vxEPA
            United States      Solid Waste And
            Environmental Protection  Emergency Response/
            Agency        Research And Development
            	(OS-420) WF	
EPA/530/UST-90-004
March 1990
            Standard Test Procedures
            For Evaluating Leak
            Detection Methods
           Volumetric Tank Tightness
           Testing Methods
                                 Printed on Recycled Paper

-------
      Standard Test Procedures for
Evaluating Leak Detection Methods:
         Volumetric Tank Tightness
                   Testing Methods
                          Final Report
      U.S. Environmental Protection Agency
       Office of Underground Storage Tanks
                          March 1990

-------
                                 FOREWORD

How to Demonstrate That Leak Detection Methods Meet EPA's Performance
Standards

The Environmental Protection Agency's (EPA's) regulations for underground storage
tanks require owners and operators to check for leaks on a routine basis using one of a
number of detection methods (40 CFR Part 280, Subpart D).  In order to ensure the
effectiveness of these methods, EPA set minimum performance standards for equipment
used to comply with the regulations.  For example, after December 22, 1990, all tank
tightness testing methods must be capable of detecting a 0.10 gallon per hour leak rate
with a probability of detection of at least 95% and a probability of false alarm of no more
than 5%.  It is up to tank owners and operators to select a method of leak detection that
has been shown to meet the relevant performance standard.

Deciding whether a method meets the standards has not been easy, however.  Until
recently, manufacturers of leak detection methods have tested their equipment using a
wide variety of approaches, some more rigorous than others.  Tank owners and
operators have been generally unable to sort through the conflicting  sales claims that
are made based on the results of these evaluations. To help protect consumers, some
state agencies have developed mechanisms for approving leak detection methods.
These approval procedures vary from state to state, making it difficult for manufacturers
to conclusively prove the effectiveness of their method nationwide. The purpose of this
policy is to describe the ways that owners and operators can check that the leak
detection equipment or service they purchase meets the federal regulatory
requirements. States may have additional requirements for approving the use of leak
detection methods.

EPA will not test, certify, or approve specific  brands of commercial leak detection
equipment. The large number of commercially available leak detection methods makes
it impossible for the Agency to test all the equipment or to review all the performance
claims. Instead, the Agency is describing how equipment should be tested to prove that
it meets the standards.  Conducting this testing is left up to equipment manufacturers in
conjunction with third-party testing organizations. The manufacturer will then provide a
copy of the report showing that the method meets EPA's performance standards.  This
information should be provided to customers or regulators as requested. Tank owners
and operators should keep the evaluation results on file to satisfy EPA's record keeping
requirements.

-------
EPA recognizes three distinct ways to prove that a particular brand of leak detection
equipment meets the federal performance standards:

   1.     Evaluate the method using EPA's standard test procedures for leak detection
          equipment;

   2.     Evaluate the method using a national voluntary consensus code or standard
          developed by a nationally recognized association or independent third-party
          testing laboratory; or,

   3.     Evaluate the method using a procedure deemed equivalent to an EPA
          procedure by a nationally recognized association or independent third-party
          testing laboratory.

The manufacturer of the leak detection method should prove that the method meets the
regulatory performance standards using one of these three approaches. For regulatory
enforcement purposes, each of the approaches is equally satisfactory.  The following
sections describe the ways to prove performance in more detail.

EPA Standard Test Procedures

EPA has developed a series of standard test procedures that cover most of the methods
commonly used for underground storage tank leak detection. These include:

   1.     "Standard Test Procedures for  Evaluating  Leak Detection Methods:
          Volumetric Tank Tightness Testing Methods"

   2.     "Standard Test Procedures for  Evaluating  Leak Detection Methods:
          Nonvolumetric Tank Tightness  Testing Methods"

   3.     "Standard Test Procedures for  Evaluating  Leak Detection Methods:
          Automatic Tank Gauging Systems"

   4.     "Standard Test Procedures for  Evaluating  Leak Detection Methods:
          Statistical Inventory Reconciliation Methods"

   5.     "Standard Test Procedures for  Evaluating  Leak Detection Methods:  Vapor-
          Phase Out-of-tank Product Detectors"

   6.     "Standard Test Procedures for  Evaluating  Leak Detection Methods:  Liquid-
          Phase Out-of-tank Product Detectors"

   7.     "Standard Test Procedures for  Evaluating  Leak Detection Methods:  Pipeline
          Leak Detection Systems"

-------
Each test procedure provides an explanation of how to conduct the test, how to perform
the required calculations, and how to report the results. The results from each standard
test procedure provide the information needed by tank owners and operators to
determine if the method meets the regulatory requirements.

The EPA standard test procedures may be conducted directly by equipment
manufacturers or may be conducted by an independent third party under contract to the
manufacturer.  However, both state agencies and tank owners typically prefer that the
evaluation be carried out by an independent third-party in order to prove compliance with
the regulations. Independent third-parties may include consulting firms, test
laboratories, not-for-profit research organizations, or educational institutions with no
organizational conflict of interest.  In general, EPA believes that evaluations are more
likely to be fair and objective the greater the independence of the evaluating
organization.

National Consensus Code or Standard

A second way for a manufacturer to  prove the performance of leak detection equipment
is to evaluate the system following a national voluntary consensus code or standard
developed by a nationally recognized association (e.g., ASTM, ASME, ANSI, etc.).
Throughout the technical regulations for underground storage tanks, EPA has relied on
national voluntary consensus codes  to help tank owners decide which brands of
equipment are acceptable.  Although no such code presently exists  for evaluating leak
detection equipment, one is under consideration by the ASTM D-34 subcommittee. The
Agency will accept the results of evaluations conducted following this or similar codes as
soon as they have been adopted. Guidelines for developing these standards may be
found in the U.S. Department of Commerce "Procedures for the Development of
Voluntary Product Standards" (FR, Vol. 51, No. 118, June 20, 1986) and OMB Circular
No. A-119.

Alternative Test Procedures Deemed Equivalent to EPA's

In some cases,  a specific leak detection method may not be adequately covered by EPA
standard test procedures or a national voluntary consensus code, or the manufacturer
may have access to data that makes it easier to  evaluate the system another way.
Manufacturers who wish to have their equipment tested according to a different plan (or
who have already done so) must have that plan developed or reviewed by a nationally
recognized association or independent third-party testing laboratory (e.g., Factory
Mutual, National Sanitation Foundation, Underwriters Laboratory,  etc.). The results
should include an accreditation by the association or laboratory that the conditions under
which the test was conducted were at least as rigorous as the EPA standard test
procedure. In general this will require the following:

-------
1.      The evaluation tests the system both under the no-leak condition and an
       induced-leak condition with an induced leak rate as close as possible to (or
       smaller than) the performance standard.  In the case of tank testing, for
       example, this will mean testing under both 0.0 gallon per hour and 0.10 gallon
       per hour leak rates. In the case of ground-water monitoring, this will mean
       testing with 0.0 and 0.125 inch of free product.

2.      The evaluation should test the system under at least as many different
       environmental conditions as the corresponding EPA test procedure.

3.      The conditions under which the system is evaluated should be at least as
       rigorous as the conditions specified in the corresponding EPA test procedure.
       For example, in the case of volumetric tank tightness testing, the test should
       include a temperature difference between the delivered product and that
       already present in the tank, as well as the deformation caused by filling the
       tank prior to testing.

4.      The evaluation results must contain the same information and should be
       reported following the same general format as the EPA standard results
       sheet.

5.      The evaluation of the leak detection  method must include physical testing of a
       full-sized version of the leak detection equipment, and a full disclosure must
       be made of the experimental conditions under which (1) the evaluation was
       performed, and (2) the method was recommended for use. An evaluation
       based solely on theory or calculation is not sufficient.
                                    IV

-------
                           ACKNOWLEDGMENTS

This document was written by Jairus D. Flora Jr., Ph.D., and Karin M. Bauer for the U.S.
Environmental Protection Agency's Office of Underground Storage Tanks (EPA/OUST)
under contract No. 68-01-7383.  The Work Assignment Manager for EPA/OUST was
Thomas Young and the EPA/OUST Project Officer was Vinay Kumar. Technical
assistance and review were provided by the following people:

      Russ Brauksieck -  New York  Department of Environmental Conservation
      Tom Clark - Minnesota Pollution Control Agency
      Allen Martinets - Texas Water Commission
      Bill Seiger - Maryland Department of Environment

      American Petroleum Institute
      Leak Detection Technology Association
      Petroleum Equipment Institute

-------
                                  CONTENTS

Foreword	i

       1.     I ntroduction	1
                    1.1     Background	1
                    1.2     Objectives	1
                    1.3     Approach	2
                    1.4     Effects of high ground-water level	2
                    1.5     Organization of this document	4
       2.     Scope of Applications	5
       3.     Summary	6
       4.     Safety	7
       5.     Apparatus and Materials	8
                    5.1     Tanks	8
                    5.2     Test equipment	9
                    5.3     Leak simulation equipment	9
                    5.4     Product	10
                    5.5     Miscellaneous equipment	11
       6.     Testing Procedure	12
                    6.1     Environmental data records	13
                    6.2     Induced leak rates and temperature differentials	13
                    6.3     Testing schedule	17
                    6.4     Testing problems and solutions	19
       7.     Calculations	21
                    7.1     Basic statistics	21
                    7.2     False alarm rate, P(FA)	24
                    7.3     Probability of detecting a leak rate of
                           0.10 gallon per hour, P(D)	26
                    7.4     Other reported calculations	26
                    7.5     Supplemental calculations and data analyses
                           (optional)	28
       8.     I nterpretation	36
       9.     Reporting of Results	37

Appendices

A.     Definitions and Notational Conventions	A-1
B.     Reporting Forms	B-1
                                       VI

-------
                                  SECTION 1

                                INTRODUCTION

1.1     BACKGROUND

The regulations on underground storage tanks (40 CFR Part 280, Subpart D) specify
performance standards for leak detection methods that are internal to the tank. For tank
tightness testing, the tests must be capable of detecting a leak of 0.10 gallon per hour
with a probability of (at least) 95%, while operating at a false alarm rate of 5% or less.

A large number of test devices and methods are reaching the market, but little evidence
is available to support their performance claims. Advertising literature for the methods
can be confusing. Owners and operators need to be able to determine whether a
vendor's tank tightness test method meets  the EPA performance standards.  The
implementing agencies (state and local regulators) need to be able to determine whether
a tank facility is following the LIST regulations, and vendors  of tank tightness test
methods need to know how to evaluate their systems.

Presently, there are two major categories of tank tightness testing methods on the
market: (a) volumetric testing methods, which measure directly the leak rate in gallons
per hour, and (b) nonvolumetric testing  methods, which report only the qualitative
assessment of leaking or not leaking. These two testing methods require different
testing and statistical  analysis procedures to evaluate their performance.  The protocol  in
this document should be followed when the method is a volumetric one. The evaluation
of the performance of nonvolumetric tank tightness testing methods is treated in a
separate protocol. To simplify the terminology throughout this document, volumetric
tank tightness testing methods are referred to as tank tightness testing  methods.

1.2    OBJECTIVES

The objectives of this protocol are twofold.  First, it provides  a procedure to test tank
tightness testing methods in a consistent and rigorous manner.  Secondly, it allows the
regulated community  and regulators to verify compliance with regulations.

This protocol provides a standard  method that can be used to estimate the performance
of a tank tightness testing method. Tank owners and operators are required to
demonstrate that the method of leak detection they use meets the EPA performance
standards of operating at (no more than) a  5% false alarm rate while having a probability
of detection of (at least) 95% to detect a leak of 0.10 gallon per hour. This
demonstration must be made no later than  December 22,  1990.  The test procedure
described in this protocol is one example of how this level of performance can be
proven.  The test procedure presented here is specific, based on reasonable choices for

-------
a number of factors. Information about other ways to prove performance is provided in
the Foreword of this document.

It should be noted that this protocol does not address the issue of safety testing of
equipment or operating procedure. The vendor is responsible for conducting the testing
necessary to ensure that the equipment is safe for use with the type of product being
tested.

1.3    APPROACH

In general, the protocol calls for using the method on a tight tank and estimating the leak
rate both under no-leak conditions and under induced leak conditions. The leak rate
measured by the method is then compared with the  induced leak rate for each test run.
To estimate the performance of the method, the differences are summarized  and used
with the normal probability model for the measurement errors. The results are
applicable to tanks of the size used in the evaluation or tanks of no more than 25%
greater capacity than the test tank.

The testing also includes conditions designed to check the method's ability to deal with
some of the more important sources of interference. A number of cycles of filling and
partially emptying the tank are incorporated to  test the method's ability to deal with tank
deformation. During some of the cycles of filling and partially emptying the tank, the
product used to refill the tank is  conditioned to  have  a temperature different from that of
the product in the tank. This allows a check on the adequacy of the method's
temperature compensation.  Four different nominal leak rates (including the no-leak
condition) are used. This demonstrates how closely the method can actually measure
leak rates as well as demonstrates the size of the measurement error for a tight tank.
The complete experimental design is given in Section 6 of this document.

1.4    EFFECTS OF HIGH GROUND-WATER LEVEL

The ground-water level is a potentially important variable in tank testing.  Ground-water
levels are above the bottom of the tank at approximately 25% of the tank sites
nationwide, with higher proportions in coastal regions.  Also, tidal effects may cause
fluctuations in the ground-water level during testing in some coastal regions.  If the
ground-water level is above the bottom of the tank, the water pressure on the exterior of
the tank will tend to counteract the product  pressure from the inside of the tank. If the
tank has a leak (hole) below the ground-water  level,  the leak rate in the presence  of the
high ground- water level will be  less than it would be with a lower ground-water level.  In
fact, if the ground-water level is  high enough, water  may intrude into the tank through the
hole.

