Baireiie
The Business of Innovation EPA/600/R-15/254
Environmental Technology
Verification Program
Advanced Monitoring
Systems Center
Quality Assurance Project Plan
Suitability of Leak Detection Technology
for Use In Ethanol-Blended Fuel Service
-------
Baltelle
The Business of Innovation EPA/600/R-15/254
Quality Assurance Project Plan
Suitability of Leak Detection Technology for Use In
Ethanol-Blended Fuel Service
April 17, 2013
Prepared by:
Anne Marie Gregg
Brian Yates
Amy Dindal
Battelle
505 King Ave.
Columbus, OH 43201-2693
SECTION A
PROJECT MANAGEMENT
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 3 of 184
Al TITLE AND APPROVAL PAGE
Quality Assurance Project Plan
for
Biofuel Properties and Behavior Relevant to
Underground Storage Tank Leak Detection
System Performance
Doug Grosse Date
EPA Project Officer
Amy Dindal Date
Battelle AMS Center Manager
Anne Marie Gregg Date
Battelle Testing Coordinator
Rosanna Buhl Date
Battelle Quality Assurance Manager
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 4 of 184
A2 TABLE OF CONTENTS
SECTION A: PROJECT MANAGEMENT 2
Al TITLE AND APPROVAL PAGE 3
A2 TABLE OF CONTENTS 4
A3 LIST OF ABBREVIATIONS/ACRONYMS 7
A4 DISTRIBUTION LIST 9
A5 VERIFICATION TEST ORGANIZATION 10
A5.1 Battelle 10
A5.2 EPA AMS Center 14
A5.3 Underground Storage Tank Leak Detection Stakeholder Committee 14
A5.4 Analysis Laboratory 15
A6 BACKGROUND 15
A6.1 Research Need 16
A6.2 Technology Description 17
A7 TEST DESCRIPTION AND SCHEDULE 18
A7.1 Test Description 18
A7.2 Schedule 21
A7.3 Health and Safety 21
A8 QUALITY OBJECTIVES AND CRITERIA FOR MEASUREMENT DATA 22
A9 SPECIAL TRAINING/CERTIFICATION 23
A10 DOCUMENTS AND RECORDS 23
SECTIONS: DATA GENERATION AND ACQUISITION 26
Bl EXPERIMENTAL DESIGN 26
Bl.l Preparation of Test Blends 26
B1.2 Preparation of BFW Mixtures 27
B1.3 Bench-scale Testing 29
B 1.3.1 Test Procedures 31
Bl.3.1.1 Intrinsic Properties of BFW Mixtures 31
Bl.3.1.2 Coefficient of Thermal Expansion 32
Bl.3.1.3 Non-additive Volume Changes 32
Bl.3.1.4 Interface Determination 34
Bl.3.2 Statistics for Bench-scale Test Sets 37
Bl.3.3 Precision 39
B1.4 Laboratory-scale Testing 40
Bl.4.1 Test Procedures 41
Bl.4.1.1 Continuous Water Ingress Series 41
Bl.4.1.2 QuickDump Series 43
Bl.4.2 Statistics for Laboratory-scale Test Sets 45
B 1.4.2.1 Accuracy 46
El.4.2.2 Sensitvity 46
Bl.4.2.3 Precision 48
B1.5 Full-scale Testing 48
Bl.5.1 Test Procedures 49
Bl.5.2 Statistics for Full-scale Test Set 51
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 5 of 184
B1.6 Reporting 54
B2 SAMPLING METHODS 55
B2.1 Sample Collection, Storage and Shipment 55
B2.2 Digital Video Recording 56
B3 SAMPLE HANDLING AND CUSTODY 56
B4 REFERENCE METHODS 56
B5 QUALITY CONTROL 57
B6 INSTRUMENT/EQUIPMENT TESTING, INSPECTION, AND
MAINTENANCE 59
B7 INSTRUMENT/EQUIPMENT CALIBRATION AND FREQUENCY 60
B8 INSPECTION/ACCEPTANCE OF SUPPLIES AND CONSUMABLES 61
B9 NON-DIRECT MEASUREMENTS 62
BIO DATA MANAGEMENT 62
SECTION C: ASSESSMENT AND OVERSIGHT 64
Cl ASSESSMENT AND RESPONSE ACTIONS 64
Cl.l Performance Evaluation Audits 64
C1.2 Technical Systems Audits 65
C1.3 Data Quality Audits 66
C1.4 QA/QC Reporting 67
C2 REPORTS TO MANAGEMENT 67
SECTION D: DATA VALIDATION AND USABILITY 69
Dl DATA REVIEW, VERIFICATION, AND VALIDATION 69
D2 VALIDATION AND VERIFICATION METHODS 69
D3 RECONCILIATION WITH USER REQUIREMENTS 71
SECTION E: REFERENCES 72
FIGURE
Figure 1. Project Organizational Chart 11
APPENDIX
Appendix A: UNDERGROUND STORAGE TANK LEAK DETECTION
STAKEHOLDER COMMITTEE 74
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 6 of 184
TABLES
Table 1. Leak Detection Technologies and Principles of Operation 17
Table 2. General Testing Schedule 21
TableS. Project Records Submitted to the TC 25
Table 4. Mixing Ratios of EO and Ethanol/Isobutanol for Preparation of Test Blends 27
Table 5. Water Volumes Required to Prepare BFW Mixtures from Each Test Blend 28
Table 6. Acceptance Criteria for Test Blend Preparation 29
Table 7. Summary of the Bench-scale Test 35
Table 8. Data Collection Quality Control Assessments of the Fuel Properties 36
Table 9. Summary of Continuous Water Ingress Runs 43
Table 10. Summary of Lab oratory-Scale Test Set 45
Table 11. Summary of Full-scale Test Sets 52
Table 12. List of ASTM Standards and Assessment of Data Quality 57
Table 13. Measurement Quality Objectives for Analytical Methods 59
Table 14. Data Verification Checks 63
Table 15. Analytical Methods and PEA Materials 65
Table 16. Summary of Assessment Reports 68
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 7 of 184
A3 LIST OF ABBREVIATIONS/ACRONYMS
ADQ audit of data quality
AMS Advanced Monitoring Systems
ANSI American National Standards Institute
ASTM ASTM International
ATG automatic tank gauge
BFW biofuel-water
COC chain of custody
CV coefficient of variation
DI deionized
DQO data quality objective
EISA Energy Independence Security Act
EPA Environmental Protection Agency
EO-E85 ethanol blended into gasoline at nominal value of number, from 0% up to 85%
ETV Environmental Technology Verification
gal/hr gallon per hour
g/L gram per liter
116 isobutanol blended into gasoline at 16%
ICFTL Iowa Central Fuel Testing Laboratory
ID identification
JHA Job Hazard Analysis
L liter
LD leak detection
LRB Laboratory Record Book
mL milliliter
MLC minimum water level change
MQO Measurement quality obj ectives
NACE National Association of Corrosion Engineers
NIST National Institute of Standards and Technology
NWGLDE National Work Group on Leak Detection Evaluations
PD probability of detection
PEA Performance Evaluation audit
PFA probability of false alarm
PO project officer
QA quality assurance
QAPP Quality Assurance Project Plan
QC quality control
QAM quality assurance manager
QMP quality management plan
RPD relative percent difference
RMO Records Management Office
SD standard deviation
SIR Statistical Inventory Reconciliation
SOP Standard Operating Procedure
SRM standard reference material
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 8 of 184
TA technology assessment
TC testing coordinator
TL tolerance limit
TSA Technical Systems Audit
UST underground storage tank
jiL microliter
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 9 of 184
A4 DISTRIBUTION LIST
EPA
Doug Grosse
U.S. Environmental Protection Agency
26 West Martin Luther King Drive
Mail code: 489
Cincinnati, OH 45268
Andrea Barbery
Tim Smith
U.S. EPA - Headquarters
OSWER/OUST/RPD (MC-5402P)
1200 Pennsylvania Avenue, NW
Washington, DC 20460
Peer Reviewers
Earle Drack
DirAction, LLC
111 Bertolet School Road
Spring City, PA 19475
Loraine Sabo
Franklin Fueling Systems
3760 Marsh Road
Madison, WI 53718
Ken Wilcox
Ken Wilcox Associates, Inc.
1125 Valley Ridge Drive
Grain Valley, MO 64029
Battelle
Anne Gregg
Brian Yates
Melissa Kennedy
Rosanna Buhl
Amy Dindal
Battelle
505 King Avenue
Columbus, OH 43201
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 10 of 184
A5 VERIFICATION TEST ORGANIZATION
Oversight of this investigation will be provided by the U.S. Environmental Protection
Agency (EPA) through the Environmental Technology Verification (ETV) Program. This
project will be performed by Battelle, which manages the ETV Advanced Monitoring Systems
(AMS) Center through a cooperative agreement with EPA. The scope of the AMS Center covers
monitoring technologies for contaminants and natural species in air, water, and soil to protect
human health and ecological resources by reducing or preventing environmental risks.
The daily operations associated with this testing will be coordinated and supervised by
Battelle. Testing will be performed using Battelle's laboratory facilities under highly-controlled
conditions and selected field sites (e.g., existing distribution stations) under real-world
conditions. Expert peer reviewers and EPA AMS Center Management will review the Quality
Assurance Project Plan (QAPP) (this document) and final report named "The Suitability of Leak
Detection Technology for Use in Biofuel Service" (henceforth referred to as the Technology
Assessment [TA]). A draft TA exists1 and data generated following the approved QAPP will be
used to inform the revisions to and finalization of the TA, which is a deliverable associated with
this project. The QAPP and TA will be approved by the EPA AMS Center Management.
The organization chart presented as Figure 1 identifies the organizations and individuals
associated with the testing. Roles and responsibilities are defined further below. Quality
assurance (QA) oversight will be provided by both the Battelle Quality Assurance Manager
(Battelle QAM) and by EPA, at its discretion. This testing is Quality Category II, which requires
a QA review of 25% of the test data (see Section Cl).
A5.1 Battelle
Ms. Anne Marie Gregg is the AMS Center's Testing Coordinator (TC) for this project. In
this role, Ms. Gregg will have overall responsibility of ensuring that the technical, schedule, and
cost goals established for the testing are met. Specifically, Ms. Gregg will:
• Prepare and oversee review and approval of the QAPP and TA;
• Establish a budget for the testing and manage staff to ensure the budget is not
exceeded;
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 11 of 184
Battelle
Management
Rosanna Buhl
Battelle
Quality Assurance
Manager
USTLD
Stakeholder
Committee
Amy Dindal
Battelle AMS
Center Manager
Anne Marie Gregg
Testing
Coordinator
Battelle
Technical Staff
Doug Grosse
EPA AMS Center
Project Officer
EPA
Quality
Representative
Analytical
Laboratories
Figure 1. Project Organizational Chart
Revise the draft QAPP and draft TA in response to reviewers' comments;
Assemble a team of qualified technical staff to conduct the testing;
Direct the team in performing the testing in accordance with this QAPP;
Ensure Battelle and subcontracted analytical laboratories perform the analyses
according to the specified method requirements.
Independently acquire technologies for testing, if necessary;
Hold a kick-off meeting approximately one week prior to the start of the testing to
review the critical logistical, technical, and administrative aspects of the testing.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 12 of 184
• Responsibility for each aspect of the testing will be reviewed to ensure each
participant understands his/her role;
• Ensure that all quality procedures specified in this QAPP and in the AMS Center
Quality Management Plan2 (QMP) are followed;
• Serve as the primary point of contact for underground storage tank (UST) leak
detection (LD) stakeholders;
• Ensure that confidentiality of sensitive information regarding tested technologies is
maintained;
• Become familiar with the operation and maintenance of tested technologies;
• Respond to QAPP deviations and any issues raised in assessment reports, audits, or
from test staff observations, and institute corrective action as necessary; and
• Coordinate distribution of the final QAPP and TA.
Ms. Amy Dindal is Battelle's Manager for the AMS Center. As such, Ms. Dindal will
oversee the various stages of testing. Ms. Dindal will:
• Review and approve the draft and final QAPP;
• Attend the proj ect kick-off meeting;
• Ensure that necessary Battelle resources, including staff and facilities, are committed
to the testing;
• Ensure that confidentiality of sensitive information regarding tested technologies is
maintained;
• Support Ms. Gregg in responding to any issues raised in assessment reports and
audits;
• Maintain communication with EPA's technical and quality managers;
• Issue a stop work order if Battelle or EPA QA staff discover any situation that will
compromise test results; and
• Review the draft and final TA.
Battelle Technical Staff will support Ms. Gregg in planning and conducting the testing.
The technical staff will:
• Assist in planning for training and testing as necessary;
• Attend the proj ect kick-off meeting;
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 13 of 184
• Arrange for and/or acquire adequate fuel supplies, equipment, and facilities/locations
for performing testing and disposal of generated wastes;
• Conduct testing and collect data and samples according to this QAPP;
• Conduct analytical methods and coordinate with analytical labs to determine select
physical and chemical properties of the fuel;
• Conduct and observe testing as appropriate;
• Support Ms. Gregg in responding to any issues raised in assessment reports and audits
related to statistics and data reduction as needed;
• Immediately report deviations from this QAPP to the TC; and
• Provide results of statistical calculations and associated discussion for the TA as
needed.
Ms. Rosanna Buhl is the Battelle QAM for the AMS Center. Ms. Buhl will:
• Review and approve the draft and final QAPP;
• Prior to the start of testing, verify the presence of applicable training records,
including any technology training, as necessary;
• Conduct a technical systems audit (ISA) at least once during the testing.
• Conduct audits of data quality;
• Prepare and distribute an audit report for each audit;
• Verify that audit responses for each audit finding and observation are appropriate and
that corrective action has been implemented effectively;
• Provide a summary of the QA/quality control (QC) activities and results for the TA;
• Communicate to the TC and/or technical staff the need for immediate corrective
action if an audit identifies QAPP deviations or practices that threaten data quality;
• Delegate QA activities to other Battelle quality staff as needed to meet project
schedules;
• Review and approve any QAPP amendments, deviations and audit reports, if
necessary;
• Work with the TC and Battelle's AMS Center Manager to resolve data quality
concerns and disputes;
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 14 of 184
• Recommend a stop work order if audits indicate that data quality or safety is being
compromised; and
• Review the draft and final TA.
A5.2 EPA AMS Center
EPA's responsibilities in the AMS Center are based on the requirements stated in the
"Environmental Technology Verification Program Quality Management Plan" (ETV QMP)3.
The EPA's quality representative will:
• Review the draft and final QAPP;
• Perform at his/her option one external TSA during the testing;
• Prepare and distribute an assessment report summarizing results of the external audit;
• Perform audits of data quality;
• Notify the EPA AMS Center Project Officer (PO) of the need for a stop work order if
the audit of data quality indicates that data quality is being compromised; and
• Review the draft and final TA.
Mr. Doug Grosse is EPA's PO for the AMS Center. Mr. Grosse or designee will:
• Review and approve the draft and final QAPP;
• Oversee the EPA review process for the QAPP and TA;
• Be available during the testing to review and authorize any QAPP deviations by
phone and provide the name of a delegate to the Battelle AMS Center Manager
should he not be available during the testing period;
• Approve decisions based on recommendations from the UST LD stakeholders;
• Review and approve the draft and final TA;
• Coordinate the submission of the TA for final EPA approval; and
• Post the QAPP and TA on the ETV Web site.
A5.3 Underground Storage Tank Leak Detection Stakeholder Committee
An UST LD stakeholder committee was specifically assembled for the execution of this
project, including the preparation and revision of this QAPP. Appendix A presents a list of
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 15 of 184
committee members. Committee members represent industry associations, technology vendors,
technology users and state and federal regulatory agencies including the National Work Group
on Leak Detection Evaluations (NWGLDE). The UST LD stakeholders and/or peer reviewers
will:
• Participate in technical panel discussions (when available) to provide input to the test
design;
• Review and provide input to the QAPP; and
• Review and provide input to the TA.
Finally, this QAPP and TA based on testing described in this document will be reviewed
by experts in the fields related to UST LD. The following experts have agreed to provide peer
review:
• NWGLDE (member names and affiliations are presented in Appendix A)
• Earle Drack, DirAction, LLC
• Lorraine Sabo, Franklin Fueling Systems
• Ken Wilcox, Ken Wilcox Associates, Inc.
A5.4 Analytical Laboratory
In addition to analytical method support, which will be provided by Battelle, Iowa
Central Fuel Testing Laboratory (ICFTL) will be contracted to provide chemical measurements
defined later in this QAPP. The laboratory is ISO-9001:2008 and BQ-9000 accredited.
A6 BACKGROUND
Currently, approximately 584,000 USTs4 containing petroleum products in service in the
United States have the potential for contaminating groundwater and subsequently drinking water
should they fail. UST LD regulations were put in place to specify monitoring requirements for
detecting leaks. To ensure protection of human health and the environment, the EPA established
minimum performance criteria for equipment used for LD and promulgated these specifications
in 40 CFR 280. For example, all tank tightness testing equipment must be capable of detecting a
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 16 of 184
0.10 gallon per hour (gal/hr) leak rate with a probability of detection of at least 95% and a
probability of false alarm of no more than 5%.
Biofuels contribute an increasing portion of the fuel supply in the United States due to the
enactment of the Renewable Fuel Standard established by the Energy Policy Act of 2005 and
amended by the Energy Independence and Security Act (EISA) of 2007. These federal mandates
have spurred increased production, distribution, dispensing and use of biofuels, particularly in
the transportation sector where the use of ethanol-blended gasoline has become common.
Ethanol is currently blended into 90% of all gasoline consumed in the United States at low and
high percentages from less than 10% (E10) and approximately 85% (E85) ethanol5. Biofuel
consumption is expected to increase in response to upcoming EISA requirements for biofuel
production and use. Because petroleum and ethanol have specific differences in their chemical
and physical characteristics, LD technologies operating based on or affected by density,
conductivity, coefficient of thermal expansion, and other properties may not function properly in
the new biofuels environment. Questions have been raised about the long-term performance of
new and existing LD devices due to the corrosive nature of ethanol, although long-term material
compatibility will not be directly evaluated in this QAPP.
A6.1 Research Need
The ETV Program's AMS Center conducts third-party performance testing of
commercially available technologies that monitor, sample, detect, and characterize contaminants
or naturally occurring species across all matrices. The purpose of ETV is to provide objective
and quality-assured performance data on environmental technologies so that users, developers,
regulators, and consultants can make informed decisions about purchasing and applying these
technologies. Stakeholder committees of buyers and users of such technologies recommend
technology categories, and technologies within those categories, as priorities for testing. The
research described in the QAPP is focused on evaluating LD technologies in general to produce a
TA that is not specific to a vendor or LD technology category. The purpose of this QAPP is to
specify procedures for gathering data to inform the TA and the UST LD community as a whole.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Page 17 of 184
A6.2 Technology Description
Several different categories of LD technologies are used to monitor USTs for possible
fuel leaks. UST LD technologies may be classified broadly as volumetric or non-volumetric
approaches. Volumetric technologies are those that measure a specific quantity which can then
specify a value of leak rate while non-volumetric technologies yield qualitative results. Non-
volumetric technologies report only the presence or absence of a leak. Table 1 describes the
variety of categories of UST LD systems and their principles of operation. Appropriately
installed and operated technologies of either type may be used to satisfy requirements of 40 CFR
280.41.
