%EPA
United States
Environmental Protection
Agency
Solid Waste And
Emergency Response/
Research and Development
EPAS30AJST-90/OOS
March 1990
Standard Test Procedures
for Evaluating Leak
Detection Methods:
Nonvolumetric Tank Tightness
Testing Methods
Sj?f$f'
/ "'-5 7 Vi^^-* ''Xl35^
" ,";,,,',;' 'y,,ff,," ,t ,,,,, ^ ,,,,,,,}.
"""<"" " ' "^t',^',:;;^' ' ' -'
Printed on Recycled Paper
-------
-------
Standard Test Procedures for
Evaluating Leak Detection Methods:
Nonvolumetric 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 bperators to select a method of leak detection that has been shown to
meet the relevant performance standards. ,
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 Teak detection equip-
ment or service they purchase meets the federal regulatory require-
ments. 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. '
ii.i
-------
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 EPAls 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
iv
-------
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 equip-
ment 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 regula-
tions. 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 volun-
tary 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 evalua-
tions 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 EPA1s
In some cases, a specific leak detection method,may not be ade-
quately 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 accredita-
tion 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 corre-
sponding EPA test procedure. For example, in the case of volu-
metric 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 experi-
mental 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.
vi
-------
ACKNOWLEDGMENTS
This document was written by Jairus D. Flora Jr., Ph.D., Karin M.
Bauer, and H. Kendall Wilcox, Ph.D., for the U.S. Environmental Protec-
tion Agency's Office of Underground Storage Tanks (EPA/OUST) under Con-
tract 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 Yqrk Department of Environmental Conservation
Tom Clark - Minnesota Pollution Control, Agency
Allen Martinets - Texas Water Commission
Bill Seiger-Mary land, Department of Environment ,
* ' ' ' '
American Petroleum Institute ,
Leak Detection Technology Association .
Petroleum Equipment Institute
vii
-------
-------
CONTENTS
Foreword i i i
Acknowledgments....i vi i
1. Introduction....... 1
1.1 Background 1
1.2 Objectives....... 2
1.3 Approach. 2
1.4 Effects of high ground-water level 5
1.5 Organization of this document..... 6
2. Scope and Applications... 7
3. Summary............. 9-
4. Safety............. ..11
5. Apparatus and Materials 13
5.1 Tanks . .......... 13
5.2 Test equipment 14
5.3 Leak simulation equipment 15
5.4 Product.................. 16
5.5 Tracers and carriers.,.. 16
5.6 Water sensor equipment.. 17
5.7 Miscellaneous equipment........................... 17
6. Testing Procedure....' 19
6.1 Environmental data records...... 21
6.2 Induced leak rates and temperature differentials.. 21
6.3 Testing schedule.. 26
6.4 Testing problems and solutions.................... 34
6.5 Method evaluation protocol for water detection.... 35
7. Calculations........................ ............' 37
. 7.1 Estimation of the method's performance
parameters. '.. 37
7.2 Water detection mode.... 40
7.3 Other reported calculations................ 45
7.4 Supplemental calculations and data analyses
(optional) ...............r..... 47
8. Interpretation ...51
8.1 Basic performance estimates.... *.... 51
8.2 Limitations -....,*... 52
v . ' 8.3 Water level detection function.... 52
8;4 Minimum water level change measurement.. *.... 53
8.5 Additional calculations................ 53
9, Reporting of Results...... 55
Appendices
A. Definitions and notatiojial conventions.... 1 A-l
B. Reporting forms ..,.T................. B-l
IX
-------
-------
SECTION 1
INTRODUCTION
1.1 BACKGROUND
The regulations on underground storage tanks (40 CFR Part 280, Sub-
part 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 UST regulations, and vendors of
tank tightness test methods need to know how to evaluate their systems.
/ - .... *
-Bresently, there are two 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
nonvolumetric one. The evaluation of the performance of volumetric tank
tightness testing methods is treated in a separate protocol. To simplify
the terminology throughout this documentj nonvolumetric tank tightness
testing methods are referred to as tank tightness testing methods.
The use of tracers for leak detection purposes is one of the
approaches permitted by the regulations. While the approach has been
classified by some as an external (out-of-tank) method, it has several
characteristics that are common to nonvolumetric internal methods. In
particular, the type and amount of data collected and the statistical
analysis of the data are. nearly identical to those used for other
honvolumetric methods. Also, the tracer is internal to the tank,=r
although the sensors are external to the tank. This protocol includes
Conceivably, a "nonvolumetric method" could utilize some measure of
volume change, but in a qualitative manner.
-------
procedures for determining whether the performance of a method using
tracers meets the performance requirements for tank tightness testing
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 regu-
lators to verify compliance with regulations.
This protocol provides a standard method that can be used to
estimate the performance of a tank tightness test 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 mor,e 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 j.s
specific, based on reasonable choices for a number of factors. Informa-
tion about other ways to prove performance is provided in the Foreword of
this document.
This protocol does not address the issue of safety testing of equip-
ment 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
under no-leak conditions and under induced-leak conditions, producing
leak rates of 0.10 gallon per hour or less. The nonvolumetric test
method being evaluated determines whether the tank is leaking or not
during each test. This reported result is compared with the actual.con-
dition of the tank during testing to estimate the false alarm rate and
probability of detection. Once these probabilities have been estimated,
the estimates are compared with the EPA performance standards to deter-
mine whether the method meets the EPA performance standards.
The companion evaluation protocol for volumetric tank tightness
tests ("Standard Test Procedures for Evaluating Leak Detection Methods:
Volumetric Tank Tightness Testing Methods," March 1990) requires testing
under different conditions that simulate interferences likely to be
encountered in actual test conditions. For volumetric methods these
include adding product at temperatures different from that of the product
in the tank and filling the tank prior to some of the .tests., Such tests
address temperature effects and tank deformation effects that can affect
measurements of level or volume change. If the nonvolumetric method
being tested uses physical principles that might be affected by
-------
temperature or tank'deformation effects, then the test series should
account for these. If the evaluatorrdetermines that the physical princi-
ples of the test are not affected by these variables, then the tempera-
ture; and tank deformation parameters need not be varied during the test
series. Conversely, if the evaluator determines that other sources of
interference (e.g., background vapor concentrations, external acoustical
noise) might affect the performance of the method, then conditions to
test for these effects must be included in the design. For purposes of
illustration, this protocol assumes that temperature and tank deformation
effects are important, unless the evaluator determines otherwise.
__^ Some nonvblumetric test methods use more than one approach to
detecting a leak. In this event, each approach must be tested and
evaluated to determine whether or under what conditions the system meets
rthe EPA performance standards. For example, some nonvolumetric methods
rely on detection of water incursion during the test to detect a leak in
the presence of a high.ground-water level. If this is part of the
standard operating procedure, the water detection sensor needs to be
evaluated as part of the evaluation procedure. In addition to deter-
mining the performance of the water detection sensor as a leak indicator,
the performance parameters (minimum detectable water level and minimum
detectable level change) must be related to the size of the test tank to
determine whether the water detector could sense water incursion at the
rate of 0.10 gallon per hour under the test conditions with a probability
of at least 95%, while operating at a false alarm rate of 5% or less.
That is, each mode of leak detection must be evaluated and compared to
the EPA performance standards. -
It is emphasized that testing must include conditions designed to
test the ability of the method to correctly detect a leak of the speci-
fied size (0.10 gallon per hour) in the presence of sources of interfer-
ence. Sources of interference, such as product temperature changes, that
do not affect the physical principles of operation of a method do not
need to be included in the testing. However, the evaluating organization
must consider what alternative sources of interference might affect the
operation of the method and must include tests to determine whether the
method successfully overcomes these sources of interference. The testing
conditions should be designed to cover the majority of cases; that is,
interference conditions as extreme as would be encountered in approxi-
mately 75% of real world tests. Testing need not include extreme cases
that are rarely encountered.
This document addresses two general types of nonvolumetric tank
tightness testing methods. One type is internal to the tank. A probe
with sensors is placed in the tank and senses whether some physical
characteristic associated with a leak is present. The second type
introduces a tracer material into the tank. The method then detects.
leaks by monitoring the exterior of the tank for the presence of the
tracer. Since the only source of the tracer is from the tank, the
presence or absence of tracer in the external environment is taken to be
conclusive evidence that the tank is either-leaking or tight.
-------
.The technical requirements for the use of tracers are described in
the release detection section of the regulations on vapor monitoring (40
CFR 280.43[e]). The major requirements which must be considered in
evaluating the tracer method are therefore:
1. The backfill where the sampling is conducted must be porous
enough to readily allow diffusion of vapors to the sensor.
2. The tracer must be volatile enough to produce vapor levels which
are detectable by the monitoring device.
3. Ground water, rain, or soil moisture must not interfere with the
operation of the monitor.
4. Background contaminations must not interfere with the detection
of releases from the tank. :
5. The number and positioning of the monitoring wells must be
optimized for the detection of leaks from any part of the
system. .
Although these requirements are for continuous vapor monitoring devices,
they apply to the use of a tracer technique when it is used as a tank
tightness test. Accordingly, the present protocol takes these factors
into account when evaluating tracer techniques.
r ' ''.'. -,- i '','
Two types of tracer techniques have been developed: those which add
tracer to the fuel and can perform a leak test with product in the tank;
and those which place a gas into an empty tank. The former typically
uses halogenated hydrocarbons as the tracer material while the latter may
use sulfur hexafluoride or helium as the tracer material. In both cases,
the tracer 1s placed in the tank and samples are collected outside the
tank. Depending upon the specific method, or variation thereof, the time
to detect a leak may vary from a few minutes to several days. Estimates
of"the leak rate can be obtained from methods which add tracer to the
product, for example, by using a spiked sample to produce a known
concentration which can be compared to the observed concentration of
tracer found at a leaking tank. Methods which use gases in an empty tank
are usually limited to pass/fail conclusions since it is difficult to.
relate the loss of a gas through a hole to an equivalent amount of
product through the same hole. The tracer techniques may also be used to
test the product lines or any other part of the system which is exposed
to the tracer.
The application of a single protocol to the various tracer tech-
niques may present some practical problems. The use of a tracer in an
actual test situation will contaminate the environment with the tracer,
rendering the site unsuitable for replicate testing, at least, for some
period of time. For methods which rely on halogenated compounds, it may
be possible to use several different tracers at the same site. For
methods which rely on a single tracer, the tracer must either be removed
from the site using techniques such as forced ventilation, another site .
-------
must-be selected for the replicate testing tracer, or the replicate tests
must wait until the tracer has dissipated. Since several replications
are required for satisfactory statistical analysis, the procedures can
prove to be cumbersome, > ^
It is recognized that new nonvolumetric methods may be developed
after this document is published. These new methods could be based on
different physical principles from those employed by currently available
methods. The detailed test methods described in this document may not be
entirely appropriate for new methods in that they may not" address these
new approaches. To allow for such contingencies, it will be the respon-
sibility of the evaluating organization to determine whether a new method
,can be evaluated with the current protocol or whether the new method has
aspects that require additional or different testing. In the latter
case, it is the responsibility of the evaluating organization to devise
an appropriate test series and conduct the testing needed to evaluate the
method in a manner such that its performance can be compared to the EPA
performance standards.. See the Foreword for a description of alternative
approaches.
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 approx-
imately 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 take the ground-water level into
account is not adequatei If the ground-water level is determined to be
above t.he bottom of the tank, a method that tests in this situation must
include a means of compensating for the high ground-water level. Accept-
able means of compensating are to either ensure that the tank has an out-
ward pressure.throughout or that the groundwater exerts an inward pres-
sure at all levels in the tank. If an alternative approach to, compensate
ing for ground-water effects is used, the evaluating organization must
perform an engineering evaluation of the approach to ensure that it is
adequate. If in doubt, the evaluating organization may require tests in
addition to those detailed in this document.
-------
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 mate-
rials.needed to conduct the evaluation are discussed iin Section 5. The
step-by-step procedure, adapted for two existing types of nonvolumetric
test methods, is presented in Section 6. Section 7 describes the data
analysis and Section 8 provides some interpretation of results. Sec-
tion 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 com-
pendium of forms: a standard reporting form for the evaluation results,
a standard form for describing the operation of the method, data report-
ing forms, and an individual test Tog. Appendix B thus forms the basis
for a standard evaluation report.
-------
SECTION 2
SCOPE AND APPLICATIONS
/ .' _ .
This document presents a standard protocol for evaluating nonvolu-
metric tank tightness testing methods* The protocol is designed to
evaluate methods that test a tank at a specific point in time. The
methods determine a yes or no answer to the question: "Is the tank leak-
ing?" The nonvolumetric methods currently commercially available use
some physical result from a leaking tank to make this determination.
Some may use more than one characteristic of.a leaking tank in making
their determination. This protocol is designed to evaluate the method's
ability to detect a leak of 0.10 gallon per hour with a probability of at
least 95%, while operating at a false alarm rate of no more than 5%, as
specified in the performance standard in the UST regulations. .
The protocol also provides tests to determine the minimum water
level that the method can detect. In addition, the protocol tests the
ability of the water sensor to measure changes in the water level. These
are evaluated over a range of a few inches in the bottom of the tank. '.
The minimum water level and minimum water level change that the method
can detect are converted to gallons using the geometry of the tank. From
that, the minimum time it would take the sensor to detect a 0.ID-gallon
per hour leak is calculated. These tests are only performed if the
method uses a water sensor to detect leaks in situations such as a high
ground-water level.
The document also presents a protocol for evaluating tracer methods
at actual tank installations. The protocol does not include laboratory
testing of components such as vapor sensors. It is designed to be used
for tracer methods that are applied to a tank at a specific point in
time. . ' ' j '
.Subject to the limitations listed on the Results of U.S. EPA
Standard Evaluation form (Appendix B), the results of this evaluation can
be used to prove that a nonvolumetric tank tightness testing method meets
the requirements of 40 CFR Part 280, Subpart D. The Results of USEPA
Standard Evaluation form lists the limitations on the method. For
example, a minimum time for the test may be required in order for the
physical characteristic of. a leak to be sensed or for the tracer to reach
the sampling ports. The performance results are valid provided the test
,is conducted for at least the specified time.
-------
-------
SECTION 3
- SUWARY
The evaluation protocol for nonvblumetric 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 reported on the data report 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 tank9 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 testing equipment is installed at the tank site
to be tested following the method's standard operating procedure. A
minimum of 21 independent tests of the tank-under the no-leak condition
are performed. The results of these tight tank tests will be used to
estimate the false alarm rate, P(FA). In addition, induced leaks at
rates not to exceed 0.10 gallon per hour are simulated. Again, a minimum
of 21 independent tests are performed with these induced leaks. The.
results of these tests will be used to estimate,the probability of
detecting a leak of the magnitude used, P(D). The simulation condition
(tight tank or induced leaks) is kept blind to the vendor.