The means by which the method deals with the ground-water level must be documented.
A method that  does not determine the ground-water level and does not take it into

-------
account is not adequate. If the ground-water level is determined to be above the bottom
of the tank, an acceptable method must include a means of compensating for its
presence. Three possible methods are overfilling, water level monitoring, and testing at
two (or more) filling levels.

Overfilling involves determining the ground-water level and then filling the tank,  perhaps
using extensions to the risers, to produce a net positive  pressure at the bottom of the
tank comparable to the pressure that the normal product volume would produce in the
absence of a high ground-water level in the tank.  Water-level monitoring relies  on
detecting an increase in product level caused by an influx of high ground water  into the
tank.  Finally, some methods have procedures for testing at two different product levels,
on the theory that different estimated leak rates (because of two different differential
pressures) will be observed.  The procedure for testing methods with each of these
approaches is described below.

A method that compensates for high ground-water levels by overfilling the tank to ensure
that there is an outward pressure throughout the tank can be tested  by the procedure
described in this document.  The method should determine the ground-water level at the
test tank and use the appropriate product level.

A method that relies on detecting  an increase in volume or product level from water
incursion for tanks with a high ground-water level can also be tested with this procedure.
The evaluator should determine that the test method does check for the ground-water
level, and should determine that the test method indicates a leak if an increase  in
volume is seen. The evaluator should also determine how the test method ensures that
there will be net flow either into or out of the tank  in the presence of a high ground-water
level. If parts of the tank are  subject to inward pressure and parts to outward pressure
during the test, there might be no net volume change during the test even though the
tank had one or more holes.  If the method ensures a flow by determining the ground-
water level and calculating the pressure, then testing for product loss with a high product
level and for water incursion with a low product level, the test matrix (in Section 6) needs
to be augmented to test at both product levels.

A method that attempts to test in the presence of a high ground-water level by testing at
two different product levels needs to be evaluated using its standard operating
procedure. The evaluator must determine whether the method is sensitive enough to
determine the change  in flow resulting from the different heads.  If the method does not
determine the ground-water level  and ensure that the pressure is in a constant direction
throughout the tank, it  is not clear that the compensation for high ground-water levels is
adequate and the evaluator must test the  method's approach under varying ground-
water conditions (real or simulated).

-------
The evaluation should include all modes of testing that the method uses.  This may
require the evaluating organization to develop and carry out an additional test series if
the tank test method uses a mode of leak detection besides volumetric measurements.
A method for evaluating the operation of a water sensor is described in the EPA
"Standard Test Procedure for Evaluating Automatic Tank Gauging Systems" and is not
repeated here because no volumetric methods are currently known that use water
sensors.

In summary, the evaluating organization should make an  engineering judgment about
the method's approach to adjusting for the ground-water level.  If in doubt, the evaluating
organization may require tests in addition to those detailed in this protocol.

1.5    ORGANIZATION OF THIS DOCUMENT

The next section presents the scope and applications of this protocol. Section 3
presents an overview of the approach, and Section 4 presents a brief discussion of
safety issues. The apparatus and materials needed to conduct the evaluation are
discussed in Section 5.  The step-by-step procedure  is presented in Section 6. Section
7 describes the data analysis, and Section 8 provides some interpretation of the results.
Section 9 describes how the results are to be reported.

Two appendices are included in this document. Definitions of some technical terms are
provided in Appendix A.  Appendix B presents a compendium of reporting forms:  a
standard reporting form for the evaluation results, a standard form for describing the
operation of the tank tightness testing method, data reporting forms, and an individual
test log.  Appendix B thus forms the basis for a standard report.

-------
                                  SECTION 2

                          SCOPE AND APPLICATIONS

This document presents a standard protocol for evaluating volumetric tank tightness
testing methods. The protocol is designed to evaluate methods that test a tank at a
specific point in time by monitoring product volume changes in the tank during the test
period. The protocol is designed to evaluate the method's ability to detect a leak of 0.10
gallon per hour with probability of 95% or higher while operating at a false alarm rate of
5% or less, as specified in the performance standards in the LIST regulations.

Subject to the limitations listed on the Results of U.S. EPA Standard Evaluation form
(see Appendix B), the results of this evaluation can be used to prove that a volumetric
tank tightness testing method meets the requirements of 40 CFR  Part 280,  Subpart D.
The Standard Evaluation Results form lists the test conditions. In particular, the results
reported are applicable for the stabilization times (or longer) used in the tests and for
temperature conditions no more severe than those used in the tests.

-------
                                   SECTION 3

                                   SUMMARY

The evaluation protocol for volumetric test methods calls for conducting the testing on a
tight tank. The organization performing the evaluation should have evidence that the
tank used for testing is tight independent of the system currently being tested.  The
evidence that the tank is tight may consist of any of the following:

   1.   At least three automatic tank gauging system (ATGS) records within a 3-month
       period with inventory and test modes indicating a tight tank.

   2.   A tank tightness test by another test method in the 6 months preceding testing
       that indicates a tight tank.

   3.   Continuous vapor or liquid monitoring system installed that indicates a tight tank.

Any of the above, verified by a tight test result on the initial test (trial run) of the method
under investigation, constitutes acceptable evidence.  This information should be
recorded on the data reporting form (see Appendix B).

The protocol calls for an initial test (trial run) under stable conditions to ensure that the
equipment is working and that there are no problems with the tank, associated piping,
and the test equipment.  If the tank fails the trial run test, however, then testing should
not proceed until the problem is identified and corrected. Only if the evaluating
organization has strong evidence that the tank is tight, should testing proceed.

The tank tightness test device is installed in the test tank and is used to measure a leak
rate under the no-leak condition and with three induced leak rates of 0.05, 0.10, and
0.20 gallon per hour.  A total number of at least 24 tests is to be performed. The tank
must be partially emptied (to half full or less) and then  refilled to the test level for at least
every other test.  When filling the tank to the test level, product at least 5°F warmer than
that in the tank is used for one third of the fillings and  product at least 5°F cooler than
that in the test tank is used for one third of the fillings.  The other third of the fillings uses
product at the same temperature.  The volumetric test method's ability to track actual
volume change is determined by the difference between the volume change rate
measured by the test device and the actual, induced,  volume change rate for each test
run. These differences are then used to calculate the performance of the method.
Performance results are reported on the Results of U.S.  EPA Standard Evaluation form
included in Appendix B of this document.

-------
                                   SECTION 4

                                    SAFETY

This discussion does not purport to address all the safety considerations involved in
evaluating leak detection equipment and methods for underground storage tanks.  The
equipment used should be tested and determined to be safe for the products it is
designed for. Each leak detection method should have a safety protocol as part of its
standard operating procedure. This protocol should specify requirements for safe
installation and use of the device or method. This safety protocol will be supplied by the
vendor to the personnel involved in the evaluation.  In addition, each institution
performing an evaluation of a leak detection device should  have an institutional safety
policy and procedure that will be supplied to personnel on site and  will be followed to
ensure the safety of those performing the evaluation.

Since the evaluations are performed on actual underground storage tanks, the area
around the tanks should be secured.  As a minimum, the following  safety equipment
should be available at the site:

    •   Two class ABC fire extinguishers
    •   One eyewash station (portable)
    •   One container (30 gallons) of spill absorbent
    •   Two "No Smoking" signs

Personnel working at the underground storage tank facility  should wear safety glasses
when working with product and steel-toed shoes when handling heavy pipes or covers.
After the safety equipment has been placed at the site and  before any work can begin,
the area should be secured with signs that read "Authorized Personnel Only" and "Keep
Out."

All safety procedures appropriate for the product in the tanks  should be followed. In
addition, any safety procedures required for a particular set of test  equipment should be
followed.

This test procedure only addresses the issue of the method's ability to detect leaks.  It
does not address testing the equipment for safety hazards. The manufacturer needs to
arrange for other testing for construction standards to ensure that key safety hazards
such as fire,  shock, intrinsic safety, product compatibility, etc., are considered. The
evaluating organization should check to see what safety testing has been done before
the equipment is used for testing to ensure that the test operation will be as safe as
possible.

-------
                                   SECTION 5

                         APPARATUS AND MATERIALS

5.1     TANKS

The evaluation protocol requires the use of an underground storage tank known to be
tight.  A second tank or a tank truck is required to store product for the cycles of
emptying and refilling. As discussed before, the tank should have been tested and
shown to be tight by any of the three methods described in Section 3.  The tank should
not have any history of problems. In addition, the protocol calls for an initial trial run with
the test equipment under stable conditions. This test should indicate  that the tank is
tight; if it does not, there may be a problem with the tank and/or the test equipment that
should be resolved before proceeding with the evaluation.

The tank facility used for testing is required to have at least one monitoring well. The
primary reason for this is to determine the ground-water level.  The presence of a
ground-water level above the bottom of the tank would affect the leak rate in a real tank,
that is, the flow of product through an orifice. The flow would be a function of the
differential pressure between the inside and outside of the tank.  However, in a tight tank
with leaks induced to a controlled container separate from the environment, the ground-
water level will not affect the evaluation testing.  Consequently, it is not necessary to
require that testing against the evaluation protocol be done in a tank entirely above the
ground-water level. The monitoring well can also be used for leak detection at the site,
either through liquid monitoring (if the ground-water level is within 20 feet of the surface)
or for vapor monitoring.

Because performance of internal tank test  methods is generally worse for large tanks,
the size of the test tank is important.  An 8,000-gallon tank is recommended because
this appears to be the most common tank in use. However, testing may be done  in
tanks of any size.  The results of the evaluation would be applicable to all smaller tanks.
The results are also applicable to larger tanks with the restriction that the tanks be no
more than 25% larger in capacity than the test tank.  That is, results from a 6,000-gallon
tank can also be applied to tanks of up to 7,500 gallons in capacity. Results from 8,000-
gallon tanks can be applied to tanks up to 10,000 gallons, those from  10,000 gallons to
up to 12,500 gallons,  etc.  If the method is  intended to test larger tanks, e.g., 20,000
gallons, it must be evaluated  in a tank within 25% of that size.

Because the protocol calls for filling the tank a number of times, a second tank or a tank
truck is needed to hold reserve product. A pump and associated hoses or pipes to
transfer the product from the test tank to the reserve product tank or truck are also
needed.
                                       8

-------
5.2    TEST EQUIPMENT

The equipment for each tank test method will be supplied by the vendor or manufacturer.
Consequently, it will vary by method.  In general, the test equipment will consist of some
method for monitoring product volume or level and for compensating for temperature. It
will also typically include instrumentation for collecting and recording the data and
procedures for using the data to calculate a leak rate and interpret the result as a pass
or fail for the tank.

It is recommended that the test equipment for the method being tested be operated by
trained personnel who regularly use the equipment in commercial tests. This should
ensure that the vendor's equipment is correctly operated and will eliminate problems that
newly trained or untrained individuals might have with the equipment.  On the other
hand, if the equipment is normally operated by the station owner, then the evaluating
organization  should provide  personnel to operate the equipment after the customary
training.

5.3    LEAK SIMULATION EQUIPMENT

The protocol calls for inducing leaks in the tank. The method of inducing the leaks must
be compatible with the leak detection method being  evaluated.  This is done by
removing product from the tank at a constant rate, measuring the amount of product
removed and the time of collection, and calculating the resulting induced leak rate. The
experimental design in Section 6 gives the  nominal leak rates that are to be used.
These leak rates refer to leak rates that would occur under normal tank operating
conditions. Test methods that use increased product head to increase flow rates to
make leaks easier to detect  should be tested with induced leaks at the higher flow rates
that would occur under the test conditions.  An approach to this is described below.

The actual change in the leak rate of a tank in response to a change in pressure is not
known and may vary with tanks.  For the purposes of the evaluation test, assume that
the flow rate  through an orifice is proportional to the square  root of the pressure. To
convert the nominal leak rates to leak rates under increased pressure  head, determine
the depth of  product under the test conditions and form the ratio of the test depth to 7/8
of the tank diameter. Take the square root of this ratio and multiply it by the nominal
leak rate.  The result is the leak rate that would be expected under the test conditions
corresponding to the nominal leak rate under operating conditions. For example, a
method that tests one foot above grade on  an 8-foot diameter tank buried 3 feet would
have a factor of             . This factor will multiply each nominal leak rate.  Methods
that do not overfill the tank are tested with the nominal leak rates.

A method  that has been successfully  used for inducing leaks in previous testing is based
on a peristaltic pump. An explosion-proof motor is used to drive a peristaltic pump head.

-------
The sizes of the pump head and tubing are chosen to provide the desired flow rates.  A
variable speed pump head can be used so that different flow rates can be achieved with
the same equipment. The flow is directed through a rotameter so that the flow can be
monitored and kept constant.  One end of the tubing is inserted into the product in the
tank.  The other end is placed in a container. Typically, volatile products  are collected
into a closed container in an ice bath.  The time of collection is monitored, the amount of
product weighed, and the volume at the temperature of the product in the tank is
determined to obtain the induced leak rate.  While it is not necessary to achieve the
nominal leak rates exactly, the induced leak rates should be within ±30% of the  nominal
rates. The induced leak rates should be carefully determined and recorded. The leak
rates measured by the test method will be compared to the induced leak  rates.  The
method of inducing the leak must be compatible with the leak detection method  under
test.  For example, one (nonvolumetric) tank tightness testing method uses the
characteristic acoustical signal produced when air drawn into a liquid through a  hole in
the tank wall produces a bubble to detect a leak.  Such a signal obviously cannot be
simulated by pumping product out of a tank.