Table 1. Leak Detection Technologies and Principles of Operation
Technology
Principle of Operation
VOLUMETRIC TESTING TECHNOLOGIES
Automatic Tank Gauge (ATG) Systems
Magnetostrictive Probes
Ultrasonic or Acoustic
Methods (speed)
Capacitance Probes
Mass
Buoyancy/Measurement
Systems
Wire sensor inside a stainless steel rod detects presence of magnetic field, which indicates height
of float.
Sensor detects changes in fluid levels detecting a sound wave echo reflected from the interface of
water/fuel or fuel/air and calculates level based on speed of sound in the product.
Detection is based on dielectric property of the stored liquid.
Buoyancy of probe is detected on a load cell and compared to tank geometry to calculate fuel liquid
level.
Statistical Inventory Reconciliation (SIR) Technologies
SIR -Manual
SIR - Data from ATG system
An SIR vendor performs analysis of manually collected product level data for evidence of tank
tightness.
Computer software is used to perform analysis of inventory records to determine tank tightness.
Interstitial Integrity Monitoring Technologies
Vacuum
Pressure
System uses an integral vacuum pump or pressurized system to continuously maintain a partial
vacuum within the interstitial space of double-walled tanks and double-walled piping. System is
capable of detecting breaches in both the inner and outer walls of double-walled tanks or double-
walled piping.
Liquid-Phase Interstitial Technologies
Liquid Filled
Sensor -liquid ingress
A liquid solution is used to fill the tank or piping interstice. The dual-point level sensor system
monitors the liquid level in the interstitial reservoir and sounds an alarm if the liquid level is either
too high (ingress of liquid) or too low (egress of liquid).
Varies depending on the type of sensor and comes in multiple forms. Most examples include use of
refractive index or float switch.
Methods of Release for Piping
Pressure Decay
Measures the change in pressure between the atmosphere and the pressurized product in the line
overtime.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 18 of 184
Table 1. Leak Detection Technologies and Principles of Operation (Continued)
Technology
Principle of Operation
VOLUMETRIC TESTING TECHNOLOGIES
Constant Pressure
Mechanical Leak Detectors
Sensors monitor change in volume at constant pressure.
Permanent installation on piping. Conducts leak tests every time the pump engages.
NON-VOLUMETRIC TECHNOLOGIES
Fuel Sensitive Polymers
Tracers
Acoustic Precision Test -
Tanks
Acoustic Precision Test-
Piping
Fiber optic cable is coated with a polymer that interacts with fuel. When fuel is present, the light
passing through the cable will be affected.
Chemical tracer is added to the product and the surrounding soil is monitored for the chemical
tracer.
Detected sounds are used to identify potential leaks; an orifice is used to simulate the sound
produced as liquid or air leaks out of a tank and associated piping. This is accomplished using
acoustic sensors and microphones, and ultrasonic sensors and hydrophones.
Water Detection Technologies
Water Float
Density Float
Conductivity Meter
Buoyancy of float allows the signal generated (magnetic field or capacitance) to coincide with the
top of the liquid layer based on the liquid density in comparison to the float density. These floats
are specifically designed for water detection and the density difference between water and
gasoline.
Buoyancy of a float signals changes in product that compares density data changes over time to
assess the change in product quality due to water ingress. This float is sensitive to the aqueous
phase detection.
Operates on the principle of conductivity. Sensors are mounted on the bottom of a probe that is
positioned just above water level. After negative pressure has been applied to the tank, and if there
is water intrusion, water will "short out" the sensor causing conductivity.
A7 TEST DESCRIPTION AND SCHEDULE
A7.1 Test Description
Specific procedures described herein are based on input received from the UST LD
stakeholders, the procedures described in the 1990 EPA protocol for automatic tank gauge
(ATG) systems6 and the performance requirements found in 40 CFR 280. This QAPP is
organized as three main test sets. The three test sets are:
1. Bench-scale test set for the determination of select physical and chemical properties
of biofuels and biofuel-water (BFW) mixtures;
2. Laboratory-scale test set for the detection and quantification of BFW mixture
processes affecting performance of UST LD systems (i.e., water ingress and mixing)
to inform operation and predict performance of full-scale UST LD systems; and
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 19 of 184
3. Full-scale field demonstration test set of UST LD systems as they perform under real
world conditions with ethanol-blended gasoline.
The first test set is only investigating the fuel mixtures and their properties. UST LD
technologies are not involved. It was identified by the UST LD Stakeholder Committee that the
properties of the different ethanol-blended fuels need to be evaluated to understand their
respective behavior when at equilibrium with water (the extreme scenario for water intrusion).
The laboratory-scale test set integrates water detection/mixing and the UST LD technologies.
The water ingress tests (bench- and laboratory-scale) will be performed in laboratories at
Battelle's facility in Columbus, OH. The fuel and water interactions and the technology
responses during these two test sets will be video recorded and the fuel properties will be
measured. The full-scale demonstrations involve LD capabilities of the technologies only. They
will be conducted in the field in an UST at a service or blending station and may be conducted
upon review of the data from the bench- and lab oratory-scale test sets with the UST LD
stakeholders and EPA PO. Since the technologies have been operating in these fuels for many
years, how many and which technologies used for the full-scale demonstration will be
determined with input from UST LD stakeholders once the data from the bench- and laboratory-
scale test sets are reviewed.
The performance of the UST LD technologies will be evaluated based on the following
parameters.
• Bench-scale
o The test blends prepared and their accuracy will be verified with respect to target
values of water and ethanol content. To be considered acceptable for testing, the
target blend level will be within 15% relative percent difference (RPD) of the
target concentration.
o To be considered acceptable for data reporting the resulting triplicate data points
on each blend properties will be <15% coefficient of variation (CV). LD
technologies will not be used in this test scale.
• Laboratory-scale
o Accuracy
o Sensitivity
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 20 of 184
o Precision
• Full-scale
o Probability of false alarm (PFA)
o Probability of detection (PD)
The responses for these parameters will be collected from the technologies as either a
"detect" or "non-detect" or if determined by the technology, as a nominal leak rate. An
independent comparison to metered rates will be used to confirm the true water ingress rates and
simulated fuel leak rates established during the bench- and full-scale test sets.
The tests will be performed with the technologies operating in accordance with the
vendor's recommended procedures as described in the user's instructions/manual or during
training provided to the operator. Similarly, calibration and maintenance of the technologies will
be performed by a trained vendor representative or a trained UST service company technician.
Details of the technology training, if not in a user's instructions/manual and just provided onsite,
will be documented in writing. Results from the technologies will be recorded electronically by
the technology display/recording console and/or manually in laboratory record books (LRBs)
and test data sheets.
A TA report using the results to understand the overall performance of LD systems will
be prepared with the obtained data from these tests and comparison to similar, previously-
reported values. The testing details and QC information will be reported either within the body
of the TA report or as an appendix. The TA will be reviewed by EPA and the peer reviewers. In
performing the testing, Battelle will follow the technical and QA procedures specified in this
QAPP and comply with the data quality requirements in the AMS Center QMP2.
Quality procedures include a TSA and audits of data quality (ADQs). The Battelle QAM
or her designee will perform the TSA and ADQs. All data collected during the first two weeks
of testing will be considered the first batch of data. The first batch of data will be delivered to
EPA within 30 days of test initiation. Unaudited data will include the disclaimer "have not been
reviewed by Battelle QAM." The first ADQ will review the first batch of data delivered. A
second ADQ will be performed once all data are collected, and a final ADQ will be performed
on the TA. More detail is provided in Section C.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 21 of 184
A7.2 Schedule
Table 2 shows the schedule of testing and data analysis/reporting activities to be
conducted in this testing.
Table 2. General Testing Schedule
Approximate
Months in
2013
March
April-May
May-June
July- August
September
October
Testing Activities
• Set up of bench-scale and laboratory-scale test
sets
• Conduct pre-test checks and dry runs for bench
and laboratory-scale test sets
• Perform performance evaluation audit (PEA)
• Complete PEA report
• Conduct bench
• Preparation and approval of laboratory test sets
QAPP addendum
• Perform ISA
• Perform initial ADQ (first batch, see Section A6.1 ;
C1. 3) (25% of all data)
• Coordinate full-scale test set by identifying sites
and technologies available for testing at the sites
• Preparation and approval of full-scale test sets
QAPP addendum
• Coordinate for testing supplies to be delivered to
testing sites
• Install necessary equipment and ensure
technologies are installed and operating
appropriately according to the vendor or LIST
service company
• Conduct full-scale testing
• Complete testing
• Perform second ADQ (25% of all data)
• Prepare TA
• Perform third ADQ of TA
• Coordinate reviews of TA
• Prepare final TA
Data Analysis and Reporting
• Not Applicable
• Compile PEA results
• Compile bench and laboratory
data
• Compile TSA results
• Compile first ADQ results
• Document field demonstration
activities
• Compile data
• Review and summarize data
• Compile second ADQ results
• Begin TA revision
• Complete TA
• Compile third ADQ results
• Complete internal review of TA
• Complete peer review of TA
• Revise TA per review
comments
• Submit final TA for EPA
approval
A7.3 Health and Safety
Battelle will conduct all testing following the safety and health protocols in place for the
locations used for testing. In addition, a job hazard analysis (JHA) will be prepared to describe
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 22 of 184
the specific hazards associated with transportation, handling and testing of gasoline, ethanol and
isobutanol, as well as the use of administrative and engineering controls, personal protective
equipment and other procedures required to reduce the possibility of potential mishaps. Specific
required training will be described in the JHA and completed by all employees conducting
testing. These include maintaining a well-ventilated, explosion-proof work environment,
providing secondary containment for all storage vessels, and promoting a current awareness of
safe chemical and waste handling methods. Proper personal protective equipment will be worn,
and safe laboratory practices will be followed. Standard Battelle JHA forms will be completed
once the hazardous activities are defined and before testing begins. The JHA form will include
the following topics, in addition to others:
• Fuel handling and safety procedures;
• Ventilation procedures;
• Waste handling and labeling; and
• Use of explosion-proof equipment.
The JHA form will be physically present at the testing locations. All test participants will
be required to review and understand the JHA form prior to initiating laboratory or field work
and adhere to its procedures during conduct of all testing.
A8 QUALITY OBJECTIVES AND CRITERIA FOR MEASUREMENT DATA
The overall data quality objectives (DQOs) of this study are to select and measure physical and
chemical properties of biofuels and identify and quantify the applicable processes (e.g., mixing)
affecting the performance of UST LD systems on three scales: (1) bench-scale test set for the
determination of select physical and chemical properties of biofuels and BFW mixtures (no
technologies will be studied at this scale); (2) laboratory-scale test set for the identification and
quantification of BFW mixture processes (i.e., water ingress and mixing) affecting performance
of UST LD systems; and (3) full-scale field demonstration test set of UST LD systems as they
perform under real world conditions with ethanol-blended gasoline. Sample measurements will
follow standard analytical methods that have been published and accepted by ASTM
International, American National Standards Institute (ANSI), National Association of Corrosion
Engineers (NACE), or EPA.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 23 of 184
A9 SPECIAL TRAINING/CERTIFICATION
The staff who will be performing the laboratory and field activities will have documented
training pertinent to their activities. Prior to testing, each staff member will be required to
review the applicable ASTM methods and have experience or become adequately trained with
the equipment employed during testing. This training/experience will be documented in the
project records. Analytical laboratories will be required to provide documented support for their
proficiency in performing the required analyses in a thorough and safe manner with proper
attention to QC samples and waste disposal via standard operating procedures (SOP) or a
laboratory QA manual. An initial demonstration of capability will be provided with the results
of the PEA. Laboratory compliance with the measurement quality objectives (MQOs; Table 13)
will be demonstrated by the results of QC samples; data flags will be applied to any sample data
where QC failures occurred. If an amount of sample remains, the QC failures will be
investigated and remedied, then the samples with data flags will be reanalyzed. If sufficient
sample does not remain, the data will be flagged and discussed in the TA.
A10 DOCUMENTS AND RECORDS
Project staff both internal and external to Battelle will record all relevant and significant
aspects of this project in LRBs, electronic files (both raw data produced by applicable analytical
methods and spreadsheets containing various statistical calculations), audit reports, and other
project reports.
Table 3 includes the records that each organization will include in their project records to
be submitted to the TC. The TC or designee will review all of these records within two weeks of
receipt and maintain them in her office during the project. At the conclusion of the project, the
TC will transfer the records to permanent storage at Battelle's Records Management Office
(RMO). The Battelle QAM will maintain all quality records. All Battelle LRBs and reports are
stored for at least 20 years by Battelle's RMO; all raw data are stored for at least 10 years. The
TC will distribute the final QAPP and any revisions to the distribution list given in Section A4.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 24 of 184
Electronic records will be maintained on Battelle's internal ETV SharePoint site. Each
electronic file will be named uniquely such that the file content is clear. Section BIO further
details the data management practices and responsibilities.
All data generated during the course of this project must be able to withstand challenges
to their validity, accuracy, and legibility. To meet this objective, data will be recorded in
standardized formats (i.e., forms or spreadsheet templates) and in accordance with the
procedures defined below, which must be implemented for the documentation of all data
collection activities:
• Data must be documented directly, promptly, and legibly. All reported data must be
uniquely traceable to the raw data. All data reduction formulas must be documented
and sample calculations must be carried out and recorded so that the accuracy and
validity of any derived or calculated value is not in question.
• Handwritten data must be recorded in dark (blue or black) ink. All original data
records include, as appropriate, a description of the data collected, units of
measurement, unique sample identification (ID) and station or location ID (if
applicable), name (signature or initials) of the person collecting the data, and date of
data collection.
• Any changes to the original (raw data) entry must not obscure the original entry and
must be made with a single line cross out. The change must be initialed and dated by
the person making the change.
• The use of pencil, correction fluid, and erasable pen is prohibited.
• Data entered into spreadsheets will be traceable to the original records (e.g.,
laboratory notebook). Traceability may be established using unique sample ID
numbers or unique test numbers, distinctive treatment codes, etc.
• In the QAPP addendum, field sites, specific USTs, and specific technologies will be
referenced either by the full name or by unique abbreviations defined in the field
records and used consistently when data are transcribed from one location to another.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 25 of 184
Table 3. Project Records Submitted to the TC
Organization
Battelle
Subcontractors
(LIST service
company, if
necessary)
Analysis
laboratories
Records
LRBs, data recording forms, electronic
data compilations (i.e., Excel
spreadsheets)
Site protocol checklist, site protocol
data forms, sample chain of custody
forms, training documentation
LRBs, result raw data spreadsheets,
QC and calibration data, chain of
custody forms, training documentation
Submission Deadline
Within one week of completion of
generation of record
Scanned copy of documents e-mailed
to the TC within one week of
generation of record
Copies of all records e-mailed to the
TC within one week of analysis
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 26 of 184
SECTION B
DATA GENERATION AND ACQUISITION
Bl EXPERIMENTAL DESIGN
Testing will be conducted as three distinct sets of tests. Each test set is designed to
acquire specific data with respect to fuel properties, fuel mixing, or leak detection technology
performance. The three sets are:
1) Bench-scale studies for the determination of select physical and chemical properties
of biofuels and BFW mixtures (bench-scale testing);
2) Laboratory-scale studies for the identification and quantification of specific biofuel
and BFW mixture processes affecting performance of UST LD systems (laboratory-
scale testing); and
3) Full-scale field demonstrations of UST LD systems as they perform under real world
conditions with ethanol-blended gasoline (full-scale testing).
Each of these test sets aims at selecting and quantifying different properties (both
extensive and intensive) and behaviors of biofuels at different scales; however, the three tests
should not be seen as independent, as one of the major goals of this project is to integrate the
data collected at all scales into a coherent and defensible understanding of biofuels and how they
may affect UST LD system performance in the TA.
Bl.l Preparation of Test Blends
All test blends will be prepared in an identical manner. All petroleum products will be
sampled, mixed and handled according to ASTM standards D40577 and D58548; volumetric
blend stocks of ethanol (or isobutanol) and gasoline will be prepared according to ASTM
D77179. In addition to ethanol blends, an isobutanol blend containing 16% (v/v) isobutanol (116)
will also be included in the list of test blends. Test blends will be prepared by mixing different
concentrations of ethanol-free gasoline (EO) with either denatured ethyl alcohol (ethanol; >97%
purity) in the case of ethanol blends or isobutyl alcohol (isobutanol; >98% purity) in the case of
116. EO will be purchased from Marble Cliff Oil (Columbus, OH) and will be approved for sale
as automotive fuel. Information such as Material Safety Data Sheets and Bills of Lading will be
collected and recorded during fuel delivery. Proposed test blend compositions have been
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 27 of 184
selected based on those that are currently available on the market or are anticipated to be
available on the market. Test blends for the bench-scale test sets will include gasoline (EO) and
be prepared to simulate low ethanol blends (EO, E10, El5, and E30), flex fuels (E50, and E85)
and an isobutanol blend (116). Test blends for the laboratory-scale test sets will be prepared at
EO, E15, E85, and 116. Before preparation of the test blends, the water and ethanol content of
the EO gasoline will be determined by ASTM D203 and ASTM D4815, respectively. In the case
that ethanol or water is measured above the limit of detection of the appropriate method, EO will
be discarded and re-collected, or the initial water and/or ethanol content of EO will be accounted
for when formulating test blends and subsequent BFW mixtures. Table 4 indicates the mixing
ratios of EO and ethanol or isobutanol to achieve the desired test blend composition assuming EO
contains no ethanol or water. Test blends will be sampled and mixed in two 4 liter (L) batches
and used as soon as possible for the bench-scale experiments. Test blends which are not used
immediately will be capped and stored at room temperature for no more than 21 days before use.
Test blends for the laboratory-scale testing will be prepared in unknown volumes as it is
uncertain until the dry run tests how large of a volume will be able to be prepared safely and
reasonably for the testing.
Table 4. Mixing Ratios of EO and Ethanol/lsobutanol for Preparation of Test Blends
Test Blend
EO
E10
E15
E30
E50
E85
116
Volume Fraction
EO
1.0
0.90
0.85
0.70
0.50
0.15
0.84
Volume Fraction
Ethanol/
Isobutanol
0.0
0.10
0.15
0.30
0.50
0.85
0.16
B1.2 Preparation of BFW Mixtures
Test blends are intended to be representative ethanol- or isobutanol-containing gasoline
which are either currently on the market or anticipated to be on the market. On the other hand,
BFW mixtures simulate test blends that have been impacted by water — either through water
ingress to USTs or during manufacturing or transport. The BFW mixtures described here
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 28 of 184
contain much higher concentrations of water and are meant not only to include composition of
typical as-received gasoline but also simulate worst case scenario fuels which have been heavily
impacted by water from the environment. After preparation of test blends, BFW mixtures will be
prepared by adding known amounts of deionized (DI) water to appropriate volumes of test
blends. DI water was selected as the aqueous-phase source instead of natural or synthetic
groundwater because of its use in the literature for similar experiments10 and because a single
groundwater would not be representative of all groundwater that may be encountered in the field;
therefore, DI water is used as a baseline for the aqueous phase. BFW mixtures will contain
0.25%, 0.50%, 2.50% and 5.00% DI (v/v). Only experiments for interface determination will
use a 50% DI mixture. In addition, test blends with no added water will also be investigated
(i.e., 0% water). BFW mixtures will be prepared according to Table 5 in separate 500 mL Class
A glass volumetric flasks, closed with a ground glass stoppers and inverted a minimum of 15
times to completely mix the contents. Similar to test blends, BFW mixtures not immediately
used for testing will be stoppered and stored at room temperature for no more than 21 days
before use.