-------
-If sources of interference are to be evaluated, test conditions
including these interferences are set up in a balanced experimental
design. The conditions that may interfere with the method are applied to
both tight and induced leak tests. The order of the tests is randomized
to ensure that the conditions are kept blind to the vendor. The order of
both the interfering conditions (if used) and the leak conditions are
randomized. The proportion .of tests under the tight tank condition that
incorrectly indicate a leak is used.to estimate the probability of a
false alarm, while the proportion of induced leak tests correctly iden-
tified is used to estimate the probability of detection. Thus, each per-
formance parameter, P(FA) and P(D), is estimated based on at least
21 tests. . .
For tracer methods, the protocol calls for the use of the method on
a tank environment which is representative of a typical UST installa-
tion. It is not necessary for the tank to be in service to be acceptable
for the evaluation process. The type of backfill around the tank,
however, should be known and should be either sand, pea gravel, crushed
rock, or other material which is commonly used as backfill material. If
the monitoring is conducted in areas other than the backfill, the charv.
acteristies of the soil at the sampling location should also be known.
The testing of a nonvolumetric method based on tracer technology
also involves a minimum of 42 tests. At least 21 tests are done under
the tight tank condition and are used to estimate the probability of a
false alarm. At least 21 tests are done with an induced or simulated
leak and are used to estimate the probability of detection. As before,
if interfering conditions are to be incorporated into the experimental
design, these are established for tests in a random order. To estimate
P(FA), the tracer is introduced into the product in the tank. After
mixing and after the appropriate waiting time determined by the method's
standard operating procedure has elapsed, the sample ports are sampled to
determine if the tracer is detected. False alarms could occur if tracer
is accidentally released during the process of adding it to the product
or mixing it with the product. Consequently, the steps of adding the
tracer and mixing the product in the tank should be repeated for each
tight tank test. " -
For tracer methods, induced leaks are simulated by spiking the soil
with a sample of nonregulated material containing the tracer. For
example, a vegetable'oil containing the tracer at the working concentra-
tion (e.g., 10 ppm) could be used to spike the soil at 0.10 gallon per
hour. This would be continued for the specified test duration and the
results recorded. To keep the process blind to the vendor, randomized
samples of spiking solution, some with and some without; tracer, could be
used and spiking done for each test.
10
-------
SECTION 4
SAFETY
This discussion does not purport to address all the safety consider-
ations 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 detec-
tion 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:
. . -s- 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 con-
struction 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.
11
-------
-------
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 needed to store
product for the cycles of emptying and refilling, if required. As dis-
cussed 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 differ-
ential 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 detec-
tion at the. site, either through liquid monitoring (if the ground-water
level is. within 20 feet of the surface) or for vapor monitoring.
Volumetric methods that measure volume or level changes of liquid
product that occur as a result of a leak generally have worse performance
as the size of the tank increases. However, the tank size does not
affect the performance of existing nonvolumetric test methods to the same
extent, since they are based on different physical principles. Con-
sequently, it is not necessary to restrict the application of these test
results to tanks with a volume equal to, or some arbitrary fraction
larger than, the test tank. The evaluating organization should determine
the appropriate size limit based on their testing, physical principles
involved, and other available data, and state the limit on the results
form (Appendix B). For example, tanks larger than 50,000 gallons have a
different construction and geometry than the standard horizontal cylin-
drical tanks used for tanks up to this size. It may be the tank geometry
and construction that impose limits rather than the size. '
' ' ' ' '' 13 ". ." '
-------
The test plan may require 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. In the case of a tracer or.
acoustical method, the evaluating organization may eliminate the tempera-:
ture and filling conditions if they are not relevant. The total number
of tests to be performed remains the same, however. The temperature and
filling conditions would obviously be inoperative if a gaseous tracer
were to be used in an empty tank.
If the protocol or the method requires that the tank be filled or
emptied 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 pr truck are
also needed.
For tracer methods, the characteristics of a tank ,are less
important. However, the test tank must be tight. The primary purpose of
the tank is to provide an environment which, is representative of typical
tank installations. The tank is important for testing for false
alarms. The procedure of adding and mixing tracer to the product is a
potential source of false alarms from inadvertent release of the tracer
into the environment.
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
the tank for the effect used by the method to indicate a leak. For
tracer methods, the equipment will also include some method for intro-
ducing the tracer(s) into the tank or the backfill. The test equipment
also typically includes instrumentation for collecting and recording the
data and procedures for using the data to 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.
14
-------
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. 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.
For volumetric methods, leak simulation can be accomplished 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. An explosion-proof motor can be 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 cari
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.
Although this leak simulation approach may work for some
nonvolumetric methods, most of these methods will require a method of
simulating leaks that is adapted to their specific principle of opera- .
tion. Examples of leak simulation methods for two nonvolumetric methods
follow. ' *
5.3.1 Leak Simulation Approach for Acoustical Methods
t ' -~'-'..
T-wo methods commercially available at the present time are based on
acoustical signals generated when product flows'through an orifice or
when air is drawn through an orifice or hole in the tank that would allow
it to leak. In order to simulate a leak condition for such a method, an*
orifice must be introduced into the tank so that product or air can flow
through it during the test. A simulator of this type has been developed
and is in the patent process. Its principle is described below. The
size and location in the tank of the orifice must be determined so that
it would represent a leak rate of 0.10 gallon per hour or less if it were
present under norma-l operating conditions in the tank. One approach is
to insert a pipe into the product in the tank through one of the openings
in the top of the tank. The pipe has an orifice of the required size,
allowing product to leak from the tank into the pipe, where it can be
removed and measured. Likewise, if a partial vacuum is applied, air
could be drawn into the tank through the orifice in the pipe. The
orifice in the pipe can be calibrated by allowing product to flow into
the pipe and measuring the flow rate.
15
-------
5.3.2 Leak Simulation Approach for Tracer Methods
Two types of leak simulation equipment are required, depending upon
the type of tracer technique in use. For methods which rely on detecting
the loss from the tank of product containing tracer, the simulation
equipment must be capable of delivering a liquid containing the tracer
into the backfill close to the tank. The rate of delivery is used to
control the volume of product introduced in the backfill. For methods
which rely on detecting the loss of gaseous tracer from the tank, the
simulation equipment must be capable of delivering the tracer gas into
the backfill in known quantities so that the ability of the system to
detect the tracer in the backfill can be evaluated. In either case, the
amount of tracer introduced into the backfill should reflect the amount
that would be released if the tank were leaking at a rate of 0.10 gallon
per hour or less. To do this, the rate of delivery is used to control
the amount of material introduced into the backfill. To simulate a zero
leak rate, the tracer material is introduced into the test tank and mixed
with the product as appropriate. However, a blank spike (without a
tracer) would be introduced into the backfill.
Other nonvolumetric methods may use principles different from those
of the methods in these examples. The evaluating organization will need
to develop .a method of leak simulation that is appropriate for a specific
test method.
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 dif-
ference was attributed to better test conditions, longer stabilization
times, and better cooperation from tank owners. .
, V ; .'
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.
5.5 TRACERS AND CARRIERS
When testing tracer methods, additional considerations apply. While
use of petroleum products spiked with tracer would be ideal, the intro-
duction of regulated products into the ground is prohibited in almost all
situations. Therefore, for test purposes, the carrier used for liquid
tracers should be of some nonregulated liquid such as mineral oil or
vegetable oil. The concentration of tracer can be elevated in the
16
-------
carrier to reduce the actual volume of material to be introduced into the
ground. if ; . t .
Direct injection of the tracer gas diluted in air carl be used to
evaluate methods which rely on the loss of tracer gases from the tank.
The concentrations of tracers injected during the simulation process
should approximate those contained in the tank during an actual test.
5.6 WATER SENSOR EQUIPMENT
_:- The equipment to test the water sensor consists of a vertical cylin-
der with an accurately known (to ±0.001 inch) inside diameter. This
cylinder should be large enough to accommodate the water sensor. Thus,
it should be approximately 4 Cinches in diameter and 8 or more inches
high. The probe is mounted so that the water sensor is in the same rela-
tion to the bottom of the cylinder as it would be to the bottom of a
tank. In addition, a means of repeatedly adding a small measured amount
of water to the cylinder is needed. This can be accomplished by using a-
pipette.
5.7 MISCELLANEOUS EQUIPMENT
As noted, the test procedure may require 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. Some test
methods require some reserve product for calibration or establishing a
specified product level. 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. '
17
-------
-------
SECTION 6
/
TESTING PROCEDURE
The overall performance of the method is estimated by comparing the
method's results, leaking or tight tank, to whether a leak was actually
induced. Performance is measured over a.variety of realistic conditions
including temperature changes and filling effects, if applicable. The
evaluating organization is responsible for adding any other variables
that may affect a specific nonvolumetric method. The range of conditions
need not represent the most extreme cases that might be encountered,
.because extreme conditions can cause any method to give misleading .
results. If the method performs well under various test conditions, then
it may be expected to perform well in the field.
The test procedures have been designed so that additional statisti-
cal analyses can be done to determine whether the method's performance is
affected by the size of the leak or other factors. These additional
analyses can only be done if the method makes a substantial number of
mistakes so that the proportion of errors is between zero and one for
some subsets of the data. Thus, they are only relevant if the method
does not meet the performance standard. ,
For illustrative purposes, the basic 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 detect leaks of
specified sizes. If a method cannot detect a leak rate of 0.10 gallon
per hour or if the method identifies too many leaks when no leak is
induced, then its performance is not adequate.
A second consideration might be the temperature of the product added
to fill a tank to the level needed for testing. Three conditions could
be used: added product at the same temperature as the in-tank product,
'added product that is wanner than that already in the tank, and added
product that is cooler. The temperature difference should be at least
5°F and should be measured and reported to the nearest degree F. For
some methods, 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 method's
performance. Note that some nonvolumetric methods require an empty tank
19
-------
or do not require a "specific product level. If the principle of the
nonvolumetric method is not affected by product temperature as determined
by the evaluating organization, the test need not include this set of
conditions, although the total number of tests must not be decreased.
Another consideration might be the tank deformation caused by pres-
sure 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 time after filling specified by the test
method. A second test follows without any change in conditions (except,
possibly, leak rate). Comparison of the order of the test pairs can
determine whether the additional time improves the method's perfor-
mance. Again, if, as determined by the evaluating organization, the
operating procedure of the method is not affected by pressure changes,
this aspect of testing need not be included.
Nonvolumetric test methods operate on a wide variety of princi-
ples. Consequently, each method may have a different set of sources of
interference related to its operating principle. .The evaluating organi-
zation should consider possible sources of interference for the method...
being evaluated. The list of these sources considered and the conclu-
sions reached should be reported. The considerations do not need to
include the most extreme possible conditions, but should include condi-
tions expected to be encountered in a large majority (e.g., 75%) of the
normal tests cases.
In addition to varying these factors, environmental data are
recorded to document the test conditions. These data may help to explain
one or more anomalous test results. .
The ground-water level is a potentially important variable in tank
testing, and the system's means of dealing with it is to be documented.
A system that does not determine the ground-water level and take it into
account'is not adequate. Ground-water levels are above the bottom of the
tank at approximately 25% of underground storage tank sites -nationwide,
with higher proportions in coastal regions. .
If the method uses water incursion to account for high ground-water
levels, this protocol evaluates two aspects of the system's water sensing
function: the minimum detectable water level and the minimum detectable
change in water level. Together, these can be used with the dimensions
of the tank to determine the ability of the system's water sensing device
to detect inflows of water at various rates.
20
-------
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 condi-
tions, the foilowing-measurements should be reported (see the Individual
Test Log forms 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
When testing tracer methods, the tank environment should also be
documented as completely as possible. A detailed site diagram should be
prepared which identifies the positions of the tanks, piping, and other
features which are present at the site. The type of backfill and soil at
the site should be verified, at the minimum, t.o be porous enough to allow
migration of vapors from the leak to the sensors. The evaluation should
not be run under backfill conditions outside the range suggested by the
vendor.
Both normal,and "unacceptable" test conditions for each method
should bedescribed 1n the, operating manual for the method arid 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 if applicable:
type of product in tank
type of tracer(s) (liquid or gas)
tank volume
tank dimensions and type
amount of water in tank (before and after each test)
if applicable, temperature of product in tank before filling
if applicable, temperature of product added each time the tank
is filled
if applicable, temperature of product in tank immediately after
filling
if applicable, temperature of product in tank at start of test
6.2 INDUCED LEAK RATES AND TEMPERATURE DIFFERENTIALS
Following a trial run in the tight tank, a minimum of 42 tests must
be performed according to an experimental design illustrated in
Table 1. (As _discussed 1n Section 7, a larger number of tests could be
used.) For Illustrative purposes, this table presumes that temperature
and tank deflection effects could interfere with the method.
21
-------
Table 1. LEAK RATE AND TEMPERATURE DIFFERENTIAL
TEST SCHEDULE (Example)
Test
No. | Set No.
Tjrjal fiirt \ - --- "' ' - - -
Nominal
Leak Rate
(gal/h)
-0
Nominal
Temperature
Differential *1
(degree, F)
' 0 - '
Empty/Fill cycle *2
1 1
2 1
LR2
LR1
T3
T3
Empty/Fill cycle
32
4 2
LR1
LR1
T2
T2
Empty/RII cycle .
5 3
6 3
LR1
LR3
T1
T1
Empty/Fill cycle
7 4
8 . 4
LR3
LR1
T3
T3 , .
Empty/RII cycle
9 5
10 5
LR4
LR1
T1
T1 "
Empty/RII cycle
11 6
12 6
LR2
LR3
T2
T2 .
Empty/Fill cycle
13 7
14 7
LR4
LR1
in .,
in '
Empty/RII cycle
15 8
16 8
LR3
LR1
T3
T3
Empty/RII cycle
17 9
18 9
LR4
LR1
T3
T3
Empty/Fill cycle
'
19 10
20 10
LR1
LR3
T2
T2
Empty/Fill cycle
-
21 11
22 11
LR3
LR1
T1
T1
Note 1: The temperature differential is calculated as the temperature of
the product added minus the temperature of the product in the tank.
i ,. ' . * - -
Note 2: Empty/RII cycles and temperature differentials may not be required.
22
-------
Table 1. LEAK RATE AND TEMPERATURE DIFFERENTIAL
TEST SCHEDULE (Example) (Continued)
Test
No. | Set No.