5.4    PRODUCT

The most common products in underground storage tanks are motor fuels, particularly
gasoline and diesel fuel. Analysis of tank test data based on tanks containing a variety
of products has shown no evidence of difference in test results by type of product, if the
same size tank is considered. The only exception to this observation is that one tank
test method did produce better results when testing tanks with pure chemicals (e.g.,
benzene, toluene, xylene) than when testing gasoline. This difference  was attributed to
better test conditions, longer stabilization times, and better cooperation from tank
owners.

Any commercial petroleum product of grade number 2 or lighter may be used for testing,
depending on the availability and restrictions of the test tanks.  The choice of the product
used is left to the evaluating organization, but it must be compatible with the test
equipment.

The test plan requires some testing with addition of product at a different  temperature
from that of the fuel already in the tank. This requirement is to verify that the method
can accommodate the range of temperature conditions that routinely occur.  The
procedure requires that some tests begin by the tank being filled from about half full to
the test level with fuel that is 5°F warmer than the product in the tank, and some tests
using fuel 5°F cooler than the product in the tank. This procedure requires that  some
method of heating and cooling the fuel be provided, such as pumping the fuel through a
heat exchanger, or by placing heating and cooling coils in the supply tank or tank truck
before the fuel is transferred to the test tank.
                                       10

-------
5.5    MISCELLANEOUS EQUIPMENT

As noted, the test procedure requires the partial emptying and filling of the test tank.
One or more fuel pumps of fairly large capacity will be required to accomplish the filling
in a reasonably short time.  Hoses or pipes will also be needed for fuel transfer.  Many
test methods require some reserve fuel for filling a standpipe or topping off a fill pipe. In
addition, containers will be necessary to hold this product as well as that collected from
the induced leaks. A variety of tools need to be on hand for making the necessary
connections of equipment.
                                      11

-------
                                  SECTION 6

                            TESTING PROCEDURE

The overall performance of the method is estimated by a-comparison of the method's
measured (or detected) leak rates and the actual induced leaks. Performance is
measured over a variety of realistic conditions, including temperature changes and filling
effects.  The range of conditions does not represent the most extreme cases that might
be encountered.  Extreme conditions can cause any method to give misleading results.
If the method performs well overall, then  it may be expected to perform well in the field.
The test procedures have been designed so that additional analysis can be done to
determine whether the method's performance is affected by the stabilization time,
different temperature effects, or the size of the leak.

The test procedure introduces three main factors that may influence the test: size of
leak, temperature effects,  and tank deformation.  The primary  consideration is the size of
the leak.  The method is evaluated on its ability to measure or detect leaks of specified
sizes. If a method cannot closely measure a leak rate of 0.10  gallon per hour or if the
method demonstrates excessive variability on a tight tank, then its performance is not
adequate. The ability of the method to track the leak rate can  be compared for the
different leak rates.

The second consideration is the temperature of product added to fill a tank to the level
needed for testing.  Three conditions are used:  added product at the same temperature
as the in-tank product, added product that is warmer than that already in the tank, and
added product that is cooler. The temperature difference is set at 5°F. The difference
should be at least 5°F and should be measured and reported to the nearest degree F.
The temperature difference is needed to  ensure that the method can adequately test
under realistic conditions.  The performance under the three temperature conditions can
be compared to determine whether these temperature conditions have an effect on the
performance.

The third consideration is the tank deformation caused by pressure changes that are
associated with product level changes. This consideration is addressed by requiring
several empty-fill cycles. One test is conducted at the minimum stabilization time
specified by the test method. A second test follows to test without any change in
conditions (except possibly leak rate). Comparison of the order of the test pairs can
determine whether the additional stabilization improves the performance. The actual
times between completing the fills and starting the tests are recorded and reported.

In addition to these factors, environmental data are recorded to document the testing
conditions. These data may explain one or more anomalous test results.
                                       12

-------
6.1     ENVIRONMENTAL DATA RECORDS

In general, the evaluation protocol requires that the conditions during the evaluation be
recorded.  In addition to all the testing conditions, the following measures should be
reported (see the Individual Test Log form in Appendix B):

          ambient temperature, monitored hourly throughout each test
          barometric pressure, monitored hourly throughout each test
          weather conditions such as wind speed; sunny, cloudy, or partially cloudy
          sky; rain;  snow; etc.
          ground-water level if above bottom of tank
          any special conditions that might influence the results

Both normal and "unacceptable" test conditions for each method should be described in
the operating  manual for each method and should provide a reference against which the
existing test conditions can be compared.  The evaluation should not be done under
conditions outside the vendor's recommended operating conditions.

Pertaining to the tank and the product, the following items should be recorded on the
Individual Test Log (see Appendix B):

          type of product in tank
          tank volume
          tank dimensions and type
          amount of water in tank (before and after each test)
          temperature of product in tank before filling
          temperature of product added each time the tank is filled
          temperature of product in tank immediately after filling
          temperature of product in tank at start of test
          If tank is overfilled, height of product above tank.

6.2     INDUCED LEAK RATES AND TEMPERATURE DIFFERENTIALS

Following a trial  run in the tight tank, 24 tests will be performed according to the
experimental design exemplified in Table 1.  In Table 1, LRj denote the nominal leak
rates and Tj denote the temperature differential conditions to be used in the testing.
These 24 tests evaluate the method under a variety of conditions.
                                      13

-------
         Table 1. LEAK RATE AND TEMPERATURE DIFFERENTIAL
                           TEST SCHEDULE

Trial run
Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Empty/Fill cycle


Test
No.
-

1
2

3
4

5
6

7
8

9
10

11
12

13
14

15
16

17
18

19
20

21
22

23
24
_ . Nominal leak
Set . . „
rate (gallon per
hour)
-

1
1

2
2

3
3

4
4

5
5

6
6

7
7

8
8

9
9

10
10

11
11

12
12
0

LR2
LRi

LR3
LR2

LR,
LR4

LR3
LR,

LR2
LR4

LR4
LR,

LR,
LR4

LRi
LR2

LR3
LR2

LR4
LR3

LR2
LR3

LR4
LR3
Nominal
temperature
differential*
(degree F)
0

T2
T2

T3
T3

T3
T3

Ti
Ti

Ti
Ti

T3
T3

T2
T2

Ti
Ti

T2
T2

T2
T2

T3
T3

Ti
Ti
Note:  The temperature differential is calculated as the temperature of the product
      added minus the temperature of the product in the tank.
                                 14

-------
Leak Rates

The following four nominal leak rates will be induced during the procedure:

               English units         Metric units
             (gallon per hour)   (milliliters per minute)
                   0.00                0.00
                   0.05                3.2
                   0.10                6.3
                   0.20                12.6


Temperature Differentials

In addition, three nominal temperature differentials between the temperature of the
product to be added and the temperature of the product in the tank during each fill cycle
will be used. These three temperature differentials are -5°, 0°, and +5°F (-2.8°, 0°, and
+2.8°C).

Randomization

A total of 24 tests consisting of duplicates of the  12 combinations of the four leak rates
(|_R!, LR2, LR3, and LR4) and the three temperature differentials (Ti, T2, and T3) will be
performed. The 24 tests have been arranged in  pairs  (sets), each pair consisting of two
tests performed at the same temperature differential.  However, the leak rates within a
pair have been randomly assigned to the first or  second position in the testing order. An
example test schedule is outlined in Table 1 above.

The randomization of the tests is achieved by randomly assigning the nominal leak rates
of 0,  0.05, 0.10, and 0.20 gallon per hour to I_R!, LR2,  LR3, and LR4 and by randomly
assigning the nominal temperature differentials of 0°, -5°, and +5°F to TI, T2, and T3,
following the sequence of 24 tests as shown  in Table 1. The organization performing the
evaluation is responsible for randomly assigning the four leak rates to I_R!, LR2,  LR3, and
LR4and the three temperature conditions to T^ T2, and T3. The results of the
randomized sequence should be kept blind to the vendor. That is, the vendor should not
know which induced leak rate is used or which temperature condition is present in
advance. The vendor should measure the leak rate and compensate for temperature
based on his instrumentation and standard operating procedure without knowledge of
the induced conditions.  Randomization should be done separately for each method
evaluated.  The order of performing  set numbers should also be randomized or varied as
needed for practicality and kept blind to the vendor.

Each test set consists of two tests performed using two induced leak rates and one
induced temperature differential (temperature of product to be added - temperature of
product in tank).  Each set indicates the sequence in which the product volumes (in
                                       15

-------
gallons per hour) will be removed from the tank at a given product temperature
differential.

Note: The tests are given in pairs for economy.  An acceptable alternative is to fill and
empty the tank with product at the indicated temperature difference before each of the
24 tests.  A minimum of 12 empty/fill cycles must be done, with a minimum of 24 tests.

Notational Conventions

The nominal leak rates to be induced, that is 0, 0.05, 0.10, and 0.20 gallon per hour,
after randomizing the order, are denoted by LR!,  LR2, LR3, and LR4. It is clear that these
figures cannot be achieved exactly in the field.  Rather, these numbers are targets that
should be achieved within ±30%.

The leak rates actually induced for each of the 24 tests will be measured during each
test.  They will be denoted by 81,  S2, ..., S24. These are the leak rates against which the
leak rates obtained by  the vendors performing their tests will be compared.

The leak rates measured by the vendor's equipment during each of the 24 tests will be
denoted by LR^ LR2, ..., L24 and correspond to the induced leak rates SL S2, ...,S24.

The subscripts 1, ..., 24 correspond to the order in which the tests were performed (see
Table 1). That is, for example, S5 and L5 correspond to the test results from the fifth test
in the test sequence.

Optional Experimental Design

The experimental design given in Table 1 is one example of a possible design. It has
been set up so that each leak rate occurs twice with each temperature differential
condition. In addition,  each leak rate occurs once as the first  of a pair and once as the
second.  Maintaining this balance is desirable for the supplemental data calculations.
However, it is not necessary for different leak rates to be used in each pair. In fact, from
a randomization point of view it is desirable if some pairs include the same leak rates.  It
would also aid in keeping the experimental conditions, especially the induced leak rates,
blind to the tester.

As an illustration of other randomization designs, consider the arrangement in Table 2
below. In Table 2, the combinations of leak rate with one temperature condition, say T^
are shown. Each option column indicates a different possible arrangement of the leak
rates that maintains the property that each leak rate appears once as the first of a pair
and once as the second, while also appearing exactly twice with the temperature
condition. The first option includes four pairs, each with different leak rates; the second
option includes one pair in which both tests use the same leak rate; the third option
includes two pairs in which both tests use the same leak rate; and the fourth option
includes all pairs with the same leak rate for each member of the pair.  For
                                       16

-------
randomization, one of the four possible options in Table 2 can be selected at random for
each temperature condition. The four leak rates can be assigned at random to LR!, LR2,
LR3, and LR4 in each temperature condition. Finally, the four pairs of tests for each
temperature condition can be interspersed in random order with the pairs from the other
two temperature conditions.

       Table 2. OPTIONAL EXPERIMENTAL DESIGN FOR RANDOMIZATION
Pair No.
1
1
2
2
3
3
4
4
Nominal Nominal
temperature leak rate
differential Option 1
TI LR2
T! LR,
Ti LR4
Ti LR3
Ti LR2
T! LR4
T! LR,
Ti LR3
Nominal
leak rate
Option 2
LR,
LR,
LR4
LR3
LR3
LR2
LR2
LR4
Nominal
leak rate
Option 3
LR,
LR,
LR4
LR4
LR2
LR3
LR3
LR2
Nominal
leak rate
Option 4
LR,
LR,
LR2
LR2
LR3
LR3
LR4
LR4
6.3    TESTING SCHEDULE

The first test to be done is a trial run.  This test should be done with a tight tank in a
stable condition and this should be known to the vendor. The results of the trial run will
be reported along with the other data, but are not explicitly used in the calculations
estimating the performance of the method.

There are two purposes to this trial run. One is to allow the vendor to check out the tank
testing equipment before starting the evaluation. As part of this check, any faulty
equipment should be identified and repaired.  A second part is to ensure that there are
no problems with the tank and the test equipment.  Such practical field problems as
loose risers, leaky valves, leaks in plumbers plugs, etc., should be identified and
corrected with this trial run. The results also provide current verification that the tank is
tight and so provide a baseline for the induced leak rates to be run in the later part of the
evaluation.

The testing will be performed  using a randomized arrangement of nominal leak rates and
temperature differentials as listed in Table 1 above. The time lapse between the two
                                       17

-------
tests in each set should be kept as short as practical. It should not exceed 30 minutes
and preferably should be held to 15 minutes or less. The date and time of starting each
test are to be reported on the test log. Twelve sets of two tests each will be carried out.
After each set of two tests, the test procedure starts anew with emptying the tank to half
full, refilling, stabilizing, etc. The details of the testing schedule are presented next.

Step 1:       Randomly assign the nominal leak rates of 0, 0.05, 0.10, and 0.20 gallon
              per hour to LR!, LR2, LR3, and LR4.  Also, randomly assign the
              temperature differentials of 0°, -5°, and +5°F to T^ T2, and T3. This will
              be done by the organization performing the evaluation and needs to be
              kept blind to the crew performing the testing.

Step 2:       Follow the vendor's instructions and install the leak simulation equipment
              in the tank if this has not already  been done, making sure that the leak
              simulation equipment will not interfere with the test equipment.

Step 3:       Trial run.  Following the test method's standard operating procedure, fill
              the tank to the recommended level, and allow for the stabilization period
              called for by the method or longer. The product added should be at the
              same temperature as that of the in-tank product. Conduct a test on the
              tight tank to check out the system (tank, plumbing, etc.) and/or the
              method. Perform any necessary repairs or modifications identified by the
              trial run.

