v>EPA
             United States
             Environmental Protection
             Agency
             Solid Waste And
             Emergency Response/
             Research And Development
             5403W
EPA/530/UST-90/006
March 1990
Standard Test Procedures
for Evaluating Leak
Detection Methods

Automatic Tank
Gauging Systems
               s\% ^f^t ^N.W^*'^'^'^"

             -0'-X.X^ V'-'S^'X.X:
             s   •• ».s s; * ^ ^\^<:\-.s ^
              ''" '-  >S";>" -\ -"-"" ^s^\>./
                           ss -K\Os% X f s •" s * "S >** j. S f

                                  --v^x -5?-" -'4,^
                                  -s.^s^-  ^
                                  **' *^xv \ ^
                                         *< S *X»S XV. S ' *A*f y*Si>\ * \ff -f' ff
                                         v-f**->f $ *^^x A *& ^y&?'» s»
                                         s '•'•/•X;-' < •• «• sW''?^ sy^ s *\ *$ff*-
-------

-------
       Standard Test Procedures for
Evaluating Leak Detection Methods:
  Automatic Tank Gauging Systems
                           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
automatic tank gauging (ATG) systems must be capable of detecting a
0.20 gallon per hour leak rate with a probability of detection of at
least 95% and a probability of false alarm of no more than 5%.  It is up
to tank owners and operators to select a method of leak detection that
has been shown to meet the relevant performance standard.

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

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

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

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

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

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


EPA Standard Test  Procedures

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

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

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

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

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

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

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

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

Each  test procedure provides an explanation of how to conduct the test,
how to perform the required calculations, and how to report the
results.   The results from each standard test procedure provide the
                                    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  EPA's

      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 condi-
     tion and an induced-leak condition with an induced leak rate as
     close as possible to (or smaller than) the pesrformance stan-
     dard.  In the case of ATG systems., for example, this will mean
     testing under both 0.0 gallon per hour and 0,20 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 ATGS
     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., and
Karin M. Bauer for the U.S. Environmental Protection Agency's Office of
Underground Storage Tanks (EPA/OUST) under contract No. 68-01-7383.  The
Work Assignment Manager for EPA/OUST was Thomas Young and the EPA/OUST
Project Officer was Vinay Kumar.  Technical assistance and review were
provided by the following people:

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

     American Petroleum Institute
     Leak Detection Technology Association
     Petroleum Equipment Institute
                                  vii

-------

-------
                                 CONTENTS


 Foreword	
     )wle

      1.   Introduction
Acknowledgments	!!!!!!!!!!	 «i}
               1.1  Background.	       	    i
               1.2  Ob ject i ves	„	!!!!!!!!!!    2
               1.3  Approach	„	!!!!!!!!!!!!!    2
               1.4  Organization of this document	!!!!!!!!!    3
      2.   Scope and  Applications	!*"*    4
      3.   Summary	„.                	    7
      4.   Safety	!!!!!!!!!!!!!!!!!"**    9
      5.   Apparatus  and Materials			!!!!!!!!!!!!!!  11
               5.1  Tanks	!.!!!!!!!!!!!!  n
               5.2  Test equipment	!!!!!!!!!!!!!  12
               5.3  Leak simulation equipment	!!!!!  12
               5.4  Product	!!!!!!!  13
               5.5  Water sensor equipment	!!!!!!!  13
               5.6  Miscellaneous equipment	!  13
      6.   Testing Procedure	!!!  15
               6.1  Environmental data records	.....!!!!!!!!!*  17
               6.2  ATGS leak detection mode	!!!!!!  17
               6.3  Testing problems and solutions	'.'.'.'.  23
               6.4  ATGS evaluation protocol for water detection	!  23
      7.   Cal cu 1 ati ons	  29
               7.1  ATGS leak detection mode	!!!!!!!!!!!!!!!!!!!!!  29
               7.2  ATGS water detection mode	'.'.'.'.'.  36
               7.3  Supplemental calculations and data analyses
                    (optional)	  41
               7.4  Outline of calculations for alternative approach!!  51
     8.   Interpretation	  55
              8.1   Leak test function evaluation	!!!!!!!!!!!!!!!  55
              8.2  Water level detection function	  55
              8.3  Minimum water level change measurement	  56
     9.  Reporting of Results	  59
Appendices
     A.  Definitions and Notational Conventions	A-l
     B.  Reporti ng Forms	! g-1

-------

-------
                                 SECTION 1

                                INTRODUCTION
 1.1  BACKGROUND
      The regulations on underground storage tanks (40 CFR Part 280
 Subpart D)  specify performance standards for leak detection methods that
 are  internal  to  the tank.   For automatic tank gauging (ATG) systems  the
 system must be capable  of  detecting a leak of 0.20 gallon per hour with a
 probability of (at least)  95%, while operating at a false alarm rate of
 5% or less..

      The regulations for ATG  systems require (1)  that automatic product
 level  monitor test be able to detect a 0.20 gallon per hour leak from any
 portion of  the tank that routinely  contains product and  (2)  that its
 automatic inventory function  meet the requirements for inventory con-
 trol.   That is,  the equipment must  be capable of:

      •    measuring the  height of the liquid to the nearest  one-eighth of
          an inch.

          measuring any  water  in the bottom  of the  tank at least  once  a
          month to  the nearest one-eighth of an inch.

          conducting  daily  reconciliation of the inventory.

          declaring  a leak  on  the basis of the  inventory reconciliation  if
          the discrepancy exceeds 1% of the  flow-through plus 130 gallons
          on a monthly basis.

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

-------
1.2  OBJECTIVES

     The objectives of this protocol are twofold.  First,  it provides  a
procedure to test ATG systems in a consistent and objective manner.
Secondly, it allows the regulatory community and regulators to verify
compliance with regulations.  This protocol provides a standard method
that can be used to estimate the performance of an ATGS.   Tank owners  and
operators are required to demonstrate that the method of leak detection
they use meets the EPA performance standards of operating  at (no more
than) a 5% false alarm rate while having a probability of  detection of
(at least) 95% to detect a leak of 0.20 gallon per hour.   This demon-
stration must be made no later than December 22, 1990.  The test
procedure described in this protocol is one example of,how this level of
performance can be proven.  The test procedure presented here is
specific, based on reasonable choices for a number of factors.
Information about other ways to prove performance is provided in the
Foreword of this document.

     It should be noted that this protocol only evaluates  the leak test
function and the water sensing function of the ATGS since  they are
considered the primary leak detection modes.  The protocol does not
address the inventory function of the ATGS.  Also, this protocol does not
address the issue of safety testing of equipment or operating
procedure.  The vendor is responsible for conducting the testing
necessary to ensure that the equipment is safe for use with the type of
product being tested.


1.3  APPROACH

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

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

-------
      An  important  function of an ATGS is its ability to detect water in
 the  product  and  to track  the  water level in the tank as a means of
 detecting  leaks  when  a high water table is present.   Since the ATGS  acts
 as a continuous  monitor with  the tank in a normal  operating condition
 the  relation of  the product height to the height of  the ground-water
 level outside the  tank varies,-.producing different relative pressures as
 the  product  level  changes during use.   One part of most ATG systems  is to
 detect the possible incursion of water.   In evaluating  the water sensor
 the  minimum  water  level that  the system can detect,  and the smallest
 change in water  level  that the system can reliably measure,  are deter-
 mined.  The  performance of the ATGS  is evaluated on  its ability to detect
 a hole in the tank  by  measuring  the  incursion of water  into  the
 product.


 1.4  ORGANIZATION OF THIS DOCUMENT

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

     Two appendices are included in this document.   Definitions of some
technical terms are provided in Appendix A.  Appendix B  presents a com-
pendium of forms:  a standard reporting form for the evaluation results
a standard form for describing the operation of the ATGS, data reportina
forms, and individual  test logs.

-------

-------
                               .  SECTION 2

                          SCOPE AND APPLICATIONS


     This document presents  a  standard  protocol for evaluating ATG
 systems.  It  is designed to  evaluate  systems  that  are  installed in the
 tank and monitor product volume  changes on  a  continuous  basis  during  the
 test period.  The protocol is  designed  to evaluate the leak  detection
 functions of  an ATGS.  These functions  are  the test mode, water detec-
 tion, and water level monitoring.  The  evaluation  will estimate the
 performance of the system's  test mode and compare  it with the  EPA
 performance standards of a false alarm  rate of (no more  than)  5% and  the
 Probability of detecting a leak of 0.20 gallon per hour  of (at least)
     The protocol provides tests to determine the threshold of water
detection for the ATGS.  In addition, the protocol tests the ability of
the water sensor to measure changes in the water level and compares the
results to the EPA performance standard of 0.125 inch.  These are
evaluated over a range of a few inches in the bottom of the tank.  The
threshold and height resolution of the water detector are converted to
gallons using the geometry of the tank.

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

-------

-------
                                 SECTION 3

                                  SUMMARY


     The evaluation protocol  for ATG  systems  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.  A tank tightness test in the 6 months preceding  testing  that
         indicates a tight tank.

     2.  At least three ATGS  records with a different AT6S than that
         being tested within  a 3-month  period with  inventory and  test
         modes indicating a tight tank.

     3.  A 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 system under  investigation, constitutes acceptable
evidence.  This information should be reported on the  data reporting  form
(see Appendix B).

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

     The ATGS is installed in the test tank and used to measure a leak
rate under the no-leak condition and with three induced leak rates of
0.10, 0.20, and 0.30 gallon per hour.   A total number of at least
24 tests is to be performed.  The tank must be 50% full for half the
tests.   It is refilled to about 90% to 95% full for the other 12 tests.
When filling the tank, product at least 5°F warmer than that in the tank
is used for one third of the fillings  and product at least 5°F  cooler
than that in the test tank is used for one third  of the fillings.   The
other third of the fillings uses product at the same temperature.   The
ATG system's ability to track actual volume change is determined by the

-------
difference between the volume change rate measured by the test device and
the actual, induced, volume change rate for each test run.  These
differences are then used to calculate the performance of the method.
Performance results are reported on the Results of U.S. EPA Standard
Evaluation form included in Appendix 8 of this document.

     The ability of the system to measure water in the bottom of a tank
is tested by placing the system in a standpipe containing product.  Mea-
sured amounts of water are added and the ability of the system to sense
the water at given depths is determined experimentally.  These results
are also reported on the standard form in Appendix B.
                                    8

-------
                                 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 system 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:

         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 system'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.
                                    9

-------

-------
                                 SECTION 5

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

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

     Because performance of internal  tank test methods is generally worse
for large tanks, the size of  the test tank is important.  An 8,000-gallon
tank is recommended because this appears to be the most common tank  in
use.  However, testing may be done in tanks of any size. The results of
the evaluation would be applicable to all  smaller tanks. The results are
also applicable to larger tanks with  the restriction  that the tanks  be no
more than 25% larger in capacity than the test tank.   That  is, results
from a 6,000-gallon tank can also be  applied  to tanks of up to
7,500 gallons in capacity.  Results from 8,000-gal tanks can  be applied
to tanks up to 10,000 gallons, those  from 10,000  gallons to up to
12,500 gallons,  etc.   If the method is intended to test  larger tanks,
e.g., 20,000 gallons, it must be evaluated  in a tank  within 25% of that
size.
                                   11

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


 5.2  TEST  EQUIPMENT

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

      Since ATG systems are installed permanently and left to the tank
 owner to be operated, it  is recommended  that the ATGS equipment being
 tested be  operated by the  evaluating organization  personnel.  The ATGS
 equipment  is normally operated by  the station  owner, so the evaluating
 organization should provide personnel  to operate the equipment after the
 customary  training.

 5.3  LEAK  SIMULATION EQUIPMENT

      The protocol calls for inducing leaks in  the  tank.  The method of
 inducing the leaks must be compatible with the leak detection  system
 under test.  This is done  by  removing product  from the tank at a constant
 rate, measuring the amount of product removed  and  the time of  collection,
 and calculating the resulting induced leak rate.   The experimental  design
 described  in Section 6 gives  the nominal leak  rates that are to be
 used.

      A method that has been successfully used  for  inducing leaks in pre-
 vious testing is based on  a peristaltic  pump.   An  explosion-proof motor
 is  used  to drive a peristaltic pump head.  The sizes of the pump head and
 tubing are chosen to provide  the desired flow  rates.  A variable speed
 pump  head  is used so that  different flow rates can be achieved with the
 same  equipment.  The flow is  directed through  a rotameter so that the
 flow  can be monitored and  kept constant. One  end  of the tubing is
 inserted into the product  in  the tank.  The  other  end is placed in a con-
 tainer.  Typically, volatile products are collected into a closed
 container  in an ice bath.   The time of collection  is monitored, the
 amount of  product weighed, and the volume at the temperature of the tank
 is  determined to obtain the induced leak rate. While 1t is not necessary
 to  achieve the nominal leak rates  exactly, the induced leak rates should
 be  within  ±30% of the nominal rates.  The induced  leak rates should be
 carefully  determined and  recorded.  The  leak rates measured by the ATGS
 will  be  compared to the induced  leak rates.
                                    12

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

      Any commercial petroleum product of grade number 2 or liqhter mav 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.

 j-^ The test plan recluires 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
 I?5V??tS.^9ln ?y the tank bei"9 filled from about ha^ full  to  90% to
 95% full  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  placinq
 heating  and cooling coils in the supply tank  or tank truck before  the
 fuel  is  transferred to the test  tank.