Empty/Fiji cycle *2
23 12
24 12
Nominal
Leak Rate
(gal/h)
LR1
LR2
Nominal
Temperature
Differential *1
(degree F)
T3
T3
Empty/Fill cycle
25 13
26 13
LR2
LR4
T2
T2
Empty/Fill cycle . .
27 14
28 14
LR3
LR1
T3
T3
Empty/Fill cycle
29 15
30 15
LR1
LR2
T1
T1
Empty/Fill cycle
31 16
32 16
LR1
LR1
T2
T2
Empty/Fill cycle
33 1.7
34 17
LR1
LR4
T3
T3
Empty/Fill cycle v
35 '18
36 18
LR1
LR4
T2
T2
Empty/Fill cycle .
* '
37 19
38 19
; LR2
LR1
T1,
T1
Empty/Fill cycle
39 20
40 20.
LR1
LR2
T2
T2
Empty/Fill cycle , ,
41 21
42 21
LR1
LR4
T1
T1
Note 1: The temperature differential is calculated as the temperature of
the product added minus the temperature of the product in the tank.
Note 2: Empty/Fill cycles and temperature differentials may not be required.
23
-------
In Table 1, LR^ denote the nominal leak rates and T^ denote the
temperature differential conditions to be used in the testing. These
42 tests evaluate the method under a variety of conditions.
The 42 tests are arranged in 21 sets of two tests each. Table 1
shows a possible ordering of the 21 sets. In practice, the evaluating
organization should randomly rearrange the order of the sets so that the
leak rates are blind to the vendor.
Leak Rates
Of the 42 tests, half will be performed under tight-tank conditions,
that is, at a leak rate of 0.0 gallon per hour. The remaining 21 tests
will be performed under induced leak conditions with leak rates not
exceeding 0.10 gallon per hour. Typically, all of these.induced leak
rates would be the same.. Alternatively, different non-zero leak rates
could be used and the results analyzed with a logistic model, as
described in Section 7.4.2. The test schedule in Table 1 is an example
of 21 tests at a 0.0 gallon per hour leak rate (LRj) and 3 groups of
7 tests at non-zero leak rates of LR2, LR3, and LR^, which may all be
equal. .
The most direct evaluation of a nonvolumetric method uses only the
zero and 0.10 gallon per hour leak rates. This, assuming that the test
results had at most one error at each leak rate, would provide the needed
performance evaluation. However, a vendor may want to claim that his
method exceeds the EPA performance standards and establish that the prob-
ability of detecting a smaller leak (e.g., 0.01 rather than 0.10 gallon
per hour) is at least 95%. In that case, two approaches are possible.
One is to use the smaller leak rate as the induced leak rate. Again,
this is straightforward. However, if the nominal leak rate selected is
close to or less than the leak rate that the method can actually detect
with 95% reliability, the testing may result in too many detection errors
at that reduced leak rate, in order to demonstrate that the method meets
the performance standards, the 21 induced leak rate tests would have to
be run again using a nominal leak rate larger than the example of
0.01 gallon per hour (e.g., 0.05 gallon per hour), with additional costs
for the evaluation.
.Another approach is to induce three non-zero leak rates and estimate
the probability of detection as a function of the leak rate. In this
case, the method would demonstrate that it meets the EPA performance
standards, provided that the probability of detection at a zero leak rate
(a false alarm) is less than 5%, and the detectable Teak rate that could
be claimed by the method is the leak rate at which the function first
exceeds 95%. If this option is chosen, a single test series of 42 tests
could demonstrate that the method meets the EPA performance standards at
the smaller leak rate determined by the evaluation. In order for this
approach to work, the probability of detecting a leak must increase
steadily with the leak rate. In addition, the non-zero leak rates must
be selected so that the observed results (proportions of tests where a
24
-------
leak-is detected) also increase with the induced leak rate. There must
be very few detections (zero or one) at zero, some missed detections at
the smaller leak rates, and very few at,the larger leak rates.
Temperature Differentials (1f applicable)
If temperature differential is important for the test method, three
nominal temperature differentials between the temperature of the product
t,o .be added and the temperature of the product in the tank during each
fill cycle should be used. These three temperature differentials are
-5°, 0°, and +5°F (-2.8°, 0°, and +2.8°C). The temperature differential
of 5°F is a minimum. Larger differences may be used. If temperature
differences are used, the actual differences are to be calculated and
reported.
Randomization
A total of 42 tests consisting of combinations of the four leak
rates (U^ = 0.0 gallon per hour, LR2, LR3, and LRJ and the three
temperature differentials (Ti, T2, and T3) will be performed. LR2, LR3,
and LR^ may all be the same, depending on the analysis method to be
used. The 42 tests have been arranged 1n 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. The test schedule is
outlined in Table 1.
A randomization of the test schedule is required to ensure that the
testing is done blind to the vendor. The randomization of the tests is
achieved by the evaluating organization by randomly assigning three
nominal leak rates below 0.10 gallon per hour to LR2, LR3, and LRH and by
randomly assigning the nominal temperature differentials of 0°, -5°, and
-+5°F to Tj, T2, and T3, following the sequence of 42 tests as shown in
Table 1. In addition, the evaluating organization should randomly assign
the set numbers (1 through 21) to the 21 pairs of tests. 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 tempera-
ture condition is present in advance. The vendor should test for the
induced leak rate based on his instrumentation and standard operating
procedure without knowledge of the induced conditions. /Randomization
should be done separately for each method evaluated.
In summary, 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 the tank). Each .set
indicates the sequence in which the induced rates are used to remove the
product volumes (in gallons per hour) from the .tank at a given product
temperature differential. In some cases, e.g., when a partial vacuum is
applied to the tank, the simulated leak will not actually remove product
from the tank. In this case, the indicated rates are those at which :
-."' ' " 25 ': . :' '" - '
-------
product would escape" or be removed from the tank if the: induced condition
were present under normal tank operating conditions.
Notational Conventions
The nominal leak rates to be induced are denoted by LRX = 0.0 gallon
per hour, LR2, LR3, and LR,f. It is clear that the nominal leak rates
selected by the evaluating organization cannot be achieved exactly in the
field. .Rather, these numbers are targets that should be established by a
calibration process. The maximum must be no more than 1035 greater than
the nominal 0.10 gallon per hour.
The leak rates actually induced for each of the 42 tests will be
calibrated for each test series. They will be denoted by Slt S?,..., ,
St.,. The results of each test will be denoted by Llt..'.-,L%2» with each
LJ being either "tight" or "leaking." The L1 may be coded numerically,
e.g., LJ - "0" for tight and "1" for leaking, for convenience.
The subscripts 1,...,42 correspond to the order in which the tests
were performed (see Table 1). That is, for example, Ss and L5 correspond
to the test results from the fifth test in the test sequence.
6.3 TESTING SCHEDULE
The first test to be done is a trial run. This test should be done
with a tight tank 1n 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 method's.performance.
There are two purposes to this trial run. One is to allow the
vendor to check out the tank testing equipment before starting the eval-
uation. 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 or the test equipment. Such practical field problems as loose
risers, leaky valves, leaks in plumber's plugs, etc., should be identi-
fied and corrected with this trial run. The results also provide addi-
tional 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 illustrated in
Table 1 above, unless the.evaluating organization determines that the
filling and/or temperature changes are irrelevant for the particular
nonvolumetric method. The time lapse between'the two tests in each set
should be kept as short as practical. It should not exceed 30 min, and
preferably should be held to 15 min or less. Twenty-one sets of two
26
-------
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., as necessary. The details of the testing schedule are
presented next, in accordance with the example ordering shown in Table 1.
Step 1: Randomly assign nominal leak rates not to exceed O-.lO gallon
per hour to LR2, LR3, and LR4. Note that LRj is identified
with the zero leak or tight tank condition as 21 trials are run
in this condition. 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 to 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. Any 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 recom-
mended temperature. The temperature differential will be T3
(Table 1, Test No. 1). Record the date and time at the comple-
tion of the fill. Allow for the recommended stabilization
period, but not longer. Induce the appropriate .leak condition.
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 of the
first set in Table 1. This nominal leak rate to be .induced is
LK2* ,
When the first test is complete, determine and record the calibrated
induced leak rate, Slt and the method's reported leak condition, Llf If
possible, also record the data used to determine the leak condition 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 Individ-
ual 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 (if done; if not, document why
not on the log). 'After the product has been added to fill'the test tank,
'. ,'''' 27 ' ./"-.'
-------
record the average temperature in the test tank. Measuring the tempera-
ture 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 LRX (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, induced
.leak rate and test result, temperatures, calculations, etc.).
Step 7: Repeat Step 4. The temperature differential will be changed to
T2- ' '.-. . ; '
Step 8: Change the nominal leak rate to the first in the second set.
In this example, the rate is unchanged at LRj. Repeat
Step 5. Record all results.
Step 9: Change the nominal leak rate to the second in the second set if
it is different. In this example the second leak rate is
LRx. 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 will be changed
to Tx.
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 42 tests are performed.
Normal and "unacceptable" test conditions for each method should be
described in the owner operating manual for each method and should pro-
vide a reference against which the existing test conditions are com-
pared. The evaluation should not be done under conditions outside the
vendor's recommended operating conditions.
28
-------
6i3.1 Application of the Protocol to Acoustical Methods
One class of commercially available nonvolumetric test methods is
based on acoustical principles. This section describes the application
of the protocol to this type of method. A basic description of the
method is needed to understand the application of the protocol.
Acoustical methods use sensitive hydrophones to detect an acoustical
signal from the tank. This signal is recorded and is analyzed to iden-
tify a specific characteristic associated with a leak. One such method
places the tank under a partial vacuum and investigates the acoustical
signal for a characteristic "bubble" signature .induced when air bubbles
are drawn from outside the tank (in an unobstructed backfill zone) into a
liquid through a hole in the tank. Leaks in the ullage are identified by
a particular frequency or "whistle" of air ingressing into the ullage
space. Another approach analyzes the acoustical signal for a character-
istic sound of fluid flowing out of an orifice in the tank.
While these methods have been called "acoustical," they typically
have additional modes of detecting leaks that are used in conditions of a
high ground-water level. Generally they rely on identification of water
ingress to detect leaks in the presence of a high ground-water level.
The evaluation must test all modes of leak detection used by the method
to "detect leaks from any portion of the tank that normally contains
product." Section 6.5 contains a protocol to evaluate a water sensor
used to detect inflow of water during a test period.
Acoustical methods can be used with a fairly wide range of product
levelsHn the tank. The deformation caused by filling the tank would not
affect these methods, nor would the temperature-'of the product in the
tank. Consequently, the sequence of temperature and filling conditions
does not need to be considered with these tests. The tank should be
filled to a level in the range specified by the method.
To induce a leak for the acoustical methods, it is necessary to use
a device that will create the same signal that a real leak would cre-
ate. One way to do this is to use an orifice-type leak simulator, this
consists of a pipe inserted into the tank through one of the tank open-
ings. The pipe is sealed to the tank. The bottom of the pipe is fitted
with a cap that contains a calibrated orifice to allow product to leak
into the pipe at the desired leak rate under a standard head. This
simulator will work for either type of acoustical signal. Flow of liquid
through the orifice would produce the signal typical of liquid'flow. If
the tank ,1s under partial vacuum, air will be'drawn into the tank through
the orifice below the liquid level and will produce bubbles. % means of
closing the orifice is needed so that a zero leak rate can be induced and
kept blind to the vendor.
Since neither temperature differential nor tank deformation should
affect the acoustical methods, the approach discussed earlier in this
29
-------
subsection is simplified as follows. The steps refer to Table 1, with
the understanding that there are no differences among Tlt T2, T3, and the
partial emptying and refilling is not necessary.
Step 1: Decide whether one or three nonzero leak rates will be used.
(The use of three may allow one to fit a model relating prob-
ability of detection to leak rate, but if this is not important
to the vendor, it is sufficient to use a single non-zero leak
rate (less than or equal to 0.10 gallon per hour), which may be
the preferred approach.)
Step 2: Decide what leak rates will be used. If only a single non-zero
leak rate is used, it can be selected between zero and 0.10 gal-
lon per hour. If the vendor wants to establish a smaller
detectable leak rate, a value of less than 0.10 gallon per hour
may be used. (The risk of doing this is that if the system does
not pass, more.testing with larger leak rates below 0.10 gallon
per hour may be needed.)
Step 3: If only two leak rates (0 and one other) are used, randomly
assign one of them to LRt and the other to all cases where LR2,
LR3, or LRn are listed. If four leak rates are to be used,
assign LRX to zero and randomly assign the other three to LR2,
LR3, and LR^.
Step 4: Randomly rearrange the order of the 21 pairs of tests listed in
Table 1. (This allows for additional randomization and provides
better control on keeping the induced leak rates blind to the
vendor.)
Step 5: Have the'vendor install the test equipment in the tank.
Step 6: Trial run. Following the test method's standaird operating
procedure, fill the tank to the recommended level. Have the
vendor conduct a test with a known zero leak rate and verify
that the equipment has been installed and is functioning cor-
rectly. This also provides confirmation that the tank is still
tight and is compatible with the test method.
Step 7: Induce the leak rate called for in the randomization developed
. above. Have the vendor test the tank with this induced leak
rate and report the results. Record the calibrated induced leak
rate and the vendor's results (tight or leaking). Record the
environmental conditions data and other ancillary data on the
test logs (see Appendix B).
30
-------
Step 8: When the first test is completed, change the leak rate to estab-
, lish the second leak rate called for in the randomized series
(Table 1). When this induced rate has been established, have
the vendor test the tank. Record the environmental conditions
data. When the vendor has completed the test, record his
reported result and the induced leak rate.
Step 9: Repeat step 8 until all 42 tests have been completed.
As will be described in Section 7, the system can produce no more
than one false alarm and still pass. Thus, if a second, false alarm
occurs in the test series, the system will not pass, and testing could be
terminated. Similarly, if only one non-zero,leak rate is used, and if a
second mistake is made with that non-zero leak rate, the system will not
pass. At the point where the evaluating organization determines that the
system will not pass, it might be desirable to conclude testing. The
series could be completed to provide added information to the vendor. If
a leak rate of less than 0.10 gallon per hour was used, starting the test
series again with a leak rate closer to 0.10 gallon per hour might be
done since the method might pass at that rate but not at the smaller leak
rate. If no errors have occurred when 20 tight tank or 20 induced leak
tests have been done, the system will pass. Since only one more test is
needed, it probably would not effect much savings to stop at this point.