Step 4:       Empty the tank to half full. Fill with product at the recommended
              temperature. The temperature differential will be T2 (Table 1, Test No. 1).
              Record the date and time at the completion of the fill.  Allow for the
              recommended stabilization period, but not longer.

Step 5:       Continue with the method's standard operating procedure and conduct a
              test on the tank, using the method's recommended test duration.  Record
              the date and time of starting the test. This test will be  performed under
              the first nominal leak rate, LR2, of the first set in Table 1.

When the first test is complete, determine and record the actual induced leak rate, 81,
and the method's measured leak rate, LI. If possible, also  record the data used to
calculate the leak rate and the method of calculation. Save all data sheets,  computer
printouts, and calculations.  Record the dates and times at which the test began and
ended. Also record the length of the stabilization period. The Individual Test Log form in
Appendix B is provided for the purpose of reporting these data and the  environmental
conditions for each test.

Record the temperature of the product in the test tank and that of the product added to
fill the test tank. After the product has been added to fill the test tank, record the
                                       18

-------
average temperature in the test tank. Measuring the temperature of the product in the
tank is not a trivial task. One suggested way to measure the temperature of the product
in the tank is to use a probe with five temperature sensors spaced to cover the diameter
of the tank. The probe is inserted in the tank (or installed permanently), and the
temperature readings of those sensors in the liquid are used to obtain an average
temperature of the product. The temperature sensors can be spaced to represent equal
volumes or the temperatures  can be weighted with the volume each represents to obtain
an average temperature for the tank.

Step 6:      Change the nominal leak rate to the second in the first set, that is I_R!
             (see Table 1).  Repeat Step 5.  Note that there will be an additional
             period (the time taken by the first test and the set-up time for the second
             test) during which the tank may have stabilized.  When the second test of
             the first set is complete, again record all results (times and dates,
             measured and induced leak rates, temperatures, calculations, etc.).

Step 7:      Repeat Step 4.  The temperature differential will be changed to T3.

Step 8:      Change the nominal leak rate to the first in the second set, that is LR3.
             Repeat Step 5.  Record all results.

Step 9:      Change the nominal leak rate to the second in the second set, that is LR2.
             Repeat Step 6.  Record all results.

Step 10:     Repeat Step 4.  The temperature differential will be changed to the
             following one in Table 1.  In this case, it remains unchanged at T3.

Step 11:     Repeat Steps 5 through 9, using each of the two nominal leak rates of
             the third set, in the order given in Table 1.

Steps 4 through 9, which correspond to two empty/fill cycles and two sets of two tests,
will be repeated until all 24 tests are performed.

6.4    TESTING PROBLEMS AND SOLUTIONS

Inevitably, some test runs will be inconclusive due to broken equipment, spilling of
product used to measure the  induced leak rate, or other events that have interrupted the
testing procedure. It is assumed that, in practice, the field personnel would  be able to
judge whether a test result is  valid. Should a run be judged invalid during testing, then
the following rules should apply.

Rule No. 1   The total number of tests must be at least 24. That is, if a test is invalid, it
             needs to be rerun.  Report the test results as invalid together with the
             reason and repeat the test.
                                       19

-------
Rule No. 2   If equipment fails during the first run (first, test of a set of two) and if the
             time needed for fixing the problem(s) is short (less than 20% of the
             stabilization time or less than 1 hour, whichever is greater), then repeat
             that run. Otherwise, repeat the empty/fill cycle, the stabilization period,
             etc. Record all time periods.

             Note: The average stabilization time will be reported on the Results of
             U.S. EPA Standard Evaluation form in Appendix B. If the delay would
             increase this time noticeably, then the  test sequence should be redone.

Rule No. 3   If equipment fails during the second run (after the first run in a set has
             been completed successfully), and if the time needed for fixing the
             problem(s) is less than 4 hours, then repeat the second run.  Otherwise,
             repeat the whole sequence of empty/fill cycle, stabilization, and test at the
             given conditions.
                                       20

-------
                                   SECTION 7

                                CALCULATIONS

From the results obtained after all testing is completed, a series of calculations will be
performed to evaluate the method's performance.

The evaluation of the method as a whole is presented first.  These calculations compare
the method's measured leak rate with the induced leak rate under a variety of
experimental conditions. The probability of false alarm and the probability of detection
are estimated using the difference between these two numbers. If the overall
performance of the tank tightness testing method is satisfactory, analysis and reporting
of results could end at this point.  However, the experimental design has been
constructed so that the effects of stabilization time and temperature can be tested to
provide additional  information to the vendor. This is described in Section 7.5.

After all tests are performed according to the schedule outlined in Section 6, a total of at
least n = 24 pairs (4 leak rates x 3 temperature differentials x 2 testing orders within a
set) of measured leak rates  and  induced leak rates will be available. These data form
the basis for the performance evaluation of the test  method. The measured leak rates
are denoted by Li,...,L24 and the associated induced leak rates by Si, ...,S24. The leak
rates are numbered in chronological order.  Table 3 summarizes the notation used
throughout this protocol.

7.1     BASIC STATISTICS

The n = 24 pairs of data are used to calculate the mean squared error, MSE, the bias, B,
and the variance of the  method as follows.

Mean Squared Error, MSE
where L\ is the measured leak rate obtained from the ith test at the corresponding
induced leak rate, Si, with i=1, ...,24.
                                       21

-------
Table 3. NOTATION SUMMARY
Test
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Nominal
Set temperature
No. differential
(degree F)
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
11
11
12
12
T2
T2
T3
T3
T3
T3
Ti
Ti
Ti
Ti
T3
T3
T2
T2
Ti
Ti
T2
T2
T2
T2
T3
T3
Ti
Ti
Nominal
leak rate
(gallon per
hour)
LR2
LR,
LR3
LR2
LR,
LR4
LR3
LR,
LR2
LR4
LR4
LR,
LR,
LR4
LR,
LR2
LR3
LR2
LR4
LR3
LR2
LR3
LR4
LR3
Induced
leak rate
(gallon per
hour)
Si
S2
S3
S4
S5
S6
S7
S8
S9
Sio
Sn
Sl2
Sl3
Sl4
815
Sl6
817
Sl8
Sl9
S2Q
S2i
S22
S23
S24
Measured
leak rate
(gallon per
hour)
Li
L2
L3
L4
L5
L6
L7
L8
L9
Lio
Ln
Ll2
Us
Lu
Ll5
Us
Ll7
Us
Ll9
L20
L2i
L22
L23
L24
Absolute
leak rate
difference
IL-SI
(gallon
per hour)
di
d2
d3
d4
d5
d6
d7
d8
d9
dio
dn
d12
dia
d14
di5
die
di7
dis
dig
d20
d21
d22
d23
d24
           22

-------
Bias
The bias, B, is the average difference between measured and induced leak rates over
the number of tests. It is a measure of the accuracy of the test method and can be
either positive or negative.

Variance and Standard Deviation

The variance is obtained as follows:
Denote by SD the square root of the variance.  This is the standard deviation.

NOTE: It is recommended that the differences between the measured and induced leak
rates be plotted against the time or the order in which they were performed. This would
allow one to detect any patterns that might exist indicating possibly larger differences in
the first test of each set or among the three temperature differentials. This could
suggest that the method calls for an inadequate stabilization time after filling or that the
method does  not properly compensate for temperature differences between in-tank
product and product to be added. (See Sections 7.5.3 and 7.5.4 for appropriate
statistical tests.)

Test for Zero Bias

To test whether the method is accurate-that is, the bias is zero-the following test on the
bias calculated above is performed.

Compute the  t-statistic
From the t-table in Appendix A, obtain the critical value corresponding to a t with (24-1) =
23 degrees of freedom and a two-sided 5% significance level.  This value is 2.07. Note:
If more tests are done, replace 24 with the number of tests, n, throughout. A  larger
number of tests will change the t-value.

Compare the absolute value of tB, abs(tB), to 2.07 (or to the appropriate t-value if more
than 24 tests were done). If abs(tB) is less than 2.07, conclude that the bias is not
                                       23

-------
statistically different from zero, that is, the bias is negligible. Otherwise, conclude that
the bias is statistically significant.

The effect of a statistically significant bias on the calculations of the probability of false
alarm and the probability of detection is clearly visible when comparing Figures A-1 and
A-2 in Appendix A.

7.2    FALSE ALARM  RATE, P(FA)

The normal probability model is assumed for the errors in the measured leak rates.
Using this model together with the statistics estimated above, allows for the calculation
of the predicted false alarm rate and the probability of detection of a leak of 0.10 gallon
per hour.

The vendor will supply the criterion for interpreting the results of his test method.
Typically, the leak rate measured by the method is compared to a threshold and the
results interpreted as indicating a leak if the measured leak rate exceeds the threshold.
Denote the method's criterion or threshold  by C. The false alarm  rate or probability of
false alarm, P(FA), is the probability that the measured leak rate exceeds the threshold
C when the tank is tight.  Note that by convention, all leak rates representing volume
losses from the tank are treated as positive.

P(FA) is calculated by one of two methods, depending on whether the bias is statistically
significantly different from zero.

7.2.1   False Alarm Rate With Negligible Bias

In the case of a nonsignificant bias (Section 7.1), compute the t-statistic
where SD is the standard deviation calculated above and C is the method's threshold.
Using the notational convention for leak rates, C is positive. P(FA) is then obtained from
the t-table, using 23 degrees of freedom.  P(FA) is the area under the curve to the right
of the calculated value ti.

In general, t-tables are constructed to give a percentile, ta, corresponding to a given
number of degrees of freedom, df, and a preassigned area, a or alpha, under the curve,
to the right of ta (see  Figure 1  below and Table A-1 in Appendix A).  For example, with 23
degrees of freedom and a = 0.05 (equivalent to a P(FA) of 5%), ta = 1.714.
                                        24

-------
                   Figure 1.  Student's t-Distribution Function.

In our case, however, we need to determine the area under the curve to the right of the
calculated percentile, t^ with a given number of degrees of freedom. This can be done
by interpolating between the two areas corresponding to the two percentiles in Table A-1
on either side of the calculated statistic, t]. The approach is illustrated next.

Suppose that the calculated ti = 1.85 and has 23 degrees of freedom.  From Table A-1,
Appendix A, obtain the following percentiles at df = 23:
                    1.714
                    1.85
                    2.069
a (alpha)
0.05
X to be determined
0.025
Calculate X by linearly interpolating between 1.714 and 2.069 corresponding to 0.05 and
0.025, respectively.
Thus the probability of false alarm corresponding to a ti of 1.85 would be 4%.

A more accurate approach would be to use a statistical software package (e.g., SAS or
SYSTAT) to calculate the probability.  Another method would be to use a nomograph of
Student's t such as the one given by Lloyd S. Nelson in Technical Aids, 1986, American
Society for Quality Control.

7.2 2   False Alarm Rate With Significant Bias   .

The computations are  similar to those in the case of a nonsignificant bias with the
exception that the bias is included in the calculations, as shown next. Compute the t-
statistic
                                       25

-------
P(FA) is then obtained by interpolating from the t-table, using 23 degrees of freedom.
P(FA)is the area under the curve to the right of the calculated value t2. (Recall that C is
positive, but the bias could be either positive or negative.)

7.3    PROBABILITY OF DETECTING A LEAK RATE OF 0.10 GALLON PER HOUR,
       P(D)

The probability of detecting a leak rate of 0.10 gallon per hour, P(D), is the probability
that the measured leak rate exceeds C when the true mean leak rate is 0.10 gallon per
hour.  As for P(FA), one of two methods is used in the computation of P(D), depending
on whether the bias is statistically significantly different from zero.

7.3.1   P(D) With Negligible Bias

In the case of a nonsignificant bias-that is, the bias  is zero-compute the t-statistic
Next, using the t-table at 23 degrees of freedom, determine the area under the curve to
the right of t3. The resulting number will be P(D).

7.3.2   P(D) With Significant Bias

The procedure is similar to the one just described, except that B is introduced in the
calculations as shown below. Compute the t-statistic
Next, using the t-table at 23 degrees of freedom, determine the area under the curve to
the right of t4. The resulting number will be P(D).

7.4    OTHER REPORTED CALCULATIONS

This section describes other calculations needed to complete the Results of U.S. EPA
Standard Evaluation form (Appendix B).  Most of these calculations are straightforward
and are described here to provide complete instructions for the use of the results form.

Size of Tank

The evaluation results are applicable to tanks up to 50% larger capacity than the test
tank and to all smaller tanks.  Multiply the volume of the test tank by 1.50.  Round this
number to the nearest 100 gallons and report the result on page 1 of the results form.
                                      26

-------
Maximum Allowable Temperature Difference

Calculate the standard deviation of the 12 temperature differences actually achieved
during testing. Multiply this number by the factor ±1.5 and report the result as the
temperature range on the limitations section of the results form.

The nominal temperature difference of 5°F used in the design was obtained from data
collected on the national survey (Flora, J. D., Jr., and J. E.  Pelkey, "Typical Tank
Testing Conditions," EPA Contract No. 68-01-7383, Work Assignment 22, Task 13,  Final
Report, December 1988). This difference was approximately the standard deviation of
the temperature differences observed in the tank tests conducted during the national
survey.  The factor 1.5 is a combination of two effects. One effect results from scaling
up the standard deviation of the design temperature differences to 5°F.  The second
effect results from using the rule that about 80% of the temperature differences on tank
tests are expected to be within ± 1.282 times the standard deviation.

Average Waiting Time After Filling

Calculate the average of the time intervals between the end of the filling cycle and start
of the test for the 12 tests that started immediately after the specified waiting time.
(Note: if more than 12 tests are done immediately after the filling, use all such tests.
However, do not use the time to the start of the second test in a pair as  this would give a
misleading waiting time.)  Report this average time as the waiting time after adding
product on the results form.  Note: The median may be used as the average instead of
the mean if there are atypical waiting times.