Table 5. Water Volumes Required to Prepare BFW Mixtures from Each Test Blend
Final Test Blend
Volume (mL)
500
500
500
500
500
Required Water
Content (%)
0
0.25
0.5
2.5
5
Required Water
Addition (mL)
0
1.25
2.50
12.50
25.00
A total of 35 different BFW mixtures will be prepared each in triplicate (three each of
seven test blends at five water concentrations). After preparation of the test blends, water and
ethanol or isobutanol content will be verified by ICFTL by either ASTM E20311 (for water) and
ASTM D550112 or ASTM D481513 (for ethanol and isobutanol), depending on their anticipated
water and ethanol contents (see Table 6). Some of the BFW mixtures will have a separated
phase once the test blend is saturated with water. In these cases, only the non-separated BFW
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 29 of 184
mixtures will be analyzed according to Table 6. In order to be considered acceptable for testing,
the RPD between the target and measured concentrations of water and ethanol content must be
Table 6. Acceptance Criteria for Test Blend Preparation
Measured
Parameters
Water content
Ethanol or
Isobutanol
content
Method of
Assessment
ASTM E20311
ASTM 481 513
(<20% ethanol)
and ASTM
D550112(>20%
ethanol)
Frequency
After preparation
of each test blend
and BFW mixture
if not phase
separated. After
collection of EO.
After preparation
of each test blend
and BFW mixture
if not phase
separated. After
collection of EO.
Acceptance
Criteria
RPD < 15%
between target
and measured
water
concentrations.
Water content of
EO non-detect.
RPD < 1 5%
between target
and measured
ethanol
concentrations.
Ethanol
concentration of
EO non-detect.
Corrective Action
Discard and re-prepare
Discard and re-prepare
B1.3 Bench-scale Testing
The bench-scale testing aims at determining several fundamental properties of biofuels
and BFW mixtures under typical conditions encountered during operation of UST LD systems.
This will differentiate whether the range of ethanol blends have properties that behave
significantly different from each other, thereby being the evidence that the technologies may or
may not function properly when used in the different blends. Then during subsequent
experiments in the laboratory and field, the type and number of ethanol blends are limited due to
waste generation and blend availability, respectively. Bench-scale testing is divided into four
series of tests:
a) Intrinsic Properties of BFW Mixtures: The properties studied in the first series of
bench-scale tests are common to all biofuels and will be referred to herein as intrinsic
properties because they belong to the biofuel due to its very nature. The intrinsic
properties evaluated in the first series of tests include pH, density, electrical
conductivity and viscosity. These are intensive intrinsic properties (i.e., do not
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 30 of 184
change with sample size) and were identified as important factors that may affect the
performance of UST LD systems while operating in BFW mixtures.
b) Coefficient of Thermal Expansion: The second series will determine the coefficient
of thermal expansion of different BFW mixtures within a temperature range that is
typically experienced in field applications of UST LD systems. The density of
biofuels, like all materials, is temperature dependent and the volume of a mass of
biofuel will change with temperature in a predictable (anticipated linear) fashion. In
the field, temperature fluctuations will cause expansion and contraction of BFW
mixtures which must be accurately predicted and accounted for by UST LD systems.
c) Non-additive Volume Changes: The third series of tests will determine the volume
effect of water addition on the test blends. When two polar solvents are combined (as
in water and ethanol in a biofuel) the resultant volume of the mixture is less than the
additive volume of the two components as water is accommodated into the ethanol
polar structure. This information is particularly applicable in the situation of water
ingress into USTs containing biofuels in that the ethanol in the gasoline will
accommodate the water in the gasoline and if the water is in high enough
concentration, phase separation will occur.
d) Interface Determination: The final series of bench-scale tests is focused on the
development of a method to optically determine the phase separation of the different
BFW mixtures. Once above the saturation level (<1% [v/v]), water separates from an
ethanol blend by pulling some of the ethanol into a denser separated phase at the
bottom of an UST. It is important that the location and properties of these layers be
able to be independently and objectively identified including not only pure water and
hydrocarbon phases, but also the colloidal mixed layers of gasoline/ethanol and
water/ethanol.
Each series of the bench-scale testing will be executed separately and sequentially in a
Battelle laboratory in Columbus, Ohio under ambient laboratory conditions unless otherwise
specified. Laboratory temperature will be measured with a glass thermometer at the beginning
and end of each testing day. For tests requiring strict temperature limits, a New Brunswick
Series 25 Incubator Shaker and a Lauda Proline Low Temperature Thermostat will be employed.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 31 of 184
Except when specific temperatures are required, all tests will be carried out at ambient laboratory
temperature (approximately 15 to 20 °C). Class A volumetric glassware and calibrated micro-
pipettes (within the last 6 months) will be used for all experiments and the accuracy of pipettes
will be determined gravimetrically at the beginning of each test day when anticipated to be used
that day. Glassware will be used as received, rinsed with EO and allowed to air dry overnight
before next use. All experiments will be carried out in triplicate to facilitate statistical
comparisons between BFW mixtures (see Section B 1.3.2).
Bl.3.1 Test Procedures
Bl. 3.1.1 Intrinsic Properties of BFW Mixtures
• Dependent Variables
o Acidity to nearest 1 mg/L as acetic acid
o Density to nearest degree API (American Petroleum Institute) and 0.0001 g/mL
o Dynamic viscosity to nearest 0.1 mm2/s
o Electrical conductivity to nearest 10 |iS/m
• Independent Variables
o Fuel concentration of ethanol to nearest 0.1% (v/v)
o Fuel concentration of water to nearest 0.1% (v/v)
This first test set aims at determining the pertinent intrinsic properties of BFW mixtures
at different ethanol or isobutanol and water contents. After preparation (Section B1.2), the BFW
mixtures will be poured into a 250 mL graduated cylinder and mixed using a magnetic stir bar.
During mixing, samples will be taken from the middle of the cylinder using a glass pipette and
sent to ICFTL for measurement of acidity by ASTM D161314, density by ASTM D405215,
viscosity by ASTM D44516, and water and ethanol content by either ASTM E20311 (for water)
and ASTM D550112 or ASTM D481513 (for ethanol) depending on their anticipated water and
ethanol contents. Where appropriate, samples will be analyzed for isobutanol concentration by
ASTM D481513. After sampling, conductivity will be measured by ASTM D262417 and density
will be measured by ASTM D28718 directly in the graduated cylinder. Each intrinsic property
will be measured in triplicate on the same sample.
Some of the BFW mixtures will have separated phases. In this case, the interest in
intrinsic properties is in the bulk fuel phase and as such, aliquots sent for analytical analysis will
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 32 of 184
be the bulk fuel samples. Where possible, the dense phase (i.e., water-ethanol separated phase)
will be archived should the analysis of this phase be performed. At this time, it has been
determined to only analyze the fuel phase because of the relevance to technology performance
for LD, the potential non-availability of enough volume for the analyses, and to minimize
extraneous analytical costs.
B 1.3. 1.2 Coefficient of Thermal Expansion
• Dependent Variable
o Volume change after temperature equilibration to nearest 10 jiL
• Independent Variable
o Water bath temperature to the nearest 0. 1 °C
In order to determine how temperature will affect the volume of specific BFW mixtures,
a series of experiments will be conducted in 10 mL-capacity glass graduated cylinders
(±0.1 mL). 5 mL of each of the 35 different BFW mixtures will be measured by pipette to
individual graduated cylinders at ambient temperature and capped with a ground-glass stopper.
Actual mass of BFW mixture will be determined gravimetrically. The BFW mixtures will then
be allowed to equilibrate for 60 minutes to 5.0°C, 10.0°C, 15.0°C, 20.0°C, 25.0°C and 30.0°C in a
Lauda Proline Low Temperature Thermostat. After each 60-minute equilibration time, the
volume of the graduated cylinder will be recorded before it is returned to the thermostat.
The coefficient of thermal expansion will be calculated using Equation 1:
Equation 1
where a is the coefficient of thermal expansion, V25 is the volume of the individual BFW mixture
at 25°C (normalization temperature) and (3V/3T) is the partial derivative (i.e., slope) of the
volume vs. temperature line as calculated by linear regression (see below).
B 1.3. 1.3 Non-additive Volume Changes
• Dependent Variable
o Total volume change after dye solution (water) addition to nearest 10 jiL
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 33 of 184
o Volume change of dense phase after dye solution addition to the nearest 10 jiL
• Independent Variable
o Volume of dye solution added to test blend to nearest 250 jiL
o Mass of dye solution added to nearest 10 mg
Because of the varying miscibility of gasoline, water and ethanol, it is expected that as an
aliquot of water is added to each of the test blends, the total volume change of the resulting BFW
mixture will be less than the volume of that aliquot, and the separated, dense phase will grow
disproportionately to the added volume of water. The relative total volume decrease is due to
accommodation of polar water molecules into the structure formed by the polar ethanol
molecules (degree of accommodation).
This experiment aims at quantifying this effect. 5 mL of each test blend (no water) will
be added separately by pipette to 10 mL (±0.1 mL) glass-graduated cylinders; the actual mass of
the test blend will be determined gravimetrically. The graduated cylinders will be placed in the
thermostat at 25°C for 15 minutes for initial temperature equilibration. After equilibration, the
cylinders will be removed from the thermostat and a dye solution consisting of water and
McCormick Blue Food Dye (1:2,000 dilution) will be added in 250 jiL increments using a
micro-pipette. The actual mass of added dye solution will be determined gravimetrically. After
the addition of each 250 jiL increment of water, the graduated cylinder will be sealed with a
ground glass stopper and mixed using a Baxter Scientific S/P Vortex Mixer. Intensity of mixing
has been determined to be large enough to ensure complete mixing of hydrocarbon, water and
ethanol phases but appropriate to reduce volatilization during mixing. The graduated cylinder
will be replaced to the thermostat for 5 minutes at 25°C, after which the total volume and the
volume of the dense phase will be measured. At the time of volume measurement, a photograph
of the cylinder will be taken to qualitatively record the interface. A total of 5 mL of dye solution
will be added in this way to each sample (total of twenty 250 jiL additions) with measurement of
volume change made after each increment.
The effect of fuel:ethanol ratio on relative volume decrease will be determined by
calculating the following using Equation 2:
= 9Vrn
Y dVa
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 34 of 184
Equation 2
The parameter y will be referred to as the degree of accommodation, AVm is the measured
incremental change in total volume with incremental dye solution addition and AVa is the
incremental volume addition of dye solution. In this way, y can be seen as the measure of the
amount of ethanol accommodated within the polar water structure which results in relative
volume reduction with the addition of water to the test blends. In practice, y will be defined as
the slope of the Vm vs. Va curve as calculated by linear regression (see below).
Bl. 3.1.4 Interface Determination
• Dependent Variables
o Absorption of light at 630 nm to the nearest 0.001 absorption units.
o Depth of sample to the nearest 0.1 cm
• Independent Variables
o Fuel concentration of ethanol to nearest 0.1% (v/v)
As water separates from pure gasoline, a well-defined interface is formed which can be
visually determined relatively easily and objectively; however, the interface becomes less
defined when water separates from an ethanol-gasoline mixture as the water can be absorbed by
both the gasoline and ethanol phases forming a hazy suspension. Gaining an understanding of the
separated phase in different ethanol blends is important for identifying and measuring water at
the bottom of an UST. This last series of bench-scale tests aims at establishing a method for
determination of a water interface in different test blends and mathematically defining the
vertical position of the interface.
A sample of 70 mL of each test blend and 70 mL of dye solution consisting of water and
McCormick Blue Food Dye (1:2,000 dilution) will be measured by glass volumetric pipette into
three individual 160 mL glass serum bottles (triplicate samples of each test blend/dye solution
mixture). Serum bottles will be sealed with Teflon® septa and aluminum caps. The 160 mL
serum bottles will be agitated with a New Brunswick Series 25 Incubator Shaker at 300 rotations
per minute for 60 minutes to ensure mixing. After the mixing period, the septa will be pierced
with a thin needle protruding to the bottom of each of the serum bottles. The needles will be
equipped with a Luer-Lok fitting able to be attached to a 10 mL syringe. The serum bottles will
be left to rest in the incubator at 25 °C for 24 h to reach equilibrium. After equilibration, each
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 35 of 184
serum bottle septum will be pierced with a second needle only to the headspace to allow 10 mL
of sample to be carefully extracted through the first needle using a 10 mL syringe. 10 mL
corresponds to approximately 1 cm liquid height which will be subsequently measured to the
nearest 0.1 cm. The absorbance of the 10 mL sample will then be measured at 630 nm using a
Hach DR5000 UV-Vis Spectrophotometer previously zeroed with EO. Following ASTM
D745119 for mixing and measurement, the cells will be briefly and vigorously shaken to ensure
homogeneity immediately before absorbance measurements are taken. Triplicate measurements
will be taken and to be considered acceptable, measurements must display a coefficient of
variation of less than 10%.
This extraction and measurement procedure will be repeated until the full contents of
each serum bottle have been removed (approximately 14 data points per serum bottle). In this
way, the transition from water to gasoline can be plotted using visible absorbance of the dye
solution as a designation of where the water is located in the sample. Each test blend will
undergo the same procedure.
Table 7 summarizes the series of tests to be performed on the bench scale. Table 8 presents
the data collection QC assessments for the fuel properties being measured in the bench-scale
testing. A table similar to Table 8 will be included in the QAPP addenda to cover QC related to
the specific technologies and their use in the laboratory- or full-scale test sets.
Table 7. Summary of the Bench-scale Test Set
Test Series
Intrinsic
Properties of
BFW Mixtures
Coefficient of
Thermal
Expansion
Description
Preparation of 35 different test
blends and BFW mixtures and
analysis of their intrinsic
properties including ethanol
concentration, water
concentration, acidity, density,
viscosity, and electrical
conductivity
Preparation of 35 different test
blends and BFW mixtures and
measurement of their volume
at different temperatures from
5.0 to 30.0 °C
Precision
Requirements
• CV<15%for
measurements on
triplicate samples
• r2 > 0.90 for volume
vs. temperature
curve
measurements on
triplicate samples
Independent
Variables
• Water
concentration
• Ethanol
concentration
• EO
concentration
• Water
concentration
• Ethanol
concentration
• EO
concentration
• Temperature
#of
Replicates
3 each
3 each
-------
Biofuel Properties and Behavior Relevant to UST LD System Performance
Date: 4/17/2013
Version 1.0
Page 36 of 184
Table 7. Summary of Bench-scale Test Set (Continued)
Test Series
Non-Additive
Volume
Changes
Determination
of Interface
Description
Preparation of eight test blends
and measurement of volume
changes with known addition of
aqueous dye solution
Mixing 50% of the eight test
blends individually with 50%
aqueous dye solution and
measuring the height-
dependent absorbance of the
resulting mixture resulting in a
height vs. absorbance curve
which can be used as a
designation of water location
Precision
Requirements
• r2 > 0.90 for volume
measured vs.
volume added
curve
• CV<15%for
single
measurements on
triplicate samples
• CV<10%for
triplicate
measurements of
optical absorbance
on the same
sample
• CV < 25% for
single, depth-
dependent
measurements on
triplicate samples
of optical
absorbance
Independent
Variables
• Ethanol
concentration
• EO
concentration
• Dye solution
added
• Fthanol
^11 ICII I\JI
concentration
• EO
concentration
#of
Replicates
3 each
3 each
Table 8. Data Collection Quality Control Assessments of the Fuel Properties
Measured
Fuel
Property
Ethanol
Concentration
Water
Concentration
Acidity
Density
Viscosity
Electrical
Conductivity
Method of
Assessment
ASTM D5501 and
D4815
ASTM E203
ASTMD161314
ASTM D28718
ASTM D4052
ASTM D44516
EMCEE Model
11 52; ASTM
D262417
Frequency
Once per
unique BFW
mixture, once
per unique test
blend and once
per collection
ofEO
Once per
unique BFW
mixture during
determination
of intrinsic
properties
Laboratory
ICTFL
ICTFL
ICFTL
Battelle
ICFTL
ICFTL
Battelle
Acceptance
Criteria
RPD< 15%
between result
and target.
Non-Detect for
EO
RPD< 15%
between result
and target.
Non-Detect for
EO
CV<15%for
triplicate
measurements
Corrective Action
Discard test blend or
BFW mixture and re-
prepare
Discard test blend or
BFW mixture and re-
prepare
First unacceptable
result: Re-test BFW
mixture. Second
unacceptable result:
Discard and re-
prepare BFW mixture
and retest. Third
unacceptable result:
trouble shoot the
instrumentation*3'
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 37 of 184
Table 8. Data Collection Quality Control Assessments of the Fuel Properties (Continued)
Measured
Fuel
Property
Absorbance
Temperature
(incubator)
Temperature
(water bath)
Method of
Assessment
Hach DR5000 UV-
Vis
Spectrophotometer
Glass thermometer
Built-in resistance
probe
Frequency
Zero
instrument
between test
blend
replicates
Once each at
the beginning
and end of
each testing
day and once
during testing
Immediately
after
temperature
equilibration
and every 30
minutes after
equilibration
Laboratory
Battelle
Battelle
Battelle
Acceptance
Criteria
CV<10%for
triplicate
measurements
±1 °C from
target,
monitored with
an audible
alarm when
out of range
±0.1°Cfrom
target,
monitored and
logged with a
calibrated
electronic
thermometer
Corrective Action
First unacceptable
result: Re-test
samples. Second
unacceptable result:
trouble shoot the
instrumentation
Replace thermometer
First unacceptable
result: trouble shoot
the instrumentation.
Second unacceptable
result: record
temperature using
external thermometer
(a) Note that BFW mixtures that do not meet acceptance criteria for one measured parameter may be tested for
other measured parameters.
Bl.3.2 Statistics for Bench-scale Test Sets
All BFW mixtures will be prepared in triplicate and measurements made on each of the
triplicate BFW mixtures will be carried out once. Statistics will be calculated on each of the
measurements as follows:
• Average: The average value ( * ) of the single measurements made on the triplicate
BFW mixtures will be calculated using Equation 3 as follows:
Equation 3
where * is the average value of n number of measurements, x; (i = 1,2,3)
• Standard Deviation: The standard deviation (SD) of a set of triplicate measurements
made on BFW mixtures will be calculated using Equation 4 as follows:
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 38 of 184
SD =
3.^—r -
Equation 4
where * and X; are defined above.
• Coefficient of Variation: The CV of a set of measurements is defined as the quotient
of the SD of that set of measurements and the average of that same set of
measurements and will be calculated using Equation 5 as follows:
SD
CV = —
x
Equation 5
where CV is the coefficient of variation and SD and * are defined above.
• Relative Percent Difference: The RPD between a measured (or calculated) value and
a target value will be calculated using Equation 6 as follows:
RPD =
x _ 71
T
Equation 6
where RPD is the relative percent difference between a calculated mean, * ancj a target
value, T.