6.3.2 Application of the Protocol to Tracer Methods
There are many variables present in external monitoring that are
difficult to predict or control. These include the nature of the back-
fill material, moisture content of the soil, size of the excavation, type
of soil surrounding the excavation, the ground-water level, position of a
leak relative to the sampling locations, and whether the method is aspi-,
rated or passive. In general, some minimum threshold concentration of
tracer must be reached before a signal is generated. The lower the
threshold, the more sensitive the method, but the more susceptible it
will be to false alarms.
For test methods that involve the loss of product from the tank, the
induced leak .rates should be designed to introduce the amount of tracer
materialinto the soil that would be released by leak rates of the speci-
fied size over the test period. Methods that add liquid tracer to the
product specify a concentration of the tracer in the product. Using this
concentration (e.g., 10 ppm), a leak rate (e.g., 0.10 gallon per hour)
and a test and waiting time after introducing the tracer into the tank
(e.g., 24 hours), one can calculate the amount of tracer that would be
released. This is the amount that should be released during the leak
simulation. A suggested way to accomplish this is to make up sampTes of
a carrier that can be introduced into the environment, say vegetable oil/
with tracer added in the appropriate concentrations. These samples can
be used to spike the ground at small rates, giving the same amount of
tracer that would be released by the specified leak rates.
31
-------
.If the method uses gas tracers, they can be introduced into the
ground to simulate leaks by using a flowmeter to allow the gas to flow at
the rate that would occur under the testing conditions,, e.g., in a tank
at 2 PSI and through a small orifice, representing a hole that would leak
liquid product at the designated leak rates (less than 0.10 gallon per
hour).
Note that once a tracer, gas or liquid, has been introduced into the
soil in a test, the tracer must be eliminated before the next test.
Forced air may be used to disperse the tracer to levels that will not be
detected and interfere with the method; the next test may be conducted
with a different tracer; or a different site may be used.
The following steps assume that multiple tracers are available, one
of which is used in the tank to investigate the false alarm possibili-
ties, and others that are used in leak simulations.
Neither the temperature conditioning nor tank stabilization is an
issue with tracer methods. Consequently, it is not necessary to change
fuel temperatures and fill and empty the tank frequently as part of the .
evaluation. At least 21 tests of the tank in the no-leak condition are
required, as are at least 21 tests using the induced leaks.
Step 1: Decide whether a .single non-zero leak or three non-zero leak
rates will be used and select these leak rates.
Step 2: Identify the zero leak rate with LRl in Table 1. Randomly
assign the other leak rate(s) to LR2, LR3, and LR%..
Step 3: Randomly rearrange the order of the 21 pairs of tests in
Table 1'that result from the assignment of the leak rates.
Step 4: Determine the rate of introducing tracer (if a gas) or liquid
carrier and tracer (if a liquid) into the backfill to simulate
the selected leak rates. If a liquid tracer is used, prepare
samples with the carrier and tracer in the needed concentra-
tions, label these"with the randomized test sequence, and
provide them to the test crew. The crew should not know
whether or in what concentration the tracer is in the leak
. simulation samples*
Step 5: Prepare the tank. If a liquid tracer is used, have the vendor
introduce it at the desired concentration into the test tank
and fill the tank to the desired level following normal oper-
ating procedures for the method. If a gas tracer is used,
empty the tank and have the vendor introduce the gas to the
tank. The tank thus prepared will serve to provide the data on
the zero leak rates.
32
-------
Step.6: Have the vendor locate the sampling ports. Also locate a spik-
ing port for leak simulation as far from the sampling ports and
as close to the tank as possible. Be careful not to damage the
tank in installing the ports in the backfill.
Step 7: Conduct the trial run. For tracer methods, the trial run will
be of a different nature than for other methods. The trial run
for a. tracer usually consists of verifying that the site condi-
tions allow the use of a tracer method. A compound is intro-
duced at the spiking port. The test locations are sampled to
determine whether the compound is detected at the sampling
locations. The trial run accomplishes two purposes. First, it
verifies that the soil or backfill conditions are such that the
tracer can migrate from the tank to the sensors. Second, it
determines the time needed for the migration and so establishes
a test time.
Step 8: Have the vendor conduct a test of the tank (zero leak rate).
Step9: Begin testing using the first non-zero leak rate. Have the
vendor conduct a test. Note: If two different tracers are
used, it may be possible for the vendor to conduct the test on
the tank (zero leak rate) and the induced leak test at the same
time.
Step 10: When the test in step 8 and/or 9 is completed, record the .
induced leak rate, the vendor's determination (tight or leak-
ing), and the environmental conditions data on the test log
(see Appendix B). ' ,
Step 11: Ensure that the test site can be used for a second leak test
(by removing the current tracer or using a different one).
Start the next induced leak rate as in steps 8 and 9 and have
the vendor conduct another test.. Record all results.
Step 12: Repeat step 11 until the test series is completed.
It should be possible.for the vendor to conduct tests on the tank
containing the tracer repeatedly for the zero leak rate tests. In con-
ducting.the repeated tests on the tight tank to estimate the false alarm
rate, the steps of adding tracer to the product and mixing the tracer in
the product should be repeated. The process of adding and mixing tracer
is a likely cause of false alarms as it could lead to inadvertent release
of tracer into the environment that could be mistaken for a leak. It
should be possible to simulate the addition and mixing of the tracer,by
using tracer-containing product and handling it in the same manner as the
tracer solution. ;
Assuming that at least two tracers are available, the tight tank
tests and the simulated leak tests can be run simultaneously. For each
test, the carrier sample is introduced in the spiking port. The con-
tainers, of carrier,'are made up in advance and coded. Half of them .
-------
contain tracer and half do not. Each test would consist of introducing
one tracer (say type A) into the tank and another sample (either a blank
'or containing tracer type B) into the spiking port. The testing company
samples the soil gas and reports on the presence of any detected
tracer. A finding of tracer A would be a false alarm. A finding of
tracer B (when it was spiked) would be a correct detection. If
additional distinct tracer compounds are used, this process could
continue spiking tracer C, etc. A finding of both tracer B (from a
previous spike) and tracer C from the current spike would be a correct
detection. . .
As will be described in Section 7, the system can record only one
false alarm and still pass. Thus, if a second false alarm occurs in the
test series, the system will not pass, and the evaluating organization
may recommend to the vendor that testing might be terminated. Similarly,
if only one non-zero leak rate is used, and if a second mistake is made
with that non-zero, leak .rate, the system will not pass. At the point
where the evaluating organization determines that the system will not
pass, it-might be desirable to conclude testing. If a leak rate of less
than 0.10 gallon per hour was used, starting the test series again with a
leak rate closer to 0.10 gallon per hour might be done since the method
might pass at that rate but not at the smaller leak rate. .
6.4 TESTING PROBLEMS AND SOLUTIONS
Inevitably, some test runs will be inconclusive due to broken equip-
ment, 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 42. 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.
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
less than 4 hours, then repeat that run. Otherwise, repeat
the empty/fill cycle, the stabilization period, etc. Record
.all time periods.
Note: The average stabilization time or average time after
introducing the tracer 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
34
-------
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.
Rule numbers 2 and 3 are only applicable if the testing schedule
. requires temperature conditioning and tank deformation effects. Other-
wise, the time between tests is not an important limitation.
. Note that an acceptable alternative to conducting the tests in pairs
is to set up the tank conditions (as required) for each test. Thus,
while the protocol allows for the tests to be run in pairs for economy,
they may all be run individually.
6.5 METHOD EVALUATION PROTOCOL FOR WATER DETECTION
Some, methods rely on detection of water incursion to identify leaks
.in the presence of a high ground-water level. These often use a water
sensor installed at the bottom of the tank. A standpipe device to test
the function of the water sensor consists of a cylinder with an accu-
rately known (to ±0.001 inch) inside diameter attached to the bottom of
a pipe of 4- to 6-inch diameter pipe. The probe is mounted so that the
sensor is in the same relation to the bottom of the cylinder as to the
bottom of a tank when installed in the field. Enough product is put into
f^ ' the cylinder and pipe so that the product level sensor is high enough so
as not to interfere with the water sensor. A measured amount of water is
then added to the cylinder until the water sensor detects it, at which
time the water level is calculated and recorded. Additional measured
amounts of water are added to produce calculated level changes. The
amount of water added, the calculated level change, and the level change
measured by the method are recorded. This is done over the range of the
water sensor or 4 inches, whichever is less. When testing is complete,
the product and water are removed, separated, and the process is
repeated. The testing procedure is given in detail next.
Step 1: Install the probe temporarily in a test standpipe*, The bottom
section of about 1 foot should have an accurately known (to
±0.001 inch) inside diameter. The diameter must be large
- enough to accommodate the probe and must be known accurately so
that the volume of water added can be used to calculate the
water level.
Step 2: Fill the bottom section of the standpipe with the product
(typically this will require a gallon or less). Enough product
needs to be added so that the product level sensor is high
enough not to interfere with the water sensor. .
35
-------
Step .3: Add water'to the cylinder .with, a pipette until the sensor
detects the presence of the water. Record the volume of water
. added and the sensor reading at each increment. The sensor
reading will be zero until the first sensor response. At that
point, total the water increments and calculate the correspond-
ing level, Xls. of water detected. Record all data on page l.of
the Reporting Form for Water Sensor Evaluation Data in
Appendix B.
Step 4:. Add water to the cylinder with a pipette in increments to
produce a height increment, h, of approximately l/20th inch.
At each increment, record the volume of water added and the
water height (denoted by W1 ^ in Table 2 of Section 7.2)
measured by the sensor. Use1 pages 2 to 4 as necessary of the
Reporting Form for Water Sensor Evaluation Data in Appen-
dix B. Repeat the incremental.addition of water 60 times until
a total height of about 3 inches (or the range limit of the
sensor, if less) has been reached.
Step 5: Empty the product and water from the standpipe, refill with
product (the same product can be used after separating the
water) and repeat Steps 2 through 4 20 times to obtain
20 replications. .
Record all data using the Reporting Form for Water Sensor Evaluation Data
in'Appendix B. The 20 minimum detectable.water levels are denoted by Xj,
j=l,...,20. The sensor reading at the itn increment of the 3 test isj
denoted by W1 -j as described in Section 7.2 and Table 2.
36
-------
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.
If the method has more than one mode of .leak .detection, then the, perfor-
mance of the method must be evaluated and the results reported for each
testing mode separately. If the performance is different for different
modes, this may limit the conditions under which the method can be used
and these should be reported under the limitations section of the results
form.
The evaluation of the nonvolumetric test method is presented
first. A separate section (7.2) presents the calculations to estimate
the minimum water level and the minimum water level change that the water
sensor can detect* Section 7.2 is only needed if the method measures or
detects water incursion as one mode of its leak detection.
The performance of the nonvolumetric test method is judged on the
basis- of the percentage of false alarms and the percentage of correctly
identified leaks. The performance standards specify that the false alarm
rate must be no more than 5% and. that the probability of detecting a leak
rate of 0.10 gallon per hour must be at least 95%. The test procedure
includes ,21 tests of the tank in the no-leak condition and 21 tests of.,
the tank with leaks induced at rates of 0.10 gallon per hour or less.
These data are used to estimate the probability of false alarm and
probability of detection directly.
7.1 ESTIMATION OF THE METHOD'S PERFORMANCE PARAMETERS
After ail tests are performed according to the schedule outlined in
Section 6, a total of at least 42 test results will be available. Of
these, 21 will have been obtained under tight tank conditions, and 21
under induced leak conditions. The probability of false alarm, P(FA),
and the probability of detection, P(D), are' calculated next.
7.1.1 False Alarm Rate, P(FA)
The results obtained from the tests performed under tight tank
conditions will be used to calculate P(FA). Let Nj denote the number of
these tests, normally 21. (Note: This number must be at least 21, but
could be larger if more tests are called for in the experimental plan set
-:... . ! . 37 :' ' " ' ' .
-------
up at the beginning of the testing.) Let m denote the number of cases
where the method indicated a leak. If the test results, U9 are coded as
zero when no leak is Indicated and 1 when.a leak is indicated, then
TL'=
where the sum is taken over the
estimated by the ratio
tests at zero leak rate. The P(FA) is
P(FA) =
In order for the system to meet the performance standards, the estimated
P(FA) must be less than or equal to,5%. Thus, in order for the system to
meet the performance standards, 11i must be no more than 1 if the .
standard 21 tests are performed.' ; .
If the method did not identify the tank to be leaking when it was
tight, that is, TL, = 0, then the proportion of false alarms becomes
0%. However, this does not mean that the method is perfect. The
observed P(FA) of 0% is an estimate of the false alarm rate based on the
evaluation test results and the given test conditions.
One can calculate an upper confidence limit for P(FA) in the case of
no mistakes. Let Na be the number of tests performed under the tight
tank condition. Choose a confidence coefficient, (1 - a), say 95% or
90%. Then the upper confidence .limit, UL, for P(FA) is calculated as:
UL for P(FA) =l-a
In the case of 0 false alarms out of 21 tests, the upper limit to P(FA)
becomes 0.133 or 13.3% with a 95% confidence coefficient. That is, P(FA)
is estimated at 0%, and with a confidence of 95%, P(FA) is less than or
equal to 13.3%. In general the confidence interval for P(FA) can be
calculated from the binomial distribution with Nx trials. The 95%
confidence interval must be calculated and reported on the results form
in Appendix B (see page 48).
7.1.2 Probability of Detecting a Leak, P(D)
The probability of detection, P(D), is calculated for a'specific
size of leak. The size of leak that can be detected with this proba-
bility is also to be reported. Normally this will be 0.10 gallon per
hour, as required by the performance standards. The exception to this
38
-------
would occur if . a method is tested using induced leak rates smaller than
0.10 gallon per hour, for example, 0;05 gallon per 'hour. Report the
probability of detection, P(D), together with the maximum leak rate used
in the evaluation testing. The leak rate corresponding to the P(D) will-
be 0,10 gallon per hour or less.
The results obtained from the tests performed under induced leak
conditions (leak rates less than or equal to 0.10 gallop per hour) will
be used to calculate P(D). Let N2 be the number of such tests. Typi-
cally, N2 will also be 21, but could be larger if the evaluation was
initially "set up to include more tests. Let TL2 be the number of cases
where the method indicated a leak. As before, the test results, L1§ are
coded as zero when the tank is declared to be tight and 1 when the tank
is declared to be leaking. Thus, TL2 is calculated as
where the sum is taken. over the N2 tests with induced leaks. The P(D) is
then estimated by the ratio
'.'.". P(P) = TL2/N2. ;.
' '
^The estimated P(D) must be at least 95% for the system to meet the per-
formarice standards. Thus, TL2 must be either 20 or 21 (out of 21 tests)
for the estimated probability of detection to be at least 95%.