Average Waiting After "Topping Off"

If the  method fills the tank up into the fill pipe,  calculate the average time interval
between the time when the final topping off was completed and the start of the test.
Calculate this average using data from all tests when this step  was performed.  Report
the result on the results form  as the waiting time after "topping  off" to the final testing
level.  If this step is not performed (e.g., for a test with the tank at 95% of capacity),  enter
NA (not applicable) in the appropriate space on the results form. Note:  The median
may be used instead of the mean if there are some atypical waiting times.

Average Data Collection Time  Per Test

Use the duration of the data collection phase of the tests to calculate the average data
collection time for the total number (at least 24) of tests.  Report this time as the average
data collection time per test.
                                       27

-------
Product Level

If all tests are done at the same product level, report that level on the results form. If
testing was done at different levels, report the applicable product level as the acceptable
range (e.g. from 60% to 90% full) used in the testing.

Minimum Total Testing Time

Finally, calculate an average total test time from the test data. This is the time it would
take from the time the test crew arrives at the site until a test is completed, the
equipment dismantled and the tank returned to service. Typically, it will be the time
required for stabilization after the tank is filled, plus the time from initial setup of
equipment through the first test data collection, plus the time required to dismantle the
equipment.  Report this total time lapse on the results form as the minimum time that the
tank can be expected to be out of service for a test of this type.

7.5   SUPPLEMENTAL CALCULATIONS AND DATA ANALYSES (OPTIONAL)

Other information can be obtained from the test data.  This information is not required for
establishing that the method meets the federal EPA performance requirements, but may
be useful to the vendor of the test method.  The calculations described in this section are
therefore optional.  They may be performed and reported to the vendor, but are not
required and are not reported on the results form. These supplemental calculations
include determining a minimum threshold, minimum detectable leak rate, and  relating
the performance to factors such as temperature differential, waiting time,  and relation to
leak size. Such information may be particularly useful to the vendor for future
improvements of his system.

7.5.1  Minimum Threshold

The 24 test results can also be used to determine a threshold to give a specified false
alarm rate of say 5%. This threshold  may not be the same as the threshold, C,
pertaining to the method  as reported by the vendor. Denote by C5% the threshold
corresponding to a P(FA) of 5%.  The following demonstrates the approach for
computing C5%.

Solve the equation
for C5%. If the bias is not statistically significant (Section 7.1), then replace B with 0.
From the t-table (Appendix A) with 23 degrees of freedom obtain the 5th-percent!le. This
value is 1.714. Solving the equation above for C5% yields
                                      28

-------
In the case of a nonsignificant bias, this would be C5% = 1.714 SD.

7.5.2  Minimum Detectable Leak Rate

With the data available from the evaluation, the minimum detectable leak rate, R5%,
corresponding to a probability of detection, P(D), of 95% and the calculated threshold,
C5%, can be calculated by solving the following equation for R5%:
where C5% is the threshold corresponding to a P(FA) of 5%, as calculated in 7.5.1.

At the P(FA) of 5%, solving the equation above is equivalent to solving
                                        or
which, after substituting 1.714 SD for (C5% - B), is equivalent to
Substitute 0 for B in all calculations when the bias is not statistically significant.
Otherwise, use the value of B estimated from the data.

Thus, the minimum detectable leak rate with a probability of detection of 95% is twice
the threshold, C5%, determined to give a false alarm of 5%, minus twice the bias if the
bias is statistically significant.

In summary, based on the 24 pairs of measured and induced leak rates, the minimum
threshold, C5%,  and the minimum detectable leak rate, R5%,  are calculated as shown
below.

If the bias is not statistically significant:

       For a (FA) of 5%                   C5% =  1.714SD

       For a P(D(R)) of 95%               R5% = 2C5%
                                        29

-------
If the bias is statistically significant:

       For a (FA) of 5%                  C5% =  1.714 SD + Bias

       For a P(D(R)) of 95%              R5% = 2C5% - 2Bias

       Remark: Other significance levels can also be used by substituting the
       appropriate values from the statistical table.

7.5.3  Test for Adequacy of Stabilization Period

The performance estimates obtained in Sections 7.2 and 7.3 will indicate whether the
method meets the EPA performance standards. The calculations in this section allow
one to determine whether the method's performance is affected by the additional
stabilization  time the tank has experienced by the start of the second test after each
empty and fill cycle. These tests are designed primarily to help determine why a method
did not meet the performance standard.

The experimental design tests the method under a variety of conditions chosen to be
reasonably representative of actual test conditions. The tests occur in pairs within an
empty-fill cycle. A comparison of the results from the first test of the pair with the second
of that pair allows one to determine if the additional stabilization time improved the
performance. Similarly, comparisons among the tests at each temperature condition
allows one to determine whether the temperature conditions affected the performance.
Finally, the performance under the four induced leak conditions can be compared to
determine whether the method performance varies with leak rate.

The factors can be investigated simultaneously through a statistical technique called
analysis of variance.  The detailed computational formulas for a generalized analysis of
variance are beyond the scope of this protocol.  For users unfamiliar with analysis of
variance, equations to test for the effect of stabilization period and temperature
individually are presented in detail, although the evaluating organization  should feel free
to use the analysis of variance approach to the calculations if they have  the computer
programs and knowledge available.

The procedure outlined in Section 6 allows the amount of time specified  by the
manufacturer for the tank to stabilize after fuel is pumped into the tank prior to the first
test in each set.  Additional stabilization takes place between the first and second tests
of each set.  The actual length of the stabilization period following refueling as well as
the time between tests are recorded for each tank test.  The following statistical test is a
means to detect whether the additional stabilization period for the second test improves
performance. If the stabilization period prior to the first test is too short, then one would
expect larger discrepancies between measured and induced leak rates for the first tests
in a set as compared to those for the second tests.
                                       30

-------
Step 1:        Calculate the absolute value of the 24 differences, dj, i=1, ..., 24, between
              the measured (L) and induced (S) leak rates for all tests (last column in
              Table 3).

Step 2:        Calculate the average of the absolute differences for the first and second
              test in each set separately.

                                                (all odd subscripts)

                                                 (all even subscripts)

Step 3:        Calculate the variances of the absolute  differences from the first and
              second tests in each set separately.
Step 4:       Calculate the pooled standard deviation.
Step 5:       Calculate the t-statistic:
Step 6:       From the t-table in Appendix A, obtain the critical value corresponding to
              a t with (12+12-2) = 22 degrees of freedom and a two-sided 5%
              significance level (a = 0.025 in the table).  This value is 2.074.

Step 7:       Compare the absolute value oft, abs(t),  to 2.074.  If abs(t) is less than
              2.074, conclude that the average difference between measured and
              induced leak rates obtained from the first tests after stabilization is not
              significantly different (at the 5% significance level) from the average
              difference between measured and induced leak rates obtained from the
              second tests after stabilization.  In other words, there has not been a
              significant additional stabilization effect  between the beginning and the
              end of a test.  Otherwise, conclude that the difference is statistically
              significant, that is, the method's performance is different with a longer
              stabilization period.
                                        31

-------
If the results are statistically significant, then the performance of the method is different
for the tests with the additional stabilization period. If the performance is better, that is, if
the absolute differences for the testing with additional stabilization are smaller than those
for the tests with the minimum stabilization period, then the method would show
improved performance if it increased its required stabilization period.  If the method's
overall performance did not meet the EPA performance standard, performance
estimates with the additional stabilization can be calculated using only the 12  test results
with the additional stabilization.  If the estimates obtained by applying the calculations in
Sections 7.2 and 7.3 to the 12 tests with the longer stabilization indicate that the method
does not meet the EPA performance standard but could meet the EPA performance
standard with the additional stabilization,  that finding should be reported. Note that the
evaluation procedure would still  need to conduct a full 24-test series at the longer
stabilization time before claiming to meet the performance standard.

7.5.4 Test for Adequate Temperature Compensation

This section allows one to test whether the method's performance is different  for various
temperature conditions. A total of eight tests will have been performed with each of the
three temperature differentials, T^ T2, and T3 (the nominal values of 0°, -5°, and +5°F
will have been randomly assigned to TI, T2, and T3). The 24 tests have been  ordered by
temperature differential and test number in Table 4 for the example order of sets from
Table 1.  In general, group the tests by temperature condition.

The test results from the three temperature conditions are compared to check the
method's performance in  compensating for temperature differentials.  If the temperature
compensation of the method is adequate, the three groups should give comparable
results.  If temperature compensation is not adequate, results from the conditions with  a
temperature differential will be less reliable than results with no temperature difference.

The following statistical procedure (Bonferroni t-tests) provides a means for testing for
temperature effect on the test  results. With three temperature differentials considered  in
the test schedule, three comparisons will  need to be made: T1 vs. T2,Ti vs. T3, and T2 vs.
T3.
                                       32

-------
Table 4. ORGANIZATION OF DATA TO TEST FOR TEMPERATURE EFFECT
Test
No.
7
8
9
10
15
16
23
24
1
2
13
14
17
18
19
20
3
4
5
6
11
12
21
22
Set No.
4
4
5
5
8
8
12
12
1
1
7
7
9
9
10
10
2
2
3
3
6
6
11
11
Nominal
temperature
differential
(degree F)
Ti
Ti
Ti
Ti
Ti
Ti
Ti
Ti
T2
T2
T2
T2
T2
T2
T2
T2
T3
T3
T3
T3
T3
T3
T3
T3
Absolute leak
rate difference
IL-SI
(gallon per hour)
d7
d8
d9
dio
di5
die
d23
d24
di
d2
di3
du
di7
dis
dig
d2o
d3
d4
d5
d6
du
d12
d21
d22




Group 1







Group 2







Group 3




                           33

-------
Step 1.       Calculate the average of the absolute differences in each group.
Step 2.       Calculate the variance of the absolute differences in each group.
Step 3.       Calculate the pooled variance of Van, Var2, and Var3.
             or
Step 4.       Compute the standard error, SE, of the difference between each pair of
             the means, ML M2, and M3.
             or
Step 5.       Obtain the 95th percentile of the Bonferroni t-statistic with (24-3) = 21
             degrees of freedom and three comparisons. This statistic is t = 2.60.
                                       34

-------
             (Reference:  RuppertG.Miller, Jr., 1981. Simultaneous Statistical
             Inference, Second Edition. Springer-Verlag, New York, New York.)

Step 6.       Compute the critical difference, D, against which each pairwise difference
             between group means will be compared.
Step 7.       Compare the absolute difference of the three pairwise differences with D.
If any difference in group means, in absolute value, exceeds the critical value of SE x
2.60, then conclude that the method's performance is influenced by the temperature
conditions.

If the results are significant, the method's performance is affected by the temperature
conditions. If the overall performance evaluation met the EPA standards, the effect of a
5°F temperature difference on the method does not degrade performance severely.
However, this does not eliminate the possibility that larger differences could give
misleading results.  If the overall performance did not meet the EPA performance
standards, and the temperature effect was significant, then the method needs to improve
its temperature compensation and/or stabilization time in order to meet EPA
performance standards. Again, an evaluation testing the modified method would need to
be conducted to document the performance before the method could claim to meet the
performance standards.
                                       35

-------
                                  SECTION 8

                               INTERPRETATION

The results reported are valid for the experimental conditions during the evaluation,
which have been chosen to represent the most common situations encountered in the
field. These should be typical of most tank testing conditions, but extreme conditions
can occur and might adversely affect the performance of the method. The performance
should be at least as good for tanks smaller than the test tank. However, the
performance evaluation results should only be scaled up to a tank of 25% greater
capacity than the test tank.  It should be emphasized that the performance estimates are
based on average results. An individual test may not do as well.  Some individual tests
may do better. Vendors are encouraged to provide a measure of the precision of each
test, such as a standard error for their calculated leak rate at that site, along with the
leak rate and test results.

The relevant performance measures for proving that a tightness test method meets EPA
standards are the P(FA) and P(D) for a leak rate of 0.10 gallon per hour. The estimated
P(FA) can be compared with the EPA standard of P(FA) not to exceed 5%.  In general, a
lower P(FA) is preferable, since it implies that the chance of mistakenly indicating a leak
on a tight tank is less.  However, reducing the false alarm rate may also reduce the
chance of detecting a leak.  The probability of detection generally increases with  the size
of the leak. The EPA standard specifies that P(D) be at least 95% for a leak of 0.10
gallon per hour.  A higher estimated P(D) means that there is less chance of missing a
small leak.

If the estimated  performance of the method did not meet the EPA performance
requirements, the vendor may want to investigate the conditions  that affected the
performance as  described in Section 7.5, Supplemental Calculations and Data Analyses.
If the stabilization time  or temperature can be shown to affect the performance of the
method, this may suggest ways to improve the method. It may be possible to improve
the performance simply by changing the procedure (e.g., waiting  longer for the tank to
stabilize) or it may be necessary to redesign the hardware. In either case, a new
evaluation with the modified system is necessary to document that the method does
meet the performance standards.

The relationship of performance to test conditions is primarily of interest when the
method did not meet the EPA performance standards. Developing these relationships is
part of the optional or supplementary data analysis that may be useful to the vendor, but
is not of primary interest to many tank owners or operators.
                                      36

-------
                                   SECTION 9

                           REPORTING OF RESULTS

Appendix B is designed to be the framework for a standard report.  There are four parts
to Appendix B, each of which is preceded by instructions for completion. The first part is
the Results of U.S. EPA Standard Evaluation form.  This is basically an executive
summary of the findings. It is designed to be used as a form that would be provided to
each tank owner/operator that uses this method of leak detection. Consequently, it is
quite succinct.  The report should be structured so that this results form can be easily
reproduced for wide distribution.