• Coefficient of Determination: The coefficient of determination (r2) of several
calculated dependent variables with respect to their associated independent variables
will be calculated according to Principles and Procedures of Statistics20 and the
formulae will not be repeated here. In all cases, r2 will be calculated based on
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 39 of 184
calculated average values of both measured dependent and independent variables by
Microsoft® Excel.
Bl.3.3 Precision
Intensive Properties: Acidity (pH), Viscosity, Density, Electrical Conductivity, and Optical
Absorbance
Measured triplicate values of acidity (i.e., pH), density, viscosity and electrical
conductivity measured as part of the intrinsic properties of BFW mixtures experiments will be
subjected to statistical analysis. The average value, SD and CV will be calculated and recorded
separately for each set of measured intrinsic properties. Calculated average values will be
compared to applicable literature values and discussed in the TA; however, no specific value will
be taken as the accepted value, thus no RPD calculations will be made. With respect to
precision, for single measurements taken on triplicate samples to be considered acceptable for
reporting, the CV for each set of triplicate measurements of pH, density, viscosity, electrical
conductivity and optical absorbance must be less than 15%.
The single depth-dependent optical absorbance measurements of samples collected
during the interface determination experiments will be considered acceptable for reporting when
triplicate measurements on one test blend in three separate serum bottles display a CV less than
15%. No accuracy criterion is established for depth-dependent measurements taken during the
interface determination experiment as this experiment aims at determining properties heretofore
undefined.
Extensive Properties: Volume Change
Single volume measurements taken on triplicate samples for the non-additive volume and
coefficient of thermal expansion experiments will be subjected to statistical analysis. The
average value, SD and CV will be calculated and recorded separately for each triplicate
measurement of volume change. Calculated average values will be compared to applicable
literature values and discussed in the TA; however, no specific value will be taken as the
accepted value, thus no RPD calculations will be made. With respect to precision, for single
measurements taken on triplicate samples to be considered acceptable for reporting, CV for each
set of triplicate measurements of volume must be less than 15%.
Calculated Properties: Coefficient of Thermal Expansion and Degree of Accommodation
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 40 of 184
The coefficient of thermal expansion (Equation 1) and degree of accommodation
(Equation 2) will be calculated from the appropriate equations and results reported with
appropriate significant figures. Calculated values will be compared to applicable literature
values and discussed in the TA; however, no specific value will be taken as the accepted value,
thus no RPD calculations will be made. In contrast, within the experimental parameters set forth,
the slopes of volume vs. temperature curve (for coefficient of thermal expansion) and measured
volume vs. added volume curve (for degree of accommodation) are expected to be linear.
Therefore, in order to be considered acceptable, the coefficient of determination calculated from
the average values (i.e., volume and temperature) must be greater than 0.90.
B1.4 Laboratory-scale Testing
The purpose of the lab oratory-scale test set is to evaluate the water ingress and the
potential effect ingress method/rate has on the detection abilities of various LD technologies in
biofuels. A similar approach was used on a large scale during the ETV testing of ATGs5 and is
being scaled down for performance comparison. Mixing conditions in laboratory studies will be
recorded by calculating total energy imparted on the laboratory reactor due to fuel and water
additions under the conditions tested. Based on comparison of data collected during ETV
testing, fundamental scaling or energy-balance arguments will be used to modify results of
laboratory-scale testing for comparison of data sets. This will elucidate the applicability of
laboratory-scale results to inform operation and predict performance of full-scale UST LD
systems. Laboratory tests will be performed in a glass laboratory test column that is
approximately 8 inches in diameter and 5 feet in height. This column has a 13-gallon capacity
and will be filled to 50% full for all runs during this set of tests. Any column adaptations and
procedures on where and how the bottom of the tank will be simulated will be determined during
preliminary experiments. This set of testing on the laboratory scale contains two series as
follows:
a) Initial water ingress detection of continuous water ingress with a splash or without a
splash and the smallest increment of water detection (continuous ingress) and
b) Water ingress detection of quick water dump, then a fuel dump (quick dump).
Each series of the laboratory-scale testing will be executed separately on each technology
under standard laboratory conditions of temperature and pressure unless otherwise specified.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 41 of 184
Technologies used in this testing will include one ATG and up to two other technologies
belonging to different LD technology categories. It is anticipated that one or both of the other
technologies will be a sensor. A sensor technology would operate similarly to the ATG in that it
would continuously be monitoring the testing condition and collecting data and it would report
the water either in height or concentration. The reaction of when the technology detects the
water and how the increments are measured will be captured in the electronic download of the
sensor. All experiments will be carried out in triplicate to facilitate statistical comparisons
between treatments (see Section Bl.3.2). The independent variables included in this set of tests
will be the test blends (three levels of ethanol content, EO, El 5, and E85, and 116) and water
ingress methods (with splash, without splash, and dump). The preparation of the test blends will
follow the procedure established in Section B1.1. Prior to testing, the percent ethanol or
isobutanol will be verified analytically using the appropriate ASTM methods (Table 6). While
water for the bench-scale test sets will be DI water for control, water for the laboratory-scale
testing will be groundwater taken from the Battelle groundwater tap for closer simulation of
operation for the technologies. At the conclusion of the test runs, the test blends and the
separated phases will be analyzed for water content and density using the appropriate ASTM
standards. Details of chosen technologies for this testing will be prepared as an addendum to this
QAPP per the AMS QMP2 and approved by the EPA PO, or his designee before testing begins.
Bl.4.1 Test Procedures
Bl. 4.1.1 Continuous Water Ingress Series
• Dependent Variables
o Detection of the water ingress
• Independent Variables
o Ethanol or isobutanol content to nearest 1% (v/v) of EO, El 5, E85, and 116
o Water mixing method (with and without a splash)
The continuous ingress series are focused on the mixing method of water addition into
the test vessel. In the first test, a continuous stream of water will be introduced into the
laboratory test column to produce a splash on the surface of the fuel or to not produce a splash by
trickling the water along a surface of the test column to slowly meet the surface of the fuel.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 42 of 184
The amount of water, introduced via either method, will be a fixed water ingress rate that
will be specifically determined during the preliminary dry run experiments. Although the rate
used may be larger than an expected ingress rate in the field, the time it takes to detect the water
is wait time to collect to the technology's threshold detection height. In the interest of
conducting testing in a reasonable amount of time and for safety purposes, the rate will be set to
establish a response from the technologies within 1 hour for these experiments. The rate will
also be presented in the TA and converted to quantified ingress rate during reporting. When the
technology detects the water, the water height will be measured using a ruler installed into or
onto the test column. Following the initial experiments under ETV, the use of visual height
measurements will introduce error that will be mitigated by installing a stationary ruler in the test
column and having the same staff take all the measurements.
• A continuous water ingress that causes a splash on the surface of the fuel. The rate
will be established such that the vendor-stated threshold height of water that can be
detected (absent any adsorption) will be produced within approximately 1 hour. This
water addition rate will be continued beyond 1 hour until a response in the water
detection technology is observed. If no response is observed in 2 hours, the test will
be terminated. With this method of water ingress, some mixing may occur due to
splash mixing and some mixing may occur by diffusion. The extent of mixing by
these two mechanisms may be influenced by independent variables and may cause
adsorption of water into the ethanol along with subsequent phase separation of the
mixture.
• A continuous water ingress that follows along the inside wall of the test column with
minimal agitation to the surface of the fuel. The rate will be established using the
same procedure as above. The test condition will be maintained until a response in
the water detection technology is observed, or terminated after 2 hours if there is no
response. With this method of water ingress, most of the mixing is expected to occur
by diffusion. The run termination times are established to be the same because it is
expected that this time interval encompasses the potential for the technology to detect
the water with both ingress rates.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 43 of 184
To address the second requirement of water detection, once the water detection
technology has reacted to the minimum water height, the smallest increment in water height that
can be measured will be determined. The ingress rate will be adjusted to produce a calculated
height increase at the bottom of the column of 1/16th of an inch in 5 minutes. After 5 minutes the
technology reading and the height of the water level will be measured and recorded. Ten 5-
minute increments will be measured for each of the eight unique run conditions of the continuous
ingress test series (to produce approximately 80 measurements). This same flow rate will be
used for all runs regardless of the initial flow rates of with or without splashing. The true
increase of the water level will be measured using a stationary ruler and recorded. Table 9
presents the 24 runs to be conducted under this testing series.
Table 9. Summary of Continuous Water Ingress Runs
Runs
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Fuel Type
EO
E15
E85
116
Water Ingress Method
With splash
Without splash
With splash
Without splash
With splash
Without splash
With splash
Without splash
B'1.4.1.2 Quick Dump Series
• Dependent Variables
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 44 of 184
o Detection of the water ingress
• Independent Variables
o Ethanol or isobutanol concentration to nearest 1% (v/v) of EO, E15, E85, and 116
o Water ingress method (with and without a splash)
The quick dump series of tests focuses on the potential to detect phase separation in an
UST. A water ingress method with a high degree of mixing will simulate addition of water in a
manner that might occur if a spill bucket is dumped into a tank, followed by a fuel delivery. The
test column will be filled at 25% fill height, then water will be dumped into the column, after
which the column will be filled to the 50% fill height with fuel. The amount of water quickly
dumped into the test column will be determined during preliminary dry run experiments. This
test is mainly observational in that the test column will be disturbed quickly with water then fuel
and the response of the technology will be recorded throughout the test. There will be 12 runs
with four test blends evaluated in triplicate. The EO run will be run first and used as the baseline
for the technology's response to establish the minimum wait time for the other test runs. If the
technology being tested is not recommended with E85 or if it is not designed to detect water that
is not at the bottom of a UST, E85 will not be used in this test series.
Table 10 presents a summary of the designs for the lab oratory-scale test series. The
associated performance parameters for each test are provided as well as the variables and number
of runs. Preliminary dry runs will be performed to establish the laboratory procedures to conduct
testing in an efficient and safe way. These will include, for example, establishing a procedure
for water introduction techniques and/or mixing methods, maintaining and monitoring
temperature, establishing a fuel blending and transferring procedure, and discerning the best
vantage point to video record the tests.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 45 of 184
Table 10. Summary of Laboratory-Scale Test Set
Test Series
Water
Ingress Test
Quick Dump
Description
Minimum detection height:
Water ingress detection of
continuous water ingress
with a splash or without a
splash to determine the
minimum water level that
the technology can detect
Smallest detection
increment: Water ingress
detection of continuous
water ingress with a splash
or without a splash to
determine the smallest
change in water level that
the technology can detect
Water ingress detection of
a quick water dump, then a
fuel dump to induce and
observe phase separation
Performance
Parameter
• Accuracy
• Sensitivity
• Precision
• Sensitivity
• Observation
Independent
Variables
• \A/fltpr
ingress
method/rate
• Fuel type
• Water
ingress
method/rate
• Fuel type
• Water dump
• Fuel dump
• Fi IP! tvnp
# of Runs
24 runs (8 run
conditions in
triplicate)
Continuation of runs
in Test 1a while
collecting 10
incremented
measurements
during one replicate
condition (80
measurements)
12 runs (4 test
blends in triplicate)
Bl.4.2 Statistics for Laboratory-scale Test Sets
All eight run conditions will be performed in triplicate. Basic statistics will be calculated
on each of the measurements following Equations 3, 4, and 5.
• The minimum height of water that the technology reliably detects will be assessed
using the methodology from the 1990 EPA ATG protocol6, with some updates to
account for different variables and subsequently the different number of test runs.
The bias, variance and SD (the square root of the estimated variance) of the results
will be reported along with a tolerance limit (TL) of water that is 95% likely to cause
the technology to detect water.
• The minimum increase in water that can be detected will be assessed using the
methodology from the 1990 EPA ATG protocol6 where the minimum water level
change (MLC) will be reported as with the increment of water that is 95% likely to
cause the technology to report a water depth estimate.
Given these calculations for water detection, the following performance parameters will
be evaluated.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 46 of 184
• Accuracy, expressed in terms of whether the depth at which water is detected is less
than or equal to the height stated by the vendor. (This analysis assumes that the depth
stated by the vendor is claimed to be a height at which their technology would detect
water at least 95% of the time.) Also whether the estimated minimum increase that
can be detected is less than or equal to the detectable increase stated by the vendor or
to the nearest l/8th of an inch (whichever is smaller).
• Sensitivity, expressed as the minimum value for water height at which the probability
is at least 95% that the water detection technology detects the presence of water in the
bottom of the tank.
• Precision, calculated as the ratio of the mean technology-measured water height or
leak rate at the specified end point of a test to the SD of that same quantity.
B1.4.2.1 Accuracy
If the estimated minimum amount of additional water that is detected in an increase is
less than or equal to the amount specified by the vendor, then the vendor-stated smallest change
in the water level that the technology can detect will be reported. The bias will be calculated as
below in Equation 7 as an estimate of accuracy.
V ' i •-<;•
Bias = > JsJi
^—1^1 n
Equation 7
where n is the number of runs, L is the technology measured increase in water height, and
S is the independently measured increase in water height.
B 1.4.2.2 Sensitivity
Sensitivity is a measure of the extent to which the methods and instrumentation
associated with a given technology are actually able to detect the event of interest when in fact
the event has occurred. A technology is determined to have higher sensitivity as the event
becomes more difficult to detect with a certain degree of probability. Sensitivity is quantified by
the minimum value for water depth at which the probability is at least 0.95 (95%) that the water
detection technology will detect the presence of water in the bottom of the tank given the true
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 47 of 184
water depth (tolerance limit). In addition, sensitivity is quantified by the smallest detectable
change in the water height once water is detected with at least a 95% probability of detecting the
change (minimum water level change).
If the estimated minimum height that would be detected at least 95% of the time is less
than or equal to the height specified by the vendor, then the vendor-stated height will be reported
as the minimum height for the technology to detect water ingress. The TL will be used for this
comparison. To calculate the TL follow the below calculations.
1. Calculate the mean ( *) of the observations as in Equation 3.
2. Calculate the SD of the observations as in Equation 4.
3. Find k from a table of tolerance coefficient for one-sided normal tolerance interval
with a 95% probability level and a 95% coverage for the number of observations.21
4. Calculate the TL as in Equation 8.
TL = ( x ) + k SD Equation 8
where * is the mean of the observations, k is the tolerance coefficient, and SD is
the standard deviation of the observations.
The estimated minimum height that would be detected at least 95% of the minimum
detectable change of the water height, the MLC will be calculated by following the steps below.
1. Calculate the difference (d) between the technology observation and the
independently-measured water increment heights for all observations as in Equation
9, noting the group of observations from each run during the continuous ingress test.
d-ir = wtr ~ wmr Equation 9
where wtr is the technology measured increment of the rth run and wmr is the
independently measured water increment of the rth run.
2. Calculate the average of the differences (D) for each group of observations from the
Test 1 runs as in Equation 10, where nr is the total number of runs
Dr = ZF=i— Equation 10
nr
3. Calculate the variance (Varr) of the differences separately for each group of
observations from the Test 1 runs as in Equation 11.
Varr = £?=1 (di"^)2 Equation 11
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 48 of 184
4. Calculate the pooled variance (Varp) of the groups as in Equation 12.
ir (nrl-l)l/arrl+ ...+(nr#-l')Varr# _ . n^.
varp = -5— Equation 12
The nr# designates the subsequent individual run data.
5. Calculate the pooled standard deviation (SDP) as in Equation 13.
Equation 13
6. Find the tolerance coefficient (k), for two-sided tolerance intervals with 95%
probability and 95% coverage from a tolerance factor table20.
7. Calculate the MLC that the technology can detect using Equation 14.
MLC = k SDp Equation 14
where k is the tolerance coefficient and SDP is the pooled standard deviation of the
observations.
Bl.4.2.3 Precision
Precision is a measure of the extent to which the methods and instrumentation associated
with a given technology yield results that are reproducible. For a given set of test conditions,
precision is characterized by the ratio of the ^ of a technology-measured value to its SD.
Precision corresponds to the ratio of the * associated with the technology-measured water
height at the specified end point of a test to the SD of water heights measured at that same point
in the test.
B1.5 Full-scale Testing
The purpose of the full-scale testing is to evaluate LD data collected under real world
conditions with ethanol-blended gasoline as a field demonstration. A similar approach is
presented as the alternative approach in the 1990 EPA ATG protocol6. Once the bench- and
laboratory-scale testing is complete, the data will be reviewed and the need for the execution of
the full-scale testing will be evaluated by UST LD stakeholders. Technologies used in this
testing will include one ATG and up to two other technologies of different LD technology
categories (see Table 1). The field sites chosen will already have a LD system installed and in
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 49 of 184
use. Once field sites are chosen, the vendor of the LD system used on site will be contacted. The
vendor will be asked to check that the system is operating properly and is properly installed. If
the vendor is not available or willing to confirm the technology's status, a UST service company
will be contacted to perform this technology check.
In addition to the technology being set up correctly, the UST also needs to be tight. A
service company will be contacted to perform a tank tightness test using a different method than
the one being tested during this project. To the extent possible, USTs with groundwater below
the bottom of the tank will be used for testing.
The full-scale demonstrations involve LD capabilities of the technologies only. They
will be conducted in the field in an UST at a service or blending station and may be conducted
upon review of the data from the bench- and lab oratory-scale test sets with the UST LD
stakeholders and EPA PO. Details determined about the number and types of technologies tested
at what locations and by what criteria will be documented as a QAPP addendum per the AMS
QMP2 and approved by the EPA PO, or his designee before testing begins.
Bl.5.1 Test Procedures
The field demonstration will have two components of gathering data under normal
operating conditions without a leak and gathering data when leaks are simulated. These
conditions are described below and summarized in Table 11.
• Non leak: For the test for false positives, technology data will be collected under non-
leak conditions. Since the technologies operate automatically, they can be
programmed to perform a test whenever the tank is out of service for long enough
periods, typically each night. This approach will provide test data under a variety of
actual operating conditions, including a wide variety of temperature conditions,
product levels in the tank as well as wait times after the tank receives a fuel delivery.
The number of runs necessary is based on the confidence bounds for the estimated
proportion of false alarms. If 59 runs when the tests in a tight tank produce 59 passes,
then the estimated false alarm rate is 0% and the exact 95% upper confidence bound
for the rate is 4.95%, so it is reasonable to conclude that the false alarm rate is below
5%. If 93 runs when the true leak rate is 0 produce one false leak detection, then the
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 50 of 184
upper confidence bound is 4.99%. Any fewer than 59 runs will not yield a strong
conclusion that the false alarm rate is 5% or lower using logic associated with a
binomial confidence interval.
• Simulated leak: Next, the database of technology test results on tight tanks will be
supplemented with a limited number of tests using a simulated leak. This is to
determine that the system can identify a leak and if the technology is quantitative, can
adequately calculate the leak rate per the EPA regulation. The combined data sets will
then be analyzed to estimate the performance of the technology.
This field demonstration will produce a large number of tests under tight conditions, and
relatively few tests under simulated leak rate conditions. A suggested sample size is more than
59 tight tank tests and 10 simulated leak rate tests for each LD technology separately; however,
the sample size will depend on the technologies, the testing schedule, and the site constraints. It
might also be necessary to exclude some results from the analysis, 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 additional detail of the full-scale testing.
1. Once a site has been identified, work with the vendor and/or a service company to
ensure the technology is installed and operating correctly as well as verifying that the
UST is tight.