If the method identified the tank to be leaking in all tests where a
leak was simulated, then the proportion detected becomes 100%. However,
this does not mean that the method is perfect. The P(D) of 100% is an
estimate of the probability of detection, based on the evaluation test
results and the given test conditions.
One can calculate a lower confidence limit for P(D) in the case of
no mistakes. Let N2 be the number of tests performed under the induced
leak conditions. Choose a confidence coefficient, (1 - a), say 95% or
90%. Theh the lower confidence limit, LL, for P(D) is calculated as:
1/N2
LL for P(D) = a . .
In the case of correct identification of 21 tests performed under
leak conditions, the lower limit to P(D) becomes 0.867 or 86.7% with a
95% confidence coefficient. That 'is i P(D) is estimated at 100%, ancT-wlth
a confidence of 95%, P(D) is greater than or equal to 86.7%. The 95%
confidence interval for P(D) must be calculated based on the binomial
distribution with N2 trials and reported on the results form in
Appendix B (see page 48).
' 39 . ' ' - - ' '
-------
7.2 WATER DETECTION MODE
This section is only applicable if the method being evaluated uses
detection of water incursion as a leak detection mode.
Two .parameters will be estimated for the water detection, sensor:
the minimum detectable water level or threshold that the sensor, can
determine, and the smallest change in water level that the device can
record. These results- will also be reported on the Results of U.S. EPA
Standard Evaluation form in Appendix B. These parameter estimates will
then be used to calculate the minimum time needed to detect water
incursion at 0.10 gallon per hour for various tank sizes.
. ' f '
7.2.1 Minimum Detectable Water Level
The data obtained consist of 20 replications of a determination of
the minimum detectable water level (see test schedule, Section 6.5).
These data, denoted by X,-, j=l,...20, are used to estimate the minimum
j ' , ' r '
water level, or threshold, that can be detected reliably.
Step 1: Calculate the mean, X, of the 20 observations:
20
X =
X../20
Step 2: Calculate the standard .deviation, SD, of the 20 observations:
SD =
20
o
. -X)
20-1
1/2
Step 3: From'a table of tolerance coefficients, K, for one-sided normal
* tolerance intervals with a 95% probability level and a 95%
coverage, obtain K for a sample size of 20. This coefficient is
K = 2.396. (Reference: Lieberman, Gerald F. 1958. "Tables
for One-Seided Statistical Tolerance Limits." Industrial Quality
Control. Vol. XIV, No. 10.) -
40
-------
Step 4: Calculate the upper tolerance limit, TL, for 95% coverage with
.tolerance coefficient 95%: f
TL = X + K SD9
°r ' '.-.
TL = X + 2.396 SD
TL estimates the minimum level of water that the sensor can
detect. That is, with 95% confidence, the method should detect water at
least 95% of the time when the water depth in the tank reaches TL.
7.2.2 Minimum Water Level Change
This following statistical procedure provides a means of estimating
the minimum water level change that the water sensor can detect, based on
the schedule outlined in Section 6.5.
Denote by W^j the sensor reading (in inches) at the jth replicate
(j=l,...,20) and the ith increment (i=l,...,n,-, with n,- being 60 in each
. ... M W
replicate). Note that the number of steps in each replicate need not be
the same, so the sample sizes are denoted by n.». Denote by X,- the water
level detected for the first time by the sensor at the jth replicate.
Denote by h the level change induced at each increment. The level
change, h, should be chosen to be consistent with the system's claimed
resolution. That is, the increments should be about half (or less) of
the method's claimed resolution.
»
Step 1: Calculate the differences between consecutive sensor readings.
The first increment will be Wlti-X]_ for the first replicate
(j=l); more generally, W^j-X-j, for the jth replicate. The
second increment will be W2ji-Wljl for the first replicate; more;
generally, W2 i-W1 ,- for the jth replicate, etc.
*vA*J , ^ ,
Step 2: Calculate the difference, at each incremental step, between h,
the level change induced during testing, and the difference
obtained in Step 1. Denote these differences by dn- .,, where i
1 > J
and j represent increment and replicate numbers, respectively.
Table 2 below summarizes the notations.
41
-------
Table 2. NOTATION SUMMARY FOR WATER SENSOR READINGS
At THE jth REPLICATE
Increment
No.
1
2
3
Calculated
1 evel
change
(inch)
A
+ h
+ h
+ h
o
0
0
o
o
+ h
Measured
Sensor sensor
reading increment
(inch) (inch)
B C
W2,j W2,o"wl,j
W3 , W3 ,-W2 ,
O,J O,J t,J
«
-
W W j-W
11^,3 n4.,j n. 1,
o j j
Increment
difference
calculated-meas.
(inch)
C-A
* . -
.*
'
_X,- is the water level (Inches) detected for the first time
j '
by the sensor during the jth replication of the test.
Note that using the first sensor reading, Xit may vary from replicate to
J
replicate, so that the number of differences d^ j will also vary. Let n
be the number of increments necessary during replicate j.
Step 3: Calculate the average, Dj, of the differences d^ j, -i-l,...,hj,
. separately for each replicate j, j=l,...,20.
o
j-'2
42
-------
Step -4: Calculate the variance of the differences d, ,, i=l ..... TI .
separately for each replicate j, j=l,..., 20.' J J
Step 5: Calculate the pooled variance, Var_, of the 20 Variances
'».,- _
-------
7.2.3 Time to Detect an Increase In Water Level
The minimum detectable water level and the minimum detectable change
can be used to estimate the minimum time needed to detect water incursion
into the tank at a specified rate. This time is specific to each tank
size and geometry and depends on specific assumptions. The calculations
are illustrated for an 8,000-gallon steel tank that is 96-inch diameter
and 256 inches long.
Suppose.there are x inches of water in the tank. The tank is made
of quarter-inch steel, so the inside diameter is 95.5 inches, giving a
radJus, r, of 47.75 inches and a length of 255.5 inches. The water sur-
face will be 2d wide, where d, in inches, is calculated as
d = r2 - (r - x)2,
where x is the water depth. The area of the water surface at depth of
x inches of water is then given by 255.5 x 2d inch2. Multiplying, this by
the minimum level change and dividing the result by 231 inch3 per gallon
gives approximately the volume change in gallons that the sensor can
detect reliably. This differs with the level of water in the tank.
For these calculations, the following as'sumptions are used. The
probe is assumed to be inserted at the midpoint of the tank length and to
rest on a striker plate the top of which is 0.63 inch above the bottom of
the tank. The initial water depth is taken as the minimum .depth the
sensor can detect with 95% probability plus the striker plate depth of
0.63 inch, rounded up to the next quarter inch. The tank is assumed
level. (Calculations show that if the tank is tilted, the cross-
sectional area of the water surface will be slightly less for the same
water depth at this location, so these calculations slightly 'overestimate
the volume.)
To determine how long the method will take to detect a water incur-
sion at the rate of 0.10 gallon per hour, divide the minimum volume
change that the water sensor can detect by O.JLO gallon per hour. As a
numerical example, suppose the minimum depth of the water detectable is
0.3 inch and the minimum detectable change is 0.02 inch. This. gives
x ~ 0.95 inch (0.3 * 0.625 rounded up). In an 8,000-gallon tank with
inside diameter 95.5 inches and length 255.5 inches, the water surface
width, d, is calculated as
d =^(47.75)2 - (46.8)2 = 9.43 inches
The volume, in gallons, corresponding to a 0.02-inch increase is
V = 2(9.48) x 255.5 x (0.02)/231 -
or
V = 0.42 gallon
'44
-------
",
The time that the sensor will take to detect water incursions at the rate
of 0.10 gallon per hour will be ;
time = 0.42 gallon/0.10 gallon per hour = 4.2 hours
Thus,, the sensor would detect water coming in at the rate of 0.10 gallon
per/hour after about 4 hours 15 minutes. The incursionvof the water into
the tank should be obvious under these conditions if the test is run for
at least 4 hours 15 minutes. .
The minimum amount of water in a tank that can be detected by a
sensor depends on the placement ;of the sensor, any tilt, of the tank, the
tank size, and the sensor threshold. This, minimum amount varies from
about 2 gallons to 10 or 15 gallons, depending on the combination of .
these factors. If water enters at a rate of 0.10 gallon per hour, it
would require anywhere from a day to a week for enough water to be
detected, starting with no water in the tank.
7.3 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 com-
plete instructions for the use of the results form.
" '-'. -' " ' '-'
-These sections are only required if they are applicable to the
particular nonvolumetric method being evaluated. If a section is not
applicable, skip the calculations and report "not applicable" on the
results form.
Size of Tank \
The evaluation results are applicable to tanks up to at most 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 2 of the results form.
45
-------
Maximum Allowable Temperature Difference
This section is only applicable if temperature conditioning was
needed and used as part of the evaluation procedure. If temperature does
not affect-the operation of the method, ignore this section and indicate
"not applicable" on the results form.
Calculate the Standard deviation of the 21 temperaiture 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 Conditionsj" 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 effec.ts. 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 the start of the test for the 21 tests that started
immediately after, the specified waiting time. (Note: If more than
21 tests are dbnfe 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.
For tracer methods, the average waiting time may more appropriately -
be the time from adding the tracer to the tank until the completion of
the test. " .
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
46
-------
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
42) of tests* Report this time as the average data collection time per
test.
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 6Q% 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 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.
The intent of this is to provide an estimate of the time that the
testing will interfere with normal operation of the tank. The nohvolu-
metric methods will differ in those parts of their operation that require
the tank to be out of service. Consequently, the time that should be
reported here is the estimated time for which testing with this method
will interfere with the use of the tank by requiring that it be out of
service. ,
7.4 SUPPLEMENTAL CALCULATIONS AND DATA ANALYSES (OPTIONAL)
This section discusses some additional data analyses that may be
possible with the data, depending on the actual results. It also pro-
vides some rationale for the sample size selection.
47
-------
7.4.1 One-Sided Confidence Limits on P(FA) and P(D)
«-- - - ' j
It is-possible to estimate the false alarm rate and probability of
detection directly as done in Section 7.1 with any sample size. However,
for fewer than 20 tests, the estimate of P(FA) will be zero or will
exceed 5%, depending on whether any false alarms are found. Similarly,
P(D) will be 100% or less than 95% for sample sizes less than 20 depend-
ing on whether any leaks are missed .or not. Thus, the sample size of 20
is the smallest that allows for one mistake in each case and still pro-
vides estimated performance meeting the EPA standards. The sample size
of 21 was chosen from experimental design considerations to balance the
different conditions.
Confidence limits for P(FA) and P(D) can be calculated based on the
observed results and the sample sizes. The formulas for perfect scores
were given in Section 7.1.1 for P(FA) and in Section 7.1.2 for P(D).
These also depend on the selected confidence coefficient. Table 3 below
gives 90% and 95% one-sided confidence limits for P(FA) and P(D) based on
samples-of 21 tests for the case of no mistakes and one mistake, the two
conditions under which the method meets the EPA performance standards if
evaluated with the minimum 21 tests.
Table 3. ONE-SIDED CONFIDENCE LIMITS
FOR P(FA) AND P(D)
Field test
results
0
1
Error
Error
out
out
of
of
21
21
Confidence
P(FA)
P(FA)
90%
< 0.
< 0.
coefficient
95%
104
173
P(FA)
P(FA)
<
<
0
0
.133
.207
0
1
Error
Error
out
out
of
of
21
21
P(D)
P(D)
> 0
> 0
.896
.827
P(D) i
P(D) ;
> 0
> 0
.867-
.793
Table 3 shows that the confidence limits start to become fairly large for
high .confidence with even one error. Using a larger sample size would
improve the confidence limits, but would add significantly to the cost of
testing. The sample sizes were selected as a compromise to provide
reasonable estimates while not requiring excessively expensive testing.
7.4.2 Alternative Statistical Model
If the evaluation uses three non-zero leak rates and if the method
fails to detect some of the induced leaks, an alternative statistical
analysis may be possible. This alternative'statistical method fits a
logistic model to the data, assuming that the probability-of'detecting a
48
-------
JSJTT8"?! w1th"tJe Si2e of ««: !«* I* one assumes that the logis-
tic model with parameters A and B holds, then the probability of detect-
ing a leak can be expressed as: ucueut
PfDetecting a leak given a leak of size S] = l/[l+exp(A+BS)]
That is, the probability that the test method will indicate a leak when
there is an actual induced leak rate, S, is given by the logistic func-
I H RTh| ftta fr01?.a11 ^ests' Can be used to es*1mate the parameters
?hS*iS ;H?*MqUatl0n- T?1S reqUlres an 1terative estimationV technique
* a£ 1S ayailab.l£ ln several commercial statistical software packages
such as SAS, BMDP, or SYSTAT. The estimation will not converge if no
mistakes are made, and it may not converge if only a few mistakes are
I?!??,** I J es|imates do converge, then the function with the estimated
values of A and B can be used to estimate the P(FA) of the method bv sub-
stituting S - - 0. The P(D).can be estimated for any leak rate S by sub-
stituting^S into the equation. Specifically, S = 0.10 gallon, per hour
anKbK-?^Stltuted to con>Pare with the EPA performance standards for
probability of detection.' '
7.;4.3 Estimation of Temperature Effect
' . JI the temperature and stabilization time variables influence the
operat on of thetest and testing is done according to the full set of
conditions in Table 1, the logistic model can also be used to test
whether the additional variables did have a significant effect on the
S5J« nf ;»« 9 1 fiwh,e!her th}! 1s Possible depends on the number and
pattern of the actual data results. The approach is to add one or more
indicator variables to the logistic model to estimate the effect of thl
additional factor. The model would become
P[Detecting a leak given a leak of size S] = l/[l+exp(A+BS+C-T-)]
where the three temperature conditions were identified by T, and coded
appropriately. This modeling becomes rather involved. The evaluating
organization should involve statistical support if these additional cal-
culations are warranted. Note that this modeling will generally hot be
p??!?b]VLn?VyStem Performs so well that the direct estimates of
P(FA) and P(D) described in Section 7.1 meet the EPA performance stan-
dards. Thus, this approach is supplemental to provide information for a
vendor to use in improving a method by identifying factors that sianifi-
cantly affect the system's performance. »i-gm.n
49
-------
-------
SECTION 8
INTERPRETATION
The results reported are valid for the experimental conditions dur-
ing the evaluation, which have been chosen to represent situations com-
monly 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. It should be emphasized that the
performance estimates are based on average results obtained in the
tests. An individual test may not do as'well. Some individual tests may
do better. .