The second part of the standard report consists of the Description of the Tank Tightness
Testing Method. A description form is included in Appendix B and should be completed
by the evaluating organization assisted by the vendor.

The third part of the standard report contains a Reporting Form for Leak Rate Data, also
described in Appendix B. This table summarizes the test results and contains the
information on  starting dates and times, test duration, leak rate results, etc.

The fourth part of Appendix B contains a blank Individual Test Log. This form  should be
reproduced and used to record data in the field. Copies of the completed daily test logs
are to be included in the standard report.  These serve as the backup data to document
the performance estimates reported.

If the optional calculations described in Section 7.5 are performed, they should be
reported to the vendor. It is suggested that these results be reported in a separate
section of the report, distinct from the standard report.  This would allow a user to
identify the parts of the standard report quickly while still having the supplemental
information available if needed.

The limitations on the results of the evaluation  are to be reported on the Results of U.S.
EPA Standard  Evaluation form.  The intent is to document that the results are valid
under conditions represented by the test conditions.  Section 7.4 describes the summary
of the test conditions that should be reported as limitations on the results form. These
items are also discussed below. The test conditions have been chosen to represent the
majority of testing situations, but do not include the most extreme conditions under which
testing could be done. The test conditions were also selected to be practical and not
impose an undue burden for evaluation on the  test companies.

One practical limitation of the results is the size of the tank.  Volumetric tests generally
perform less well as the size of the tank increases.  Consequently, the results of the
evaluation may be applied to tanks smaller than the test tank. The results may also be
extended to tanks of 25% larger capacity than the test tank. Thus,  if testing is done in a
                                       37

-------
10,000-gallon tank, the results may be extended to tanks up to 12,500 gallons in size. If
a company wants to document that it can test large tanks, the evaluation needs to be
done in a large tank.

A second limitation on the results is the temperature differential between the product
added to the tank and that of the product already in the tank. Often, product must be
added to bring the product level  to the test level required by the method.  The reported
results apply provided the temperature differential is no more than that used in the
evaluation. Testing during the EPA national survey (Flora, J. D., Jr., and J. E. Pelkey,
"Typical Tank Testing Conditions," EPA Contract No. 68-01-7383,  Work Assignment 22,
Task 13, Final Report, December 1988) found that temperature differentials were no
more than 5°F for at least 60% of the tests.  However, it is clear that larger differences
could exist. The evaluation  testing may be done using larger temperature differentials,
reporting those actually used. The results cannot be guaranteed for temperature
differentials larger than those used in the evaluation.

A third limitation  on the results is the stabilization time needed by the method.  The
Individual Test Logs call for recording the actual  stabilization time used during the
testing.  The mean of these  stabilization times is reported.  The results are valid for
stabilization times at  least as long  as those used in the evaluation.  This is viewed as an
important limitation, since shorter stabilization times can adversely affect the
performance. Also, there may be a market pressure to shorten the times in the field. In
practice, many methods may require the tank to  be filled the day before  the test starts,
allowing an overnight stabilization time, resulting in somewhat longer times in the field
than used in the  evaluation. Similarly, the time after "topping off" the tank for a test that
overfills the tank can  be important.  If applicable, this is also reported as a limitation.

The duration of the data collecting phase of the test is another limitation of the method.
If a test shortens the  data collection time and so  collects less data, this may adversely
affect its performance.  As a consequence, the results do not apply if the data collection
time is shortened. This is primarily of concern in documenting that a tank is tight. If
results clearly indicate a leak, this  may sometimes be ascertained  in less time than
needed to document  a tight tank, particularly if the leak rate is large. Thus, while the
false alarm rate may  be larger if the test time is shortened, this is not usually a problem
in that if test results indicate a leak, efforts are usually made to identify and correct the
source of the leak.

The presence of a high ground-water level can interfere with many tank  tests.  The
organization performing the evaluation must consider the method's approach to testing
for and dealing with a high ground-water level. On the basis of the method's approach
to adjusting for high ground-water  levels,  the determination of whether the method can
successfully test in high ground-water level situations made and reported by checking
the appropriate box at the end of the "Limitations on the Results" section of the Results
                                       38

-------
of U.S. EPA Standard Evaluation form. If the method cannot be used in a high ground-
water level situation, then the method must determine the ground-water level and state
that it is not to be used when the ground-water level is above a specified level on the
tank.

The product level in the tank during the test is reported to let the user know at what
levels this method can be used to give a valid test.  Finally, the average time for the total
testing process is reported as a guide for users as what to realistically expect when their
tank is tested with the method.
                                       39

-------
             APPENDIX A




DEFINITIONS AND NOTATIONAL CONVENTIONS
                 A-1

-------
In this protocol leaks are viewed as product lost from the tank. As a convention, leak
rates are positive numbers, representing the amount of product loss per unit time. Thus
a larger leak represents a greater product loss. Parts of the leak detection industry
report volume changes per unit time with the sign indicating whether product is lost from
the tank (negative sign) or is coming into the tank (positive sign).  We emphasize that
here, leaks refer to the direction out of the tank and the rate to the magnitude of the flow.

The performance of a leak detection method is expressed in terms of the false alarm
rate, P(FA), and the probability of detecting a leak of specified size, P(D(R)), where R is
the leak rate. In order to understand these concepts, some explanation is helpful.
Generally, the volumetric leak detection method, either a precision tank test or the leak
test function of an automatic tank gauging system (ATGS), estimates a leak rate. This
calculated rate is compared to a criterion or threshold,  C, determined by the
manufacturer. If the calculated rate is in excess of the  criterion, the tank is declared to
be leaking, otherwise, the tank is called tight.

Figure A-1  represents the process of determining whether a tank is leaking or not. The
curve on the left represents the inherent variability of the measured leak rate on a tight
tank (with zero leak ate).  If the measured leak rate  exceeds C, the tank is declared to
leak, a false alarm.  The chance that this happens is represented  by the shaded area
under the curve to the right of C,  denoted a (alpha).

The variability of the measured leak rates for a tank that is actually leaking at the rate R
is represented by the curve on the right in Figure A-1.  Again, a leak is declared if the
measured rate exceeds the threshold,  C. The probability that the leaking tank is
correctly identified as leaking is the area under the right hand curve to the right of C.
The probability of mistakenly declaring the  leaking tank tight is denoted by p (beta), the
area of the left of C  under the leaking tank  curve.

Changing the criterion, C, changes both a and  p for a fixed leak rate, R. If the leak rate
R is increased, the curve on the right will shift further to the right, decreasing p and
increasing the probability of detection for a fixed criterion, C.  If the precision of a method
is increased, the curve becomes  taller and  narrower, decreasing both a and p, resulting
in improved performance.

A bias is a  consistent error  in one direction. This is illustrated by Figure A-2. In it, both
curves have been shifted to the right by an amount  of bias, B. In this illustration, the bias
indicates a greater leak rate than is actually present (the bias is positive in this case).
This has the effect of increasing the probability of a false alarm, while reducing the
probability of failing  to detect a leak. That is, the probability of detecting a leak of size R
is increased, but so is the chance of a false alarm. A bias toward underestimating the
leak rate would have the opposite effect. That is, it  would decrease both the false alarm
rate and  he probability of detecting  a leak.
                                        A-2

-------
Definitions of some of the terms used throughout the protocol are presented next.
Nominal Leak Rate:
Induced Leak Rate:
Measured Leak Rate:
Critical Level, C:
False Alarm:

Probability of False
Alarm, P(FA):
Probability of
Detection, P(D(R)):
Method Bias, B:
Mean Squared Error,
MSE:

Root Mean Squared
Error, RMSE:

Precision:
Variance:
The set or target leak rate to be achieved as closely as
possible during testing. It is a positive number in gallon per
hour.

The actual leak rate, in gallon per hour, used during testing,
against which the results from a given test device will be
compared.

A positive number,  in gallon per hour, measured by the test
device and indicating the amount of product leaking out of the
tank.  A negative leak rate would indicate that water is leaking
into the tank.

The leak  rate above which a method declares a leak.  It is also
called the threshold of the method.

Declaring that a tank  is leaking when in fact it  is tight.

The probability of declaring a tank leaking when it is tight.  In
statistical terms, this is also  called the Type I error,  and  is
denoted by alpha (a).  It is usually expressed in percent, say,
5%.

The probability of detecting a leak rate of a given size, R, gallon
per hour. In statistical term, it is the power of the test method
and is calculated as one minus beta (P), where beta is the
probability of not detecting (missing) a leak rate R.  Commonly,
the power of a test  is  expressed in percent, say, 95%.

The average difference between measured and induced
(actual) leak rates,  in  gallon per hour. It is an  indication of
whether the test device consistently overestimates (positive
bias) or underestimates (negative bias) the actual leak rate.

An estimate of the overall performance of a test method.
The positive square root of the mean squared error.
A measure of the test method's ability in producing similar
results (i.e., in close agreement) under identical test conditions.
Statistically, the precision of repeated measurements is
expressed as the standard deviation of these measurements.

A measure of the variability of measurements.  It is the square
of the standard deviation.
                                       A-3

-------
Accuracy:             The degree to which the measured leak rate agrees with the
                      induced leak rate on the average.  If a method is accurate, it
                      has a very small or zero bias.

Resolution:           The resolution of a measurement system is the least change in
                      the quantity being  measured which the system is capable of
                      detecting.
                                      A-4

-------
                   Tight Tank
                      Leaking Tank
                                                       Measured Leak Rate, L,
                                                          Gallons Per Hour
          C   =
          a
          P
        1-P
          R
Criterion or Threshold for declaring a leak
(a leak is declared if the measured rate exceeds C)
Probability of False Alarm,  P(FA)
Probability of not detecting a leak rate R
Probability  of detecting a leak rate R, P(D(R))
Leak Rate
Figure A-1.  Distribution of measurement error on a tight and leaking tank.
                                   A-5

-------
                 Tight Tank
Leaking Tank
                  0    B   C
  R R
                                                         Measured Leak Rate, L,
                                                            Gallons Per Hour
            C   =   Criterion or Threshold for declaring a leak
                     (a leak is declared if the measured rate exceeds C)
            a   =   Probability of False Alarm,  P(FA)
            P   =   Probability of not detecting a leak rate R
          1- P   =   Probability of detecting a leak rate R, P(D(R))
            R   =   Leak Rate
            B   =   Bias

Figure A-2.  Distribution of measurement error  on a tight and leaking tank in the
                           case of a positive bias.
                                      A-6

-------
Table A-1.   PERCENTAGE POINTS OF STUDENT'S t-DISTRIBUTION


V
0
df
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
40
60
120
inf.
a =.10
3.078
1.886
1.638
1.333
1.476
1.440
1.415
1.397
1.383
1.372
1.363
1.356
1.350
1.345
1.341
1.337
1.333
1.330
1.328
1.325
1.323
1.321
1.319
1.318
1.316
1.315
1.314
1.313
1.311
1.310
1.303
1.296
1.289
1.282
a =.05
6.314
2.920
2.353
2.132
2.015
1.943
1.895
1.860
1.833
1.812
1.796
1.782
1.771
1.761
1.753
1.746
1.740
1.734
1.729
1.725
1.721
1.717
1.714
1.711
1.708
1.706
1.703
1.701
1.699
1.697
1.684
1.671
1.658
1.645
a =
12
4.
3.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
2.
1.
1.
4
.025
.706
303
182
776
571
447
365
306
262
228
201
179
160
145
131
120
110
101
093
086
080
074
069
064
060
056
052
048
045
042
021
000
980
960

a = .010
31.821
6.965
4.541
3.747
3.365
3.143
2.998
2.896
2.821
2.764
2.718
2.681
2.650
2.624
2.602
2.583
2.567
2.552
2.539
2.528
2.518
2.508
2.500
2.492
2.485
2.479
2.473
2.467
2.462
2.457
2.423
2.390
2.358
2.326
t
a = .005
63.657
9.925
5.841
4.604
4.032
3.707
3.499
3.355
3.250
3.169
3.106
3.055
3.012
2.977
2.947
2.921
2.898
2.878
2.861
2.845
2.831
2.819
2.807
2.797
2.787
2.779
2.771
2.763
2.756
2.750
2.704
2.660
2.617
2.576
                             A-7

-------
   APPENDIX B



REPORTING FORMS
       B-1

-------
Appendix B provides four sets of blank forms. Once filled out, these forms will provide
the framework for a standard report. They consist of the following:

   1.     Results of U.S. EPA Standard Evaluation-Volumetric Tank Tightness
          Testing Method (two pages)

   2.     Description-Volumetric Tank Tightness Testing Method (seven pages)

   3.     Reporting Form for Leak Rate Data-Volumetric Tank Tightness Testing
          Method (two pages)

   4.     Individual Test Log-Volumetric Tank Tightness Testing Method (five
          pages)

Each set of forms is preceded by instructions on how the forms are to be filled out and
by whom.  The following is an overview on various responsibilities.

Who is responsible for filling out which form?

   1.     Results of U.S. EPA Standard Evaluation. The evaluating organization is
          responsible for completing this form at the end of the evaluation.

   2.     Description of Volumetric Tank Tightness Testing Method. The
          evaluating organization assisted by the vendor (or his field crew) will
          complete this form at the end of the evaluation.

   3.     Reporting Form for Leak Rate Data. This form is to be completed by the
          evaluating organization. In general, the  statistician analyzing the data will
          complete this form.  A blank form can be developed on a personal computer,
          the data base for a given evaluation generated, and the two merged on the
          computer.  The form can also be  filled out manually.  The input for that form
          will consist of the field test results recorded by the evaluating organization's
          field crew on the Individual Test Logs (below) and the vendor's test results.