 5.5  WATER SENSOR EQUIPMENT

      The  equipment  to  test the water  sensor consists of a vertical
 cylinder  with  an  accurately  known  (to ±0.001  inch) inside diameter.  This
 S  !S5c  sh°uld be  larqe  enough  to  accommodate the water sensor part of
 the ATGS.   Thus,  it  should be approximately 4 inches in diameter and 8 or
 more  inches high.  A means of mounting the  ATGS so that its water  sensor
 is  in the  same relation to the bottom of the cylinder as it would  be to
 the bottom of a tank is needed.  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.


 5.6  MISCELLANEOUS EQUIPMENT

     As noted, the test procedure requires the partial emptying  and
tilling of the test tank.  One or more fuel pumps of fairly larqe
capacity will be required to accomplish the filling in a reasonably short
time.  Hoses or pipes will be needed for fuel  transfer.   In addition,
containers will be necessary to hold the product collected from  the
induced leaks.  A variety of tools need to be on hand for  makinq the
necessary connections of equipment.
                                   13

-------

-------
                                SECTION  6

                            TESTING PROCEDURE


     The evaluation protocol for AT6 systems consists of two parts.  The
first evaluates the leak detection function of the ATGS.  The second
evaluates its water detection function and the system's resolution of
water sensing.

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

     The test procedure introduces four main factors that may influence
the test:  size of leak, amount of product in the tank, temperature dif-
ferentials, and tank deformation.  An additional  factor is the method's
ability to deal with ground-water level effects.   This factor is
evaluated when determining the system's water sensing threshold and
resolution.

     The primary consideration is the size of the leak.  The system is
evaluated on its ability to measure or detect leaks of specified sizes.
If a system cannot closely measure a leak rate of 0.20 gallon per hour or
if the system demonstrates excessive variability  on a tight tank, then
its performance is not adequate.  The ability of  the system to track the
leak rates can be compared for the different leak rates.

     The second consideration is the temperature  of product added to fill
a tank to the level needed for testing.  Three conditions  are used:
added product at the same temperature as  the in-tank product,  added
product that is warmer than that already  in the tank,  and  added  product
that is cooler.  The temperature difference should be  at  least 5°F and
should be measured and recorded to the nearest degree  F.   The temperature
difference is needed to ensure that the system can adequately test under
                                   15

-------
realistic conditions.  The performance under the three temperature condi-
tions can be compared to determine whether these temperature conditions
have an effect on the system's performance.

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

     The fourth consideration is the amount of product in the tank.
Since ATG systems work at different levels of product in the tank, the
required monthly test may be done at various levels.  Two levels have
been chosen to represent these product levels.  One is half full, which
requires the most sensitive level measurement.  The other is 90% to 95%
full, which requires the most sensitive temperature compensation.

     In addition to varying these factors, environmental data are
recorded to document the test conditions.  These data may 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.  The water sensing function
of the ATGS is used to detect leaks in the presence of a ground-water
level above the bottom of the tank.  If the ground-water level is high
enough so that there is an inward pressure through most levels of product
in the tank, then water will come into the tank if there is a hole below
the ground-water level.  Since an ATGS must operate at normal operating
levels of product in the tank, it uses water incursion to detect leaks if
there is a high ground-water level.  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.
                                    16

-------
 6.1  ENVIRONMENTAL DATA RECORDS

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

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

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

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

          type of  product in tank
          tank volume
          tank dimensions and type
          amount of water in tank (before  and  after each  test)
          temperature  of  product in  tank before  filling
          temperature  of  product added each time the  tank  is filled
          temperature  of  product in  tank immediately  after  filling
          Temperature  of  product in  tank at start of  test


 6.2 ATGS LEAK  DETECTION MODE

     The  following presents the  test conditions and  schedule to determine
 the performance of the ATGS.


 6.2.1  Induced  Leak Rates, Temperature Differentials, and Product
       Volume

     Following a trial run in the tight tank, 24 tests will be performed
according to the experimental design exemplified in Table 1.  The actual
design will be randomized for each system.  In Table 1, LRn- denote the
nominal leak rates and T^ denote the temperature differentials to be used
in the testing.  These 24 tests evaluate the method under a variety of
conditions.
                                   17

-------
          Table 1.  PRODUCT VOLUME, LEAK RATE, AND TEMPERATURE
                       DIFFERENTIAL TEST SCHEDULE




Trial run
Empty to 50% full (if
Fill to 90-95% full


Empty to 50% full


Fill to 90-95% full


Empty to 50% full


Fill to 90-95% full


Empty to 50% full


Fill to 90-95% full


Empty to 50% full


Fill to 90-95% full


Empty to 50% full


Fill to 90-95% full


Empty to 50% full




Test
No.
-


Pair
; NO.
-


Set
No.
-
Nominal
leak rate
(gallon
per hour)
0.00
Nominal
temperature
differential*
(degree F)
0
applicable)

1
2

3
4

5
6

7
8

9
10

11
12

13
14

15
16

17
18

19
20

21
22

23
24

1
1

2
2

3
3

4
4

5
5

6
6

7
7

8
8

9
9

10
10

11
11

12
12

1
1

1
1

2
2

2
2

3
3

3
3

4
4

4
4

5
5

5
5

6
6

6
6

LRi
LR2 •

LR*
LR3

LRt
LR^

LR2
LR3

LR»
LRi
1
LR3
LR2

LR3 i
LR^

LR2
LRi

LR2
LR3

LR*
LRi

LR3
LR2

LR^
LRi

T2
T2

T2
T2

T!
T!

Tx
' T!

T3
T3

T3
T3

T2
T2

T2
T2

T!
Tt

Tx
Tj

T3
T3

T3
T3
*Note:  The temperature differential is calculated as.the temperature of
        the product added minus the temperature of the; product in the
        tank.

                                   18

-------
 Leak Rates

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

                English  units  .           Metric units
             (gallon per hour)  '      (milliliters per minute)

                    0.00                        0.00
                    0.10                        6.3
                    0.20                       12.6
                    0.30                       18.9

 Temperature  Differentials

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


 Product Volumes

      The  tests  will  be  run  in  sets of two pairs,  holding  the  temperature
 differential constant within a set of four tests  but  changing the  leak
 rate within  each pair.   The product  volume will alternate from pair to
 pair.  The first pair of tests within a  set  will  be run with  the  tank
 filled to 90% to 95% capacity.   Then the tank  will  be emptied to  50% full
 and  the second  pair  of  tests in  the  set  will be run.


 Randomization

      A total of 24 tests  will  be performed by  inducing the  12  combina-
 tions of  the four  leak  rates (LRlt LR2,  LR3, and  LRJ and the  three tem-
 perature  differentials  (Tr, T2,  and  T3)  at the two  product  volumes (50%
 full  and  90% to 95%  full) as outlined in Table 1.

     The  randomization  of the  tests  is achieved by  randomly assigning the
 nominal leak rates of 0,  0.10, 0.20  and 0.30 gallon per hour to LR1S LR2,
 LR3,  and  LR^ and by  randomly assigning the nominal temperature differen-
 tials of  0°, -5°,  and +5°F to T1$ T2, and T3, following the sequence of
 24 tests  as shown  in Table  1.  The organization performing the evaluation
 is responsible  for randomly assigning the four leak rates to LR^ LR2,
 LR3, and  LR% and the three temperature conditions to Tlt T2, and T3.  In
 addition, the evaluating organization should randomly assign the groups
 of four tests to the set numbers 1 to 6, without disturbing the order of
 the four tests within a set.

     The vendor will install the ATGS and train the evaluating organiza-
 tion to operate it.  After the trial  run the ATGS will be operated as it
would be  in a commercial establishment.   The evaluating organization will


                                   19

-------
operate the ATGS and record its data.  Note that since an ATGS operates
automatically, it is not necessary to keep the induced leak rates blind
to the operator.  The operator merely starts the leak detection function
of the ATGS at the appropriate time and records the results.  The random-
ization is used to balance any unusual conditions and to ensure that the
vendor does not have prior knowledge of the sequence of leak rates and
conditions to be used.

     In summary, each test set consists of two pairs of tests.  Each pair
of tests is performed using two induced leak rates, one induced tempera-
ture differential (temperature of product to be added •• temperature of
product in tank), and one in-tank product level.  Each pair of tests
indicates the sequence in which the product volumes (in gallon per hour)
will be removed from the tank at a given product temperature
differential.
Notations! Conventions

     The nominal leak rates, that is 0, 0.10, 0.20, and 0.30 gallon per
hour, after randomizing the order, are denoted by LRj.'LRj., LR3, and
LRjj.  It is clear that these figures cannot be achieved exactly in the
field.  Rather, these numbers are targets that should be achieved within
±30%.

     The leak rates actually induced for each of the 24 tests will be
measured during each test.  They will be denoted by S11( S2,».., S2tf.
These are the leak rates against which the leak rates obtained by the
vendors performing their tests will be compared.

     The leak rates measured by the ATGS during each of the 24 tests will
be denoted by Llf L2,...,L2i, and correspond to the induced leak rates Slf
^      ^
O2, ... ,O2I|..

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

                                                       I
6.2.2  Testing Schedule

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

     There are two purposes to this trial run.  One is to allow the
vendor to check out the ATGS equipment and provide instructions to the
operators before starting the evaluation.  As part of this check, any
faulty equipment should be identified and repaired.  A second part is to
ensure that there are no problems with the tank and the test equipment.


                                    20

-------
Such  practical  field  problems  as  leaky valves or plumbing problems should
be  identified and  corrected with  this trial  run.  The results also pro-
vide  current verification  that the tank is tight and so provide a base-
line  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, temperature differentials,  and in-tank product levels
as  shown  in Table  1 above.  The time  lapse between the two tests in each
pair  should be  kept as short as practical.  The date and time of starting
each  test are to be recorded and  reported in the test log.  Twelve pairs
of  tests  will be carried out.   After  each pair  of tests, the test proce-
dure  starts anew with  either emptying the tank  to half full  or filling  it
up  to 90% to 95% capacity, stabilizing,  etc.  The details of the testing
schedule  are presented next.

Step  1:   Randomly assign  the  nominal  leak rates of  0,  0.10,  0.20,  and
          0.30  gallon  per  hour to LR1S  LR2,  LR3,  and LRH.  Also,  randomly
          assign the temperature  differentials  of 0°,  -5°, and  +5°F to
          TU T2,  and  T3.  Randomly assign the  groups  of four tests to
          the 6 sets.   This will  be done  by  the evaluating organization
          supervising  the  testing.

Step  2:   Follow the vendor's  instructions and  install  the ATGS  in  the
          tank.  Also  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 ATGS.  Perform
          any calibration or operation checks needed with the
          installation of the  ATGS.

Step  3:   Trial run.   Following the test  system's standard operating
          procedure, fill  (if  needed) the tank to the recommended level
          for operation in the  Teak detection mode, and allow for the
          stabilization period  called for by the system 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 ATGS equipment.
          Perform  any  necessary repairs or modifications identified by
          the trial run.

Step  4:   Empty the tank to 5Q% full if the product volume was above that
          level  during the trial run.

Step  5:   Fill  the tank to 90% to 95% capacity.   Fill with product at the
          temperature required by the randomized test schedule.  The
          temperature differential will be T2 (Table 1, Test No. 1).
          Record the date and time at the completion of the fill.  Allow
          for the recommended stabilization period, but not longer.

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


                                    21

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

Step 6:   Continue with the system's standard operating procedure and
          conduct a test on the tank, using the system'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
          L.RX.

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

Step 7:   Change the nominal leak rate to the second in the first set,
          that is LR2 (see Table 1).  Repeat Step 6.  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 (dates and times, measured
          and induced leak rates, temperatures, calculations, etc.).

Step 8:   Empty the tank to 50% capacity (to within ±6 inches of the tank
          midpoint).  The temperature of the in-tank product will remain
          unchanged.

Step 9:   Change the nominal leak rate to the third in the first sest,
          that is LR^.  Repeat Step 6.  Record all results.

Step 10:  Change the nominal leak rate to the fourth in the first set,
          that is LR3.  Repeat Step 7.  Record all results.

Step 11:  Repeat Step 5.  The temperature differential will be changed to

          T»-
Step 12:  Repeat Steps 6 through 10, using each of the four nominal leak
          rates of the second set, in the order given in Table 1.
                                    22

-------
      Steps 5 through 10, which correspond to a fill and empty cycle and
 one set of two pairs of tests, will be repeated until all 24 tests are
 performed.


 6.3  TESTING PROBLEMS AMD SOLUTIONS

      Inevitably,  some test runs will  be inconclusive due to broken equip-
 ment,  spilled 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  rule applies.

 Rule 1:    The total  number of tests must be  at least 24.   That  is, if a
           test is invalid, it needs to be rerun.  Report  the test results
           as  invalid together with the reason  and repeat  the test.

 Rule 2:    If  equipment fails  during the first  run (first  test of a set of
           four tests)  and  if  the  time needed for fixing the  problem(s)  is
           short (less  than 20% of the stabilization time  or  less than
           1 hour, whichever is greater),  then  repeat that  run.   Other-
           wise, repeat the empty/fill  cycle, the stabilization period,
           etc.  Record all  time periods.