8.1 BASIC PERFORMANCE ESTIMATES
.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
O.lO^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. For a concern with many tanks, there will
be fewer false.alarms. 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 .spec-
ifies 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. ,
The discrete nature of the data implies that only a few values of
P(FA) or P(D) are possible. With the standard 21 tests for each test
condition (tight or .leaking tank), the possible values are 0, 1/21, 2/21
etc. Consequently, the reported estimates are only precise to about
5%. The confidence limits reported in the case of a perfect score
indicate the range in which the true P(FA) or P(D) is expected to be.
For example, a method that achieved zero false alarms out of 21 would not
be expected to have a zero false alarm rate. Instead,'its false alarm
rate should_be less than 10.4$ with 95% confidence.
If testing is done at an induced lak rate less than 0.10 gallon per
hour, the P(D) may be reported at the smaller leak rate actually used.
The standard test, using an induced leak rate of 0.10 gallon per hour
would report P(D) for the rate of 0.10 gallon per hour. In general, a
51
-------
method that can detect a smaller leak with high probability ,i's to be
preferred because it will identify a potential problem earlier. This may
reduce the amount of pollution and the cost of remedial action.
8.2 LIMITATIONS .
Nonvolumetric tank tightness testing methods that are based on
different operating principles will have different factors that can
interfere with their performance. Consequently, the limitations on the
applicability of-the performance estimates will also vary with the
method. If a factor, for example temperature, does not affect the
principle of operation, it should not be reported as a limitation.
However, there may be interfering factors other than those listed in the
experimental plan that affect a particular test method. If so, those
additional factors might limit the applicability of the method. The
reporting form provides a place to identify other sources of interference
and to state the test conditions for them.
Some nonvolumetric test methods use more than one mode of
operation. If so, different limitations may apply to each mode of leak
detection. It is possible that one mode of operation may be unaffected,
by size of tank, but that another may depend strongly on tank size. For
example, a water sensor may be used to .test for leaks in the presence of
a high ground-water level. It may do so by sensing water incursion, in
which case it must be able to detect water incursion at the rate of
0.10 gallon per hour. Since the time required for the water level to be_
.detectable at a fixed rate of incursion will be a function of.the size of
the tank, this mode of leak detection is dependent on tank size.
8.3 HATER LEVEL DETECTION FUNCTION
If the .system uses a water level sensor, the following results are
reported. . ,"',
The minimum water level detected by the sensor is; estimated from_the
average threshold of detection, and the variability of the water level
threshold is estimated' b'y the standard deviation of the test data. The
minimum water level that will be detected at least 95% of the time is the
level to be reported. Statistically, this is a one-sided tolerance
limit. .
The'tolerance limit calculated in Section 7.2.1 estimates the
minimum water level that the sensor can detect above the bottom of the
probe. If the installation of the sensor leaves the probe at a specified
distance above the bottom of the tank (for example, 1 inch), then this
minimum distance needs to'be added to the reported minimum detectable
water level. ,
52
-------
8.4 MINIMUM WATER LEVEL CHANGE MEASUREMENT
-' c -: - .
_ The water sensor may be used to test for leaks in the event of a
h]9h ground-water level. If the ground-water level is above the bottom
of the tank, there will be an inward pressure when the product level is
sufficiently low, and if there is a hole in the tank, water will flow
into the tank under these conditions. Based on the'ability of the water
sensor to detect a change in the level of water in the product, one can
determine how much water must enter the tank in order for an increase in
the water level to be detected. From this information, in turn, one can
determine the size of a leak of water into the tank that the system can
detect at a given time.
The standard deviation of the differences between the change in
water level measured by the sensor and the change induced during the
tests is used to determine the ability of the water level sensor to
detect changes in the water level. A two-sided 95% tolerance interval is
then calculated for this detection ability (Section 7.2.2).
The minimum change in water level that can be detected is used to
compute a minimum change in water volume in the tank. This conversion is
specific to the tank size. Using the minimum change in water volume that
the sensor can detect, the time needed for the method to detect an incur^
sion of water at the rate of 0.10 gallon per hour is calculated (Sec-
tion 7.2.3). This calculation indicates the minimum time needed for the
water detector to identify an inflow of water at the minimum leak rate
and to alert the test operator that the water level has increased.
8.5 ADDITIONAL CALCULATIONS
.If the performance estimates do not meet the performance require-
ments, the vendor may want to investigate the conditions under which
errors occurred. Calculating the percent of errors by size of leak by
temperature condition, and by length of stabilization time as applicable
may suggest .ways to improve the method. This may be as straightforward
as identifying conditions that lead to popr performance and revising the
operating procedure to avoid those, or it may require redesign of the
method. (. . .
The relationship of performance to test conditions is primarily of
interest when the method does 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 not to many tank
owners or operators.
53
-------
-------
SECTION 9
REPORTING OF RESULTS
.Appendix B is designed to be the framework for a standard report.
There are five parts to Appendix B, each of which is preceded by instruc-
tions for completion. The first part is the Results of U.S. EPA Standard
Evaluation form. This is basically an executive summary of the find-
ings. It is designed to be used as a form that would be provided to each
tank owner/operator that Uses this system of leak detection. Conse-
quently, it is quite succinct. The report should be structured so that
this results form can be easily reproduced for wide distribution.
A method that uses more than one mode of leak detection may achieve
different performance results for the different modes of operation. The
results form is structured to allow for reporting the P(FA) and P(D)
separately for different modes of leak detection. The method meets the
EPA performance requirements only if all modes of leak detection meet
those requirements. . . :
Suppose that a method had two modes of testing, a basic one and an
ancillary one for testing in the presence of a high ground-water level.
Suppose that the test method when evaluated in the case of high ground-
water level did not meet the EPA performance requirements, but the basic
one did. Then a report could be issued, stating that the method meets
the EPA performance requirements, but cannot test when the ground-water
level.is above the bottom of the tank. -
The statement of compliance with the EPA performance standards must
be consistent with stated limitations on the form and also with the
standard operation of the method as described on the Description form.
The second part of the standard report consists of the Description
of the 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 Test Results, also described in Appendix B. This table summarizes
the test results and contains the information on starting dates and
times, test duration, leak test results, etc. .
55
-------
The fourth part of Appendix B contains a blank Individual Test
Log. While the Individual Test Log has been designed to be flexible, it
may need modifications for some test methods. This form should be repro-
duced 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.
The fifth part of Appendix B provides a form to record the test
results when evaluating the system's water sensor. The data to be .,
recorded follow the testing protocol (in Section 6.5) to determine the
minimum level of water and the minimum water level change that the system
can detect. This part is only applicable if the system uses a water
sensor.
If the optional calculations described in Section 7.4 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 informa-
tion available if needed.
The limitations on the results of the evaluation tire 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 7.3 describes the summary of the test condi-
tions 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 condi-
tions 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.
Tests based on volumetric changes generally perform less well as the size
of the tank increases. However, for some nonvolumetric test methods,
size is not such a restriction. The evaluating organization must deter-
mine the extent to which tank size affects performance and report a size
limitation here.
A second potential limitation on the results is the temperature
differential between the product added to the tank and that of the
product already in the tank. 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 differ-
ences could exist. If temperature affects the method, then the tempera-
ture differences used in the evaluation must be reported. If the physi-
cal principle of the method is not affected by temperature, then report
that the method is not limited by temperature and the basis for this
conclusion. The evaluation testing may be done using larger temperature
56
-------
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 time heeded by the method
for its operation. For example, tracer methods require some time for the
tracer to move through the backfill to the sensors. The Individual Test
Logs call for recording the actual time used in the testing. The average
time is to be reported and the results should be valid for times at least
this long. It may be the case for some nonvolumetric methods that the
time for preparation does not require taking the tank out of service. If
so, this should be noted.
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 the method's perfor-
mance. 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, particu-
larly 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 iden-
tify and correct the source of the leak.
If the method uses a water detector as part of its operation, the
minimum depth of water that the sensor can detect is reported. In addi-
tion, the minimum change in water level that the sensor can detect is
reported. From this minimum detectable change in water level, a,minimum
volume change can be calculated based on the tank size and depth of the
water. A minimum time for detection is calculated and reported as the
time needed for water flowing into the tank at the rate of 0.10 gallon
per hour to increase the water volume enough to be detected by the
sensor.
/ -
It is expected that nonvolumetric methods may require some
modification of the forms. It is hoped that the forms supplied will be
flexible enough to provide for most of the data recording needs. How-
ever, if modifications are needed to accommodate a particular method, the
evaluating organization should make the required modifications and use
the resulting forms. The conditions during the evaluation tests are to
be recorded. The factors that affect the performance of the method being
evaluated must be recorded. The performance results are limited by the
test conditions actually used and reported.
57
-------
-------
APPENDIX A
DEFINITIONS AND NOTATIONAL CONVENTIONS
A-l
-------
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 under-
stand these concepts, some explanation is helpful. ,
Nonvolumetric test methods make a determination of whether a tank is
leaking or not. The false alarm rate is the proportion of times that the
method would incorrectly indicate that a tight tank is leaking. The
probability of detection is the probability that the method will cor-
rectly identify a leak of specified size, R. Usually, the larger the
leak rate, the more likely the method is to detect it, so the probability
of detection must specify the leak rate to be detected. In evaluating
nonvolumetric methods, the performance measures are generally estimated
directly from the test results. The false alarm rate is estimated by
conducting a number of trials on a tight tank and calculating the pro-
portion of those during which the method incorrectly indicates a leak.
The probability of detection is estimated by conducting a series of
trials with an induced leak rate, R, and calculating the proportion of
those trials during which the method correctly identifies the tank as
leaking. :
. Definitions of some of the terms used throughout the protocol are
presented ne'xt.
Nominal Leak Rate: The set or target leak rate to be achieved as
closely as possible during testing. It is a
positive number in gallon per hour.
Induced Leak Rate: The actual leak rate, in .gallon per hour, used
during testing, 'against which the results from
a given test device will be compared.
False Alarm: Declaring that a tank is leaking when in fact,
it is tight.
' ' '
Probability of The probability of declaring a tank leaking
False Alarm, P(FA): -when it is tight. In statistical terms, this
is also called the Type I error and is denoted
by alpha (o).. It is usually expressed in
percent, say, 5%.
A-2
-------
Probability of ' The probability of detecting a leak rate of a
Detection, P(D(R)): given sizei R gallon per hour. In statistical
terms, it is the power of the test method and
is calculated,as one minus beta (e), 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%.
Resolution: The resolution of a measurement system is the
least change in the quantity being measured
which the system is capable of detecting.
A-3
-------
-------
APPENDIX B
REPORTING FORMS
B-l
-------
-------
Appendix B provides'five 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.Nonvolumetric Tank Tight-
ness Testing Method (four pages)
2. DescriptionNonvolumetric Tank Tightness Testing Method (six pages)
3. Reporting Form for Leak Test ResultsNonvolumetrtc Tank Tightness
Testing Method (three pages)
4. Individual Test LogNonvolumetric Tank Tightness Testing Method
(five pages) .,
5. Reporting Form for Water Sensor Evaluation DataNonvolumetric Tank
Tightness Testing Method (four 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 1s responsible for filling out which form?
1. Results of UoS. EPA Standard Evaluation. The evaluating organiza-
tion is responsible for completing this form at the end of the
evaluation.
2. Description of Nonvolumetric Tank Tightness Testing Method. The
evaluating organization assisted by the vendor will complete this
form by the end of the evaluation.
3. Reporting Form for Leak Test Results. 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.
' ' - - l .
4. Individual Test Logs. These, forms are to be used and completed by
the evaluating organization's field crew. These forms ne.ed to be
kept blind to the vendor during testing. It is recommended that the
evaluating organization reproduce a sufficient number (at least 42
copies) of the blank form provided in this appendix and produce a
bound notebook for the complete test period.
It is expected that nonvolumetric methods may require some modifica-
tion of the test log. The form provided in this appendix was
designed from a volumetric test log. It is the responsibility of
B-2
-------
the evaluating organization to design the appropriate forms with the
vendor's input. It is important to include in the test logs all
parameters relevant to the evaluation of a specific method: In
particulars it is necessary to document the inducted leaks.
5. Reporting Form for Water Sensor Evaluation Data. These forms pro-
vide a template for the water sensor evaluation data if the method
includes such a leak detection.mode. The forms are to be used,and
completed by the evaluating organization's field crew. It is
recommended that the evaluating organization reproduce a sufficient
number (at least 20 copies) of the blank form provided in this
appendix and produce a bound notebook to be used in the field.
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 addi-
tional, optional calculations (see Section 7.4 of the protocol), these
results may be attacin-J to the standard report. There is no reporting
requirement for these calculations, however. :
Distribution of the Evaluation Test Results
The organization performing the evaluation will prepare a report for 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 four-page form is designed to be distributed
widely. A copy of this four-page form will be supplied to each tank
owner/bperator 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 tarik(s) passed the
tightness test. This four-page form will also be distributed to regula-
tors who must approve leak detection methods for use in their jurisdic-
tion.
The complete report, including all the forms in Appendix B, will be
submitted by the evaluating organization to the vendor of the leak detec-
tion 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 detec-
tion method. .
B-3
-------
The optional part of the calculations (Section 7,4), if done, would be
reported by the evaluating organization to the vendor of the leak detec-
tion method. This is intended primarily for the vendor's use in under-
standing 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 tp regulators is the responsibility of the vendor.
The forms, each preceded by its instructions for completion, are
presented next.
B-4
-------
-------
Results pf U.S. EPA Standard Evaluation
Nonvolumetric Tank Tightness Testing Method
Instructions for completingi the form
This 3-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 six main parts. These are:
1. Method Description ,
2. Evaluation Results .
,3. Test Conditions During Evaluation .
4. Limitations on the Results
5. Certification of Results
6. Additional Evaluation Results (if applicable)
Method Description , "
Indicate the commercial name of the method, the version, and the name,
address, and telephone number of the vendor. Some vendors might 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 evaluation results must be reported separately for each detection
mode if the method operates in different detection modes depending on
field conditions. Describe the mode of detection for which the results
are applicable. .
P(FA) is the probability of false alarm as calculated in Section 7.1.1.
Report the number of false alarms and the number of tight tank tests, and
report the 95% confidence interval based on the binomial distribution
with NH tests. Some values are tabled on page 48;
The leak rate .used in.the evaluation, is to be inserted in the blank.
This is the leak rate corresponding to the reported P(D) below.
P(D) is the probability of detecting a leak of the size induced (no more
than 0.10 gallon per hour) as calculated in Section 7.1.2.
Report the number of correct detections and the number of simulated leak
tests, and report the 95% confidence interval based on the binomial
distribution with N2 tests. Some values are tabled on page 48.