   4.     Individual Test Logs. These forms are to be used and completed by the
          evaluating organization's field  crew.  These forms need to be kept blind to
          the vendor's field crew.  It is recommended to reproduce a sufficient number
          (at least 24 copies) of the blank form provided in this appendix and produce a
          bound notebook for the complete test period.

At the completion of the evaluation, the evaluating organization will collate all the forms
into a single Standard Report in the order listed above.  In those cases where the
evaluating organization performed additional, optional calculations (see Section 7.5 of
the protocol), these results can be attached to the standard report.  There is no reporting
requirement for these calculations,  however.
                                      B-2

-------
Distribution of the Evaluation Test Results

The organization performing the evaluation will prepare a report to the vendor describing
the results of the evaluation. This report consists primarily of the forms in Appendix B.
The first form reports the results of the evaluation. This two-page form is designed to be
distributed widely.  A copy of this two-page form will be supplied to each tank
owner/operator who uses this method of leak detection.  The owner/ operator must
retain a copy of this form as part of his record keeping requirements.  The
owner/operator must also retain copies of each tank test performed at his facility to
document that the tank(s) passed the tightness test.  This two-page form will also be
distributed to regulators who must approve leak detection methods for use in their
jurisdiction.

The complete report, consisting of all the forms in Appendix B, will be submitted by the
evaluating organization to the vendor of the leak detection method.  The vendor may
distribute the complete report to regulators who wish to see the data collected during the
evaluation.  It may  also be distributed to customers of the leak detection method who
want to see  the additional information before  deciding to select a particular leak
detection method.

The optional part of the calculations (Section  7.5), if done, would be reported  by the
evaluating organization to the vendor of the leak detection method.  This is intended
primarily for the vendor's use  in understanding the details of the performance and
perhaps suggesting how to improve the method.  It is left to the vendor whether to
distribute this form, and if so, to whom.

The evaluating organization of the leak detection method provides the report to the
vendor. Distribution of the results to tank owner/ operators and to regulators is the
responsibility of the vendor.
                                       B-3

-------
                     Results of U.S.EPA Standard Evaluation
                   Volumetric Tank Tightness Testing Method

                      Instructions for completing the form

This 2-page form is to be filled out by the evaluating organization upon completion of the
evaluation of the method.  This form will  contain the most important information relative
to the method evaluation.  All items are to be filled out and the appropriate boxes
checked. If a question is not applicable to the method, write 'NA' in the appropriate
space.

This form consists of five main parts. These are:

    1.     Method Description
    2.     Evaluation Results
    3.     Test Conditions
    4.     Limitations on the Results
    5.     Certification of Results

Method Description

Indicate the commercial name of the method, the version, and the name, address, and
telephone number of the vendor. Some  vendors use different versions of their method
when using it with different products or tank sizes. If so, indicate the version used in the
evaluation.  If the vendor is not the party responsible for the development and use of the
method, then indicate the  home office name and address of the responsible party.

Evaluation Results

The method's threshold, C, is supplied by the vendor. This is the criterion for declaring a
tank to be leaking. Typically, a method declares a tank to be  leaking if the measured
leak rate exceeds C.

P(FA) is the probability of false alarm calculated in Section 7.2.  Report P(FA) in percent
rounded to  the nearest whole percent.

P(D) is the  probability of detecting a leak rate of 0.10 gallon per hour and is calculated in
Section 7.3. Report P(D)  in percent rounded to the nearest whole percent.

If the P(FA) calculated in Section 7.2 is 5% or less and if the P(D) calculated in Section
7.3 is 95% or more, then check the 'does' box. Otherwise, check the 'does not' box.

Test Conditions During Evaluation

Insert the information in the blanks provided. The nominal volume of the tank in gallons
is requested as is the tank material,  steel or fiberglass. Also,  give the tank diameter and
                                      B-4

-------
length in inches. Report the product used during the testing. Give the range of
temperature differences actually measured as well as the standard deviation of the
observed temperature differences. Also indicate the level in the tank at which the testing
was done.  Note, if more than one tank, product, or level was used in the testing,
indicate this by a footnote and refer to the data summary form where these should be
documented.

Limitations on the Results

The size (gallons) of the largest tank to which these results can be applied is calculated
as 1.50 times the size (gallons) of the test tank.

The temperature differential, the waiting time after adding the product until testing, the
waiting time after "topping off" (if applicable), the total data collection time,  and the
product level in tank,  should be completed using the results from calculations in Section
7.4.

If the method compensates for ground-water levels above the bottom of a tank, then
check the 'can' box. Otherwise, check the 'cannot' box.  (See Section 1.4.)

Certification of Results

Here, the responsible person at the evaluating  organization provides his/her name and
signature, and the name, address, and telephone number of the organization.
                                       B-5

-------
             Description of Volumetric Tank Tightness Testing Method

                       Instructions for completing the form

 This 7-page form is to be filled out by the evaluating organization with assistance from
 the vendor, upon completion of the evaluation of the method. This form provides
 supporting information on the principles behind the method or on how the equipment
 works.

 To minimize the time to complete this form, the most frequently expected answers to the
 questions have been provided. For those answers that are dependent on site
 conditions, please give answers that apply in "typical" conditions.  Please write in any
 additional information about the testing method that you believe is important.

 There are seven parts to this form.  These are:

    1.      Method Name and Version
    2.      Product
           >  Product type
           >  Product level
    3.      Level Measurement
    4.      Temperature Measurement
    5.      Data Acquisition
    6.      Procedure Information
           >  Waiting times
           >  Test duration
           >  Total time
           >  Identifying and correcting for interfering factors
           >  Interpreting test results
    7.      Exceptions

 Indicate the commercial  name and the version of the method in the first part.

NOTE: The version is provided for methods that use different versions of the equipment
       for different products or tank sizes.

 For the six remaining parts, check all appropriate boxes for each question. Check more
 than one box per question if it applies.  If a box 'Other' is checked, please complete the
 space provided to specify or briefly describe the matter.  If necessary, use all the white
 space next to a question for a description.
                                       B-6

-------
                                     Description
              Volumetric Tank Tightness Testing Method
This section describes briefly the important aspects of the volumetric tank tightness testing method. It is not
intended to provide a thorough description of the principle behind the method or how the equipment works.
Method Name and Version
Product
> Product type
For what products can this method be used? (Check all applicable)
      D gasoline
      D diesel
      D aviation fuel
      D fuel oil #4
      D fuel oil #6
      D solvent
      D waste oil
      D other (list)	
> Product level
What minimum product level is required to conduct a test?
      D above grade
      D within the fill pipe
      D greater than 90% full
      D greater than 50% full
      D other (specify)	
Is a method used to add or withdraw product to maintain a constant level of product?
      D yes
      D no
Volumetric TTT Method - Description                                                   Page 1 of 7

-------
Does the method measure inflow of water as well as loss of product (gallon per hour)?
       D yes
       D no
Does the method detect the presence of water in the bottom of the tank?
       D yes
       D no
Level Measurement
What technique is used to measure changes in product volume?
       D directly measure the volume of product change
       D changes in head pressure
       D changes in buoyancy of a probe
       D mechanical level measure (e.g., ruler, dipstick)
       D changes in capacitance
       D ultrasonic
       D change in level of float (specify principle, e.g., capacitance, magnetostrictive, load cell, etc.)

       D other (describe briefly)	
Temperature Measurement
If product temperature is measured during a test, how many temperature sensors are used?
       D single sensor, without circulation
       D single sensor, with circulation
       D 2-4 sensors
       D 5 or more sensors
       D temperature-averaging probe
If product temperature is measured during a test, what type of temperature sensor is used?
       D resistance temperature detector (RTD)
       D bimetallic strip
       D quartz crystal
       D thermistor
       D other (describe briefly)	
Volumetric TTT Method - Description                                                      Page 2 of 7

-------
If product temperature is not measured during a test, why not?
       D the factor measured for change in level/volume is independent of temperature (e.g. mass)
       D the factor measured for change in level/volume self-compensates for changes in temperature
       D other (explain briefly)	
Data Acquisition
How are the test data acquired and recorded?
       D manually
       D by strip chart
       D by computer
Procedure Information
> Waiting times
What is the minimum waiting period between adding a large volume of product to bring the level to test
requirements and the beginning of the test (e.g., from 50% to 95% capacity)?
       D no waiting period
       D less than 3 hours
       D 3-6 hours
       D 7-12 hours
       D more than 12 hours
       D variable, depending  on tank size, amount added, operator discretion, etc.
What is the minimum waiting period between "topping off" the tank (adding a small amount of product to fine
tune the desired level for testing, e.g., from 2 inches to 5 inches above grade)  and beginning the test?
       D no waiting period
       D less than 1 hour
       D 1-2 hours
       D more than 2 hours
       D variable, depending  on the amount of product added
Volumetric TTT Method - Description                                                      Page 3 of 7

-------
> Test duration
What is the minimum time for collecting data?
      D less than 1 hour
      D 1 hour
      D 2 hours
      D 3 hours
      D 4 hours
      D 5-10 hours
      D more than 10 hours
      D variable
> Total time
What is the total time needed to test with this method?
(setup time plus waiting time plus testing time plus time to return tank to service)
       hours        minutes
What is the sampling frequency for the level and temperature measurements?
       D more than once per second
       D at least once per minute
       D every 1-15 minutes
       D every 16-30 minutes
       D every 31-60 minutes
       D less than once per hour
       D variable

> Identifying  and correcting for interfering factors
How does the method determine the presence and level of the ground water above the bottom of the tank?
       D observation well near tank
       D information from USGS, etc.
       D information from personnel on-site
       D presence of water in the tank
       D other (describe briefly)	
       D level of ground water above bottom of the tank not determined

Volumetric TTT Method - Description                                                        Page 4 of 7

-------
How does the method correct for the interference due to the presence of ground water above the bottom of the
tank?
       D  head pressure increased by raising the level of the product
       D  different head pressures tested and leak rates compared
       D  method tests for changes in water level in tank
       D  other (describe briefly)	
       D  no action

How does the method identify the presence of vapor pockets?
       D  erratic temperature, level, or temperature-compensated volume readings
       D  sudden large changes in readings
       D  statistical analysis of variability of readings
       D  other (describe briefly)	
       D  not identified
       D  not applicable; underfilled test method used

How does the method correct for the presence of vapor pockets?
       D  bleed off vapor and start test over
       D  identify periods of pocket movement and discount data from analysis
       D  other (describe briefly)	
       D  not corrected
       D  not applicable; underfilled test method used

How does the test method determine when tank deformation has stopped following delivery of product?
       D  wait a specified period of time before beginning test
       D  watch the data trends and begin test when decrease in product level has stopped
       D  other (describe briefly)	
       D  no procedure
Are the temperature and level sensors calibrated before each test?
       D yes
       D no
Volumetric TTT Method - Description                                                        Page 5 of 7

-------
If not, how often are the sensors calibrated?
       D weekly
       D monthly
       D yearly or less frequently
       D never

> Interpreting test results
How are level changes converted to volume changes (i.e., how is height-to-volume conversion factor
determined)?
       D actual level changes observed when known volume is added or removed (e.g., liquid, metal bar)
       D theoretical ratio calculated from tank geometry
       D interpolation from tank manufacturer's chart
       D other (describe briefly)	
       D not applicable; volume measured directly

How is the coefficient of thermal expansion (Ce) of the product determined?
       D product sample taken for each test and Ce determined from specific gravity
       D value supplied by vendor of product
       D average value for type of product
       D other (describe briefly)	
How is the leak rate (gallon per hour) calculated?
       D average of subsets of all data collected
       D difference between first and last data collected
       D from data of last	hours of test period
       D from data determined valid by statistical analysis
       D other (describe briefly)	
What threshold value for product volume change (gallon per hour) is used to declare that a tank is leaking?
       D 0.05 gallon per hour
       D 0.10 gallon per hour
       D 0.20 gallon per hour
       D other (list)	
Volumetric TTT Method - Description                                                         Page 6 of 7

-------
Under what conditions are test results considered inconclusive?
       D ground-water level above bottom of tank
       D presence of vapor pockets
       D too much variability in the data (standard deviation beyond a given value)
       D unexplained product volume increase
       D other (describe briefly)	
Exceptions
Are there any conditions under which a test should not be conducted?
       D ground-water level above bottom of tank
       D presence of vapor pockets
       D large difference between ground temperature and delivered product temperature
       D high ambient temperature
       D invalid for some products (specify)	
       D other (describe briefly)	
What are acceptable deviations from the standard testing protocol?
       D none
       D lengthen the duration of test
       D other (describe briefly)	
What elements of the test procedure are determined by testing personnel on-site?
       D waiting period between filling tank and beginning test
       D length of test
       D determination of presence of vapor pockets
       D determination that tank deformation has subsided
       D determination of "outlier" data that may be discarded
       D other (describe  briefly)	
       D none
Volumetric TTT Method - Description                                                        Page 7 of 7

-------
                       Reporting Form for leak Rate Data
                   Volumetric Tank Tightness Testing Method

                      Instructions  for completing the form

This 1- or 2-page form is to be filled out by the evaluating organization upon completion
of the evaluation of the method. A single sheet provides for 24 test results, the minimum
number of tests required in the protocol.  Use as many pages as necessary to
summarize all of the tests attempted.

Indicate the commercial name and the version of the method and the period of
evaluation above the table. The version is provided for methods that use different
versions of the equipment for different products or tank sizes.