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

 Rule 3:    If  equipment fails  during a  later test  (after the first  run in
           a set of four  has been  completed successfully), and if the time
           needed  for fixing the problem(s) is  less than 8 hours, then
           repeat  the test.  Otherwise, repeat  the whole sequence of
           empty/fill cycle, stabilization, and test at the given
           conditions.


6.4  AT6S  EVALUATION PROTOCOL  FOR WATER DETECTION

     Typically  the ATGS probe  has a water sensor near the bottom of the
tank.  A standpipe device to test the function of the water sensor con-
sists of a  cylinder with an accurately known (to ±0.001 inch) inside
diameter attached to the bottom of a 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.  Enough product is put into
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


                                    23

-------
measured by the ATGS are recorded.  This is done over the range of the
water sensor or 6 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 is high enough not
         to interfere with the water sensor.

Step 3:  Add water in increments 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 cor-
         responding  level, Xlf of water detected.  Record all data on
         page 1 of the Reporting Form for Water Sensor Evaluation Data in
         Appendix B.

Step 4:  Add enough  water to the cylinder with a pipette to produce a
         height  increment, h, measured to the lesser of 1/16 inch or half
         of the claimed resolution.  At each increment, record the volume
         of water added and the water height (denoted by W^j in Table 3

         of Section  7.2) measured by the sensor.  Use pages 2 to 4 as
         necessary of the Reporting Form for Water Sensor Evaluation Data
         in Appendix B.  Repeat the incremental addition of water at
         least  20 times to cover the height of about 6 inches (or, the
         range  limit of the  sensor, if  less).

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  and  3 20 times to obtain 20 repli-
         cations.  Repeat Step  4  at least 3 times or as needed to obtain
          a minimum of  100  increments.

      Record  all  data using the  reporting form for ATGS water  sensor data
 in Appendix  B.   The  20 minimum  detectable water  levels are denoted by  Xj,

 0=1,..., 20.   The sensor reading at  the ith  increment of the  jth test  is
denoted by
                 as described in Section 7.2 and Table 3.
                                     24

-------
 6.5  ALTERNATIVE EVALUATION PROCEDURE

      As noted in the Foreword, EPA will accept alternative evaluation
 protocols to the specific one just described.  An overview of an
 alternative protocol is presented next.  Although it is not completely
 specified, enough detail is presented so that an evaluating organization
 should be able to set it up and carry it out.                 a«i nation

      The previous sections (6.1 to 6.4) provide a test plan that can be
 accomplished in about three calendar weeks.  The approach described there
 requires a tank that can be fully devoted to testing, which may be a
 difficult requirement.   The following alternative approach uses
 in-service tanks.  Only a limited amount of work is required that would
 prohibit using the tank for dispensing product.

      The alternative approach consists of installing the AT6S in a number
 of tanks.   Since the ATGS operates automatically, it can be programmed to
 perform a test whenever the tank is out of service for a long enough
 period, typically each  night.   With several available tanks,  a large set
 of tests could be performed in a relatively short time.   By selecting
 tanks in different climates or observing tanks over the  change of sea-
 sons, tests can be performed under a wide variety of conditions.  Thus
 with  little expenditure of effort, a large data  base of  test  results  on
 tight tanks can be obtained readily.

      The alternative approach  will provide test  data under  a  variety  of
 actual  conditions.   In  selecting  the sites and times for the  data collec-
 tion, the  evaluating organization should attempt  to  obtain  a  wide variety
 of temperature conditions  and  to  conduct the tests  at a  wide  variety  of
 product levels in the tank  as  well  as  a  variety of times after the tank
 receives a product  delivery.   This alternative approach  will  produce  data
 under conditions  as  actually observed  in the field.  The primary dif-
 ference between the  standard and  alternative procedures  is how the test
 conditions are attained.  Both approaches  attempt to conduct the evalu-
 ation testing  under  conditions representative of the real world.  The
 standard approach does  this  by controlling  the test  conditions, while the
 alternative tests under a variety  of situations and  records the test
 conditions.

      Next, the data  base of ATGS test results on tight tanks needs to be
 supplemented with a  limited number of tests using an induced leak. This
 is to demonstrate that the system can track an induced leak adequately,
 that  is, that  it will respond to and identify a loss of product from the
 tank of the magnitude specified in the EPA performance standard.  The
 combined data sets can then be analyzed to estimate the performance of
 the ATGS.  If the resulting performance estimate meets the performance
 standard for an ATGS, that would constitute demonstration that the system
meets the EPA standard.

     This alternative approach will result in a large number of tests on
tight tanks, and relatively few tests under induced leak  rate condi-
tions.  A suggested sample size is 100 tight tank tests and 10 induced


                                    25

-------
leak rate tests.  Larger numbers of either type of test can be used. It
should be easy to obtain the tight tank tests, however, some work will be
needed to prepare the data base, recording the ancillary data.  It will
also be necessary to exclude some tests, for example those that were
started, but had a delivery or dispensing operation during the test
period thus invalidating the test.

     The following steps provide an outline of this method of evaluation.

Step 1:   Identify a number of tanks for installation of the AT6 sys-
          tems.  These tanks should be known to be tight, by meeting one
          of the criteria described in Section 3.  The tanks can be of
          varying sizes, but the sizes used will limit the applicability
          of the results.  The tanks should be at several sites, with a
          suggested minimum of 5 different sites and 10 different tanks.
                                                      i
Step 2:   Install identical ATG systems in the tanks.  Arrange to collect
          and record ancillary data to document the test conditions. The
          data needed are:

               the average in-tank product temperature prior to a
               delivery.

          •    the time and date of each delivery.

               the average in-tank product temperature immediately after
               a delivery.

               the amount of product added at. each delivery.

          •    the date, time, and results of each test.

          •    the product level when the test is run.

               the tank size, type of tank, product contained, etc.,  (see
               the Individual Test Log for a form to record these data).

Step 3:   Conduct tests in each tank for at least a two-week period.
          Tests should be run approximately nightly or as frequently as
          practical with the tank's use.  Report the starting and ending
          dates of the test period.  Record the test result along with
          the data listed in Step 2.  The data above define the condi-
          tions of each test in terms of the time since the last fill
          (stabilization time), the product level, and the difference
          between the temperature of the product added and that of the
          product in the tank.  Report all test results, even if some
          tests must be discarded because of product delivery or dis-
          pensing during the scheduled test period.  Identify and report
          the reason for discarding any test data on the test log.
                                    26

-------
Step  4:    Conduct  tests with  an  induced  leak  at  the  rate  between  0.10  and
           0.20  gallon per  hour.   These induced leak  tests will  generally
           require  a person on site  to monitor the  induced leak  rates and
           measure  the rates actually achieved.   A  minimum of  10 such
           tests is suggested,  with  some  conducted  shortly after a fill
           with  a nearly full  tank,  and others conducted when  the  tank  is
           about half full.  The  induced  leak  tests should be  conducted on
           the largest available  tanks to demonstrate the  performance on
           the largest tank  that  the ATGS is intended for.

Step  5:    At some  time during  the evaluation  period, evaluate the per-
           formance of the water  sensor function.   This can be done at a
           separate site and does not require  a tank.  Follow  the
           procedure described  in Section 6.4.

Step  6:    Using  the resulting  data, analyze the differences between the
           leak rate measured by the ATGS and  the induced  leak rate
           achieved (zero for the many tests on tight tanks) for teach
           test to  estimate the performance.

     The data base can be used to investigate the relationship of the
error size (the  leak rate differences)  to each of the variables measured
for the tests.  These include tank size,  length of stabilization time
temperature differential, product level,  and presence of induced leaks
Multiple regression techniques can be used for these analyses, most of
which would fall under the category of  optional  analyses.   However,  the
data should be analyzed with the two groups of tight tank  tests and
induced leak rate tests separately to demonstrate that the system can
determine the leak rates.   Otherwise,  it  would be possible to  have such a
large number of tight tests that small  errors  on  those would obscure
large errors on the small  number of induced leak  rates tests.   An  outline
of the data analysis approach  is given  in Section 7.4.
                                   27

-------

-------
                                 SECTION 7

                               CALCULATIONS


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

      The evaluation of the ATGS  in its  leak  detection mode is presented
 first.  These  calculations compare the  system's  measured  leak rate  with
 the induced  leak rate  under  a variety of  experimental  conditions.   The
 probability  of false alarm and the probability of detection are estimated
 using the  difference between these two  numbers.   If the overall  perfor-
 mance of the ATGS  is satisfactory, analysis  and  reporting  of results
 could end  at this  point.  However, the  experimental  design has  been con-
 structed so  that the effects of  stabilization time,  product level   and
 temperature  can be tested to provide additional  information to  the
 vendor.

      A separate section  (Section 7.2) presents the  calculations to  esti-
 mate  the minimum water level (detection threshold)  and the  minimum  water
 level change that  the  sensor can detect.


 7.1  ATGS  LEAK DETECTION MODE

      After all tests are performed according  to the schedule outlined in
 Section 6, a total of  at least n = 24 pairs  (4 leak rates x 3 temperature
 differentials  x 2  product volumes) of measured leak rates and induced
 leak  rates will be available.  These data form the basis for the perfor-
mance evaluation of the system.  The measured leak rates are denoted by
 Li,...tL2ti and the associated induced leak rates by Slt...,S2u.   These
 leak rates are numbered in chronological order.  Table 2 summarizes the
notation used throughout this protocol,  using the example test plan of
Table 1.


7.1.1  Basic Statistics

     The n = 24 pairs of data are used to calculate the mean squared
error, MSE, the bias, and the variance of the method as follows.
                                   29

-------
Table 2.  NOTATION SUMMARY


Test
No.
1
2
3
4
*
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24


Pair
No.
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
11
11
12
12


Set
No.
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
Nominal
temperature
differential
(degree F)
T2
T2
T2
T2
T!
TI
T!
TI
T3
T3
T3
T3
T2
T2
T2
T2
T!
TI
T!
TI
T3
T3
T3
T3
Nominal
leak 'rate
(gallon
per hour)
LR
LR2
LR,,
LR3
LR,
LR,,
LR2
LR3
• LR,,
LR,
LR3
LR2
LR3
LRH
LR2
LRi
LR2
LR3
LR,,
LRi
LR3
LR2
LR,,
«,
Induced
leak rate
(gallon
per hour)
Si
S2
S3
s*
ss
Se
S7
Sa
S9
Sio
Si,
S12
Sis
Si*
SIB
Sis
SIT
SIB
Sis
S20
Szi
S22
S23
s.
Measured
leak rate
(gallon
(per hour)
LI
L2
L3
Ll»
L5
L6
L7
L8
L9
Lio
LH
Ll2
Ll3
Lm
LIB
Lie
LIT
LIB;
Ll9
L20
Lzi
L22
L23

Absolute
leak rate
difference
L-S|
(gallon
per hour)
d
d2
d3
d*
ds
d6
d7
d8
d9
dio
dn
d12
di3
di*
dis
die
d17
die
dig
d20
d21
d22
d23
d2i»
30
i

-------
Mean  Squared  Error, MSE

                                24

                         MSE -2
                               1-1
where  Li  is the measured  leak rate obtained from the  1th test at the cor
responding induced leak rate, S^ with  1-1, ..., 24.
Bias, B

                                24
                               1-1
     The bias, B, is the average difference between measured and induced
leak rates over the number of tests.  It is a measure of the accuracy of
the system and can be either positive or negative.
Variance and Standard Deviation

     The variance is obtained as follows:
                               24

                   Variance = ^ [(L. - S^ - B]2/23
Denote by SD the square root of the variance.  This is the standard
deviation,

NOTE:  It is recommended that the differences between the measured and
induced leak rates be plotted against the time or the order in which they
were performed.  This would allow one to detect any patterns that might
exist, indicating potentially larger differences in the results from the
first test of each set of tests, among the three temperature differen-
tials, or between in- tank product levels.  This could suggest that the
system calls for an inadequate stabilization time after filling, that the
system does not properly compensate for temperature differences between
in-tank product and product to be added, or that the system is influenced
by the product level.  (See Sections 7.3.3, 7.3.4, and 7.3.5 for
appropriate statistical tests.)


Test for Zero Bias

     To test whether the method is accurate—that is,  the bias is zero—
the following test on the bias calculated above is performed.
                                   31

-------
   , Compute the t-statist1c


                             tB = \/24~ B/SD


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

     Compare the absolute value of tB, abs(tB), to 2.07 (or to the

appropriate t-value if more than 24 tests were performed).,  If abs(tg) is

less than 2.07, conclude that the bias is not statistically different

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

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

7.1.2  False Alarm Rate, P(FA)

     The normal probability model is assumed for the errors in the
measured leak rates.  Using this model, together with the statistics
estimated above, allows for the calculation of the predicted false alarm
rate and the probability of detection of a leak of 0.20 gallon per hour.