'..-.- . B.5 - : . '- , . '' - '
-------
If the calculated P(FA) is 5% or less and if the calculated P(D) is 95%
or more, then check the "does" box. Otherwise, check the "does not1'
'box. Note: the P(FA) and P(D) requirements apply to.each leak detection
mode used by the method.
Indicate whether this method operates under mbre than one mode of detec-
tion. Check the appropriate box and complete page 4 (Additional Evalua-
tion Results) if applicable. ,
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 fiber-
glass. Also report the backfill material in the tank excavation, e.g.,
clean sand or pea gravel. Give the tank diameter and length in inches.
Report the product used in the testing. Give the range of temperature
differences actually measured as well as the standard deviation of the
observed temperature .differences. Report the ground water level for .the
test tank in inches above the bottom of the tank. Report zero for ground.
water at or below the bottom of the tank.
Other sources of interference may affect non-volumetric methods. Report
any sources of interference specific to the method on the lines pro-
vided. Also report the range of test conditions for the indicated
interference source. If no additional sources of interference were
identified, check "None."
Limitations on the Results
The size (gallons) of the largest tank to which these results can be
.applied may be calculated as 1.50 times the size (gallons) of the test
tank. ,
The temperature differential, the waiting time after adding product until
testing,-and the total data collection time should be completed using the
results from calculations in Section 7.1.4. Alternately, if the
principle of operation of the method is riot affected by product
temperature changes, check the box indicating that temperature is not a
limiting factor and give' the justification.
Certification of Results
Here, the responsible pers.on at the evaluating organization indicates .
which test procedure was followed and provides his/her name and signa-
ture, and the name, address, and telephone number of the organization.,
Additional Evaluation Results (if applicable)
If the "yes" box relating to other leak detection modes on page 1 was
checked, then provide the necessary information for the P(FA) and P(D)
for the additional leak detection mode. These probabilities will have
been calculated as described in Sections 7.1.1 and 7*1.2, based on the
evaluation results obtained in jthat detection mode. ,
B-6 ..-..
-------
out this section as described on page B-5.
If the method includes a water sensor, then complete the results for that
sensor.
The minimum detectable water level and the minimum detectable level
change that the sensor can detect will have been obtained from the
calculations in Sections 7.2.1 and 7.2.2.
The minimum time for the water sensor to detect a leak of 0.10 gallon per
hour by detecting an increase in the water level in the tank will have
been obtained from the calculations in Section 7.2.3. This time is
calculated based on a water depth equal to the striker plate height plus
the minimum detectable water level (above the striker .plate). It assumes
a level tank and that the sensor is located midway along the tank length.
The minimum detectable increase is used to calculate the volume change
needed. This volume is divided by 0.10 gallon per hour to get the time
reported. .Indicate the size of the tank on which this t/ime calculation
is based.
B-7
-------
I
*
-------
Results of U.S. EPA Standard Evaluation
Nonvolumetric 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: Nonvolumetric
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 ' , ', -x , -
.Name
Version
Vendor
(street address)
(state) - (zip) (phone)
Evaluation Results
This method, which declares a tank to be leaking when
has an estimated probability of false alarms [P(FA)] of % based on the test
results of _ false alarms out of. tests. A 95% confidence interval for P/FAV
is from .,.'-
to %.
The corresponding probability of detection [P(D)] of a ' gallon per hour leak is
__; % based on the test results of detections out of ' ,
simulated leak.-tests. A 95% confidence interval for P(D) is from ______ to %.
Does this method use additional modes of leak detection? D Yes D No. If .Yes, complete
additional evaluation results on page 3 of this form.
Based on the results above, and on page 3 if applicable, this method D does ED 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, installed in
, : - ' backfill. '
The ground-water level was inches above the bottom of the tank.
Nonvplumetric TTT Method - Results Form " Page i of 3
-------
Nonvolumetric
Version
Method
Test Conditions During Evaluation (continued)
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 jn the evaluation was ____.
This method may be affected by other sources of interference. List these interferences below
and give the ranges of conditions under which the evaluation was done: (Check None if not'
applicable.) ~~
None
Interferences
Range of Test Conditions
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 contains a product identified on the method description form.-
The tank capacity is - gallons or smaller.
* The difference between added and in-tank product temperatures
is no greater than+ or - degrees Fahrenheit.
CU Check if applicable:
Temperature is not a factor because .
The waiting time between the end of filling the test tank and the start of the test data collec-
tion is at least _________ hours. . .'.,.
The waiting time between the end of "topping off" to final testing level and the start of
the test data collection is at least hours.
The total data collection time for the test is at least hours.
The product volume in the tank during testing is ______% full.
This method Dean EH 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:
Nonvolumetric TTT Method - Results Form
Page 2 of 3
-------
Nonvolumetric TTT Method,
Version
> 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.
Additional Evaluation Results (if applicable) "~ ~ "~~" ~~~
This method, which declares a tank to be leaking when ._
has an estimated probability of false alarms [P(FA)] of - ' % based on the test
results of false alarms out of _____ tests. Note: A perfect score during testing
does not mean that the method is perfect/Based on the observed results a 95% confidence
interval for P(FA) is from 0 to %.
The corresponding probability of detection [P(D)] of a gallon per hour leak is
% based on the test results of .- detections out of - simulated
. leak tests. Note: A perfect score during testing does not mean that the method is perfect
Based on the observed results, a 95% confidence interval for P(D) is from 0 to ____%.
> Water detection mode (if applicable)
Using a false a'latm rate of 5%, the minimum water level that the water sensor can detect
with a 95% probability of detection is inches.
Using a false alarm rate of 5%, the minimum change in water level that the water sensor
can detect with _a 95% probability of detection is _; inches.
Based on the minimum water level and change in water-level that the water sensor can
detect with a false alarm rate of 5% and a 95% probability of detection, the minimum time for
the system to detect an increase in water level at an incursion rate of 0.10 gallon per hour is
; : minutes in a -gallon tank. .
Certification of Results * ' ' - ' * - .
I certify that the nonvolumetric tank tightness testing method was installed and operated
according to the vendor's instructions. I also certify that the evaluation was performed
according to the standard EPA test procedure for nonvolumetric 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)
Nonvolumetric TTT Method - Results Form .''..- Page 3 of 3
-------
-------
Description of Nonvolumetric Tank Tightness Testing Method
Instructions for completing the form
This 6-page form is to be filled out by the evaluating organization with
assistance from the vendor, as part of the evaluation of the method. '
This form.provides supporting information on the principles behind the
system 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 testinq
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. Principle of Operation
4. Temperature Measurement
5. Data Acquisition
6. Procedure Information
> Waiting times . .
> Test duration
> Total time
> Other important elements of the procedure or method
> 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.
The section "> Other important elements of the procedure or method"
should be completed carefully. List, here any other important elements of
the procedure or method that could affect its performance. For example:
- If-the pressure in the ullage space is different from atmospheric
during testing, indicate whether a negative or positive pressure was
applied. Report that pressure and its units.
- B-ll
-------
If the method used is a tracer method, clearly document the process of
.adding the1 tracer to the tank and in the spiking port.
If a tracer is added to the product in the tank, provide information on
the following items:
* type of tracer(s) ' ,
* tracer concentration in the product '.;..
* type of carri er :
* time between spiking and starting the test
* .type of sampling, e.g., whether sampling is active or passive (in
other words, how does the tracer reach the sampling ports? by
natural diffusion process? is the process enhanced by adding forced
air? etc.)
* other relevant items
When sampling ports are installed for tracer methods, measure ,the
distances between any part of the tank to its nearest sampling port.
Report the largest of these distances.
B-.12
-------
Description
Nonvoiumetric Tank Tightness Testing Method
This section describes briefly the important aspects of the nonvolumetric tank tightness testine
method. It is not intended to provide a thorough description of the principles behind the
Ttifitnnn nr nnw tnp Am-iin-me-nt m/-«rlre .
method or how the equipment works.
Method Name and Version
Product
- > Product type
For what products can this method be used? (check all applicable)
ED gasoline /
O diesel
D aviation fuel
D fuel oil #4
[H fuel oil #6 ,
IZ1 solvents
^ * ' ' ; -4 - -
EH waste oil
D other (list) '' . ' ' ,
> Product level
What product level is required to conduct a test?
EH above grade
[H within the fill pipe .
D greater than 90% full .
. CH greater than 50% full
D empty .
CH other (specify) '______^_
Nonvolumetric TTT Method - Description Page 1 of 6'
-------
Principle of Operation
What principle or principles are used to identify a leak?
D .acoustical signal characteristic of a leak
CU identification of a tracer chemical outside the tank system
D changes in product level or volume
CD detection of water inflow
HH other (describe briefly) '
Temperature Measurement
If product temperature is measured during a test, how many temperature sensors
are used? . .
CU single sensor, without circulation
EH 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?
EU resistance temperature detector (RTD)
C3 bimetallic strip ,
D quartz crystal . ,
D thermistor . . .'.".-
D other (describe briefly) _ _ _____ _ -
If product temperature is not measured during a test, why not? ,
D the factor measured for change in level or volume is independent of temperature
(e.g., mass) . ;
D the factor measured for change in level or volume self-compensates for changes in
temperature , . .
CD other (explain briefly) _ _. _ _ __ . ' - _
Data Acquisition
How are the test data acquired and recorded?
D manually ,
CD by strip chart
D by computer
Nonvolumetric TTT Method - Description , Page 2 of 6
-------
Procedure Information
> Waiting times
What is the minimum waiting period between adding a large volume of product to bririq the
level to testrequirements and the beginninQ of the test (e.g., from 50% to 95% capacity)?
D not applicable
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,
> Test duration
What is the minimum time for collecting data?
D less than 1 hour. . .
D 1 hour ', , .
-D 2 hours _.
- D 3 hours . :
Q 4 hours . - .
lU 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 :
> Other important elements of the procedure or method
List here any other elements that could affect the performance of the procedure or method
(e.g., positive or negative ullage pressure, tracer concentration, distance between tank and
sampling ports, etc.) .
Nonvolumetric TTT Method - Description ' , , , Page 3 Of 6
-------
> 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
C3 presence of water in the tank
D other (describe briefly) [ . _ ,
D Jevel of .ground water above bottom of the tank not determined
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 tests for changes in water levej in tank
D other (describe briefly) . ' ' ' . _j "'
D no action .
Does the method measure inflow of water as well as loss, of product (gallon per hour)?
Dyes .
Dno _ , . I
Does the method detect the presence of water in the bottom of the tank?
Cl yes . . ,
' Dnq .
How does the method identify the presence of vapor pockets?
CD erratic temperature, level, or temperature-compensated volume readings
D sudden large changes in readings
IH statistical analysis of variability of readings
[H other (describe briefly) . ' ':
D not identified
D not applicable; underfilled test method used
Nonvolumetric TTT Method - Description ' Page 4 of 6
-------
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) .. , .. . . ; --.
..Q not corrected
D not applicable; underfilled test method used
How does the test method determine when tank deformation has stopped followina
delivery of product? . . a
. . " : . i ,
D wait a specified period of time before beginning test
Q watch the data trends and begin test when decrease in product level has stopped
D other (describe briefly) . .
D no procedure :
LJ not applicable, does not affect principle of operation
Are the method's sensors calibrated before each test? -
D yes . - .
-." D'no -
5 , . . ... -.' ' * . '
If not, how often are the sensors calibrated? . . '
. D weekly . >
CH monthly . ' - ;
EH yearly or less frequently
\; D never
> Interpreting test results
What effect is used to declare the tank to be7leaking? (List all modes used by the.method.)
If a change in volume is used to detect leaks, 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
C] 0.10." gallon per hour
D 0:20 gallon per hour . , .
- D other '''.'.-' ' ; .. ' - . ' <_
Nonvolumetric TTT Method - Description Page 5 of 6
-------
Under what conditions are test results considered inconclusive?
CD ground:water level above bottom of tank
EH 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) ' ' .' ' ' j
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 extremely high or low ambient temperature
D invalid for some products (specify) -'
D soil not sufficiently porous ......
D other (describe briefly) ' -
What are acceptable deviations from the standard testing protocol?
HU none
D lengthen the duration of test
ID other (describe briefly) '
What elements of the test procedure are left to the discretion of the testing personnel
on-site?
D watting period between filling tank and beginning test
EH 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
CD other (describe briefly) ' .- :
none
Nonvolumetric TTT Method - Description
Page 6 of 6
-------
Reporting Form for Leak Test Results
Nonvolumetric Tank Tightness Testing Method
Instructions for completing the form
This 3-page form is to be filled out by the evaluating organization upon
completion of the evaluation of the method in each of its leak detection
modes. This form provides for 60 test results, although the minimum
number of tests required in the protocol is 42. Use as many pages as
necessary to summarize all of the tests attempted. Report the results
for each leak detection mode on separate forms.
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
might use different versions of the equipment for different products or
tank sizes. Also, indicate the leak detection mode for which these
results were obtained.
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 com-
puter. 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 10 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 Sec-
tion 6.2 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.
B-18
-------
-------
Reporting Form for Leak Test Results
Nonvolumetric Tank Tightness Testing Method
Instructions for completing the form
This 3-page form is to be filled out by the evaluating organization upon
completion of the evaluation of the method in each of its leak detection
modes. This form provides for 60 test results, although the minimum
number of tests required in the protocol is 42. Use as many pages as
necessary to summarize all of the tests attempted. Report the results .
for each leak detection mode on separate forms.
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
might use different versions of the equipment for different products or
tank sizes. Also, indicate the leak detection mode for which these
results were obtained. .
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 com-
puter. The .form can also be filled out manually. The input for that form
wfll 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 10 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 Sec-
tion 6.2 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.
B-19
-------
Column No. Input ,
1 Test number or trial run.
2 Date at completion of last fill
(if applicable)
3 Time at completion of last fill
(if applicable)
4 Date test began
5 Time test began
,6 Time test ended
7 Product temperature differential
_^ (if applicable)
8 , Nominal leak rate
9 Induced leak rate
10 Leak test result
Source ! ,
Randomization design
Individual Test Log
Individual Test Log
Individual Test Log
Individual Test Log
Individual Test Log
Individual Test Log
Randomization design
Individual Test Log
Vendor's test result
Note: the product temperature differential (column 7) is the difference
between the temperature of the product added and that of the product in
the tank each time the tank is filled. This temperature 'differential .is
the actual differential achieved in the field and not the nominal
temperature differential.
B-20
-------
m
n(^>r
Reporting?%rm for Leak Test Results
Nonvolumetric Tank Tightness Testing Method
Method Name and Version:
Leak Detection Mode:
Evaluation Period: from
to
.(Dates)
Test No.