2. Arrange to collect and record ancillary data to document the test conditions. The data
needed are:
• Average in-tank product temperature prior to a delivery;
• Time and date of each delivery;
• Average in-tank product temperature immediately after a delivery;
• Amount of product added at each delivery;
• Date, time, and results of each test;
• Product level when the test is run; and
• Tank size, type of tank, product contained, etc.
3. Conduct tests at the site for at least a 2-week period. For these tests, the technology
will be set up to automatically conduct tank tightness tests as frequently as practical
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 51 of 184
with the USTs use. The data will be collected in the technology's console then
downloaded on a weekly basis. If a remote internet connection is established with the
console, then data will be retrieved three times a week. Report the starting and
ending dates of the test period. Record the test results along with the data listed in
Step 2. The data above define the conditions 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. All test results
will be presented in the TA appendix. Results that need to be discarded because of
product delivery or dispensing will be identified as such.
4. Conduct test runs in triplicate with a simulated leak at the rates of 0.10, 0.20, and 0.30
gal/hr. These induced leak tests will require technical staff on site to monitor the
simulated leak rates and measure the rates actually achieved. For these tests, the
technology will be operating under leak test mode while the tank is undisturbed with
dispensing or accepting fuel. The simulated leak will be established by inserting
tubing into the fuel through an open riser pipe. The tubing will be used with a
peristaltic pump equipped with an explosion proof motor set at one of the three leak
rates. Each simulated leak test will be performed in triplicate and the fuel from the
simulated leak will be collected and returned to the UST once the testing for the day
is complete.
Bl.5.2 Statistics for Full-scale Test Set
Using the resulting data, analyze the differences between the leak rate measured by the
technology and the simulated leak rate achieved (zero for the many tests on tight tanks) for each
test to estimate the performance. Given the unknown technology type and data set size, the
statistical analysis approach may need to be modified. Any deviations from this approach will be
documented in the QAPP addendum.
The database will be used to investigate the relationship of the error size (the leak rate
differences) to each of the variables monitored for the tests. These include tank size, length of
stabilization time, temperature differential, product level, and detection of induced leaks.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 52 of 184
Table 11. Summary of Full-scale Test Sets
Test Series
Non-leak
Simulated
Leak
Description
Data collection under real-
world conditions under non-
leak conditions
Field demonstration while
simulating leaks at 0.1 , 0.2,
and 0.3 gal/hr
Performance
Parameter
• Probability
of false
alarm (PFA)
• Probability
of Leak
Detection
(PD)
Independent
Variables
• Temperature
conditions
• Product levels in the
tank
• Times after the tank
receives a product
delivery
• Leak rate
• Temperature
conditions
• Product levels in the
tank
# of Runs
59 without
false alarm, 93
with 1 false
alarm
10
Multiple regression techniques will be used for these analyses to determine the significance of
their effect on the error size. Because it is not possible to control the variables in the field
testing, it may not be possible to quantify the effects of these variables. Visual inspection of the
residuals and a test for consistency of the error variances will be used to assess the difference in
error variance between the results from the group with simulated leaks and the group without
leaks. It is expected that the simulated leak test results will have more variance than the non-leak
test results.
The evaluation of the technologies in LD 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 PFA and the PD are estimated using the difference between these
two numbers. In addition, maximum allowable temperature difference, average waiting time
after filling, and average data collection time per test will be calculated to inform the TA.
Probability of False Alarm and the Probability to Detect
The PFA and PD will be calculated as follows:
1. Using the leak rate reported by the technology and the actual leak rate (zero for tight
tank tests, measured for the induced leak rate tests), the differences between the
measured and actual leak rates will be calculated (similar to Equation 9).
2. Then the * (Equation 3) and SD (Equation 4) of these differences will be calculated.
3. Perform a t-test for significant bias.
4. Estimate the PFA and the PD as described below.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 53 of 184
Calculate the variances of the differences separately for the data from the tests under tight
conditions and those from the tests with simulated leak rates. This is done considering that there
are two groups defined by the leak status of the tanks and the sample sizes, although sample sizes
are not equal. Let the subscript" 1" denote the tight tank data set and "2" denote the data from
the tests with simulated leaks.
Let m be the number of test results from tight tanks and m be the number of test results
from induced leak rate tests. Denote by dji the difference between measured and induced leak
rates for each test, where j=l or 2, and i=l, ...,m or 112. Then calculate
1=1
and Equation 15
n2 - -
(d2l-
_ V (d2j ~ dz)
~ Zj fn, - 1)
Equation 16
where the summations are taken over the appropriate groups of data, and where
denotes the mean of the data in group j, and is given by
form the ratio Equation 17
s22
F = —
c 2
^1
Equation 18
and compare this statistic to the F statistic with (n2-l) and (nl-1) degrees of freedom for the
numerator and denominator, respectively, at the 5% significance level. The F statistic can be
obtained from the F-Table.20 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 LD technology to detect
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 54 of 184
leaks. Re-compute the PD using the SD calculated from just the induced leak rate tests, S2, to
verify that PD is still at least 95%.
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 _ TAVA - TBVB
T°- ^D
Equation 19
The subscript A denotes product in tank after delivery, B denotes product in tank before
delivery, D denotes product delivered, T denotes product temperature, and V denotes volume.
Calculate the SD (Equation 4) of the temperature differentials.
Average Waiting Time After Fuel Drop
Use the time interval between the most recent fuel drop and each following test run as a
stabilization time. These will be ordered from least to greatest to determine the 20th percentile.
The minimum and average (as calculated in Equation 3) stabilization times will be reported.
Average Data Collection Time Per Test Run
The tests often have a constant or nearly constant duration prescribed by the technologies.
If so, the test data collection time will be reported as it is. If the technology software determines
a test time from the data, the average test time actually taken by the test will be reported.
B1.6 Reporting
The data obtained during this testing will be reported and the statistical analyses
described above will be conducted separately for each technology being tested. Information on
the performance parameters will be compiled and presented as evidence in the body of the TA or
in an appendix of the TA. If a test is inconclusive or incomplete (due to fuel dispensing or
delivery), the result will be reported; however, the run will be excluded from the statistical
analysis.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 55 of 184
All actions taken on the technology (such as maintenance, cleaning, and calibration) will
be documented at the time of the test and reported. In addition, descriptions of the data
recording procedures, use of vendor-supplied software, and fuel supplies or other consumables
used will be presented in an appendix of the TA.
B2 SAMPLING METHODS
B2.1 Sample Collection, Storage and Shipment
Fuel ethanol content determination will be performed before testing to verify that the
ethanol concentration is within ± 15% of the target level and that the water content is < 0.01.
The test blends will be verified for each batch prepared.
For the laboratory-scale testing, ASTM E20311 or 630422 will be used to characterize the
water content of the dense phase separated layer and the fuel, respectively, due to the high
concentration of water expected in the dense separated phase. They will be sampled and
analyzed after each run is completed. These analyses of the dense phase and fuel are to
characterize the water ingress testing condition. The 1.5 to 2 mL glass sampling vials and
Teflon®-lined caps for this analysis method will be solvent washed and dried overnight in a
100°C oven11 and allowed to cool in a desiccator before filling and sealing. Syringes will be
used to draw out samples from various places in the test column. Samples collected will be
stored in desiccators before analysis and held for 14 days. The samples will be shipped to the
analytical laboratory.
The analysis methods for the fuel ethanol content and water content determinations are
described in Section B4. Duplicate samples for both analytical determinations will be collected
at a frequency of 10% of the samples into a separate sampling jar for analysis. This will evaluate
the reproducibility of the sampling method. Duplicate sample analysis from the same sampling
jar at a frequency of 10% analyzed will evaluate the reproducibility of both ASTM D481513 and
D550112. Duplicate sample analysis of every sample is specified for water determination by the
Karl-Fischer titration methods, and the sample results are acceptable when they are less than
10% different12'22.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 56 of 184
B2.2 Digital Video Recording
The laboratory-scale tests will be performed in transparent containers so that the physical
impact of adding water to the vessel can be seen and video recorded. To facilitate visualization
of the physical changes occurring within the test vessel, colored food dye will be mixed into all
water introduced to the test column in a sufficient amount so as to clearly show the water phase
of the system. In addition to dye, visualization will be enhanced by using time-lapse video to
capture subtle changes during the experiments.
B3 SAMPLE HANDLING AND CUSTODY
Each sample will be labeled with a unique sample identifier along with the date/time
collected and the name of the technical staff. Sample custody will be documented throughout
collection and analysis of the test samples following the Battelle SOP for Sample Chain of
Custody (COC)23. A COC form will include details about the sample such as the time, date,
location, and person collecting the sample. The COC form will track sample release from the
sampling location to the analytical laboratory. Each COC form will be signed by the person
relinquishing samples once that person has verified that the COC form is accurate. Upon arrival
at the analytical laboratory, COC forms will be signed by the person receiving the samples (if
different from the sample collector) once that person has verified that all samples identified on
the COC forms are present. Copies of all COC forms will be delivered to the TC and maintained
with the test records.
B4 REFERENCE METHODS
Prior to analyzing test samples, a PEA will verify the reference laboratory performance
(ICFTL) using two National Institute of Standards and Technology (NIST) standard reference
materials (SRMs; Section A9). At the beginning of the test, fuel samples will be collected from
the prepared test blends to confirm ethanol and water content. In addition, samples will be taken
from the phase separated layer on the bottom of the test column for water content determination
and from the fuel after water ingress testing for ethanol content determination. As presented in
Table 11, analytical technicians will conduct these analyses according to the QC requirements
stated in the specific analytical methods.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 57 of 184
Ethanol content will be determined by ASTM D481513 and D550112 using gas
chromatography or an equivalent method(s). Water content will be determined using an
automated Karl-Fischer titration water analysis instrument following ASTM E20311 for water
content.
B5 QUALITY CONTROL
Steps will be taken to maintain the quality of data collected during testing by
implementing acceptance criteria for assessment of data collection quality (Table 12) and MQOs
(Table 13). In addition, instruments and equipment used for this verification will operate at the
expected ranges and calibration records will be verified and kept for all monitoring
instrumentation and equipment used for establishing the variables. All data collected will be
within the accepted QC criteria (or corrective action will be taken) and the true measured value
will be reported. NIST traceable calibration standards will be used where possible.
Table 12. List of ASTM Standards and Assessment of Data Quality
Method
ID
Title
Measurement
Method QC Requirements
(Reproducibility/Repeatability/
Bias)
GENERAL PROTOCOLS
D40577
ASTM
D58548
ASTM
D77179
Standard Practice for Manual Sampling of
Petroleum and Petroleum Products
Standard Practice for Mixing and Handling of
Liquid Samples of petroleum and Petroleum
Products
Standard Practice for Preparing Volumetric
Blends of Denatured Fuel Ethanol and Gasoline
Blendstocks for laboratory Analysis
NA
NA
NA
NA
NA
NA
SPECIFIC METHODOLOGIES
ASTM
E20311
ASTM
D630422
ASTM
D550112
ASTM
D481513
Standard Test Method for Water Using
Volumetric Karl Fischer Titration (Procedure §10)
Standard Test Method for Determination of
Water in Petroleum Products, Lubricating Oils,
and Additives by Coulometric Karl Fischer
Titration (Procedure B)
Standard Test Method for Determination of
Ethanol and Methanol Content in Fuels
Containing Greater than 20% Ethanol by Gas
Chromatography
Standard Test Method for Determination of
MTBE, ETBE, TAME, DIPE, tertiary-Amyl Alcohol
and Ci to C$ Alcohols in Gasoline by Gas
Chromatography(c)
Volume percentage of
water to the nearest 0.001%
Volume percentage of
water to nearest 0.01%
Volume percentage of
ethanol to nearest 0.01%
Volume percentage of
ethanol to nearest 0.01%
SD = 0.0034% absolute at 40 df.
The 95% CL = 0.010% absolute.
The difference between two
successive results shall exceed
0.08852x°-7 in less than one case
in 20; x = mean of duplicate
measurements
The normal range between two
results, each the mean of
duplicate determinations should
be less than 2.18*the mass % -°-6
The normal range between two
results, each the mean of
duplicate determinations should
be less than 0.06*the mean
mass % -°-61
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 58 of 184
Table 12. List of ASTM Standards and Assessment of Data Quality (Continued)
Method
ID
Title
Measurement
Method QC Requirements
(Reproducibility/Repeatability/
Bias)
ASTM
D161314
Standard Test Method for Acidity in Volatile
Solvents and Chemical Intermediates Used in
Paint, Varnish, Lacquer and Related Products
Percent acetic acid to
nearest 0.0001%
The normal range between two
results, each the mean of
duplicate determinations should
be less than 0.0008% absolute
ASTM
D28718
Standard Test Method for API Gravity of Crude
Petroleum and Petroleum Products (Hygrometer
Method)
Corrected hygrometer
reading to nearest 0.1
degree API converted to
g/mL
Difference between successive
test results on same material
shall exceed 0.2 degrees API in
less than one case in 20
ASTM
044516
Standard Test Method for Kinematic Viscosity of
Transparent and Opaque Liquids (and Calculation
of Dynamic Viscosity)
Kinematic and dynamic
viscosity to four significant
figures
Difference between successive
test results on same material
shall exceed 0.0013(y+l) in less
than one case in 20; y = average
of triplicate values
ASTM
D262417
Standard Test Methods for Electrical
Conductivity of Aviation and Distillate Fuels
(Portable Meter Method)
Electrical conductivity of the
fuel to nearest |aS/m
Maximum allowable difference
between two measurements
determined by absolute
measure of average of two
measurements but in all cases
less than 175 |aS/m
(a) For the laboratory-scale testing only of the separated phase at the bottom of the test column
(b) For samples with anticipated ethanol concentrations greater than 12.0 mass percent
(c) For samples with anticipated ethanol concentrations less than 12.0 mass percent
df = degrees of freedom
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 59 of 184
Table 13. Measurement Quality Objectives for Analytical Methods
Method
Designation
Method Title
QC Procedures
Recommended MQOs
ASTMD481513
Standard Test Method for
Determination of MTBE, ETBE,
TAME, DIPE, tertiary-Amyl
Alcohol and Ci to 64 Alcohols in
Gasoline by Gas
Chromatography
Daily calibration curve and
continuing QC check
samples every 10
samples
Calibration curve r2 > 0.99
QC check samples ±0.10
ASTMD550112
Standard Test Method for
Determination of Ethanol and
Methanol Content in Fuels
Containing Greater than 20%
Ethanol by Gas Chromatography
Daily calibration curve and
continuing QC check
samples every 10
samples
Calibration curve r2 > 0.995
QC check samples ±0.10
ASTM E2031
Standard Test Method for Water
Using Volumetric Karl Fischer
Titration
Daily calibration curve
and continuing QC check
samples every 10
samples
Calibration curve r2 > 0.90
QC check samples ±0.10
ASTM D630422
Standard Test Method for
Determination of Water in
Petroleum Products, Lubricating
Oils, and Additives by
Coulometric Karl Fischer
Titration (Procedure B)
Daily calibration curve and
continuing QC check
samples every 10
samples
Calibration curve r2 > 0.90
QC check samples ±0.10
ASTMD161314
Standard Test Method for Acidity
in Volatile Solvents and
Chemical Intermediates Used in
Paint, Varnish, Lacquer and
Related Products
No calibration; however,
duplicate determinations
will be considered suspect
if they differ more than
0.0008%
< 0.0008% Repeatability of
duplicate measurements
ASTM D28718
Standard Test Method for API
Gravity of Crude Petroleum, and
Petroleum Products
Daily check
±10 kg/m3 each for two
standards of 998 kg/m3 and
749 kg/m3
ASTM D44516
Standard Test Method for
Kinematic Viscosity of
Transparent and Opaque Liquids
(and Calculation of Dynamic
Viscosity)
Daily check
0.05 mm2/s for Certified
Reference Standard S3
ASTM D2624
Electrical Conductivity
Daily instrument check of
probe
Bias: Conductivity <1% error
each for two standards in
uS/cm and mS/cm range.
Repeatability: <0.1%fortwo
standards in uS/cm and
mS/cm range
B6 INSTRUMENT/EQUIPMENT TESTING, INSPECTION, AND MAINTENANCE
The equipment needed for this project (samplers, sample containers, miscellaneous
laboratory items, etc.) will be tested, inspected, maintained and operated according to the quality
requirements and documentation of any applicable standard method or of the laboratory
responsible for its use to ensure confidence in data that they generate. Testing and maintenance
must be performed according to manufacturer instructions and analytical methods and
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 60 of 74
documented. Only properly functioning equipment will be used; any observed malfunctioning
will be documented and appropriate maintenance or replacement of malfunctioning equipment
will be performed.
B7 INSTRUMENT/EQUIPMENT CALIBRATION AND FREQUENCY
The instruments and equipment used for this study must be calibrated prior to use to
ensure that the data generated are accurate. Calibration must be performed according to
manufacturer instructions and the analytical methods. Some of the methods used during this
project require calibration each day of analysis, but some require only a QC check sample to be
analyzed to confirm the ongoing accuracy of calibration that is performed periodically (or
possibly only by the manufacturer) (see Table 13). Instrument and equipment calibration
activities must be documented by model and serial number so that activities are traceable to the
specific unit.
The analytical laboratory must have documented quality procedures for equipment and
instrument calibration. Laboratories performing chemical analysis will provide full data
packages which contain all information required for validation. Data packages must contain any
of the following elements that are applicable to the analysis:
• Title page;
• Table of contents;
• Data package QC narrative;
• Final analytical results for each sample;
• Summary of samples processed with each analytical batch, showing that QC samples
were processed at the same time as the samples with the same solvents, reagents,
standards, etc.;
• Results of quality control samples and surrogate recoveries at least as percent
recovery, percent difference, etc.;
• Instrument sequences with dates/times for initial calibration and on-going calibration
checks, samples and QC samples.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 61 of 184
• Analytical records:
o Raw data (instrument quantification reports) for initial and on-going
calibration, quality control samples, and test samples;
o Chromatograms for samples, calibrations, and QC samples;
o Mass spectra for GC/MS analyses;
o Entire package of sample custody documentation, including sample receipt
forms;
o Sample processing and spiking records; and
o Description of manual integration procedures.
B8 INSPECTION/ACCEPTANCE OF SUPPLIES AND CONSUMABLES
All materials, supplies, and consumables used to establish the test conditions will be
ordered by the TC or designee. Where possible, Battelle will rely on sources of materials and
consumables that have been used previously as part of ETV testing without problems. Battelle
will also rely on previous experience or recommendations from UST LD stakeholders to guide
selection of manufacturers and materials. E10 is currently the only ethanol-blended fuel with a
standard reference material (SRM 2297). The performance of ASTM D481513 for ethanol will
be verified with this National Institute of Standards and Technology (NIST) provided SRM for
E10 fuel. This method will also be verified for isobutanol determination using a NIST traceable
calibration standard at 15% isobutanol. The performance of ASTM D550112 will be verified
with the NIST provided SRM for ethanol (SRM 2900). To ensure that each test blend is made
with the proper ethanol/isobutanol content, the ethanol content for EO, E10, El5, E30, E50 and
E85 test blends or the isobutanol content for 116 test blends will be verified before the beginning
of testing with that fuel.
All fuel and ethanol supplies, as well as generated liquid wastes, will be stored in tanks or
containers approved for the material being stored. Fuel, ethanol, and liquid waste storage areas
will be on impermeable surfaces with adequate secondary containment. Arrangements will be
made with trained waste handling technicians for removal and disposal of wastes generated
during testing.
