A
3.366
3.300
3. 130
0.300
3.130
-3 . 223
0.363
-3 . 229
0.123
3.123
3. 360
3.123
3 . 360
3.363
-3.114
3.123
3.120
3.000
3.123
-3.114
0. 123
3-. 063
3.363
3.353
3.360
-3.114
3.123
-3 . 354
3. 123
-3.343
3.130
3.300
•3.123
0 . 300
-D.3S-i
3.129
-3.354
3.360
3.123
-3.238
3.363
3.333
3.123
0 . I 20
-3 . 343
3. 123
3.360

-------
22
12      a.sis      3. sz      t3«as     a.aos     a.asa
                  1-47

-------
                       SITE  G3TQ000464-

                               TANK t ClEScL.
     0.3 -4
                                                           ac
                                   ,uu
i    -.,  I
-•    w.^. "1
    u.ia-1

"3
•s


5    G.1 -i
'->        I
    n n* -1
        G.O.
-a
                      P

                      r
                    Expansion
I       I       I

       X


ETcpsed TTme Cr>]

        o  Carrectad
                              i

                              3
I
i

•i.
                              1-48

-------
                              SITS DATA SUMMARY
Surrey ED fto.  G310000*^

     Tank So.  2 of 3
     Tank Size
10152
                            Test Data  Mar 27.
                            Product  . Diesel
                                TSST RESTILTS
Certified

Rate

S.E.
Line I
             -0.007
line 2
                                                         Line 3
                                                     Line  4
tfotes:
        There was virtually no level change throughout the entire test.  However,
   the temperature seemed to-be slightly increasing.   The increase ia temperature
   reported oy the theraister would nean that there should have been a volume
   increase over the test due to thermal expansion.  The entire estimated'1'eak  "
   rate is due to the Lack o£ observed expansion to correspond with the
   temperature.
 Form 3^29-OOlSSS
                                     1-49

-------
02-*ci—85                         Fage !


  L3  iSite Code S3100004S4 Fuel Type QIE3SL     Oats      MAR 27,1385
 3ig? !Test Firtn OHM         Tank Vol       1 0*152 T digits      I 4310"
 N  Y Heat Crew. RS . JS .OP    API Dens        35.7 T digits/F      325
    *  MRI Crst*  SR.U8     •  £xp Caaf  0.9004S1S5 Laak Rat*    -0.10S
Tine Level ^
Hr ,1in < div )
17
17
17
17
17
17
17
17
17
13
13
13
13
13
13
fa
18
13'
13
13
!3
13
13
13
13
If
19
13
IS
13
1 3
IS
13
23
23
23
23
23
23
23
23
23
23
23
23
2!
21
2!
15
23
25
30
35
40
45
50
55
9
5
10
15
23
25
30
35
40
45
50
55
9
5
10
IS
23
25
30
35
4<3
45
50
55
3
5
10
15
23
25
30
35
40
45
53
55
3
z
13
12
12
12
12
12
12
12
12
12
12
12 .
12
12'
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
! 2
12
12
1 2
12
12
12
12
12
12
12
12
12
12
/ 3efore
(gal)
N/A
0.330
3.33
0.885
0.385
0.885
0.33
0.83
0.SS5
0.335
3.835
0.835
0.335
0.335
3.335
0.895
0.895
0.895
0.835
0.395
0.395
0.835
0.395
0.895
3.325
9.335
3.355
0.835
0.835
0.835
0.335
0.825
3.325
3.325
3,325
3.325
3.335
0.325
9.335
3.335
3.335
3.3
3.3
3.305
3.305
3.335
0.235
3.205
y After Fuel Temg

-------
32-*or-3S






21  IS         \Z       9.305      3.31       14363     3.305     3.363
                                 1-51

-------
a
                0.5 -r
                     I

                0.4. 4





                0.2 H
                a. i
           U
                     I
                     Q
                                      SI T E  Q 31Q 0 0 Q 4 6 4-
                                               TANK  . 01E3SL
           S«os«d Time (h
Beocnsion            o
                                                r-52

-------
                              SITE DATA SUMMARY
Survey LTD Ko.  G"310000*^4	i          Test Data   Mar 23.
     Tank No.  3 o.f 3                         Product   Unleaded	a

     Tank Size    10152	


                                TEST RESULTS


                                            Line  2        Line 3       Line 4
Rate

1.2.
tfotea:
        There la an ua explain ad volume increase in this test.  It appears to
   qcexir priaarily in tae first hour and a, oaif or so=  That Is, the- taaperatur*
   was constant for tae first period, then began to drift upward,-'parallelling ""
   the observed Tolurae increase.
 Fosa 3429-001SDS
                                    1-53