     The vendor will supply the criterion (threshold) for interpreting
the results of the AT6S test function.  Typically, the leak rate measured
by the ATGS is compared to that threshold and the results interpreted as
indicating a leak if the measured leak rate exceeds the threshold.
Denote the system's criterion or threshold by C.  The false alarm rate or
probability of false alarm, P(FA), is the probability that the measured
leak rate exceeds the threshold C when the tank is tight.  Note that by
convention, all leak rates representing volume losses from the tank are
treated as positive.

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


False Alarm Rate With Negligible Bias

     In the case of a nonsignificant bias (Section 7.1.1), compute the
t-statistic

                                tx = C/SD


                                    32

-------
 where SD is the standard deviation calculated above and C is the system's
 threshold.  Using the notational convention for leak rates, C is posi-
 tive.  P(FA) is then obtained from the t-table, using 23 degrees of free-
 dom.  P(FA) is the area under the curve to the right of the calculated
 value ti.

      In general, t-tables are constructed to give a percentile  t   cor
 responding to a given number of degrees of freedom, df, and a preassianed
 3r^ aor'a]Pha» under the curve, to the right of t. (see Figure 1 below
 and  Table A-l in Appendix A).  For example, with 23 Segrees of freedom
 and  a = 0.05 (equivalent to a P(FA) of 5%), ta = 1.714.
               Figure 1.  Student's t-D1stributton Function.


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

     Suppose that the calculated tx = 1.85 and has 23 degrees of
freedom,  l-rom Table A-l, obtain the following percentiles at df = 23:
    1.714
    1.85
    2.069
  a (alpha)

0.05
X to be determined
0.025
Calculate X by linearly interpolating between 1.714 and  2.069 correspond-
ing to 0.05 and 0.025,  respectively.
                                   33

-------
               °'05 -           '     X (1'714 ' 1-85) = °-
Thus the probability of false alarm corresponding to a tj of 1.85 would
be 4%.

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

     The computations are similar to those in the case of a nonsignifi-
cant bias with the exception that the bias is included in the calcula-
tions, as shown next.  Compute the t-statistic


                              tz = (C-B)/SD


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


7.1.3  Probability of Detecting a Leak Rate of 0.20 gallon per hour, P(D)

     The probability of detecting a  leak rate of 0.20 gallon per hour,
P(D),  is the probability that the measured leak rate exceeds C when the
true mean  leak rate is 0.20 gallon per hour.  As for P(FA), one of two
methods is used in the computation of P(D), depending on whether the bias
is  statistically significantly different from zero.


P(D) With  Negligible Bias

      In the case of a nonsignificant bias— that is, the bias is zero-
compute the t-statistic


                             t3 = (C-0.20)/SD


Next,  using the t-table at 23 degrees of freedom, determine the area
under the  curve to the right of t3.  The resulting  number will be P(D).
                                    34

-------
P(D) With Significant Bias

     The procedure is similar to the one just described, except that B  is
introduced in the calculations as shown below.  Compute the t-statistic


                            t,/= (C-B-0.20)/SD


     Next, using the t-table at 23 degrees of freedom, determine the area
under the curve to the right of ti,.  The resulting number will be P(D).


7.1.4  OTHER REPORTED CALCULATIONS

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


Size of Tank

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


Maximum Allowable Temperature Difference

     Calculate the standard deviation of the 6 temperature differences
actually achieved during testing (these 6 tests are the first in each of
the 6 sets).  Multiply this number by the factor ± 1.5 and report the
result as the temperature range on the limitations section of the results
form.

     The nominal temperature difference of 5°F used in the design was
obtained from data collected on the national  survey (Flora, J. D., Jr.
and J. E. Pelkey, "Typical Tank Testing Conditions," EPA Contract
No. 68-01-7383, Work Assignment 22, Task 13,  Final Report, December
1988).  This difference was approximately the standard deviation of the
temperature differences observed in the tank  tests conducted during the
national survey.  The factor 1.5 is a combination of two effects.  One
effect results from scaling up the standard deviation of the design
temperature differences to 5°F.  The second effect results from using the
rule that about 80% of the temperature differences on tank tests are
expected to be within ± 1.282 times the standard deviation.
                                    35

-------
 Average Waiting Time After Filling
      Calculate the average of the time  intervals between the end of the
            le>and fuart °f-tl?e test  for the 6 tests that started
             after the specified waiting time (first test in each set}
  lip1"6 ^ 6 teS?  *r*  d°ne 1"«d1«te^ ««er the filing, use
 all such tests.  However,  do  not  use the time to the start of the
 JJSpTTJj ^IL™ a Set as.th1s would 9ive a »1slead1ng waiting
 time.)  Report this average time  as  the waiting time after adding product
 on the results form.  Note:   The  median may be used as the average
 instead of the mean if there  are  atypical waiting times.
 Average Data Collection Time Per Test
              durat1on of the data collection phase of the tests to
          «    aDerdg? ?.a*a f?llect1on «« for the total number (at least
 24) of tests.   Report this time as the average data collection time per
 7.2  ATGS WATER DETECTION MODE

 thp Jn?mnSramf TJf111 5e eft1mated for the water detection sensor:
 dP?pJiyi!T. fj?ablei?at!r level °r threshold t^t the sensor can
 determine,  and the smallest change in water  level that the device can
 sXn^H  ?e?e :?SUlŁS w111 also be reP°rted on the Results of 5?S??PA
 Standard  Evaluation form in Appendix B.
 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  4)
 S level d50SLS i4'j;iv20\are used to esti;ate
 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
                                       _. 2
                                      - x)
                                  20-1
                                            1/2
                                  36

-------
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-Sided Statistical Tolerance Limits."  industrial
          Quality Control.  Vol.  XIV,  No. 10.)

Step 4:   Calculate the upper tolerance limit, TL, for 95% coverage with
          a tolerance coefficient of 95%:
                              TL = X + K SD,

                                   or

                              TL = X + 2.396 SD


     TL estimates the minimum level of water that the sensor can
detect.  That is, with 95% confidence, the ATGS should detect water at
least 95% of the time when the water depth in the tank reaches TL.


7.2.2  Minimum Mater Level Change

     The 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 6.4.

     Denote by W^j the sensor reading (in inches) at the jth replicate

and the ith increment (i=l,...,nj, with nj being 20 or more in each

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 h (measured to the lesser of 1/16 inch or half the claimed
resolution) the level change induced at each increment.  Let m (greater
than or equal to 3) be the number of replicates.

Step 1:   Calculate the differences between consecutive sensor read-

          ings.  The first increment will be W1 j-X-j^ for the first

          replicate (j=l); more generally, W-^j-Xj, for the jth

          replicate.  The second increment will be W2 1-W1 ± for the

          first replicate; more generally, w^j-W^j for the jth

          replicate, etc.
                                   37

-------
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 di  it where i
                                                             • v J
          and j represent increment and replicate numbers, respec-
          tively.  Table 3 below summarizes  the notations.
           Table 3.  NOTATION SUMMARY FOR WATER SENSOR READINGS
                           AT THE jth REPLICATE
Calculated
level
change
Increment (inch)
No. A
1 + h
2 + h
3 + h
* «
• •
n-j ' + h
Sensor
reading
(inch)
B
H1.J
W2,j
W3,j
*
o "
*
V
Measured Increment
sensor difference
increment calculated-meas.
(inch) (inch)
C C-A
Wi -f-X.:* di 1
i,j j i,j
W2 •«- Wi ,- dj» ^
^•jj ^iJ '••**}
Wo -J~"9 •? Q"[ •?
J»J ^»J «*»J
• a
• «
! .
Wn ,-Wn 1 4 dri ,•
n^»j n,— i,j n.-,j
\J \J J
     *  X,- is the water level  (inches) detected for the first time
        by the  sensor during the  jth replication of the test.

Note that the first  sensor reading, Xj, may vary from replicate to repli-
cate, 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, Dn-, of the differences cL  .:, i=l,...,nn-,
                                  J                       ' »v,l            J
         separately  for each replicate j, j=l,...,20.
                                    38

-------
Step 4:  Calculate the variance of the differences d^  _-,  i=l,...,n
         separately for each replicate j, j=l,...,m.
Step 5:  Calculate the pooled variance, Var_, of the m variances
         Varlt...,Varm.                    p

                         (n,-l)  Var,  + ••• + (n -1) Var
                  Var                          m       "
                     P                P.
                                      I   (nrl)
                                     3-1    J

Step 6:  Calculate the pooled standard deviation, SD_.


                               SDp = TvaTp"


Step 7:  From a table of tolerance factors, K, for two-sided tolerance
         intervals with 95% probability and 95% coverage, obtain K for
         (Zn-j-m) degrees of freedom.  For the suggested sample size, the
         value corresponding to a total of 100 degrees of freedom
         (K = 2.233) can be used unless the number of differences
         obtained is less than 100.  (Reference:  CRC Handbook of Tables
         for Probability and Statistics.  1966.  William H. Beyer (ed.).
         pp. 31-35.  The Chemical Rubber Company.)

Step 8:  Calculate the minimum water level change, MLC, that the sensor
         can detect.
                               MLC - K SD
         or                              p
                             MLC = 2.233 SD
     The result, MLC, is an estimate of the minimum water level change
that the water sensor can detect.


7.2.3  Time to Detect a 0.20-Gallon per Hour Water Incursion (Optional)

     The minimum detectable water level and the minimum detectable change
can be used to determine a minimum time needed to detect a water incur-
sion into the tank at a specified rate.  This time is specific to each
tank size and geometry.  The calculations are illustrated for an
8,000-gallon steel tank with a 96-inch diameter and 256 inches long.
                                    39

-------
     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 inchess giving a
radius, r, of 47.75 inches.  The water surface will be 2d wide, where d,
in inches, is calculated as
                                    - (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.  iMultiplying 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 Coin
detect reliably.  This differs with the level of water in the tank.  (For
a somewhat more accurate approximation, calculate d at level x and at
level x + MLC and average the two readings for the d to be used to calcu-
late the change in volume of water that can be detected.)

     To determine how long the AT6S will take to detect a water incursion
at the rate of 0.20 gallon per hour, divide the minimum volume change
that the water sensor can detect by 0.20 gallon per hour.  As a numerical
example, suppose the depth of the water were 1 inch and the minimum
detectable change were 1/8 inch.  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 = >/(47775)2 - (46.75)2 = 9.72 inches

The volume, in inch3, corresponding to a 1/8-inch increase is


                       V = 2(9.72) x 255.5 x (1/8)

                                    or

                             V = 620.94 inch3         |

                                                      i
In gallons, the volume is


                      V = 220.94/231 = 2.688 gallons


The time that the  sensor will take to detect water incursions at the rate
of 0.20 gallon per hour will be


         time = 2.688 gallons/0.20 gallon per hour = 13.44 hours
                                    40

-------
 Thus,  the sensor would detect water coming in at the rate of 0.20 gallon
 per hour after 13.4 hours,  or about half a day.   The incursion of the
 water  into the tank should  be obvious on a day-to-day basis under these
 conditions.


 7.3 SUPPLEMENTAL CALCULATIONS AND DATA ANALYSES (OPTIONAL)

     Other information can  be obtained from the  test data.   This informa-
 tion is  not required for  establishing that the AT6S meets the federal  EPA
 performance requirements, but may  be useful  to the  vendor of the ATGS
 The calculations  described  in this section are therefore optional.   They
 may be performed  and reported to the vendor,  but are not required and  are
 not reported on the results form.   These supplemental  calculations
 include  determining a  minimum threshold, a minimum  detectable leak  rate
 and relating the  performance  to factors such  as  temperature differential
 waiting  time,  and product level.   Such information  may be particularly  '
 useful to the  vendor for future improvements  of  his ATGS.

     The experimental  design  tests the system under a  variety of condi-
 tions  chosen to be reasonably representative  of  actual  test conditions
 The tests occur in pairs after each  fill  cycle.  A  comparison of the
 results  from the  first of the pair with the second  of  that  pair  allows
 one to determine  if the additional  stabilization time  improved the
 performance.   Similarly, comparisons  among the tests at  each  temperature
 condition allow one to determine whether the  temperature conditions
 affected the performance.   A  comparison among test  results  performed with
 a tank either  full  or  half  empty will  provide an assessment of the effect
 of  product  level  on the system's performance.  Finally, the performance
 under  the four induced  leak conditions  can be compared to determine
 whether  the  system performance  varies with leak rate.

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


 7.3.1  Minimum Threshold

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

The following demonstrates the approach for computing C5%.  Solve the
equation
                                   41

-------
                    P(FA)  = P(t > (C5% - B)/SD} = 0.05

for Cgeg.  If the bias is not statistically significantly different from
zero (Section 7.1.1), then replace B with 0.   From the t-table with
23 degrees of freedom obtain the 5th-percentile.  This value is 1.714.
Solving the equation above for C5^ yields
                           (C5% - B)/SD = 1.714

     In the case of a nonsignificant bias, this would be C5g =  1.714 SO.

7.3.2  Minimum Detectable Leak Rate
     With the data available from the  evaluation, the minimum detectable
leak rate, R55Ł, corresponding to a probability of detection, P(D), of 95%
and a calculated threshold, C5^, can be calculated by solving the follow-
ing equation for R$%:

               P(D(R5%)) = PŁt > (C5% - R5% - B)/SD} = 0.95

where C5%  is the threshold corresponding to  a  P(FA) of  5%  as previously
calculated.
     At  the P(FA)  of  5%,  solving the equation  above is  equivalent to
solving                                               |
                                    or
                             - 1.714 SD + C   - B
which,  after substituting 1.714 SD for (C5^ - B),  is  equivalent to

                              Rco; = 2Ccv - 2B
                               3A>     3«

Substitute 0 for B in all calculations when the bias  is not statistically
significant.  Otherwise, use the value of B estimated from the data.