Supplies must meet the following criteria:
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 62 of 184
• Solvent and reagent grades are based on the intended use. All materials must meet the
purity requirements of the method.
• Equipment used to generate data must provide appropriate sensitivity.
• A certificate of analysis must be provided and retained for reagents and standards.
• The quality and purity of expendable materials must be documented and adequate to
meet the DQOs of the client.
B9 NON-DIRECT MEASUREMENTS
No non-direct measurements will be used during the bench- and laboratory-scale testing.
Any secondary data will be collected from the field site owners and operators and will be
assumed to be accurate upon data gathering. Such information may include tank volume,
throughput, additive information, etc.
BIO DATA MANAGEMENT
Various types of data will be acquired and recorded electronically or manually by
technical staff during this testing. All data and observations for the operation of the technologies
will be documented by the verification staff on data sheets, in LRBs, or captured electronically.
Table 3, presented previously, summarizes the types of records to be collected and maintained
during the study. Results from the laboratory analytical instruments will be compiled by
laboratory staff in electronic format and submitted to the TC or other technical staff upon
obtaining results before the beginning of each test run.
Records received by or generated by any of the technical staff during the testing will be
reviewed by the TC or designee within 2 weeks of receipt or generation, respectively, before the
records are used to calculate, evaluate, or report results. The review will be documented as the
dated initials of the reviewer. Table 14 summarizes the checks to be performed. If a Battelle
staff member generated the record, this review will be performed by a Battelle technical staff
member involved in the testing, but not the staff member that originally received or generated the
record. The review will be documented by the person performing the review by adding his/her
initials and date to the hardcopy of the record being reviewed. In addition, data calculations
performed by technical staff will be spot-checked by a second technical staff to ensure that
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 63 of 184
calculations are performed correctly. Calculations to be checked include any statistical
calculations described in this QAPP. The data obtained from this testing will be compiled and
reported for each set of tests.
All electronic testing records and documents will be stored on a test-specific networked
ETV SharePoint site. This site is within the protected Battelle network; incremental back-ups
are performed nightly and full back-ups weekly by Battelle's Corporate Information Technology
group. In addition, the back-ups are also saved to a second disk storage (data domain) located in
a different data center. All back-up files are retained for nine weeks. Testing data will be
uploaded to the SharePoint site on a weekly basis. The goal of this data delivery schedule is
prompt identification and resolution of any data collection or recording issues.
In addition, once testing is complete, all testing records and documents are sent to
Battelle's RMO for archival within 2 months of project closeout.
Table 14. Data Verification Checks
Data Verification Activity
QC samples and calibration standards will be analyzed according to this document, and the
acceptance criteria will be met. Corrective action for exceedances will be taken.
100% hand-entered and/or manually calculated data will be checked for accuracy.
Calculations performed by software will be verified at a frequency sufficient to ensure that the
formulas are correct, appropriate, and consistent.
For each cut and paste function, the first and last data values will be verified against the original
source data.
Data will be reported in the units specified in the QAPP.
Results of QC will be reported.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 64 of 184
SECTION C
ASSESSMENT AND OVERSIGHT
Cl ASSESSMENT AND RESPONSE ACTIONS
One of the major objectives of the QAPP is to establish mechanisms necessary to
anticipate and resolve potential problems before data quality is compromised. Internal QC
measures described in this QAPP will yield day-to-day information on data quality. The
responsibility for interpreting the results of these checks and resolving any potential problems
resides with the TC. Technical staff has the responsibility to identify problems that could affect
data quality or usability. Any problems that are identified will be reported to the TC, who will
work with the Battelle QAM to resolve any issues. Action will be taken to identify and
appropriately address the issue and minimize losses and correct data, where possible. The TC
will also relay testing progress and data to the EPA PO, or his designee, once every 2 weeks
during testing to ensure that EPA has real-time access to the data as generated and testing
continues to fulfill the DQOs. Battelle will be responsible for ensuring that the audits described
in the following subsections are conducted as part of this testing. See Table 2 for the proposed
schedule of audits.
Any changes to the approved QAPP must be reported within 24 hours and documented in
a formal deviation submitted to the Battelle AMS Center Manager, EPA PO and EPA QAM. If
approval by EPA PO or his designee is not received within 24 hours of notification, testing will
be halted until a suitable resolution has been achieved.
Cl.l Performance Evaluation Audits
A PEA will be conducted to assess the quality of the variable measurements made in this
test. The PEA will verify that the measured water content and ethanol content of the test blends
and BFW mixtures are achievable within the stated acceptance criteria presented in Table 6.
The accuracy of the analytical methods will be evaluated in the PEA by analyzing a NIST
traceable certified standard. For the low-level ethanol content determination method D481513,
SRM 2297- Reformulated Gasoline (10% Ethanol) will be used. This method will also be
verified for isobutanol determination using a NIST traceable calibration standard at 15%
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 65 of 184
isobutanol. For the high-level ethanol content determination method D550112, SRM 2900-
Ethanol-Water Solution, (nominal 95.6%) will be used. The results of this E10 standard are
acceptable when within 10% of the target ethanol content. The water standard concentration and
source will be determined during the pre-checks and dry runs and will also be NIST traceable.
The results of the water standard are acceptable when within 10% of the target control standard
concentration. The analytical methods and their associated PEA material and acceptance criteria
are summarized in Table 15. If the results do not meet the requirements, they will be repeated.
If the outlying results persist, the TC, or designee, and the analytical laboratory representative
will discuss corrective actions, and the PEA will be repeated. The results from the PEA will be
sent to the EPA PO and EPA QAM within 10 days of receipt of the results. The PEA report will
include the raw data, performance evaluation certificate of analysis, calculations of the
comparison to the expected concentration, and a discussion of corrective action, if applicable.
Table 15. Analytical Methods and PEA Materials
Method
ID
ASTM
E20311
ASTM
D550112
ASTM
D481513
Title
Standard Test Method for Water Using
Volumetric Karl Fischer Titration (Procedure §10)
Standard Test Method for Determination of
Ethanol and Methanol Content in Fuels
Containing Greater than 20% Ethanol by Gas
Chromatography
Standard Test Method for Determination of
MTBE, ETBE, TAME, DIPE, tertiary-Amyl Alcohol
and Ci to C$ Alcohols in Gasoline by Gas
Chromatography
PEA Material
SRM 2900
SRM 2900
SRM 2297 for Ethanol;
Spectum Calibration
standard for Isobutanol
Acceptance Criteria
Within 10% of the target
concentration, repeat analysis if
out of range
C1.2 Technical Systems Audits
The Battelle QAM will perform a one-day TSA of the bench-scale test set. The purpose
of this audit is to ensure that the tests are being performed in accordance with the AMS Center
QMP2 and this QAPP. During this audit, the Battelle QAM, or designee, will review
• Documentation for the preparation of the test blends and BFW mixtures and the
results of the EO analysis;
• Testing facility equipment (calibration, maintenance, and operation);
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 66 of 184
• Actual test procedures versus those specified or referenced in this plan; and
• Data acquisition and handling procedures, including observation of testing and
records (including custody forms).
The TSA will be guided by a project-specific checklist based on this QAPP. It will be
performed during the bench-scale test sets because this is where many different steps of the
process will be performed (sample preparation, shipment to the analytical laboratory, multiple
data points collected on one test blend).
A TSA report will be prepared as a memo to the TC within 10 business days after
completion of the audit; the completed checklist will be attached. The Battelle AMS Center
Manager and EPA PO will be copied on the memo. The TC will respond to the audit within 10
business days. The Battelle QAM or designee will verify that all audit findings and observations
have been addressed and that corrective actions are appropriately implemented. A copy of the
complete TSA report with corrective actions will be provided to the EPA PO, or his designee,
within 10 business days after receipt of the audit response. At EPA's discretion, EPA QA staff
may also conduct an independent on-site TSA during one or multiple phases during the
execution of this QAPP. The TSA findings will be communicated to technical staff at the time
of the audit and documented in a TSA report.
C1.3 Data Quality Audits
The Battelle QAM, or designee, will audit at least 25% of the sample results acquired in
the verification test and 100% of the calibration and QC data per the QAPP requirements. A
checklist based on the QAPP will guide the audit. An initial ADQ will be conducted on the first
batch of test data within 10 business days of when data were posted on the project SharePoint
site to identify errors early in the data reduction process. The first batch is defined as the testing
and variable data generated over the first two weeks of testing by the TC. The remaining data
will be audited at the completion of each set of tests (i.e., bench-, laboratory- and full-scale) after
all data for that set of tests have been posted on the project SharePoint site and once all statistical
analyses for that set of tests are complete. Finally, a third ADQ, performed by the Battelle QAM
or designee, will trace the data from initial acquisition, through reduction and statistical
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 67 of 184
comparisons, to final presentation in the reports and TA. It will also confirm reconciliation of
the two ADQs.
All formulae applied to the data will be verified, and 25% of the calculations will be
checked. Data for each set of tests will be reviewed for calculation and transcription errors and
data traceability. An audit report will be prepared as a memo to the TC within 10 business days
after completion of each data audit; the completed checklist will be attached. The Battelle AMS
Center Manager, EPA PO and EPA QAM will be copied on the memo. The TC will respond to
the audit within 10 business days. The Battelle QAM or designate will verify that all audit
findings and observations have been addressed and that corrective actions are appropriately
implemented. A copy of the complete ADQ report with corrective actions will be provided to
the EPA PO, or his designee, within 10 business days after receipt of the audit response. EPA
QA staff will also conduct an independent ADQ.
C1.4 QA/QC Reporting
Each assessment and audit will be documented in accordance with Section 10.5 of the
AMS Center QMP2. The results of the PEA, including raw data and calculations, will be
reported as stated in Section Cl. 1. The results of the TSA and ADQ will be submitted to EPA.
Assessment reports will include the following:
• Identification of findings and observations;
• Recommendations for resolving problems;
• Response to adverse findings or potential problems;
• Confirmation that solutions have been implemented and are effective; and
• Citation of any noteworthy practices that may be of use to others.
C2 REPORTS TO MANAGEMENT
The Battelle QAM, during the course of any assessment or audit, will identify to the
technical staff performing experimental activities any immediate corrective action that should be
taken. If serious quality problems exist, the Battelle QAM is authorized to notify the Battelle
AMS Center Manager, who will issue a stop work order. Once the TSA or ADQ report has been
prepared, the TC will respond to each finding or observation following the timeline defined in
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 68 of 184
section Cl and will implement any necessary corrective action. The Battelle QAM will verify
that corrective action has been implemented effectively.
In addition to this QAPP, a final TA report will be prepared and reviewed. The TA
report will present the data collected as evidence for how UST LD technologies perform or are
expected to perform when employed in biofuels. The TA will be submitted to expert peer
reviewers for review. They will then be reviewed by EPA PO, or his designee. Upon final
review and approval, the document will be posted on the ETV Web site (www.epa.gov/etv). A
summary of the required assessments and audits, including a listing of responsibilities and
reporting timeframes, is included in Table 16.
Table 16. Summary of Assessment Reports*
Assessment
ISA
ADQ1
(first batch)
ADQ2
(raw data)
ADQ3
(synthesized
data and
verification
report)
Prepared By
Battelle
Battelle
Battelle
Battelle
Report Submission Timeframe
ISA response is due to QM within
10 business days
ISA responses will be verified by
the QM and provided to EPA within
20 business days
ADQ will be completed within 10
business days after receipt of first
data set
ADQ will be completed once all data
are received and analyzed
ADQ will be completed within 10
business days after completion of
the verification report review
Submitted To
EPA ETV AMS Center
EPA ETV AMS Center
EPA ETV AMS Center
EPA ETV AMS Center
(a) Any QA checklists prepared to guide audits will be provided with the audit report.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 69 of 184
SECTION D
DATA VALIDATION AND USABILITY
Dl DATA REVIEW, VERIFICATION, AND VALIDATION
Data verification and validation procedures are used throughout the data collection,
analysis, and reporting process to assess data quality. Data verification is defined as the process
of evaluating the completeness, correctness, and conformance/compliance of a specific data set
against the method, procedural, or contractual requirements. Data verification will first be
performed by the laboratory that generated the data and then by Battelle within two weeks of
receipt of the laboratory data. Table 14 summarizes the verification activities. The reviewer will
be familiar with the technical aspects of the verification test but will not be the person who
generated the data. This process will serve both as the data review and the data verification, and
will ensure that the data have been recorded, transmitted and processed properly. Furthermore,
this process will ensure that the technology data and reference method data were collected under
appropriate testing conditions and that the reference sample data meet the specifications of
analytical methods.
Data validation is an analyte- and sample-specific process that extends the evaluation of
data beyond method, procedural, or contractual compliance (i.e., data verification) to determine
the analytical quality of a specific data set. Data validation will be performed by the QAM or
designate who is independent of the data generation process. The data validation requirements
for this test involve an assessment of the quality of the data relative to conformance to the test
design specifications, QC acceptance criteria and MQOs defined in Section Bl (e.g., Tables 6, 7,
and 8) and Section B5. The QA audits described in Section C are also designed to validate the
quality of the data. Data failing to meet the QAPP DQOs and acceptance criteria will be flagged
in the data set and not used for evaluation of the monitoring systems, unless these deviations are
accompanied by descriptions of their potential impacts on the data quality.
D2 VALIDATION AND VERIFICATION METHODS
Data verification is conducted as part of the data review as described in Section BIO of
this QAPP. Data verification includes a visual inspection of hand written data to ensure that all
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 70 of 184
entries were properly recorded or transcribed and that any erroneous entries were properly noted,
as described in Sections A10 and BIO. Data verification of completeness and correctness
consists of tracing individual sample analytical results from the ETV test (bench-scale,
laboratory-scale, or full-scale testing) through the COC records, to the analytical results.
Sampling documentation is verified through the review and approval of each testing LRB or
logbook. Data verification is also accomplished by ensuring the accuracy and completeness of
data transcribed from raw data to the results report. A comparison of raw data sheets, field logs
or LRB comments against final data will be conducted to flag any suspect data and resolve any
questions about apparent outliers. Entry of data into spreadsheets from field logs and laboratory
reports is verified when the Battelle QM audits the data.
Data verification of conformance/compliance consists of reviewing the test records to
verify that the tests were conducted according to the QAPP requirements. For analytical
laboratory data, the laboratory report and supporting data will be reviewed to verify that the
calibration, analysis, detection limits, and QC sample results meet the requirements of the
methods and this QAPP.
During data verification, electronic data will be inspected to ensure proper transfer from
the data logging system. All calculations used to transform the data will be reviewed to ensure
the accuracy and the appropriateness of the calculations. Calculations performed manually will
be reviewed and repeated using a handheld calculator or commercial software (e.g., Excel).
Calculations performed using standard commercial office software (e.g., Excel) will be reviewed
by inspection of the equations used for the calculations and verification of selected calculations
by handheld calculator. Calculations performed using specialized commercial software (i.e., for
analytical instrumentation) will be reviewed by inspection and, when feasible, verified by
handheld calculator, or standard commercial office software.
Sections B and C of this QAPP provide a description of the validation safeguards
employed for this verification test. To ensure that the data generated from this test meet the
goals of the test, a number of data validation procedures will be performed. Data validation
efforts include the completion of QC activities, and the performance of ADQ and PEAs as
described in Section C. The data from this test will be evaluated relative to the measurement
criteria defined in Sections Bl, B5, and B7 and PEA acceptance criteria given in Section Cl.l of
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 71 of 184
this QAPP to ensure that the DQOs are met. Data failing to meet these criteria will be flagged in
the data set and not used for evaluation of the technologies, unless these deviations are
accompanied by descriptions of their potential impacts on the data quality.
An ADQ will be conducted by the Battelle QAM to ensure that data review, verification,
and validation procedures were completed and to assure the overall quality of the data.
D3 RECONCILIATION WITH USER REQUIREMENTS
Once data have been generated and compiled in the laboratory, the TC will review data to
identify and make professional judgments about any suspicious values. All suspect data are
reported with a qualifier and appropriate comment. These data may not be used in calculations
or data summaries without the review and approval of the TC. No data measurements are
eliminated from the reported data or database and data gaps are never filled based on other
existing data. If samples are lost during shipment or analysis, it is documented in the data
qualifiers and comments submitted to EPA. The data obtained during this project will provide
thorough documentation of the required measurements. The data review and validation
procedures described in the previous sections will determine if data meet the quality objectives.
The data generated throughout this project will be compiled into a TA report. The TA report will
present the data as evidence of how UST LD technologies perform in biofuels.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 72 of 184
SECTION E
REFERENCES
1. Carvitti, J.A.M.G.a.A.D., DRAFT Technology Assessment: Suitability of Leak Detection
Technology for Use In Ethanol-Blended Fuel Service. September 30, 2011.
2. Quality Management Plan for the ETV Advanced Monitoring Systems Center, Version 8.
U.S. Environmental Technology Verification Program, Battelle, April 2011.
3. Environmental Technology Verification Program Quality Management Plan January
2008(EPA/600/R-08/009).
4. EPA Office of UndgroundStorage Tanks. January 11, 2013 January 15, 2013]; Available
from: http://www.epa.gov/oust/.
5. RFA, Building Bridges to a More Sustainable Future: 2011 Ethanol Industry Outlook.
2011.
6. Ken Wilcox Associates, I, Standard Test Procedures for Evaluating Leak Detection
Methods: Automatic Tank Gauging Systems (ATGS). 1990, USEPA Solid Waste and
Emergency Response/Research and Development.
7. ASTM, D4057-06: Standard Practice for Manual Sampling of Petroleum and Petroleum
Products. August 2011.
8. ASTM, D5854-96: Standard Practice for Mixing and Handling of Liquid Samples of
Petroleum and Petroleum Products. May 2010.
9. ASTM, D7717-11: Standard Practice for Preparing Volumetric Blends of Denatured
Fuel Ethanol and Gasoline Blendstocks for Laboratory Analysis. August 2011.
10. E. de Oliveira, J.F.B.a.I.C. Gasoline-Water-Ethanol Inter actions and Fluid Properties, in
Groundwater: Prevention, Detection and Remediation Conference and Exhibition
Special Focus:Natural Attenuation and Gasoline Oxygenates. November 2000.
Anaheim, CA.
11. ASTM, E 203-08: Standard Test Method for Water Using Volumetric Karl Fischer
Titration. November 2008.
12. ASTM, D5501-12: Standard Test Method for Determination of Ethanol andMethanol
Content in Fuels Containting Greater than 20% Ethanol by Gas Chromatography. April
2013.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 73 of 184
13. ASTM, D4815-09: Standard Test Method for Determination ofMTBE, ETBE, TAME,
DIPE, tertiary-Amyl Alcohol and Ci to €4 Alcohols in Gasoline by Gas Chromatography.
November 2009.
14. ASTM, D1613-06: Standard Test Method for Acidity in Volatile Solvents and Chemical
Intermediates Usedin Paint, Varnish, Lacquer, and Related Products. July 2012.
15. ASTM, D4052-11 Standard Test Method for Density, Relative Density, and API Gravity
of Liquids by Digital Density Meter. 2011.
16. ASTM, D445-12: Standard Test Method for Kinematic Viscosity of Transparent and
Opaque Liquids (and Calculation of Dynamic Viscosity). May 2012.