-------
OS-Jun-83
Page 1
LR
si
N

Ti
Hr
13
13
13
13
13
13
13
12
13
13
14
14
14
14
14
14
14
14
14
14
14
14
13
15
13
15
13
IS
13
IS
IS
13
IS
15
16
16
16
16
16
16
16
g?
Y
*
me
l-Un
10
IS
20
23
3O
33
4O
45
SO
S3
0
3
10
13
20
23
30
33
40
43
SO
33
0
3
10
13
20
25
30
33
4O
43
SO
33
0
-j
.1 0
13
20
^er
— -J
30
Site Cade
last Firm
Test Craw
MR I Cr sw
Level
(div)
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
G3 10000464
QHn
RS,JS,DP
SR,WB
V Be-fore
(gal)
N/A
0 . 5-33
0 . 553
0 . 37
0.583
0.6
0.61
0.62
0.63
0.63
0.665
0.683
0.7
0.69S
0.71
0.723
0.743
0 . 763
0.78
0.793
0.81
0.83
0 . 83
0 . 363
0.3
0.31
0 . 325
0 . 33
0.33
0.37
• 0,39
0. 4
0. 415
0.433
•0.435
0. 48
0 . 495
0.515
0.535
0 . S3
0 . 353
Fuel Type
Tank Vol
API Dens
Exp Coe-f
V A*ter
(gal )
N/A
0 , 555
0 . 57
0,533
0.6
0 . 6 1
0 . 62
0. 63
0. 63
0.663
0,683
0.7
0.72
0 . 7 1
0.723
0 . 743
0 . 765
0 . 73
0 . 793
0.81
0. 33
0.83
0 . 863
0.883
0.31
0 . 325
0 . 33
0 . 33
0 . 37
0 . 39
0. 4
0.415
0 . 433
0 . 433
0. 48
0 . 493
O.S15
0 . 533
0 . 55
0 . 535
0.37
UNLEADED
10152
53.3
0 . 0(3058023
Fuel Temp
(digits)
14723
14723
14726
14727
14727
14727
1472S
14729
1 4730
14731
14731
14731
14733
14734
14733
14736
14737
14737
14738
14733
1 474O
14742
14743
14744
14744
14743
14745
14745
14744
14748
14748
14748
1 4730
1 473O
14732
14.732
14732
14734
14736
14736
14756
Data
T digits
T digits/F
Leak Rate
Tcorr dV
(gal;
N/A
-0 .016
-0.003
-O . 003
0 .015
0 . 0 1 0
-0 . OOS
-O . 008
0 . 002
-O .003
0*020
0.015
-O .016
-O . OO3
-O . 003
0 . OO2
0 . OO2
0.015
-O . 003
0.013
-O.016
-O .016
' • -0 . 003
0 . 002
0.010
-O . 003
0. 005
0 . 020
0 . 002
-O .016
0 . <".) 1 0
0 .015
-0 . 0 1 6
0 . 020
-O .011
0 . 0 1 3
n ..020
-0 .016
-0.021
0 . 003
0.015
riAR 28,83
1468O
326
0 . 0 1 0
Leak Rate
(gal/h)
N/A
-O. 194
-O . 037
-O . 037
0 . I SO
0, 120
-O . 097
-O. 097
0.023
-O . 037
0.24O
0. 18O
-0. 194
-0 . 037
-O . 037
0 . 023
0.023
0. ISO
-0 . 037
0 . i SO
-O. 194
-O. 194
-0 . 037
0 . 023
0. 120
-O . 037
0 . 06O
0.24O
0 . 023
-O. 194
0. i20
o. 130
-O. 194
0.24O
-O. 134
0. 130
0. 24O
-0. 194
-0 . 234
0 . 06O
0. 180
                              1-54

-------
£
tn
a + o
a + 4
a -r o
a + o
a -r o
a 4- o
a •»- o
a * <







>
a-t- o
a -f

0
a -r
a -r

; Cj>
3
O-
•a . <
•a
€
Gr
a •»•
CH-
s
3
3
3
1 &
\ a*
s°
-w~
TtJ
-c
i +Q
i -3
i ~£

^e
•rC
: -r-C
«•
i
! i i i :
s tO 1/3 ^ fO CN »-
**••«**
D o a o a o o


(|o6) aujn|o/\p SAi4.D|nujn3


1-55
0


p

«
c>
d
1
o

<0
I CM


m «
1
i
r- I

03
r CM
i
i
r*


(- ; TJ
! CN nj
<^WI









H
fj
O
OJ
V.
^
- CM >SO
18
Jo

• . "O
,. (0
a
a
UJ
N '.CM
!<>
!i<>
^ c
0
H L n
* - §
! ' T
> r- T
a
i
W»


-»
( <
o ^
*
O -n
i
"o

-------
TECHNICAL REPORT DATA
(Please read laarucnons an the reverse before completing
T SSPORTNO. 2.
EPA-560/5-86-OH
A. TITU2 ANO SU3TITV.S
Development of a Tank Test Method for a National Survey
of Underground Storage Tanks
7. AtjTHCR
-------
16. Abstract (Concluded)

Finally, the selected method was evaluated in a pilot study.  Some additional
method refinements resulted from the pilot study analysis.  An operating pro-
cedure was developed for use of the recommended method on the national survey.

-------

-------