                                    42

-------
      Thus, the minimum detectable leak rate with a probability of detec-

 tion of 95% is twice the calculated threshold, C5%, determined to give  a
 false alarm of 5%, minus twice the bias if .the bias is statistically
 significant.

      In summary, based on the 24 pairs of measured and induced leak

 rates, the minimum threshold, Cg%, and the minimum detectable leak rate,
 R5«g, are calculated as shown below.

      If the bias is not statistically significant:

      For a P(FA) of 5%             Ccv = 1.714 SD
      For a P(D(R)) of 95%          RjjJ = 2C5%

      If the bias is statistically significant:
      For a P(FA)  of 5%             C5% = 1.714 SD + Bias
      For a P(D(R))  of 95%          Rg = 2C5% - 2 Bias


 7.3.3  Test for Adequacy of Stabilization Period

      The performance estimates  obtained in Sections 7.1.2  and  7.1.3  will
 indicate whether  the system meets  the  EPA performance  standards.   The
 calculations in this section allow one to determine whether  the system's
 performance is  affected  by  the  additional  stabilization  time the  tank  has
 experienced by  the  second test  after each fill  cycle.  These statistical
 tests are designed  primarily to help determine  why  an  ATGS did not meet
 the performance standards.

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

Step 1:  Calculate the absolute value of the 12 differences,  d,,  between
         the measured (L) and induced  (S) leak rates for the first
         2 tests in each set (last column in Table 2).

Step 2:  Calculate the average of the absolute differences  for the first
         and second test in each set separately.
                                   43

-------
                                 dg + d13  + dl7 + dai)/6
  D2 =  (da + d6 + d
         10     llf     ie     22
                                             die + d22)/6
Step 3:  Calculate the variances of the absolute differences from the
         first and second test in each set separately.
= {(d2 -
    (d. - D2)
                                                 (d22- D2)2} /5
Step 4:  Calculate the pooled standard deviation.
                                 10
Step  5:   Calculate the t-statistic:
(D, - Da)
                                             1 - D2)
                           P \Tfi
 Step 6:   From the t-table, obtain the critical value corresponding to a t
          with (6+6-2)  =  10 degrees of freedom and a two-sided 5% signifi-
          cance level  (a  = 0.025  in the table).  This value  is 2.228.

 Step 7:   Compare  the  absolute  value  of t, abs(t), to 2.228.   If abs(t)  is
          less than 2.228, conclude that the  average difference between
          measured and induced  leak rates obtained from the  first tests
          after stabilization is  not  significantly different (at the 5%
          significance level) from the average difference  between measured
          and induced  leak rates  obtained from the second  tests after
          stabilization.   In other words, there has not been an additional
          stabilization effect  between the beginning of the  testing and
          the end.  Otherwise,  conclude that  the difference  is statisti-
          cally significant, that is, the system's performance is differ-
          ent with a longer stabilization period.

      If the results are  statistically significant, then the performance
 of the system is  different for the tests with the additional  stabiliza-
 tion period.  If  the  performance is  better,  that is, if the absolute
 differences for the testing with additional  stabilization are smaller
 than those for the tests with  the minimum stabilization period, then the
                                    44

-------
 system would show improved performance if it increased its required
 !u  rl«zat1on Per1od-  If the system's overall performance did not meet
 the EPA performance standard, performance estimates with the additional
 stabilization can be calculated using only- the 6 test results with the
 additional  stabilization.  If the results indicate that the system does
 not meet the EPA performance standard but could meet the EPA performance
 standard with the additional stabilization,  that finding should be
 reported.  Note that the system would still  need to conduct the full
 24 tests at the longer stabilization time before claiming to meet the EPA
 performance standard.


 7.3.4  Test for Adequate Temperature Compensation

 ,*•« Th1! !ection.a^°ws one to test whether the system's performance is
 different for various temperature  conditions.   A total  of eight tests
 will  have been performed with  each of the three temperature  differen-
 tials,  Tlt  T2,  and  T3 (the nominal values of 0°,  -5°, and +5°F  will have
 been  randomly assigned to TIf  T2,  and T3).   The 24  tests  have been
 ordered by  temperature differential  and test number in  Table 4  for the
 example order of sets from Table 1.   In general,  group  the tests by
 temperature condition.

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

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

Step 1.  Calculate the average of the absolute differences in each group.


        Mi - Z) d./8  where gt  denotes the 8  subscripts  in Group 1
       M2 = Z) d-j/8  where g2 denotes the 8 subscripts in Group 2
       Ms = Z) d./8  where g3 denotes the 8 subscripts in Group 3
                                   45

-------
           Table 4.  ORGANIZATION OF DATA TO TEST  FOR
                       TEMPERATURE EFFECTS
Test No,
Pair No.   Set No,
  Nominal
temperature
differential
 (degree F)
    Absolute
   leak  rate
   difference
    |L-s|
(gallon  per hour)
5
6
7
8
17
18
19
20
1
2
3
4
13
14
15
16
9
10
11
12
21
22
23
24
3
3
4
4
9
9
10
10
1
1
2
2
7
7
8
8
5
5
6
6
11
11.
12
12
2
2
2
2
5
5
5
5
1
1
1
1
4
4
4
4
3
3
3
3
6
6
6
6
TI
TI
TI
TI
TI
TI
TI
TI
T2
T2
T2
T2
T2
T2
T2
T2
T3
T3
T3
T3
T3
T3
T3
T3
cls
d.
d7
d8
d17
dI8
d«
d20
di
d2
d3
dH
; di,
di*
dis
di.
d9
die
dn
dia
d21
d22
d23
d21f



Group 1







Group 2







Group 3




                               46

-------
 Step 2.   Calculate the variance of the absolute differences  in  each
          group.
                                    (d. - ^
                                     , - M2)2/7
                          Var3 -Ł  (d. - M3)2/7
                                93

Step 3.  Calculate the pooled variance of Var^ Var2, and Var3.

                             7Varx + 7Var_ +  7Var,
                      Var  = - — -^ - - _ i
                         P          24 - 3
         or

                      Varp = (Varj + Var2 + Var3)/3


Step 4.  Compute the standard error,  SE, of the difference between each
         pair of the means,  Mls M2,  and M3.
                           KHI
1/2
         or
Step 5.  Obtain the  95th percentile of the Bonferroni  t-statistic with
         (24-3) = 21 degrees of freedom and three comparisons.  This
         statistic is t = 2.60.  (Reference:  Miller,  Ruppert G., Jr.
         1981.   Simultaneous Statistical inference.   Second  Edition.
         Springer-Verlay, New York, New York.)

Step 6.  Compute the critical difference, D, against which each pairwise
         difference  between group means will be  compared.

                         D  = SE  x  t = SE x  2.60
                                  47

-------
Step 7.  Compare the absolute difference of the three pairwise
         differences with D.
              Compare |Mj - M2| with SE x 2.60

              Compare \^ - M3j with SE x 2.60

              Compare |M2 - M3| with SE x 2.60
If any difference in group means, in absolute value, exceeds the critical
value of SE x 2.60, then conclude that the system's performance is influ-
enced by the temperature conditions.

     If the results are statistically significant, the system's perfor-
mance is affected by the temperature conditions.  If the overall perfor-
mance evaluation met the EPA standards, the effect of a 5™F temperature
difference on the system does not degrade performance severely.  However,
this does not eliminate the possibility that larger differences could
give misleading results.  If the overall performance did not meet the EPA
performance standards, and the temperature effect was significant, then
the system needs to improve its temperature compensation and/or stabi-
lization time in order to meet EPA performance standards.  Again, an
evaluation testing the modified ATGS would need to be conducted to docu-
ment the performance before the ATGS could claim to meet the performance
standards.
7.3.5  Test for Effect of In-Tank Product Volume

     The procedure outlined in Section 6 required that the tank be either
half full or filled to between 90% and 95% capacity.  As shown in
Table 1, 12 tests will have been run with the tank half full, and
12 tests with the tank full to 90% to 95% capacity.  The 24 tests have
been ordered by product volume and test number in Table* 5 for the example
order of tests from Table 1.

     The test results from the two volume levels are compared to check
for the effect of product volume on the system's performance.  If the
effect is negligible, the two groups of results should be comparable.  If
the system's performance is affected by the product level, then the ATGS
may not meet EPA performance standards at all product levels.  If it does
meet the performance standards at both levels, it can be used in the test
mode at any product level.  However, if there is a significant difference
in performance at the two levels, it might be advisable to recommend that
the ATGS be used in its test mode only for certain product levels.  If
the performance is not adequate for one of the product levels, the per-
formance of the ATGS is probably marginal.  The operation of the test
function could be restricted to the product level where the performance
was adequate.
                                    48

-------
Table 5.  ORGANIZATION OF DATA TO TEST
       FOR  PRODUCT VOLUME  EFFECT


Test No.
1
2
5
6
9
10
13
14
17
18
21
22
3
4
7
8
11
12
15
16
19
20
23
24


Pair No.
1
1
3
3
5
5
7
7
9
9
11
11
2
2
4
4
6
6
8
8
10
10
12
12


Set No.
1
1
2
2
3
3
4
4
5
5
6
6
1
1
2
2
3
3
4
4
5
5
6
6

In-tank
product
volume
90-95% full
90-95% full
90-95% full
90-95% full
90-95% full
90-95% full
90-95% full
90-95% full
90-95% full
90-95% full
90-95% full
90-95% full
50% full
50% full
50% full
50% full
50% full
50% full
50% full
50% full
50% full
50% full
50% full
50% full
Absolute
leak rate
difference
IL-SI
(gallon per hour)
dj
d,
U2
o •-
U5
df
ug
d10 Group 1
j13
d1*
j17
do,
U21
d22
d
dif

da
U8
dn
d12 Group 2
a i •-
U15
"16
Q t **
U19
f| _ _
"20
n « -»
U23
                49

-------
     One of the consequences of using an ATGS to test eit various levels
of product in the tank is that the test can only find leaks below the
product level used in the test.  The performance standard calls for
detecting a leak from any portion of the tank that normally contains
product.  Ideally, the test should be run with the tank as full as it is
filled in practice so that leaks can be detected from any part of the
tank.  If the test results were restricted to testing when the tank was
half full, for example, the test could not find leaks iin the upper half
of the tank.

     The following statistical procedure (two-sample t--test) provides a
means for testing the effect of product volume on the test results.

Step 1.  Calculate the average of the absolute differences in the two
         groups.
                      where gt  denotes  the 12 subscripts  in  Group  1
            9i
       M2 =^}d./12  where g2 denotes the 12 subscripts  in  Group 2
            92

Step 2.  Calculate the variance of the absolute differences in the two
         groups.

                                91
Var2 -
                                       - M2) /ll
Step 3.  Calculate the pooled variance of Va^ and \lar2
                                         HVar
Varp '
                                    242
         or
                         Var  =  (Var1 + Var2)/2
Step  4.   Compute  the  standard error, SE, of the difference, between Mt and
          M2.

                                              Jl/2
                         SE  =
                         SE  =-y/Var   /6
                                    50

-------
 Step 5.  Calculate the t-statistic:
                                      SE    !


 Step 6.  From the t-table in Appendix A, obtain the critical value
          corresponding to a t with (12 + 12-2) =22 degrees of freedom
          and a two-sided 5% significance level.  This value is 2.074.


 Step 7.  Compare the absolute value of t, abs(t), to 2.074.  If abs(t) is
          less than 2.074, conclude that the average difference between
          measured and induced leak rates obtained with a tank half full
          is  not significantly different (at the 5% significance level)
          from the average difference between measured and induced leak
          rates obtained with a tank filled to 90% to 95% capacity.   In
          other words, the amount  of product,  in this given range, has no
          significant impact on the leak rate results.   Otherwise, con-
          clude that the difference is statistically significant,  that is,
          the system's performance depends on  the amount of product  in the
          tank.


 7.4   OUTLINE OF CALCULATIONS FOR  ALTERNATIVE  APPROACH

      This section describes  the data  analysis  required  for the  alterna-
 tive  protocol  described in Section 6.5.

      The  water  sensor data will be identical  to  that obtained with the
 standard  protocol  outlined  in  Section 6.4.  Consequently,  the same data
 analysis  will  be  used.   Refer  to  Section  7.2  for the details.


 7.4.1  Calculation of P(FA)  and P(D)

      Using the  leak rate reported  by  the ATGS and the actual leak rate
 (zero for tight tank  tests, measured  for the induced leak rate tests),
 calculate the differences between  the measured and actual leak rates.
 Calculate the mean  and  standard deviation of these differences as in
 Section 7.1.1.  Perform the test for  significant bias and estimate the
 P(FA) and the P(D)  as described in that section.