17. ASTM, D2624-09: Standard Test Methods for Electrical Conductivity of Aviation and
Distillate Fuels. February 2010.
18. ASTM, D28 7-92: Standard Test Method for API Gravity of Crude Petroleum and
Petroleum Products (Hydrometer Method). June 2006.
19. ASTM, D7451-08a: Standard Test Methods for Water Seperation Propoerties of Light
and Middle Distillate, and Compression and Spark Ignition Fuels. January 2009.
20. Steel, R.G.D.a.T., J. H., Principles and Procedures of Statistics. 1960, New York:
McGraw-Hill.
21. Lieberman, G., Tables for One-Sided Statistical Tolerance Limits, ed. I.Q. Control. Vol.
Vol. XIV, No 10. 1958.
22. ASTM, D 6304-07: Standard Test Method for Determination of Water in Petroleum
Products, Lubricating Oils, and Additives by Coulometric Karl Fischer Titration. August
2007.
23. SOP ENV-ADM-009, Standard Operating Procedure for Sample Chain-of-Custody
Battelle, September 2007.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 74 of 184
APPENDIX A
Underground Storage Tank Leak Detection
Stakeholder Committee
-------
Appendix A
Underground Storage Tank Leak Detection Stakeholder Committee
Last Name
Barbery
Bareta*
Baustian
Boucher
Bradley*
Brauksieck
Brevard
Chapin
Cochefski
Cornett
Courville
D'Alessandro
Dockery
Drack
Emmington
Fenton
Fisher
Flora
Folkers
Geyer
Gordji
Henderson
Hoffman
Indest
Johnson*
Jones
Juranty*
Keegan
Kubinsky
Lauen
Marston
McKernan
First Name
Andrea
Greg
James
Randy
Lamar
Russ
Danny
Tom
Peter
Ken
Jamie
Tom
Howard
Earle
Dave
Charles
Laura
Jerry
Joie
Wayne
Sam
Kevin
Brad
April
Curt
Bill
Mike
Kevin
Ed
Dorcee
Dan
John
Company
US EPA OUST
Engineering Consultant Bureau of Storage Tank Regulation
(Wisconsin)
Butamax
Franklin Fueling Systems
Tennessee Dept of Environment and Conservation Division of
USTs
(New York)
ACCENT Services, Inc.
Underwriters Laboratory (U.L.)
Ryder Fuel Services
Veeder-Root
Southern Tank Testers, Inc.
OMNTEC Mfg., Inc.
Simmons
DirAction, LLC.
Veeder-Root
Hansa Consult of North America, LLC (HCNA)
UST Leak Prevention Unit (California)
JDF Consulting
NOV Fiber Glass Systems
Steel Tank Institute
SSG Associates, University of Mississippi
Kevin Henderson Consulting, LLC
Tanknology
Southern Tank Testers, Inc.
Alabama Department of Environmental Management
(Alabama)
Warren Rogers Associates, Inc.
New Hampshire Department of Environmental Services Waste
Management Division
Tanknology
Crompco, LLC
Williams & Company
Franklin Fueling Systems
US EPA
-------
Last Name
McMillan
Mills
Moore*
Moore
Moureau
Muhanna*
Neil
Nelson
Parnell
Peters*
Poxson*
Purpora
Ramshaw
Reid
Renkes
Robbins*
Rollo*
Sabo
Scheib
Smith*
Thuemling
Toms
Wilcox
Wilcox
Young
First Name
Corey
Tony
Bill
Kristy
Marcel
Shaheer
Peter
Bill
Brian
Heather
Marcia
Steve
Chris
Kent
Bob
Helen
Peter
Lorraine
Jeff
Tim
George
Patrick
Craig
Ken
Greg
Company
Ryder Fuel Services
OPW Fuel Management Systems
Utah Department of Environmental Quality
Renewable Fuels Association (RFA)
Marcel Moreau Associates
Georgia Department of Natural Resources
OPW Fuel Management Systems
Franklin Fueling Systems
MAPCO Express, Inc.
Missouri Department of Natural Resources
Michigan Department of Environmental Quality
Protanic
Purpora Engineering
Veeder-Root
PEI
Connecticut Department of Environmental Protection
Delaware Natural Resources and Environmental Conservation
Franklin Fueling Systems
Gevo
US EPA OUST
Varec, Inc.
Varec, Inc.
Ken Wilcox Associates, Inc.
Ken Wilcox Associates, Inc.
Vaporless Mfg., Inc.
"Designates members of the National Work Group on Leak Detection Evaluation (NWGLDE)
-------
Appendix A
Environmental Technology Verification
Fuel Property and Technology Testing
-------
Appendix A
Environmental Technology Verification Fuel Property and Technology Testing
Al BACKGROUND
The U.S. Environmental Protection Agency (EPA) supports the Environmental
Technology Verification (ETV) Program to facilitate the deployment of innovative environmental
technologies through performance verification and dissemination of information. The goal of the
ETV Program is to further environmental protection by accelerating the acceptance and use of
improved and cost-effective technologies. ETV seeks to achieve this goal by providing high-
quality, peer-reviewed data on technology performance to those involved in the design,
distribution, financing, permitting, purchase, and use of environmental technologies.
ETV works in partnership with recognized testing organizations; with stakeholder groups
consisting of buyers, vendor organizations, and permitters; and with the full participation of
individual technology developers. The program evaluates the performance of innovative
technologies by developing test plans that are responsive to the needs of stakeholders, conducting
field or laboratory tests (as appropriate), collecting and analyzing data, and preparing peer-
reviewed reports. All evaluations are conducted in accordance with rigorous quality assurance
(QA) protocols to ensure that data of known and adequate quality are generated and that the
results are defensible. The definition of ETV verification is to establish or prove the truth of the
performance of a technology under specific, pre-determined criteria or protocols and a strong
quality management system. The highest-quality data are assured through implementation of the
ETV Quality Management Plan. ETV does not endorse, certify, or approve technologies.
The EPA's National Risk Management Research Laboratory (NRMRL) and its
verification organization partner, Battelle, operate the Advanced Monitoring Systems (AMS)
Center under ETV.
A2 TEST DESIGN AND PROCEDURES
A2.1 Test Overview
This verification test was conducted according to procedures specified in the Quality
Assurance Project Plan1 and the Addendum2 for Biofuel Properties and Behavior Relevant to
A-l
-------
Underground Storage Tank Leak Detection System Performance (QAPP) and adhered to the
quality system defined in the ETV AMS Center Quality Management Plan (QMP)3. A
stakeholder committee was specifically assembled for the preparation of the QAPP. A list of
participants in the stakeholder committee members is presented at the end of this appendix (Table
9). The committee included representatives from industry associations, state and federal
governments, including representatives of the National Work Group on Leak Detection
Evaluations (NWGLDE), and users. The responsibilities of verification test stakeholders and/or
peer reviewers included:
• Participate in technical panel discussions (when available) to provide input to the test
design;
• Review and provide input to the QAPP; and
• Review and provide input to the verification report(s)/verification statement(s).
Battelle conducted this verification test with funding support from the EPA's Office of
Underground Storage Tanks (OUST).
Testing was conducted as three distinct sets of tests. Each test set was designed to acquire
specific data with respect to fuel properties or leak detection technology performance. The three
sets were:
1. Bench-scale studies for the determination of select physical and chemical properties
of biofuels and biofuel- water (BFW) mixtures.
2. Laboratory-scale studies for the identification and quantification of specific biofuel
and BFW mixture processes affecting performance of UST LD operating principles.
3. Pressure decay testing for the understanding of the effect of ethanol, if any, on a leak
when pressurized.
The bench-scale testing aimed at determining several fundamental properties of alcohol-blended
fuels and BFW mixtures under typical conditions encountered during operation of underground
storage tank (UST) leak detection (LD) systems. The goal of the bench-scale testing was to
differentiate whether the range of ethanol blends had properties that behaved significantly
different from each other, thereby being the evidence that the technologies may or may not
function properly when used in the different blends. Bench-scale testing was divided into four
series of tests described below and the results are presented in Appendix B.
1. Intrinsic Properties of BFW Mixtures
2. Coefficient of Thermal Expansion
3. Non-additive Volume Changes
A-2
-------
4. Interface Determination of Phase Separation
The laboratory-scale tests evaluated the performance of an optical sensor, a sensor with a
float switch and fuel sensitive polymer and a capacitance/conductance sensor (that is not yet on
the market) in ethanol blended fuels. One of the goals of this test was to provide information on
the performance of different operating principles when used with ethanol-blended fuel. To
accomplish this goal, the experimental design included the following three options for testing:
1. Initial water/test blend detection
2. High liquid detection
3. Water ingress detection when submerged in a test blend
The technologies were tested according to their abilities; therefore, not all tests were performed
for all three technologies. The testing and results for the verification testing of the sensor can be
found in Appendix C.
The pressure decay testing aimed at determining the impact of different
ethanol/isobutanol blended fuels on the functionality of pressure decay as a pipeline leak
detection method. Pressure decay relies on the concept that a pipeline containing fuel is
pressurized and sections isolate to show a loss of pressure overtime if a leak is present. This
pressure decay test is focused on whether the different blends of fuel would affect the leak rate.
The testing procedures and results for the pressure decay testing can be found in Appendix D.
A2.2 Test Site Description
The interior of existing research buildings (Building 9 and Building 1) at Battelle's
Columbus, Ohio campus was used to conduct the bench- and laboratory-scale experiments.
Building 9 contains a large, high-bay room (9-0-50) on the north end of the building. Within the
room, there is a smaller ventilated room (9-0-5OC) where experimentation took place. The
ventilated room was modified and connected to building steel to provide bonding and grounding
to eliminate risks of static build up. Fuel and waste storage areas were located outside on the
northwest side of Building 9. All experimental work on the pressure decay testing was conducted
in a fume hood in the Environmental Restoration laboratory in Building 1 (1-2-30). The fume
hood was modified and connected to building steel to provide bonding and grounding for the
pressure decay vessel. The testing occurred between May and November 2013. Analytical results
were determined by a contracted laboratory, Iowa Central Fuel Testing Laboratory (ICFTL).
A-3
-------
A2.3 Experimental Design-Preparation of Test Blends
All test blends were prepared in an identical manner for all portions of the testing. All
petroleum products were sampled, mixed and handled according to ASTM D40574 and D58545;
volumetric blend stocks of ethanol (or isobutanol) and gasoline were prepared according to
ASTM D77176. In addition to ethanol blends, an isobutanol blend containing 16% (v/v)
isobutanol (116) was included in the list of test blends. Test blends were prepared by mixing
different concentrations of ethanol-free gasoline (EO) with either denatured ethyl alcohol
(ethanol; >97% purity) in the case of ethanol blends or isobutyl alcohol (isobutanol; >98% purity)
in the case of 116. EO was purchased from Marble Cliff Oil (Columbus, OH) and was approved
for sale as automotive fuel. Information such as Material Safety Data Sheets and Bills of Lading
were collected and recorded during fuel delivery. Proposed test blend compositions have been
selected based on those that are currently available on the market or are anticipated to be
available on the market. Test blends for the bench-scale test sets included gasoline (EO) and was
prepared to simulate low ethanol blends (E10, E15, and E30), flex fuels (E50, and E85) and an
isobutanol blend (116). Test blends for the laboratory-scale test sets were EO, E15, E30, E50, E85
and 116 (only one technology was tested using E30 and E50) and groundwater. An aliquot of EO,
E15, E85, and 116 test blends for the laboratory-scale testing were used for the pressure decay
testing, as well as deionized water. E85 for the laboratory-scale and pressure decay testing was
purchased from a local Giant Eagle (Columbus, OH) gas station.
Before preparation of the test blends, the water and ethanol content of the EO gasoline
were determined by ASTM D2037 and ASTM D48158, respectively. Table 1 indicates the mixing
ratios of EO and ethanol or isobutanol to achieve the desired test blend composition assuming EO
contains no ethanol or water. Table 2 and 3 indicates the data quality objectives (DQO) that had
to be met for the test blends. Table 4 and Table 5 display the test blend results for
ethanol/isobutanol content and water content for all three sets of testing. As presented in these
tables, all of the bench scale test blends had ethanol content relative percent differences (RPDs)
<15% and less than 0.1% water content and therefore met the acceptance criteria. As well as, all
of the laboratory-scale and pressure testing test blends fell below the required 0.25% water
content, except for the E85 test blend which was purchased. In addition, they all fell within 25%
of the target alcohol value. Test blends were sampled and mixed in two 4-liter (L) batches and
used as soon as possible for the bench-scale and laboratory-scale experiments. Test blends which
were not used immediately will be capped and stored at room temperature for no more than 21
days before use.
A-4
-------
Table 1. Mixing Ratios of EO and Ethanol/Isobutanol
for Preparation of Test Blends
Test Blend
EO
E10
E15
E30
E50
E85
116
Volume Fraction
EO
1.0
0.90
0.85
0.70
0.50
0.15
0.84
Volume Fraction
Ethanol/Isobutanol
0.0
0.10
0.15
0.30
0.50
0.85
0.16
Table 2. Data Quality Objectives and Corrective Action for Bench-scale Testing
Test Blend
Purchased Gasoline
(EO)
Prepared Ethanol
Test Blends
(E10,E15,E30, E50
and E85)
Prepared Isobutanol
Test Blend (116)
Analysis
Water
Content
Ethanol
Content
Water
Content
Ethanol
Content
Ethanol
Content
Water
Content
Isobutanol
Content
Method
ASTM
E2037
ASTM
D48158
ASTM
E2037
ASTM
D48158
ASTM
D55019
ASTM
E2037
ASTM
D55019
Data Quality
Objective
Water Content
<0.1%(v/v)
Ethanol Content
< 1% (v/v)
Water Content
< 0.1% (v/v)
Ethanol Content
<15%RPD
Ethanol Content
<15%RPD
Water Content
< 0.1% (v/v)
Isobutanol
Content <15%
RPD
Corrective Action
Note discrepancy in project
files
Note discrepancy in project
files
Note discrepancy in project
files
Remake and reanalyze test
blend
Remake and reanalyze test
blend
Note discrepancy in project
files
Remake and reanalyze test
blend
A-5
-------
Table 3. Data Quality Objectives and Corrective Action for Laboratory Scale and Pressure
Decay Testing
Test Blend
Purchased Gasoline
(EO)
Prepared Ethanol
Test Blends (El 5,
E30, andE50)
Purchased Ethanol
Test Blend (E85)
Prepared Isobutanol
Test Blend (116)
Analysis
Water
Content
Ethanol
Content
Water
Content
Ethanol
Content
Water
Content
Ethanol
Content
Water
Content
Isobutanol
Content
Method
ASTM
E2037
ASTM
D48158
ASTM
E2037
ASTM
D48158
ASTM
D55019
ASTM
E2037
ASTM
D55019
ASTM
E2037
ASTM
D55019
Data Quality
Objective
Water Content <
0.25% (v/v)
Ethanol Content <
1% (v/v)
Water Content <
0.25% (v/v)
Ethanol content
11.25-18.75%
(v/v)forE15
Ethanol Content
22.5-37.5 % (v/v)
for E30. Ethanol
Content 37.5-
62.5% (v/v) for
E50
None
None
Water Content <
0.25% (v/v)
Isobutanol Content
12.00-20.00%
(v/v)
Corrective Action
Note discrepancy in project
files
Note discrepancy in project
files
Note discrepancy in project
files
Note discrepancy in project
files
Note discrepancy in project
files
Note true value in project
files
Note true value in project
files
Note discrepancy in project
files
Note discrepancy in project
files
Table 4. Test Blend Ethanol and Water Content for Bench-Scale Testing
Test Blend
E0#l
E0#2
E0#l
E10#l
E10#2
E15#l
E15#2
116 #1
I16#2
E30#l
E30#2
E30#l
E30#2
E50#l
E50#2
E85#l
E85#2
Date Prepared
4/2/2013
8/14/2013
4/22/2013
4/24/2013
8/14/2013
4/30/2013
8/14/2013
8/15/2013
5/8/2013
5/15/2013
Measured Ethanol
Content
(% volume)
0.495
0.495
0.32
10.85
10.76
14.84
15.02
17.41
17.35
28.32
28.34
29.03
28.82
45.62
45.44
78.67
78.47
Measured
Water Content
(% volume)
0.008*
0.008*
0.017
0.024*
0.037*
0.034*
0.032*
0.050
0.051
0.036
0.030
0.066
0.054
0.040
0.041
0.051
0.053
Data Quality
Objective For
Ethanol (%RPD)
< 1% ethanol
< 1% ethanol
< 1% ethanol
8.50%
7.60%
1.07%
0.13%
8.81%
8.44%
5.60%
5.53%
3.23%
3.93%
8.76%
9.12%
7.45%
7.68%
A-6
-------
*Water content was measured as % mass, not % volume
Table 5. Test Blend Analytical Results for Laboratory-Scale and Pressure Decay Testing
Test
Blend
EO
E15
116
E30
E50
E85
Water Content
%
Mass
0.013
0.011
0.038
0.041
0.038
0.029
0.095
0.054
0.068
1.111
%
Volume
0.01
0.008
0.029
0.032
0.029
0.022
0.073
0.042
0.053
0.87
Alcohol Content
%
Mass
0.32
0.33
18.05
18.04
18.20
17.85
17.84
29.62
47.81
84.41
%
Volume
0.31
0.32
17.48
17.47
17.61
17.00
17.08
28.77
46.85
83.21
Viscosity1
mm2/sec
0.555
0.5467
0.5922
0.6037
0.6001
0.648
0.6576
0.6947
0.8345
1.2206
Density2
g/mL
0.7601
0.7608
0.7659
0.7681
0.7672
0.7681
0.7699
0.7712
0.7781
0.7827
Acidity
%
Mass
0.0008
0.0008
0.0008
0.0012
0.0012
0.0008
0.0008
0.0012
0.0012
0.0031
Sample Information
Date
Prepared
8/22/2013
11/13/2013
10/21/2013
11/13/2013
11/13/2013
10/21/2013
11/13/2013
11/15/2013
11/15/2013
10/21/2013
Sample ID
54013-64-22
54013-109-14
54013-80-21
54013-108-21
54013-108-21
DUPLICATE
54013-81-21
54013-107-21
54013-111-21
54013-114-21
54013-82-2
1 Viscosity measurement was taken at 25°C
2 Density measurement was taken at 15.6°C
A3 QUALITY ASSURANCE/ QUALITY CONTROL
Quality assurance/quality control (QA/QC) procedures were performed in accordance
with the QMP3 for the AMS Center and the QAPP1 for this verification test. QA/QC procedures
and results are described in the following subsections.
A3.1 Data Collection Quality Control
The overall DQOs of this study measured physical and chemical properties of biofuels
and identified and quantified the applicable processes (e.g., mixing) affecting the performance of
UST LD systems on two scales: (1) bench-scale test set for the determination of select physical
and chemical properties of biofuels and BFW mixtures (no technologies were studied at this
scale); and (2) laboratory-scale test set for the identification and quantification of initial fuel and
water detection as well as water ingress (where applicable) affecting performance of UST LD
systems. Sample measurements followed standard analytical methods that have been published
and accepted by ASTM International, American National Standards Institute (ANSI), National
Association of Corrosion Engineers (NACE), or EPA. The QC procedures and measurement
A-7
-------
quality objectives (MQOs) for the methods utilized by ICFTL and Battelle Labs are described in
Table 6.