Trial Run
...'../......',. .'..
1
2
3
4
5
6
7
8
I 9 I
10
11
12
13
14
15
16
17
18
19
20
If applicable
Date at
Completion
of Last Fill
(m/d/y)
\
-
-
If applicable
Time at
Completion
ofLastRii
(military)
i
( , .
Date Test
Began
(m/d/y)
Time Test
Began
(military)
Time Test
Ended
(military)
If applicable
Product
Temperature
Differential
(deg F)
0
Nominal
Leak Rate
(gal/h)
0
Induced
Leak Rate
(gal/h)
0
11 < , '* MS''' , : s s ^
- ,
.
'
-
-
- -
Tank Tight?
(Yes, No, or
Test Invalid)
<.'''
.,
Nonvolumetric TTT-Data Reporting Form
Page 1 of 3
-------
ReportinSBrrm for Leak Test Results
Nonvolumetric Tank Tightness Testing Method
Method Name and Version:
Evaluation Period: from
Leak Detection Mode:
to
. (Dates)
Test No.
21
22
23
24
25
. 26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
!f applicable
Date at
Completion
of Last Fill
(m/d/y)
i
-
If applicable
Time at
Completion
of Last Fill
(military)
_
Date Test
Began
(m/d/y)
. .
Time Test
Began
(military)
.
Time Test
Ended
(military)
If applicable
Product
Temperature
Differential
(deg F)
. - - - ' - . - -
Nominal
Leak Rate
(gal/h)
induced
Leak Rate
(gal/h)
Tank Tight?
(Yes, No, or
Test Invalid)
Nonvolumetric TTT-Data Reporting Form
Page 2 of 3
-------
Reportirrtp^rm for Leak Test Results
Nonvolumetric Tank Tightness Testing Method
Method Name and Version:
Leak Detection Mode:
Evaluation Period: from
.to.
. (Dates)
Test No.
41
42
43
44
45
46
47-
48
49
. 50
51
52
53
54
55
56
57 .
58
59
60
If applicable
Date at
Completion
of Last Fill
(m/d/y)
If applicable
Time at
Completion
of Last Fill
(military)
Date Test
Began
(m/d/y)
- -i
Time Test
Began
(military)
-
Time Test
Ended
(military)'
If applicable
Product
Temperature
Differential
(deg F)
-
' . ' .
Nominal
Leak Rate
(gal/h)
Induced
Leak Rate
(gal/h)
. i,
Tank Tight?
(Yes, No, or
Test Invalid)
'-
..,-. -
,,
-
Nonvolumetric TTT-Data Reporting Form
Page 3 of 3
-------
-------
:*'
Individual Test Log
Nonvolumetric Tank Tightness Testing Method
Instructions for completing the form
This 5-page test Tog 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 including the trial run (at least 43.) The information
on these forms is to be kept blind to the vendor during the period of
evaluation of. the method. Adaptations of the form may be made as needed
to document the evaluation data.
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. For tracer methods only
6. Conditions at beginning of test
7. Conditions at completion of Resting . . . .
8. Leak rate data .
9. Additional comments, if needed
10. -Data sheet for leak simulation for tracer methods
11. Data sheet for induced leak rate calibration
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 identifi-
cation if the method uses different versions for different products or
tank sizes. The vendor's recommended stabilization period (if appli-
cable) 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.
B-24
-------
*
-------
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. , v - ..
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.
Note that the term "conditioning" refers to all activities undertaken by
the evaluating field crew to prepare for a test. As such, the term
refers to emptying or filling the tank, heating or cooling product, and
changing the leak rate. In some .cases, all of the above is performed, in
others, only one parameter might be changed. For tracers, "conditioning"
refers to preparation of the tank for testing. It includes the determin-
ation of the time to wait between spiking and testing.
Topping Off Records (if applicable)
If this step is performed, fill in the'appropriate blanks.
For Tracer Methods Only
Fill in the appropriate information. Follow the instructions and
complete the form on page 4.
Conditions at Beginning of Test
The, evaluation organization's field crew will have calibrated the leak
simulation equipment prior to the test. All leak rate calibration data
need to be documented using the form on pages 4 or 5, as appropriate.
Refer to previous calibration if this has been done. Adapt the form as
necessary.
Once the evaluating"organization's field crew is ready with the induced
.leak rate simulation, and the vendor 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. '
B-25
-------
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 (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 from the simulation data reported by
the evaluating field crew on page 4 or 5 of this form.
The test result is that obtained by the vendor for that test.
Give the mode being investigated on the line following the test answer if
the method uses more than one mode of leak detection.
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.
Leak Simulation Form for Tracer Methods (page 4)
For tracer methods, use the form on page 4 to document and measure
delivery of the carrier with the appropriate concentration of the tracer
to the spiking ports. Indicate the tracer used and the concentration of
tracer in the carrier in the appropriate spaces. Report the distances
between spiking port and all sampling ports. Record the time and amount
of material released in the spiking port to document the leak simulation
for. tracer methods. Use as many pages as needed.
Induced Leak Rate Calibration Form (page 5)
For acoustical methods, the form on page 5 may be used to calibrate the
liquid flow through the simulator under a standard set of conditions.
The induced leak rate is the rate at which the liquid will flow at a
specified head or depth of product. This rate is determined by
calibration and used as the leak rate for detection. The calibration
will have to be done at a different time, preferably before) than the
testing. A calibration is needed for each distinct leak rate. Once the
calibrations have been done, document on each daily_test log the simula-
tion conditions'and reference the appropriate calibration data sheets,
which should be attached to the daily test log that first uses the given
induced leak rate.
B-26
-------
Name of Field Operator . ;. ,'.
Signature of Field Operator . Test No.
Date of Test . ;1 .
individual Test Log
Nonvolumetric 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 feyel inches above bottom of tank
Recommended stabilization period .before test (per vendor SOP) . - -
hours minutes
Conditions Before Testing
Date and military time at start of conditioning test tank
Stick reading before partial emptying of tank
Product inches gallons s
, _ / , - - -
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
Stick reading after filling tank to test level _
Product inches gallons
Water inches gallons '
Amount of product added to fill tank (by subtraction) gallons
Nonvolumetric'TTT Method - Test Log Page 1 of 5
-------
Name of Field Operator . r___ "'
Signature of Field Operator ' Test No._ ' . ' '
DateofTest__.
Conditions Before Testing (continued)
Temperature of product added to fill tank °F D qr °C D
Temperature of product in tank immediately after filling °F EH or °G CD
Date and military time at completion of fill
Topping Off Records (if applicable)
Date . and military time__ at completion of topping off
Approximate amount of product added " . : gallons
If tank overfilled, height of product above tank inches
For Tracer Methods Only
Date " and military time tracer(s) is added to product in test tank _
Tracer used ; ,
Amount of tracer used
Amount of product in test tank . gallons
-iuir-nr- , ^ ,
> Complete the Tracer Leak Simulation form (use page 4)
Date and military time at start of test -
Date and military time at conclusion of test
Conditions at Beginning of Test
Date 1 and military time vendor began setting up test equipment
> Document induced leak rate determination (use page 5)
Date and military time at start of vendor's test data collection
Temperature of product in tank at start of test °FD or °clZl .
Weather Conditions
Temperature °FD or°cD ' -' '-...'
Barometric pressure jnm Hg D or in. Hg d
Wind NoneD Light D Moderate D Strong Q
Precipitation None CD Light HH Moderate D Heavy D
Sunny D Partly Cloudy D Cloudy D ...
Nominal leak rate gallon per hour
Nonvolumetric TTT Method - Test Log . . Page 2 of 5
-------
Narne-of Field Operator . '-_..'
Signature of Field Operator jest No.
Conditions at Completion of Testing
Pate and military time_ at completion of test data collection
Stick reading at completion of test data collection
Product inches gallons .
Water inches gallons
Date of Test_v
Conditions at Completion of Testing (continued) . ""~"~~
Weather Conditions [ ' '
Temperature °F CD or °C D . .
Barometric pressure mm HgCD or in. HgD ,
Wind NoneD , LightD Moderate CD Strong CD
Precipitation None CD Light EJ Moderate D Heavy D
Sunny D Partly Cloudy D ' Cloudy CD
Date_ and military time test equipment is disassembled (if done
for this test) and tank is ready for service .
Leak Rate Data "~ " "" " : ~ ~~
Leak detection mode__ ''-./ ' ;.'
Nominal leak rate gal/h
Induced leak rate gal/h .
"*"' ' '- * '
Findings for Tracer Methods
r .
CD No tracer found DTracer(s) found
If tracer(s) found, list '
Test answer CD leaking CD tight CD inconclusive
Additional Comments (Use back of page if needed")"
NonvolumetricTTT Method-Test Log - Page 3 of 5
-------
Name of Field Operator -
Signature of Field Operatpr.
Date of test
Test Nci.
Leak Simulation Form for Tracer Method
(Reproduce form if needed)
Tracer used.
Carrier
Concentration of tracer in carrier.
Distance from spiking port to:
Sampling port 1
Sampling port 2
Sampling port 3 ,
Sampling port 4
Sampling port 5.
Sampling port 6.
Sampling port 7.
Sampling port 8,.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Time
(military)
(
. *
Carrier amount
released in
spiking port
'
Comments
' .
, -' , . . " ~
,'/*'
Indicate all measurement units!
Nonvolumetric TTT Method-Test Log
Page 4 of 5
-------
Name of Field Operator
Signature of Field Operator.
Date of test -
Test No.
induced Leak Rate Calibration Form
(Reproduce form if needed)
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
(military)
'
Amount *
-
.-
Comments , :
' - ' -
-
' ' . ^
t -
- _ '
" - . '. - .
, . - .'."
. ^
" ' ' , ' - '
- '..'
-. '...,'
* Indicate all measurement units!
Nonvolumetric TTT Method-Test Log
Page 5 of 5
-------
-------
Reporting Form for Water Sensor Evaluation Data
Nonvolumetrtc Tank Tightness Testing Method
This 4-page form is to be filled out by..the field crew of the evaluating
organization when evaluating the performance of the method's water
sensor, if applicable. A separate form is to be filled out for each
individual test replicate (at least 20). The form provides a template to
record the data and consists of three parts. These are:
1. Header information
2. Template for recording the data obtained to determine the minimum
water level that the sensor can detect in each-replicate (page 1)
3. Template for recording the data obtained when determining the
minimum water level change that the sensor can detect in each
replicate (pages 2-4).
Header Information
The header information is to be repeated on all four pages, if used. ' If
a page is not used, cross it out and initial it.
Indicate the commercial name of the method. Include a version identifi-
cation if the method uses different versions for different products or
tank-sizes. Complete the date of test'and product type information.
Indicate the test (replicate) number on each sheet for each test. .
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.
Minimum Detectable Water Level Data
Follow the test protocol described in .Section 6.5 arid record all data on
page 1 of the form. When the sensor first detects the water, stop test-
ing for this replicate. The minimum detected water level is calculated .
from the total amount of water added until the first sensor response and
the geometry of the probe and the cylinder. This calculation can be done
after all testing is completed and is generally performed by the statis-
tician or other person responsible for data analysis.
Minimum Detectable Water Level Change
After the first sensor response, continue with the test protocol as
described in Section 6.5. Record all amounts of water added and the
sensor readings at each increment using pages 2 to 4 as necessary. The
data to be entered in the third, fifth,, and sixth columns on pages, 2, 3,
and 4 of the form will be calculated once all testing is completed.
Again, the person responsible for the data analysis will generally~~
compute these data and enter the calculated minimum water level detected
in that replicate run.
B-32
-------
-------
Reporting Form for Water Sensor Evaluation Data
Nonvolumetric Tank Tightness Testing Method
Method Name and Version:
Date of Test:
Product Type:
Name of Field Operator: _
Signature of Field Operator:
Increment
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
25
Total
Volume
(mL)
Volume of
Water Added
(mL)
.
Sensor
Reading
(inch)
,
>
-
Test No.
Calculated Minimum
Detectable Water Level (inches)
NOTE:
This form provides a template for data reporting. Since the number of
increments is not known from the start, the length of the report form
will vary from test to test.
Nonvolumetric TTT-Water Sensor
Page 1 of 4
-------
Reporting Form for Wafer Sensor Evaluation Data
Nonvolumetric Tank Tightness Testing Method
Method Name and Version:
Date of Test:
Product Type:
Name of Reid Operator: _
Signature of Field Operator:.
Test No.
Increment
No.
A
Volume of
Water Added
(mL)
B
Calculated
Water Height
Increment, h
(in)
C
Sensor
Reading
(in)
- D
Measured
Sensor
Increment
(in)
E
Increment
Difference
Caic.-Meas.
(in)
C-E
Minimum water level detected, X: inches (from page 1)
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 .
.
,
i '"
' I
' m
4
1
f
NOTE: This form provides a template for data reporting.
Use as many pages as necessary.
Nonvolumetric TTT-Water Sensor
Page 2 of 4
-------
Reporting Form for Water Sensor Evaluation Data
Nonvolumetric Tank Tightness Testing Method
Method Name and Version:
Date of Test: '
Product Type: --
Name of Reid Operator:.
Signature of Field Operator:
Test No.
Increment
No.
A
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Volume of
Water Added
(mL)
B
, P
.
;
Calculated
Water Height
increment, h
(in)
C
Sensor .
Reading
On)
D
',. ,
Measured
Sensor
Increment
(in)
E
*
' . ' ,
Increment
Difference
Caic.-Meas.
(in)
C-E
/
NOTE: This form provides a template for data reporting.
Use as many pages as necessary.,
Nonvolumetric TTT-A/Vater Sensor
Page 3 of 4
-------
Reporting Form for Water Sensor Evaluation Data
Nonvolumetric Tank Tightness Testing Method
Method Name and Version:
Date of Test:
Product Type: ;
Name of Field Operator: ,
Signature of Field Operator:.
Test No.
Increment
No.
A
51
52
53
54 ,
55
56
57
58 .
59
60
61
62
63-
64
65
66
67
68
69
70
71
72
73
74
75
Volume of
Water Added
(mL)
B
Calculated
Water Height
Increment, h
(in)
C
.
Sensor
Reading
(in)
D
. " > !
. ,-.
Measured
Sensor
. Increment
(in)
E
' ;
Increment
.Difference
Calc.-Meas.
(in)
C-E
>
'
it
H
:
_ .
NOTE: This form provides a template for data reporting.
Use as many pages as necessary.
Nonvolumetric TTT-Water Sensor
Page 4 of 4
-------
-------
------- |