     Calculate the variances of the differences separately for the data
from the tests on the tight tanks and those from the tests on tanks with
 induced leak rates.  This can be done as in Section 7.3.3, except that
the two groups are now defined by the leak status of the tanks and the
 sample sizes will not be equal.  Let the subscript "1" denote the tight
tank data set and  "2" denote the data from the tests with induced leaks.
                                   51

-------
     Let P! be the number of test results from tight tanks and n2 be the
number of test results from induced leak rate tests.  Denote by d^ the
difference between measured and induced leak rates for each test, where
j-1 or 2, and' i-l,...,n1 or n2.  Then calculate
                              I WH - di)2/K - D
and
                      s2 =
                             n.
(d2.  - d2)2/(n:
where .the summations are taken over the appropriate groups; of data,  and
where d. denotes the mean of the data in group j, and is given by
                                    n.
                              d_.  =
Form the ratio
and compare this statistic to the F statistic with (n2-l)  and (n^l)
degrees of freedom for the numerator and denominator, respectively, at
the 5% significance level.  (The F statistic can be obtained from the
F-Table found in any statistical reference book.)  If the  calculated  F
statistic is larger than the tabulated F value, conclude that the data
from the induced leak rate tests are significantly more variable than
those from the tight tanks.  If this is the case, it might impair the
ability of the AT6S to detect leaks.  Recompute the P(D) (see Sec-
tion 7.1.3) using the standard deviation calculated from just the induced
leak rate tests, S2, to verify that P(D) is still at least 95%.


7.4.2  Limitations on the Results

     The limitations on the results must be calculated from the  actual
test conditions.  Since the conditions were not controlled,, here, but
were observed, the following approach is taken to determine the  appli-
cable conditions.
                                    52

-------
 Size of Tank

      List the tank sizes of the tests in the complete data set (all valid
 tests).  Order the sizes from smallest to largest.  Determine the 80th
 percent-He of these ordered sizes.  (That is, the smallest size just
 exceeded by 20% of the tank sizes.)  Multiply that size by 1.25 and
 report the result as the maximum size to which the performance results
 can be extended.  Note that this implies that at least 20% of the tanks
 must be of at least a specified size if the ATGS is intended to work on
 en at size OT tank.


 Maximum Allowable Temperature Difference

      Calculate the temperature difference between the product in  the tank
 and that of newly added product for each delivery in  the data set.   Note
 that the temperature of the delivered  product can be  calculated from the
 temperature of the product in the  tank immediately before delivery   the
 temperature of the product in the  tank immediately after delivery  and
 the volumes of product by the following  formula:



                             T  - TA VA - TB  VB
                             TD - — V
    ^UuSKr]Pt V??°tes Product  1n tank after delivery, B denotes product
    tank  before delivery, D denotes product delivered, T denotes product
 temperature, and V denotes volume.                              yruuuct

     Calculate the standard deviation of the temperature differentials
 pn^-ai JiŁ i*MM!y Irh  RePo1l.th1s as ^ maximum temperature differ-
 ential for which the ATGS evaluation is valid.

     When the calculations are complete, enter the results on the
 standard results reporting form in Appendix B.  Also check the box on
 that form to indicate that the evaluation was done using the alternative
 approach .


 Average Waiting Time After Filling

     Use the time interval  between the most recent fill  or product
delivery and each following test as a stabilization time.   Order these
times from least to greatest and determine the 20th percentile.   Report
this as the minimum stabilization time.                           "epori
                                   53

-------
Average Data Collection Time Per Test

     The tests often have a constant or nearly constant duration
prescribed by the AT6S.  If so, simply report this as the test data
collection time.  If the ATGS software determines a test time from the
data, report the average test time actually taken by the test and note
that the ATGS software determines the applicable test time.
                                    54

-------
                                 SECTION 8

                              INTERPRETATION


      Each function of  the ATGS  is  evaluated  separately based on data
analysis of experimental test results.   This  section covers the leak
detection function, water level  detection function,  and measurement  of
minimum water  level change.  The entire evaluation process  results  in
performance estimates  for the leak detection  function of the ATGS.   The
results reported are valid for the experimental  conditions  during the
evaluation, which have been chosen to represent  the  most common situa-
tions encountered in the field.  These  should be typical  of most tank
testing conditions, but extreme  weather conditions can occur and might
adversely affect the performance of the ATGS.  The performance  of the
leak  detection function should be  at least as  good for tanks smaller  than
the test tank.  However, the performance evaluation  results  should only
be scaled up to tanks  of 25% greater capacity  than the test  tank.  The
performance of the water sensor  in terms of minimum  detectable  level  and
minimum detectable change are independent of the tank  size.   However,  the
volume that corresponds to these heights of water does depend on tank
size.  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.   Vendors  are  encouraged to
provide a measure of the precision of a  test,  such as  a  standard error
for their calculated leak rate at  that  site, along with the  leak rate  and
test results.


8.1  LEAK TEST FUNCTION EVALUATION

     The relevant performance measures for proving that an ATGS meets  EPA
standards are the P(FA) and P(D) for a  leak rate of 0.20 gallon per
hour.  The estimated P(FA) can be  compared with the EPA standard of P(FA)
not to exceed 5%.  In general, a lower P(FA) is preferable,  since it
implies that the chance of mistakenly indicating a leak on a tight tank
is less.  However, reducing the false alarm rate may also reduce the
chance of detecting a leak.   The probability of detection generally
increases with the size of the leak.   The EPA standard specifies that
P(0) be at least 95% for a leak of 0.20 gallon per hour.  A higher esti-
mated P(D) means that there is less chance of missing a small leak.

     If the estimated performance of  the ATGS did not meet the EPA
performance requirements,  the vendor  may want to investigate the condi-
tions that affected the performance as described in Section  7.3, Supple-
mental Calculations and Data Analyses.   If the stabilization time,
temperature condition,  or the product level can be shown to  affect the


                                   55

-------
performance of the ATGS, this may suggest ways to improve the ATGS.  It
may be possible to improve the performance simply by changing the proce-
dure (e.g., waiting longer for the tank to stabilize) or it may be neces-
sary to redesign the hardware.  In either case, a new evaluation with the
modified system is necessary to document that the ATGS does meet the per-
formance standards.

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

                                                      I
8.2  WATER LEVEL DETECTION FUNCTION

     The minimum water level detected by the ATGS is estimated from the
average threshold of detection, and the variability of the water level
threshold is estimated by 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 mini-
mum water level that the ATGS can detect above the bottom of the probe.
If the installation of the ATGS 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.
8.3  MINIMUM WATER LEVEL CHANGE MEASUREMENT

     Since ATG systems operate with the product at all levels of normal
tank operation, the water sensor can be used to test for leaks in the
event of a high 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 turns one
can determine the size of a leak of water into the tank that the ATGS 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


                                    56

-------
specific to the tank size.  Using the minimum change in water volume that
the sensor can detect, the time needed for the ATGS to detect an incur-
sion of water at the rate of 0.20 gallon per hour is calculated (Sec-
tion 7.2.3).  This calculation indicates the time needed for the water
detector to identify an inflow of water at the minimum leak rate and to
alert the operator that the water level has increased.  If the particular
ATGS has a water alarm, and if'the conditions for activating the water
alarm are specified, the length of time for that alarm to be activated
can be calculated.
                                   57

-------

-------
                                 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.

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

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

     The fourth part of Appendix B contains a blank  Individual Test
Log.  This form should be reproduced and used to record data in the
field.  Copies of the completed  daily test logs are  to be included in the
standard report.  These serve as the backup data to  document the perfor-
mance 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 6.4) to determine the
minimum level  of water and the minimum water level change that the system
can detect.

     If the optional calculations described in Section 7.3 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 are to be  reported
on the Results of U.S. EPA Standard Evaluation form.   The intent is to
document that the results are valid under conditions  represented by the
test conditions.  Section 7.1.4 describes the summary of the test condi-
tions that should be reported as limitations on the results form.   These


                                    59

-------
Items are also discussed below.  The test conditions hiave 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.  Consequently, the results of the evaluation may
be applied to tanks smaller than the test tank.  The results may also be
extended to tanks of 25% larger capacity than the test tank.  Thus, if
testing is done in a 10,000-gallon tank, the results may be extended to
tanks up to 12,500 gallons in size.  If a company wants to document that
it can test large tanks, the evaluation needs to be done in a large tank.

     A second limitation on the results is the temperature differential
between the product added to the tank and that of the product already in
the tank.  Often the AT6S must perform a test shortly after the tank has
been filled.  The reported results apply provided the temperature differ-
ential is no more than that used in the evaluation.  Testing during the
EPA national survey (Flora, J. D., Jr., and J. E. Pelkey, "Typical Tank
Testing Conditions," EPA Contract No. 68-01-7383, Work Assignment 22,
Task 13, Final Report, December 1988) found that temperature differen-
tials were no more than 5°F for at least 60% of the tests.  However, it
is clear that larger differences could exist.  The evaluation testing may
be done using larger temperature differentials, reporting those actually
used.  The results cannot be guaranteed for temperature differentials
larger than those used in the evaluation.

     A third limitation on the results is the stabilization time needed
by the ATGS.  The Individual Test Logs call for recording the actual
stabilization time used during the testing.  The mean of these stabiliza-
tion times is reported.  The results are valid for stabilization times at
least as long as those used in the evaluation.  This is viewed as an
important limitation, since shorter stabilization times can adversely
affect the system's performance.  In practice, an ATGS will often test
late at night when the tank is not used.  This will usually result in a
stabilization time as long or longer than that used in the evaluation.
If an ATGS is used in a very active tank and does not do daily tests,
then the required monthly test should be scheduled to have at least the
minimum stabilization time used in the evaluation.

     The duration of the data collecting phase of the test is another
limitation of the ATGS.  If a test shortens the data collection time and
so collects less data, this may adversely affect the system'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.


                                    60

-------
     The minimum depth of water that the sensor can detect is reported.
In addition, the minimum change in water level that the sensor can detect
is reported.  This minimum detectable change is compared to the EPA per-
formance standard of 0.125 inch.  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 0.20 gallon per hour to increase the water volume enough for the
sensor to detect.

     The same reporting forms can be used for the alternative evaluation
described in Section 6.5.  The data analysis for the alternative approach
is described in Section 7.4.  This analysis will result in reporting
observed average conditions during the evaluation.  The limitations are
based on the observed conditions instead of experimentally controlled
conditions, but the results are reported on the same form.  The Individ-
ual Test Log form should be applicable to the induced leak rate tests
under the alternative evaluation procedure.  However, the evaluating
organization may find it more efficient to design a different data col-
lection form for recording the data from the many tight tank  tests.
                                   61

-------

-------
              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.  Generally, the volu-
metric leak detection method, either a precision tank test or the leak
test function of an automatic tank gauging system (ATGS), estimates a
leak rate.  This calculated rate is compared to a criterion or threshold,
C, determined by the manufacturer.  If the calculated rate is in excess
of the criterion, the tank is declared to be leaking, otherwise, the tank
is called tight.

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

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

     Changing the criterion, C, changes both a and B for a fixed leak
rate, R.  If the leak rate R is increased, the curve on the right will
shift further to the right, decreasing 8 and increasing the probability
of detection for a fixed criterion, C.  If the precision of a method is
increased, the curve becomes taller and narrower, decreasing both a and
$, resulting in improved performance.

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

                               A-2

-------
      Definitions of some of the terms used throughout the protocol are
 presented  next.
 Nominal  Leak Rate:
 Induced  Leak  Rate:
Measured  Leak Rate:
Critical Level,  C:
 False Alarm:
Probability of
False Alarm, P(FA):
Probability of
Detection,, P(D(R));
Method Bias, B:
Mean Squared Error, MSE
Root Mean Squared Error„
RMSE:
 The set or target leak rate to be achieved as
 closely as possible during testing.  It is a
 positive number in gallon per hour.

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

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

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

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

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

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

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

An estimate of the overall performance of a
test method.

The positive square root of the mean squared
error.
                               A-3

-------
Precision:                A measure of the test method's ability in pro-
                          ducing similar results (i.e., in close agree-
                          ment) under identical test conditions.
                          Statistically, the precision of repeated
                          measurements is expressed as the standard
                          deviation of these measurements.

Variance:                 A measure of the variability of measurements.
                          It is the square of the standard deviation.