In general, the statistician analyzing the data will complete this form.  A blank form can
be developed on a personal computer, the data base for a given evaluation generated,
and the two merged on the computer.  The form can also be filled out manually. The
input for that form  will consist of the field test results recorded by the evaluating
organization's field crew on the Individual Test Logs and the vendor's test results.

The table consists of 11 columns. One line is provided for each test performed during
evaluation of the method.  If a test was invalid or was aborted, the test should be listed
with the appropriate notation (e.g., invalid) on the line.

The Test Number in the first  column refers to the test number from the randomization
design determined according  to the instructions in Section 6 of the protocol.  Since some
changes  to the design might occur during the course of the field testing, the test
numbers  might not always be in sequential order.

Note that the results from the trial run need to be reported here as well.

The following list matches the column input required with its source, for each column in
the table.

   Column No.     Input                                  Source
         1           Test  number or trial run                    Randomization design
         2           Date at completion of last fill               Individual Test Log
         3           Time at completion of last fill               Individual Test Log
         4           Date test began                          Individual Test Log
         5           Time test began                          Individual Test Log
         6           Time test ended                          Individual Test Log
         7           Product temperature differential            Individual Test Log
         8           Nominal leak rate                         Randomization design
         9           Induced leak rate                         Individual Test Log
        10          Measured leak rate                      Vendor's records
        11          Measured minus induced leak rate         By subtraction
                                      B-14

-------
                              Reporting Form for Leak Rate Data
                       Volumetric Tank Tightness Testing Method
      Method Name and Version:	
      Evaluation Period:     from
to
(Dates)

Test No.
Trial Run

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Date at
Completion
of Last Fill
(m/d/y)


























Time at
Completion
of Last Fill
(military)

/
























Data Test
Began
(m/d/y)


























Time Test
Began
(military)


























Time Test
Ended
(military)


























Product
Temperature
Differential
(cleg F)
0

























Nominal
Leak Rate
(gal/h)
0

























Induced
Leak Rate
(gal/h)
0

























Measured
Leak Rate
(gal/h)


























Meas.-lnd.
Leak Rate
(gal/h)

/
























Volumetric TTT Method-Data Reporting Form
                                            Page 1 of 2

-------
                              Reporting Form for Leak Rate Data
                       Volumetric Tank Tightness Testing Method
      Method Name and Version:	
      Evaluation Period:     from
to
(Dates)

Test No.
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Date at
Completion
of Last Fill
(m/d/y)
























Time at
Completion
of Last Fill
(military)
























Data Test
Began
(m/d/y)
























Time Test
Began
(military)
























Time Test
Ended
(military)
























Product
Temperature
Differential
(cleg F)
























Nominal
Leak Rate
(gal/h)
























Induced
Leak Rate
(gal/h)
























Measured
Leak Rate
(gal/h)
























Meas.-lnd.
Leak Rate
(gal/h)
























Volumetric TTT Method-Data Reporting Form
                                            Page 2 of 2

-------
                               Individual Test Log
                   Volumetric Tank Tightness Testing Method

                       Instructions for completing the form

This 5-page test log form is to be filled out by the field crew of the evaluating
organization.  A separate form is to be filled out for each individual test (at least 24). The
information on these forms is to be kept blind to the vendor's crew during the period of
evaluation of their method.

The form consists of nine parts.  These are:

    1.     Header information
    2.     General background information
    3.     Conditions before testing
    4.     Topping off records (if applicable)
    5.     Conditions at beginning of test
    6.     Conditions at completion of testing
    7.     Leak rate data
    8.     Additional comments, if needed
    9.     Induced leak rate data sheets

All items are to be filled out and the appropriate boxes checked.  If a question is not
applicable, then indicate so as "NA".  The following provides guidance on the  use of this
form.

Header Information

The header information is to be repeated on all five pages,  if used.  If a page is not used,
cross it out and initial it.  The field operator from the evaluating organization needs to
print and sign his/her name and note the date of the test on top of each sheet.

The test number is the number obtained from the randomization design.  It is not the
sequential running test number.  If a test needs to be rerun, indicate the test number of
the test being rerun and indicate that on the test log (e.g., Test No.  5 repeat).

General Background Information

Indicate the commercial name of the method.  Include a version identification if the
method uses different versions for different products or tank sizes.  The vendor's
recommended stabilization period has to be obtained from the vendor prior to testing.
This is important since it will impact on the scheduling of the evaluation. All other items
in this section refer to the test tank and product.  Indicate the ground-water level at the
time of the test.

Theoretically,  this  information would remain unchanged for the whole evaluation period.
However, weather conditions could change  and affect the ground-water level. Also, the
evaluating organization could change the test tank.
                                      B-17

-------
Conditions Before Testing

Fill in all the blanks. If the information is obtained by calculation (for example the
amount of water in the tank is obtained from the stick reading and then converted to
volume), this can be done after the test is completed. Indicate the unit of all temperature
measurements by checking the appropriate box.

Topping Off Records (if applicable)

If "topping off" is not part of the procedure, indicate so as "NA".  Fill in all the blanks.

Conditions at Beginning of Test

Indicate the date and time when the vendor began setting up his test equipment.  This is
not the start of the test data collection itself.

The evaluating organization's field crew starts inducing the leak rate and records the
time on pages 4 and 5. All leak simulation data are to be recorded using the form on
pages 4 and 5.

Once the evaluating organization's field crew is ready with the induced leak rate
simulation, and the vendor's crew starts the actual testing, record the date and time that
the vendor's test data collection starts. Also, indicate the product temperature at that
time.  Fill out the weather condition section of the form.  Indicate the nominal leak rate
which is obtained from the randomization design.

Conditions at Completion of Testing

Indicate date and time when the test is completed.

Again,  stick the tank and record the readings and the amount of water in the tank.
Record all weather conditions as requested.

Leak Rate Data

This section is to be filled out by the evaluating organization's statistician or analyst
performing the calculations. This section can therefore be filled out as the evaluation
proceeds or at the end of the evaluation.

The nominal leak rate is obtained from page 2 (Test Conditions at Beginning of Test). It
should be checked against the nominal leak rate in the randomization design by
matching test numbers.

The induced leak rate is obtained by calculation from the data reported by the evaluating
field crew on page 4 (and 5, if needed) of this form. The measured leak rate is that
reported by the vendor's crew for that test.
                                      B-18

-------
The difference is simply calculated by subtracting the induced from the measured leak
rate.

Additional Comments (if needed)

Use this page for any comments (e.g., adverse weather conditions, equipment failure,
reason for invalid test, etc.) pertaining to that test.

Induced Leak Rate Data (pages 4 and 5)

This form is to be filled out by the evaluating organization's field crew.  From the
randomization design, the crew will know the nominal leak rate to be targeted.  The
induced leak rate will be known accurately at the end of the test.  However, the protocol
requires that the induced leak rate be within 30%  of the nominal leak rate.
                                      B-19

-------
Name of Field Operator,
Signature of Field Operator	Test No.	
Date of Test  	
                               Individual Test Log
           Volumetric Tank Tightness Testing Method
Instructions:
Use one  log for each test.
Fill in the blanks and check the boxes, as appropriate.
Keep test log even if test is inconclusive.

General  Background  Information
Method Name and Version	
Product Type 	
Type of Tank  	
Tank Dimensions (nominal)
      Diameter     	inches
      Length       	inches
      Volume      	gallons
Ground-water level	inches above bottom of tank
Recommended stabilization period before test (per vendor SOP)
	hours 	minutes
Conditions Before Testing
Date and time at start of conditioning test tank 	date  	military time
Stick reading before partial emptying of tank
      Product 	inches 	gallons
      Water  	inches 	gallons
Temperature of product in  tank before partial emptying    	°F D or °C D
Stick reading after partial emptying of tank
      Product 	inches 	gallons
Amount of product removed from tank (by subtraction)	gallons
Volumetric TTT Method - Test Log                                               Page 1 of 5

-------
Name of Field Operator,
Signature of Field Operator	Test No.	
Date of Test  	


Conditions  Before Testing (continued)
Stick reading after filling to test level
      Product 	inches 	gallons
      Water 	inches 	gallons
Amount of product added to fill tank (by subtraction) 	gallons
Temperature of product added to fill tank    	°F D or °C D
Temperature of product in tank immediately after filling	°F D or °C D
Date and time at completion of fill 	date   	military time
Topping Off Records (if applicable)
Date and time at completion of topping off	date   	military time
Approximate amount of product added	gallons
If tank overfilled, height of product above tank	inches
Conditions  at Beginning of Test
Date and time vendor began setting up test equipment
      	date   	military time
> Complete induced leak rate data sheet (use attached pages 4 and 5)
Date and time at start of vendor's test data collection
      	date   	military time
Temperature of product in tank at start of test 	°F D or °C D
Weather Conditions at Beginning of Test
      Temperature	°F D or  °C D
      Barometric pressure  	mm Hg D or	in. Hg D
      Wind         None D     Light D      Moderate D         Strong D
      Precipitation  None D     Light D      Moderate D         Heavy
      Sunny D     Partly Cloudy D     Cloudy D
Nominal leak rate 	gallon per hour

Volumetric TTT Method - Test Log                                                 Page 2 of 5

-------
Name of Field Operator,
Signature of Field Operator,
Date of Test
                                Test No.
Conditions at Completion of Testing
Date and time at completion of test data collection
      	date  	military time
Stick reading at completion of test data collection
      Product  	inches 	gallons
      Water 	inches 	gallons
Weather Conditions at End of Test
      Temperature	°F D or °C D
      Barometric pressure 	mm Hg D or	in. Hg D
      Wind        None D      Light  D      Moderate D
      Precipitation  None D      Light  D      Moderate D
      Sunny D    Partly Cloudy D     Cloudy D
                               Strong D
                               Heavy
Date and time test equipment is disassembled (if done for this test) and tank is ready for service
      	date  	military time
Leak Rate Data (not to be filled out by field crew)
Nominal leak rate
Induced leak rate
_gal/h
_gal/h
Leak rate measured by vendor's method
            _gal/h
Difference (measured rate minus induced rate)
                  _gal/h
Additional Comments (Use back of page if needed)
Volumetric TTT Method - Test Log
                                              Page 3 of 5

-------
Name of Field Operator.
Signature of Field Operator,



Date of Test
Test No.
Induced Leak Rate Data Sheet

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Time at
product
collection
(military)
























Amount of
product
collected
(mL)
























Comments (if applicable)
























Volumetric TTT Method - Test Log
               Page 4 of 5

-------
Name of Field Operator,
Signature of Field Operator,



Date of Test
Test No.
Induced Leak Rate Data Sheet

25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Time at
product
collection
(military)
























Amount of
product
collected
(mL)
























Comments (if applicable)
























Volumetric TTT Method - Test Log
               Page 5 of 5

-------
                        Results of U.S. EPA Standard Evaluation

            Volumetric Tank Tightness Testing Method

This form tells whether the tank tightness testing method described below complies with the
performance requirements of the federal underground storage tank regulation. The evaluation was
conducted by the equipment manufacturer or a consultant to the manufacturer according to the U.S.
EPA's "Standard Test Procedure for Evaluating  Leak Detection  Methods: Volumetric Tank Tightness
Testing Methods." The full evaluation report also includes a form describing the method and a form
summarizing the test data.

Tank owners using this leak detection system should keep this form on file to prove compliance with the
federal regulations. Tank owners should check with State and local agencies to make sure this form
satisfies their requirements.

Method Description

Name	

Version number 	

Vendor
             (street address)

      (city)                     (state)              (zip)                (phone)

Evaluation Results

This method, which declares a tank to be leaking when the measured leak rate exceeds the threshold
of	gallon per hour, has a probability of false alarms [P(FA)] of	%.

The corresponding probability of detection [P(D)] of a 0.10 gallon per hour leak is 	%.

Therefore, this method D does  D does not meet the federal performance standards established by
the U.S.  Environmental Protection Agency (0.10 gallon per hour at P(D) of 95% and P(FA) of 5%).

Test Conditions During  Evaluation

The evaluation testing was conducted in a   	gallon D steel D fiberglass tank that was
	inches in diameter and	inches long.

The tests were conducted with the tank 	percent full.

The temperature difference between product added to fill the tank and product already in the tank
ranged from   	°F  to  	°F, with a standard deviation of	°F.

The product used in the evaluation was	.
Volumetric TTT Method - Results Form                                             Page 1 of 2

-------
Volumetric TTT Method	
Version	
Limitations on the Results
The performance estimates above are only valid when:
   •      The method has not been substantially changed.
   •      The vendor's instructions for using the method are followed.
   •      The tank is no larger than  	gallons.
   •      The tank contains a product identified on the method description form.
   •      The tank is at least  	percent full.
   •      The waiting time after adding any substantial amount of product to the tank is at least
          	hours.
   •      The temperature of the added product does not differ more than 	degrees
          Fahrenheit from that already in the tank.
   •      The waiting time between the end of "topping off," if any, and the start of the test data
          collection is at least 	hours.
   •      The total data collection time for the test is at least	hours.
   •      Large vapor pockets are identified and removed (for methods that overfill the tank).
   •      This method D can D cannot be used if the ground-water level is above the bottom of the
          tank.
   •      Other limitations specified by the vendor or determined during testing:
>  Safety disclaimer: This test procedure only addresses the issue of the method's ability to
   detect leaks. It does not test the equipment for safety hazards.
Certification of Results
I certify that the volumetric tank tightness testing method was operated according to the vendor's
instructions. I also certify that the evaluation was performed according to the standard EPA test
procedure for volumetric tank tightness testing methods and that the results presented above are those
obtained during the evaluation.
(printed name)                                       (organization performing evaluation)
(signature)                                           (city, state, zip)
(date)                                               (phone number)
Volumetric TTT Method - Results Form                                                Page 2 of 2

-------