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

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

-------
           Tight Tank
Leaking Tank
                                                   Measured Leak Rate, L,
                                                      Gallons Per Hour
       C = Criterion or Threshold for declaring a leak
            (a leak is declared if the measured rate exceeds C)

       a m Probability of False Alarm, P(FA)

       ft = Probability of not detecting a leak rate R

    I - /3 = Probability of detecting a leak rate R, P(D(R))

       R = Leak Rate
Figure A-l.  Distribution of measurement error  on  a tight
                    and leaking tank.
                           A-5

-------
         Tight Tank
Leaking Tank
                B   C
  R R + B
Measured Leak Rate, L,
   Gallons Per Hour
     C = Criterion or Threshold for declaring a leak
          (a leak is declared if the measured rate exceeds C)

     a = Probability of False Alarm, P(FA)

     yS = Probability of not detecting a leak rate R

  I - /3 = Probability of detecting a leak rate R,  P(D(R))

     R = Leak Rate

     B = Bias

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

-------
Table A-l.  PERCENTAGE POINTS OF STUDENT'S t-DISTRIBUTION
f(t)
df
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
40
60
120
inf
• = .10
3.078
1.886
1.638
1.333
1.476
1.440
1.415
1.397
1.383
1.372
1.363
1.356
1.350
1.345
1.341
1.337
1.333
1.330
1.328
1.325
1.323
1.321
1.319
1.318
1.316
1.315
1.314
1.313
1.311
1.310
1.303
1.296
1.289
1.282

0 f.
• - .05 a - .025
6.314
2.920
2.353
2.132
2.015
1.943
1.895
1.860
1.833
1.812
1.796
1.782
1.771
1.761
1.753
1.746
1.740
1.734
1.729
1.725
1.721
1.717
1.714
1.711
1.708
1.706
1.703
1.701
1.699
1.697
1.684
1.671
1.658
1.645
12.706
4.303
3.182
2.776
2.571
2.447
2.365
2.306
2.262
2.228
2.201
2.179
2.160
2.145
2.131
2.120
2.110
2.101
2.093
2.086
2.080
2.074
2.069
2.064
2.060
2.056
2.052
2.048
2.045
2.042
2.021
2.000
1.980
1.960
a = .010
31.821
6.965
4.541
3.747
3.365
3.143
2.998
2.896
2.821
2.764
2.718
2.681
2.650
2.624
2.602
2.583
2.567
2.552
2.539
2.528
2.518
2.508
2.500
2.492
2.485
2.479
2.473
2.467
2462
2.457
2423
2.390
2.358
2.326
t
• -.005
63.657
9.925
5.841
4.604
4.032
3.707
3.499
3.355
3.250
3.169
3.106
3.055
3.012
2.977
2.947
2.921
2.898
2.878
2.861
. 2.845
2.831
2.819
2.807
2.797
2.787
2.779
2.771
2.763
2.756
2.750
2.704
2.660
2.617
2.576
                          A-7

-------

-------
  APPENDIX B





REPORTING FORMS
      B-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—Automatic Tank Gauging
     System (two pages)

2.   Description—Automatic Tank Gauging System (six pages)

3.   Reporting Form for Leak Rate Data—Automatic Tank Gauging System
     (two pages)

4.   Individual Test Log—Automatic Tank Gauging System (five pages)

5.   Reporting Form for Water Sensor Evaluation Data—Automatic Tank
     Gauging System (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 is responsible for filling out which form?

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

2.   Description of Automatic Tank Gauging System.  The evaluating
     organization assisted by the vendor will complete this form by the
     end of the evaluation.

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

4.   Individual Test Logs.  These forms are to be used and completed by
     the evaluating organization's field crew.  These forms need to be
     kept blind to the vendor during testing.  It is recommended that the
     evaluating organization reproduce a sufficient number (at least 24
     copies) of the blank form provided in this appendix and produce a
     bound notebook for the complete test period.
                                                      I
5.   Reporting Form for Water Sensor Evaluation Data.  These forms pro-
     vide a template for the water sensor evaluation data.  They 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.

                                   B-2

-------
 JniiSo0???1!!:10!! °f the evaluation» the evaluating organization will
 collate all the forms into a single Standard Report in the order listed
 ?K«!i    +*   ? Ca?cs,w!!?r?-t5e evaluating organization performed addi-
 tional, optional calculations (see Section 7.3 of the protocol), these
 results can be attached to the standard report. There is no reportina
 requirement for these calculations, however.                  ^rtmy

 If the alternative EPA test procedure described in Section 6.5 was
 followed, then the reporting is essentially the same as that for the
 standard evaluation procedure.  The major difference is that the Results
 of U.S. EPA Standard Evaluation form will be completed using the results
 °f *he«lon«tions described in Section 7.4.  A box is provided to indi-
 cate which evaluation procedure was used.  Individual  test logs will  onlv
    dca  ?? V?r th??6 !65tS Performed under the induced leak rate condi-
    ns.  All data collected on the tanks under the no-leak condition need

           rTed ^a-taChl!:9 C°P.1eS Of the forms on Wh1ch the ^suHs  wire
 ratP rnnHit-" "?  M?^ the tan* *?* results ^o-^k and induced leak
 rate conditions) will  be summarized on the Reporting Form for Leak Rate
 uata.   There will  be no changes  in the reporting of the  water sensor
 performance since only one testing procedure is presented.

 Distribution of the Evaluation Test Results

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

 The  complete report, consisting of 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.

 The  optional part of the calculations (Section 7.3), 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 to regulators is the responsibility of the  vendor
                                   B-3

-------
                  Results  of U.S.  EPA Standard  Evaluation
                   Automatic Tank Gauging System (ATGS)

                   Instructions for completing the formi

This 2-page form  is to be filled  out by the evaluating organization upon
completion of the evaluation of the ATGS.  This form will! contain the
most important information  relative to the ATGS evaluation.  All items
are to be filled  out and the appropriate boxes checked.   If a question is
not applicable to the ATGS, write 'NA1 in the appropriate space.

This form consists of five  main parts.  These are:

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

ATGS Description

Indicate the commercial name of the ATGS, the version, and the name,
address, and telephone number of  the vendor.  Some vendors use different
versions of their ATGS 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
ATGS9 then indicate the home office name and address of the responsible
party.

Evaluation Results

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

P(FA) is the probability of false alarm calculated in Section 7.1.2.
Report P(FA) in percent.  P(FA) may be rounded to the nearest whole
percent.

P(D) is the probability of detecting a leak rate of 0.20 gallon per hour
and is calculated in Section 7.1.3.  Report P(D) in percent.  P(D)  may be
rounded to the nearest whole percent.

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.

If the P(FA) calculated in Section 7.1.2 is 5% or less and if the P(D)
calculated in Section 7.1.3 is 95% or more, then check the first 'does'
box.  Otherwise,  check the first  'does not'  box.  If the minimum water
level change calculated in Section 7.2.2 is less than or equal  to
1/8 inch,  then check the second 'does'  box.  If the minimum water level
change exceeds 1/8 inch, then check the second 'does not' box.


                                   B-4

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

Limitations on the Results

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

The temperature differential, the waiting time after adding the product
until testing, and the total data collection time should be completed
using the results from calculations in Section 7.1.4.

If the alternative evaluation procedures described in Section 6.5 has
been followed, then report the results obtained from the calculations in
Section 7.4.

Certification of Results

Here, the responsible person 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.
                                   B-5

-------

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

             Automatic Tank Gauging System (ATGS)

 This form tells whether the automatic tank gauging system (ATGS) 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: Automatic
 Tank Gauging Systems." 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.
 ATGS Description
 Name
 Version number

 Vendor
                     (street address)     ~

       (citv)                      (state)       (zip]           (phone)
 Evaluation Results                                     ~                  ~~	

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

 The corresponding probability of detection [P(D)] of a 0.20 gallon per hour leak is	%.

 The minimum water level (threshold) in the tank that the ATGS can detect is	inches.

 The minimum change in water level that can be detected by the ATGS is      inches
 (provided that the water level is above the threshold).                 	

 Therefore, this ATGS  D does  D does not meet the federal performance standards
 established by the U.S. Environmental Protection Agency (0.20 gallon per hour at P(D) of 95%
 and P(FA) of 5%), and this ATGS U does  D does not meet the federal performance
 standard of measuring water in the bottom of the tank to the nearest 1/8 inch.

 Test Conditions During Evaluation                        ~                  ~

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

The 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 tests were conducted with the tank product levels	and	% full.

The product used in the evaluation was	
ATGS - Results Form                                                          Page 1 of 2

-------
Name of ATGS_
Version
Limitations on the Results
The performance estimates above are only valid when:
• The method has not been substantially changed.
• The vendor's instructions for installing and operating the ATGS are followed.
• The tank contains a product identified on the method description form.
• The tank is no larger than	gallons.
• The tank is at least	percent full.
• The waiting time after adding any substantial amount of product
   to the tank is	hours.
• The temperature of the added product does not differ more than
   	degrees Fahrenheit from that already in the tank.
• The total data collection time for the test is at least	hours.
• Other limitations specified by the vendor or determined during testing:    :
 > Safety disclaimer: This test procedure only addresses the issue of the ATG system's
   ability to detect leaks. It does not test the equipment for safety hazards.

 Certification of Results
 I certify that the ATGS was installed and operated according to the vendor's instructions and
 that the results presented on this form are those obtained during the evaluation. I also certify
 that the evaluation was performed according to one of the following:
 D standard EPA test procedure for ATGS
 CD alternative EPA test procedure for ATGS
 (printed name)                                 (organization performing evaluation)
 (signature)                                    (city, state, zip)
 (date)                                        (phone number)
ATGS - Results Form                                                           Page 2 of 2

-------
                Description of Automatic Tank Gauging System

                    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 AT6S   This
 form provides supporting information on the principles behind the system
 or on how the equipment works.                                     jaw=m

 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 "tvoical"
 conditions.  Please write in any additional information about the testina
 method that you believe is important.                                   a

 There are seven parts to this form.  These are:

 1.    AT6S Name and Version
 2.    Product
      > Product type
      > Product level
 3.    Level  Measurement
 4.    Temperature Measurement
 5.    Data Acquisition
 6.    Procedure Information
      > Waiting times
      > Test duration
      > Total  time
      > Identifying  and correcting for interfering factors
      > Interpreting test results
 7.    Exceptions

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

 NOTE:  The version  is provided for AT6 systems that use different
       versions of the equipment for different products or tank sizes.

 For the six remaining parts, check all appropriate boxes for each
?XX ""?"; Check,niore 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 whfte space next
to a question for a description.                               H
                                   B-8

-------

-------
                                 Description
                Automatic Tank Gauging System
 system or how the equipment works.
 ATGS Name and Version
 Product
 > Product type
 For what products can this ATGS be used? (check all applicable)
    D gasoline
    D diesel
    D aviation fuel
    D fuel oil #4
    D fuel oil #6
    CD solvents
    CH waste oil
    D other (list)	
 > Product level
 What product level is required to conduct a test?
   CH greater than 90% full
   D greater than 50% full
   D other (specify)	
Does the ATGS measure inflow of water as well as loss of product (gallon per hour)?
    D yes
    Dno

Does the ATGS detect the presence of water in the bottom of the tank?
    D yes
    Dno
ATGS - Description                                                  Page 1 of 6

-------

Level Measurement
What technique is used to measure changes in product volume?
    D directly measure the volume of product change                  !
    D changes in head pressure
    CD changes in buoyancy of a probe
    D mechanical level measure (e.g., ruler, dipstick)
    D changes in capacitance
    D ultrasonic                                              ;
    D change in level of float (specify principle, e.g., capacitance, magnetostrictive,
       load cell, etc.)	'  .	
    CD other (describe briefly)	
Temperature Measurement                                        ,
If product temperature is measured during a test, how many temperature sensors are
used?
    EH single sensor, without circulation                              ;
    HU single sensor, with circulation
    D 2-4 sensors
    D 5 or more sensors
    D temperature-averaging probe

If product temperature is measured during a test, what type of temperature sensor is used?
    D resistance temperature detector (RTD)
    CI bimetallic strip
    D quartz crystal
    CH thermistor
    D other (describe briefly)	
 If product temperature is not measured during a test, why not?
    D the factor measured for change in level/volume is independent of temperature
       (e.g., mass)
    D the factor measured for change in level/volume self-compensates for changes in
       temperature
    D other (explain briefly)	     .	
 ATGS - Description                                                       Page 2 of 6

-------
 Data Acquisition
 How are the test data acquired and recorded?
    EH manually
    D by strip chart
    [U by computer
 Procedure Information
 > Waiting times
 What is the minimum waiting period between adding a large volume of product (i e a
 delivery) and the beginning of a test (e.g., filling from 50% to 90-95% capacity)?
    CH no waiting period
    d less than 3 hours
    D 3-6 hours
    D 7-12 hours
    D more than 12 hours
    LJ 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
    EH 2 hours
    D 3 hours
    C] 4 hours
    CH 5-10 hours
    D more than 10 hours
    D variable (explain)	
 > Total time
What is the total time needed to test with this ATGS after a delivery?
(waiting time plus testing time)
      hours        minutes
ATGS - Description                                                      Page 3 of 6

-------
What is the sampling frequency for the level and temperature measurements?
    EU more than once per second
    D at least once per minute                                       ,
    CH every 1-15 minutes                                            :
    D every 16-30 minutes
    D every 31-60 minutes
    CD less than once per hour
    EH variable (explain)	;	
 > Identifying and correcting for interfering factors
How does the ATGS determine the presence and level of the ground water above the
bottom of the tank?
    CD observation well near tank
    D information from USGS, etc.
    D information from personnel on-site
    D presence of water in the tank
    CD other (describe briefly)	
    D level of ground water above bottom of the tank not determined      '

How does the ATGS correct for the interference due to the presence of ground water
above the bottom of the tank?
    CH system tests for water incursion
    D different product levels tested and leak rates compared
    EH other (describe briefly)	
    D no action

How does the ATGS determine when tank deformation has stopped following delivery of
product?
    D wait a specified period of time before beginning test
    D watch the data trends and begin test when decrease in product level has stopped
    CD other (describe briefly)	;	
    CD no procedure                                                ,
ATGS - Description                                                       Page 4 of 6

-------
  Are the temperature and level sensors calibrated before each test?
     D yes
     Dno

  If not,  how frequently are the sensors calibrated?
     D weekly
     D monthly
     D yearly or less frequently
     CH never
  >  Interpreting test results
 How are level changes converted to volume changes (i.e., how is height-to-volume
 conversion factor determined)?                                       vuiumo
     D actual level changes observed when known volume is added or removed (e a
        liquid, metal bar)                                                   ^ a->
     D theoretical ratio calculated from tank geometry
     D interpolation from tank manufacturer's chart
     D other (describe briefly)	
    LJ not applicable; volume measured directly

 How is the coefficient of thermal expansion (Ce) of the product determined?
    D actual sample taken for each test and Ce determined from specific gravity
    D value supplied by vendor of product
    D average value for type of product
    D other (describe briefly)	
 How is the leak rate (gallon per hour) calculated?
    D average of subsets of all data collected
    D difference between first and last data collected
    D from data from last	hours of test period
    D from data determined to be valid by statistical analysis
    D other (describe briefly)	
ATGS - Description                                                        _.   _  ,
                                                                       Page 5 of 6

-------
What threshold value for product volume change (gallon per hour) is used to declare that
a tank is leaking?
    CH 0.05 gallon per hour                                      ,
    D 0.10 gallon per hour                                          ;
    D 0.20 gallon per hour
    D other (list).	         '	

Under what conditions are test results considered inconclusive?
    D too much variability in the data (standard deviation beyond a given value)
    D unexplained product volume increase
    D other (describe briefly)	
Exceptions
Are there any conditions under which a test should not be conducted?
    [U water in the excavation zone
    D large difference between ground temperature and delivered product temperature
    D extremely high or low ambient temperature                     ,  •
    D invalid for some products (specify)	    !	
       other (describe briefly)	;	
 What are acceptable deviations from the standard testing protocol?
    EH none
    D lengthen the duration of test
    D other (describe briefly)	
 What elements of the test procedure are determined by personnel on-site?
    D product level when test is conducted
    C] when to conduct test
    D waiting period between filling tank and beginning test            :
    D length of test                                                ;
    D determination that tank deformation has subsided               ;
    D determination of "outlier" data that may be discarded
    D other (describe briefly)	           '	
    C3 none

 ATGS - Description                                                        Page 6 of 6

-------
                     Reporting Form for Leak Rate Data
                   Automatic Tank Gauging System (ATGS)

                   Instructions for completing the form

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

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

 In  general, the statistician analyzing the  data  will  complete this
 form.  A blank  form  can  be developed  on a personal computer,  the data
 base for a given evaluation generated, and  the two merged  on  the 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 ATGS  test
 results.

 The table consists of 11 columns.  One line  is provided for each test
 performed during evaluation of the ATGS.  If a test was invalid  or was
 aborted,  the test should be listed with  the  appropriate notation  (e.q.
 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 6.1  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-15

-------
Column No.

   1
   2
   3
   4
   5
   6
   7
   8
   9
  10
  11
Test number or trial run
Date at completion of last fill
Time at completion of last fill
Date test began
Time test began
Time test ended
Product temperature differential
Nominal leak rate
Induced leak rate
Measured leak rate
Measured minus induced leak rate
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
AT6S records
By subtraction
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 from 50% full to between 90% :to 95% full.
This temperature differential is the actual differential achieved in the
field and not the nominal temperature differential.  The difference can
be calculated by one of two methods.  If the field crew measured the
temperature of the product added and that of the product in the tank just
prior to filling, then take the difference between these two tempera-
tures.  If the field crew measured the temperature of the product in the
tank before and after filling and recorded the amount of product added,
then calculate the temperature differential based on volumes and tempera-
tures according to the formula in Section 7.4  The data necessary for
these calculations should all be provided on the Individual Test Log.
                                   B-16

-------
                                   Reporting Form for Leak Rate Data
                             Automatic Tank Gauging System (ATGS)
         ATGS Name and Version-
         Evaluation Period:
from
to
. (Dates)

Test No.
Trial Run

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

	 y....s. 	 '
























Time at
Completion
of Last Fill
(military)


























Date Test
Began
(m/d/y)


























Time Test
Began
(military)

Time Test
Ended
(military)

Product
Temperature
Differential
(deg F)
0
Nominal
Leak Rate
(gal/h)
0
Induced
Leak Rate
(gal/h)
0
, - ' ',- ' * . i
























































































































Measured
Leak Rate
(gal/h)


























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


























ATGS-Data Reporting Form
                                                                    Pagel of 2

-------
                                 Reporting Form for Leak Rate Data
                            Automatic Tank Gauging System (ATGS)
        ATGS Name and Version:
        Evaluation Period:
from
to.
.(Dates)

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
























Time at
Completion
of Last Fill
(military)
























Date Test
Began
(m/d/y)
























Time Test
Began
(military)
























Time Test
Ended
(military)
























Product
Temperature
Differential
(degF)
























Nominal
Leak Rate
(gal/h)
























Induced
Leak Rate
(gal/h)
























Measured
Leak Rate
(gal/h) .







.
















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
























ATGS-Data Reporting Form
                                                                                                 Page 2 of 2

-------
                            Individual Test Log
                   Automatic  Tank Gauging System (ATGS)

                   Instructions  for completing the form

 This 5-page test log form is  to be filled out by the field crew of the
 evaluating organization.   A separate form is to be filled out for each
 individual test including the trial run (at least 25.)  The information
 on these forms is to be kept  blind to the vendor during the period of
 evaluation of the ATGS.

 The form consists of eight parts.  These are:

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

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

 Header  Information

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

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

 General  Background Information

 Indicate the  commercial name  of the ATGS.  Include a version identifica-
 tion  if  the ATGS uses different versions for different products or tank
 sizes.   The vendor's recommended stabilization period (if applicable) 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 qround-
water level at the time of the test.

Theoretically, this information would remain unchanged for the whole
 evaluation period.  However, weather conditions could change  and  affect
the ground-water level.   Also, the evaluating organization could  chanqe
the test tank.                                                       3
                                  B-19

-------
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.                                       j

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.               \

Special Case Reporting                                     :

Use the Individual Test Log form to record all data pertaining to the
trial run.  Next, when emptying the tank to half full and then filling to
90% to 95% capacity before performing the first test, note on the form
that this has been done.  Simply indicate on page 1 the dates and time
periods and volumes when product was removed and then added.  This is the
only case where emptying and filling are performed in sequence without a
test being performed in between.  Record all other information (e.g.,
temperature of product added) as applicable.
                                                           j
Conditions at Beginning of Test

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

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

Conditions at Completion of Testing                        !

Indicate date and time when the test is completed.

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

Leak Rate Data

This section is to be filled out by the evaluating organization's
statistician or analyst performing the calculations.  This section can
therefore be filled out as the evaluation proceeds or at the end of the
evaluation.
                                   B-20

-------
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 by calculation from the data reported
by the evaluating field crew ori page 4 (and 5, if needed) of this form.
The measured leak rate is that recorded by the ATGS for that test.

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

Additional Comments (If needed)

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

Induced Leak Rate Data (pages 4 and 5)

This form is to be filled out by the evaluating organization's field
crew.  From the randomization design, the crew will know the nominal leak
rate to be targeted. The induced leak rate will  be known accurately at
the end of the test. However, the protocol requires that the induced leak
rate be within 30% of the nominal leak rate.
                                  B-21

-------

-------
 Name of Field Operator	
 Signature of Field Operator_	      jest No.
 Date of Test
                            Individual Test Log
         Automatic Tank Gauging System (ATGS)
 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
 ATGS Name and Version	
 Product Type	
 Type of Tank  	
 Tank Dimensions (nominal)
    Diameter   	inches
    Length     	inches
    Volume    	gallons
 Ground-water level	inches above bottom of tank
 if applicable, recommended stabilization period before test (per vendor SOP)
    	hours	minutes

 Conditions Before Testing
 Date and time at start of conditioning test tank	date  	military time
 Stick reading before conditioning test tank
    Product	inches	gallons
   Water  	inches	gallons
 Temperature of product in test tank before conditioning    	   °F D or °C D
 Stick reading after conditioning test tank
   Product	inches	gallons
Amount of product (check one only):
            CH no change in product level
            D removed from tank (by subtraction)      	gallons
            D added to tank (by subtraction)          	gallons
ATGS-Test Log                                                      Page 1 of 5

-------
Name of Field Operator	
Signature of Field Operator_
Date of Test
Test No.
Conditions Before Testing (continued)
If product was added
   Temperature of product added to fill test tank to test level
   •'	°FD or°cD
   Temperature of product in tank immediately after filling	
Date and time at completion of conditioning	date
     BF I
] or°cD   .
..military time
Conditions at Beginning of Test
Date and time at start of ATGS test data collection
   	date	military time                           ;
> Complete the induced leak rate data sheet (use attached pages 4 and 5)
Temperature of product in tank at start of test	°FD or °cD  '
                                                                i
Weather conditions at beginning of test
   Temperature 	°FD or°cD                            :
   Barometric pressure	mm Hg CH or	in. Hg EU      i
   Wind              None EH    Light D     Moderate d     Strong CD
   Precipitation        NoneD    Light D     Moderate D     Heavy D
   Sunny CD          Partly cloudy D          Cloudy D
Nominal leak rate	gallon per hour                           I
ATGS - Test Log
               Page 2 of 5

-------
 Name of Field Operator	
 Signature of Field Operator	      jest No.
 Date of Test
 Conditions at Completion of Testing  .
 Date and time at completion of test data collection
    	date 	military time
 Stick reading at completion of test data collection
    Product	inches	gallons
    Water  	inches	gallons
 Weather Conditions at End of Test
    Temperature	°F D or °C D
    Barometric pressure	mm Hg D or  	in. Hg D
    w'nd             NoneD     Light D     Moderate D      Strong D
    Precipitation       NoneD     Light D     Moderate D      Heavy D
    Sunny D         Partly Cloudy D          Cloudy D

 Leak Rate Data (not to be filled out by field crew)
 Nominal leak rate	gal/h
 induced leak rate	gal/h
 Leak rate measured by vendor's method	gal/h
 Difference (measured rate minus induced rate)     	gal/h

 Additional Comments (Use back of page if needed)
ATGS- Test Log                                                        Page 3 of 5

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

Date of test	
Test No,
Induced Leak Rate Data Sheet

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
























Amount of
product
collected
(mL)
























Comments (if applicable)














,

•
•
i



i
i
;
ATGS-Test Log
                     Page 4 of 5

-------
 Name of Field Operator	
 Signature of Field Operator	
 Date of test	

 Induced Leak Rate Data Sheet (continued)
Test No.

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
























Amount of
product
collected
(mL)
























Comments (if applicable)
























ATGS-Test Log
                                                                              Page 5 of 5

-------

-------
              Reporting Form for Water Sensor Evaluation Data
                       Automatic Tank Gauging System

 This  4-page form is  to be filled out by the field crew of the evaluating
 organization when evaluating the performance of the AT6S water sensor   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  ATGS.   Include  a version  identifica-
 tion  if the ATGS  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  6.4 and 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 6.4.   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-27

-------

-------
                          Reporting Form for Water Sensor Evaluation Data
                               Automatic Tank Gauging System
 ATGS 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.
ATGS-Water Sensor
                                                                                Page 1 of 4

-------
ATGS Name and Version:
Date of Test: 	
Product Type: 	
                       Reporting Form for Water Sensor Evaluation Data
                            Automatic Tank Gauging System;
  Name of Field 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 i
Increment
Difference
Calc.-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
i
.




i




i



i
i
\
\
, I
;

























 NOTE:   This form provides a template for data reporting.
         Use as many pages as necessary.
ATGS-Water Sensor
                                       Page 2 of 4

-------
                        Reporting Form for Water Sensor Evaluation Data
                             Automatic Tank Gauging System
 ATGS Name and Version:
 Date pf Test:  	.
 Product Type: 	
  Name of Field Operator:.
Signature of Reid 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

























Calculated
Water Height
Increment, h
(in)
C

























Sensor
Reading
(in)
D

























Measured
Sensor
Increment
(in)
E

























Increment
Difference
Calc.-Meas.
(in)
C-E

























NOTE:    This form provides a template for data reporting.
         Use as many pages as necessary.
ATGS-Water Sensor
                                                                            Page 3 of 4

-------
Name of ATGS_
Version
Limitations on the Results (continued)
•  The difference between added and in-tank product temperatures is no greater
   than + or-	_degrees Fahrenheit.  '                             i
•  The waiting time between the end of filling and the start of the test data collection
   is at least	hours.

•  The total data collection time for the test is at least	hours.

>  Safety disclaimer: This test procedure only addresses the issue of th© ATG system's
   ability to detect leaks. It does not test the equipment for safety hazards.

Certification of Results
I certify that the ATGS was installed and operated according to the vendor's instructions and
that the results presented on this form are those obtained during the evaluation. I also certify
that the evaluation was performed according to one of the following:
ED standard EPA test procedure for ATGS
EH alternative EPA test procedure for ATGS
ED equivalent test procedure for ATGS (describe below or reference document)
(printed name)
(signature)
(organization performing evaluation)              (city, state)
(date)                                         (phone number)
ATGS - Results Form                                                      ,     Page 2 of 2

-------

-------

-------