Table 6. Data Collection Quality Control (QC) Procedures and Measurement Quality
Objectives (MQO) for Analytical Methods
Method Designation: Method
Title
QC Procedures
MQOs
ASTMD4815: Standard Test
Method for Determination of
MTBE, ETBE, TAME, DIPE,
tertiary-Amyl Alcohol and Ci to
€4 Alcohols in Gasoline by Gas
Chromatography8
Annual multi-point
calibration curve and with
newly installed column
and continuing QC check
samples every 10
samples*
Calibration curve r2 > 0.99
QC Check Samples:
Good: PR<4 & TPI >1.2; PR>4 &
TPI>2.4
Fair: PR<4 & TPI between 0.8-1.2;
PR>4 & TPI between 1.6-2.4
Poor: PR<4 & TPI <0.8; PR>4 &
TPK1.6
ASTMD5501: Standard Test
Method for Determination of
Ethanol and Methanol Content in
Fuels Containing Greater than
20% Ethanol by Gas
Chromatography9
Annual multi-point
calibration curve and with
newly installed column
and continuing QC check
samples every 10
samples*
Calibration curve r2 > 0.99
QC Check Samples:
Good: PR<4 & TPI >1.2; PR>4 &
TPI>2.4
Fair: PR<4 & TPI between 0.8-1.2;
PR>4 & TPI between 1.6-2.4
Poor: PR<4 & TPI <0.8; PR>4 &
TPK1.6
ASTM D5501: Modified to
analyze Isobutanol
Annual multi-point
calibration curve and with
newly installed column
and continuing QC check
samples every 10
samples*
Calibration curve r2 > 0.99
QC Check Samples:
Good: PR<4 & TPI >1.2; PR>4 &
TPI>2.4
Fair: PR<4 & TPI between 0.8-1.2;
PR>4 & TPI between 1.6-2.4
Poor: PR<4 & TPI <0.8; PR>4 &
TPK1.6
ASTME203: Standard Test
Method for Water Using
Volumetric Karl Fischer
Titration7
QC check samples every
10 samples*
Good: PR<4 & TPI >1.2; PR>4 &
TPI>2.4
Fair: PR<4 & TPI between 0.8-1.2;
PR>4 & TPI between 1.6-2.4
Poor: PR<4 & TPI <0.8; PR>4 &
TPK1.6
ASTM D1613 Standard Test
Method for Acidity in Volatile
Solvents and Chemical
Intermediates Used in Paint,
Varnish, Lacquer and Related
Products10
QC check samples every
10 samples*
Good: PR<4 & TPI >1.2; PR>4 &
TPI>2.4
Fair: PR<4 & TPI between 0.8-1.2;
PR>4 & TPI between 1.6-2.4
Poor: PR<4 & TPI <0.8; PR>4 &
TPK1.6
A-8
-------
Table 6. Data Collection Quality Control (QC) Procedures and Measurement Quality
Objectives (MQO) for Analytical Methods (Continued)
Method Designation: Method
Title
ASTM D4052: Standard Test
Method for Density, Relative
Density, and API Gravity of
Liquids by Digital Density
Meter11
ASTMD287: Standard Test
Method for AP Gravity of Crude
Petroleum, and Petroleum
Products12
ASTM D2624: Electrical
Conductivity13
ASTMD445: Standard Test
Method for Kinematic Viscosity
of Transparent and Opaque
Liquids (and Calculation of
Dynamic Viscosity)14
QC Procedures
QC check samples every
10 samples*
Daily Check
Daily instrument check of
probe
QC check samples every
10 samples*
MQOs
Good: PR<4 & TPI >1.2; PR>4 &
TPI>2.4
Fair: PR<4 & TPI between 0.8-1.2;
PR>4 & TPI between 1.6-2.4
Poor: PR<4 & TPI <0.8; PR>4 &
TPK1.6
Two standards were used to check
hygrometer. The standards ranged in
densities from 0.7788 g/mL to 0.8083
g/mL.
Probe was calibrated as per
manufacturer's specifications
Good: PR<4 & TPI >1.2; PR>4 &
TPI>2.4
Fair: PR<4 & TPI between 0.8-1.2;
PR>4 & TPI between 1.6-2.4
Poor: PR<4 & TPI <0.8; PR>4 &
TPK1.6
*Assessment of QC data compared to repeatability and reproducibility outlined in ASTM Methods.
Precision Ration (PR) =test method reproducibility/ test method repeatability
Test Performance Index (TPI) =test method reproducibility/site precision
Site precision=2.77*standard deviation
A3.2 Audits
Three types of audits were performed during the verification test: a performance
evaluation audit (PEA) of the analytical methods, a technical systems audit (TSA) of the
verification test procedures, and a data quality audit (DQA). Audit procedures are described
further below.
A3.2.1 Performance Evaluation Audits
The accuracy of the analytical methods performed by ICFTL was evaluated in the PEA
by analyzing certified standards. For the low-level ethanol content determination method
D48158, SRM 2287- Reformulated Gasoline (10% Ethanol) was used. The isobutanol method
(ICFTL In-House Modified D5501) was verified using a Spectrum Quality Standard calibration
standard at 11.37% isobutanol. For the high-level ethanol content determination method D55019,
SRM 2900-Ethanol-Water Solution, (nominal 95.6%) was used. The results of the standards
were acceptable when within 10% of the target ethanol content. For water content determination
by method E2037, the NIST traceable SRM 2287 was used. The results of the water standard
were considered acceptable because the lab results fell within the SRM certification range,
A-9
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 10 of 184
however, it was outside the QAPP acceptance criteria of being within 10% of the target control
standard concentration. The analytical methods and their associated PEA material and
acceptance criteria are summarized in Table 7. The results from the PEA were sent to the EPA
Project Officer (PO) and EPA Quality Assurance Manager (QAM). The PEA report included the
raw data, performance evaluation certificate of analysis, calculations of the comparison to the
expected concentration, and a discussion of corrective action, if applicable. A summary of the
PEA results is presented in Table 8.
Table 7. Analytical Methods and PEA Materials
Method
ID
ASTM
D4052
ASTM
D445
ASTM
D55019
ASTM
D48158
Modified
ASTM
D5501
ASTM
E2037
Title
Standard Test Method for Density,
Relative Density, and API Gravity of
Liquids by Digital Density Meter
Standard Test Method for Kinematic
Viscosity of Transparent and Opaque
Liquids (and Calculation of Dynamic
Viscosity)
Standard Test Method for Determination
of Ethanol and Methanol Content in
Fuels Containing Greater than 20%
Ethanol by Gas Chromatography
Standard Test Method for Determination
of MTBE, ETBE, TAME, DIPE,
tertiary-Amyl Alcohol and Ci to CA
Alcohols in Gasoline by Gas
Chromatography
ICFTL In-House Isobutanol Method
Standard Test Method for Water Using
Volumetric Karl Fischer Titration
(Procedure §10)
PEA Material
Fluka Standard
N.10ISO17025/ISO
Guide 34
Fluka Standard
N.10ISO17025/ISO
Guide 34
NSIT
SRM 2900
NIST
SRM 2287
Spectrum Calibration
standard for
Isobutanol
NIST
SRM 2287
Acceptance Criteria
Within 10% of the
target concentration,
repeat analysis if out of
range
The water content
range specified by the
SRMof0.04±0.02must
be met
A-10
-------
Table 8. PEA Results for Analytical Methods
Date
Completed
4/3/2013
4/3/2013
4/3/2013
8/6/2013
8/6/2013
8/13/2013
Sample ID
53972-12-15
53972-12-10
53972-12-15
54013-44-19
54013-44-19
54013-45-16
Analytical
Method
D4052
D5501
D445
D4815
E203
Modified
D5501
Determination
Density
High Ethanol
Content
Viscosity
Low Ethanol
Content
Water Content
Isobutanol
Lab Result
0.78 14 unit less
at 15.6°C
94.28 % mass
1.2 mm2/sec at
27°C
11.05% mass
0.052 %mass
11.37%mass
RPD
0.33%
1.38%
2.36%
9.73%
30%*
5.01%
*The SRM water content certification range is 0.04 ± 0.02 (0.02 - 0.06). Not considered as a failure, because the lab
result falls within the SRM range.
A3.2.2 Technical System Audits
The Battelle QAM performed a one-day TSA of the bench-scale test set on May 1, 2013.
The purpose of this audit was to ensure that the tests were being performed in accordance with
the AMS Center QMP3 and the QAPP1. During the audit, the Battelle QAM reviewed
• Documentation for the preparation of the test blends and BFW mixtures and the
results of the EO analysis;
• Testing facility equipment (calibration, maintenance, and operation);
• Actual test procedures versus those specified or referenced in the QAPP; and
• Data acquisition and handling procedures, including observation of testing and
records (including custody forms).
The TSA was guided by a project-specific checklist based on the QAPP. It was
performed during the bench-scale testing because this was where many different steps of the
process were performed (sample preparation, shipment to the analytical laboratory, multiple data
points collected on one test blend, etc.).
A TSA report was prepared as a memo to the Testing Coordinator (TC) and the
completed checklist was attached. The Battelle AMS Center Manager and EPA PO were copied
on the memo. The TC responded to the audit. The Battelle QAM verified that all audit findings
and observations were addressed and that corrective actions were appropriately implemented. A
copy of the complete TSA report with corrective actions was provided to the EPA PO. At EPA's
discretion, EPA QA staff conducted an independent on-site TSA on November 20, 2013 during
the execution of the lab-scale testing. The TSA findings were communicated to technical staff at
A-ll
-------
the time of the audit and documented in a similar TSA report following the same documentation
and dissemination procedure.
A3.2.3 Data Quality Audit
The Battelle QAM, or designee, audited at least 25% of the sample results acquired in the
testing and 100% of the calibration and QC data per the QAPP requirements. A checklist based
on the QAPP guided the audit. An initial ADQ was conducted on the first batch of test data and
the PEA data on June 26 - July 1, 2013 to identify errors early in the data reduction process. The
first batch was defined as the testing and variable data generated over the first two weeks of
testing by the TC. The remaining data were audited September 26 - October 2, 2013 at the
completion of bench-scale testing after all data for that set of tests was posted on the project
Share Point site. A third ADQ was performed on December 30, 2013 - January 6, 2014 by the
Battelle QAM. A final ADQ of this document that traced the data from initial acquisition,
through reduction and statistical comparisons, to final presentation was conducted on February
28, 2014. It also confirmed reconciliation of the first two ADQs.
All formulae applied to the data were verified, and 25% of the calculations were checked.
Data for all testing were reviewed for calculation and transcription errors and data traceability.
An audit report was prepared as a memo to the TC after completion of each data audit; the
completed checklist was attached. The Battelle AMS Center Manager, EPA PO and EPA QAM
were copied on the memo. The TC responded to the audit. The Battelle QAM verified that all
audit findings and observations were addressed and that corrective actions were appropriately
implemented. A copy of the complete ADQ report with corrective actions was provided to the
EPA PO.
A-12
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 13 of 184
Table 9. Underground Storage Tank Leak Detection Stakeholder Committee
Last Name
Barbery
Bareta*
Baustian
Boucher
Bradley*
Brauksieck
Brevard
Chapin
Cochefski
Cornett
Courville
D'Alessandro
Dockery
Drack
Emmington
Fenton
Fisher
Flora
Folkers
Geyer
Gordji
Henderson
Hoffman
Indest
Johnson*
Jones
Juranty*
Keegan
Kubinsky
Lauen
Marston
McKernan
McMillan
Mills
Moore*
First Name
Andrea
Greg
James
Randy
Lamar
Russ
Danny
Tom
Peter
Ken
Jamie
Tom
Howard
Earle
Dave
Charles
Laura
Jerry
Joie
Wayne
Sam
Kevin
Brad
April
Curt
Bill
Mike
Kevin
Ed
Dorcee
Dan
John
Corey
Tony
Bill
Company
US EPA OUST
Engineering Consultant Bureau of Storage Tank Regulation
(Wisconsin)
Butamax
Franklin Fueling Systems
Tennessee Department of Environment and Conservation Division of
USTs
(New York)
AC'CENT Services, Inc.
Underwriters Laboratory (UL.)
Ryder Fuel Services
Veeder-Root
Southern Tank Testers, Inc.
OMNTEC Mfg., Inc.
Simmons
DirAction, LLC.
Veeder-Root
Hansa Consult of North America, LLC (HCNA)
UST Leak Prevention Unit (California)
JDF Consulting
NOV Fiber Glass Systems
Steel Tank Institute
SSG Associates, University of Mississippi
Kevin Henderson Consulting, LLC
Tanknology
Southern Tank Testers, Inc.
Alabama Department of Environmental Management (Alabama)
Warren Rogers Associates, Inc.
New Hampshire Department of Environmental Services Waste
Management Division
Tanknology, Inc.
Crompco, LLC
Williams & Company
Franklin Fueling Systems
US EPA
Ryder Fuel Services
OPW Fuel Management Systems
Utah Department of Environmental Quality
A-13
-------
Table 9. Underground Storage Tank Leak Detection Stakeholder Committee (Continued)
Last Name
Moore
Moureau
Muhanna*
Neil
Nelson
Parnell
Peters*
Poxson*
Purpora
Ramshaw
Reid
Renkes
Robbins*
Rollo*
Sabo
Scheib
Smith*
Thuemling
Toms
Wilcox
Wilcox
Young
First Name
Kristy
Marcel
Shaheer
Peter
Bill
Brian
Heather
Marcia
Steve
Chris
Kent
Bob
Helen
Peter
Lorraine
Jeff
Tim
George
Patrick
Craig
Ken
Greg
Company
Renewable Fuels Association (RFA)
Marcel Moreau Associates
Georgia Department of Natural Resources
OPW Fuel Management Systems
Franklin Fueling Systems
MAPCO Express, Inc.
Missouri Department of Natural Resources
Michigan Department of Environmental Quality
Protanic
Purpora Engineering
Veeder-Root
PEI
Connecticut Department of Environmental Protection
Delaware Natural Resources and Environmental Conservation
Franklin Fueling Systems
Gevo
US EPA OUST
Varec, Inc.
Varec, Inc.
Ken Wilcox Associates, Inc.
Ken Wilcox Associates, Inc.
Vaporless Mfg., Inc.
*Designates members of the National Work Group on Leak Detection Evaluation (NWGLDE)
A-14
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 15 of 184
A4 REFERENCES
1. Quality Assurance Project Plan for Biofuel Properties and Behavior Relevant to
Underground Storage Tank Leak Detection System Performance. U.S. Environmental
Technology Verification Program, Battelle, April 2013.
2. Quality Assurance Project Plan for Biofuel Properties and Behavior Relevant to
Underground Storage Tank Leak Detection System Performance Addendum. U.S.
Environmental Technology Verification Program, Battelle, November 2013.
3. Quality Management Plan for the ETV Advanced Monitoring Systems Center, Version 8.
U.S. Environmental Technology Verification Program, Battelle, April 2011.
4. ASTM, D4057-06: Standard Practice for Manual Sampling of Petroleum and Petroleum
Products. August 2011.
5. ASTM, D5854-96: Standard Practice for Mixing and Handling of Liquid Samples of
Petroleum and Petroleum Products. May 2010.
6. ASTM, D 7717-11: Standard Practice for Preparing Volumetric Blends of Denatured
Fuel Ethanol and Gasoline Blendstocks for Laboratory Analysis. August 2011.
7. ASTM, E 203-08: Standard Test Method for Water Using Volumetric Karl Fischer
Titration. November 2008.
8. ASTM, D4815-09: Standard Test Method for Determination ofMTBE, ETBE, TAME,
DIPE, tertiary-Amyl Alcohol and Ci to €4 Alcohols in Gasoline by Gas Chromatography.
November 2009.
9. ASTM, D5501-12: Standard Test Method for Determination of Ethanol andMethanol
Content in Fuels Containting Greater than 20% Ethanol by Gas Chromatography. April
2013.
10. ASTM, Dl 613-06: Standard Test Method for Acidity in Volatile Solvents and Chemical
Intermediates Ulsed in Paint, Varnish, Lacquer, and Related Products. July 2012.
11. ASTM, D4052-11 Standard Test Method for Density, Relative Density, and API Gravity
of Liquids by Digital Density Meter. 2011.
12. ASTM, D287-92: Standard Test Method for API Gravity of Crude Petroleum and
Petroleum Products (Hydrometer Method). June 2006.
13. ASTM, D2624-09: Standard Test Methods for Electrical Conductivity of Aviation and
Distillate Fuels. February 2010.
14. ASTM, D445-12: Standard Test Methodfor Kinematic Viscosity of Transparent and
Opaque Liquids (and Calculation of Dynamic Viscosity). May 2012.
A-15
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 16 of 184
Appendix B
Fuel Property Testing Methods and Data Results
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 1 of 184
Appendix B
Fuel Property Testing Methods and Data Results
Bl BENCH-SCALE TESTING
The bench-scale testing focused on determining several fundamental properties of biofuels and
BFW mixtures under typical conditions encountered during operation of UST LD systems. This
differentiated whether the range of ethanol blends had properties that behaved significantly different from
each other, thereby being the evidence that leak detection technologies may or may not function properly
when used in the different blends. Bench-scale testing was divided into four series of tests and followed
theQAPP1:
e) Intrinsic Properties of BFW Mixtures: The properties studied in the first series of bench-
scale tests are common to all biofuels and is referred to herein as intrinsic properties because
they belong to the biofuel due to its very nature. The intrinsic properties evaluated in the first
series of tests include acidity, density, electrical conductivity and viscosity. These are
intensive intrinsic properties (i.e., do not change with sample size) and were identified as
important factors that may affect the performance of UST LD systems while operating in
BFW mixtures.
f) Coefficient of Thermal Expansion: The second series determined the coefficient of thermal
expansion of different BFW mixtures within a temperature range that is typically experienced
in field applications of UST LD systems. The density of biofuels, like all materials, is
temperature dependent and the volume of a mass of biofuel changes with temperature in a
predictable (anticipated linear) fashion. In the field, temperature fluctuations cause expansion
and contraction of BFW mixtures which must be accurately predicted and accounted for by
UST LD systems.
g) Non-additive Volume Changes: The third series of tests determined the volume effect of
water addition on the test blends. When two polar solvents are combined (as in water and
ethanol in a biofuel) the resultant volume of the mixture is less than the additive volume of
the two components as water is accommodated into the ethanol polar structure. This
information is particularly applicable in the situation of water ingress into USTs containing
biofuels in that the ethanol in the gasoline will accommodate the water in the gasoline and if
the water is in high enough concentration, phase separation will occur.
B-l
-------
h) Interface Determination: The final series of bench-scale tests focused on the development of
a method to optically determine the phase separation of the different BFW mixtures. Once
above the saturation level (<1% [v/v]), water separates from an ethanol blend by pulling some
of the ethanol into a denser separated phase at the bottom of an UST. It is important that the
location and properties of these layers be able to be independently and objectively identified
including not only pure water and hydrocarbon phases, but also the colloidal mixed layers of
gasoline/ethanol and water/ethanol.
Each series of the bench-scale testing was executed separately and sequentially in a Battelle
laboratory in Columbus, Ohio under ambient laboratory conditions unless otherwise specified.
Laboratory temperature was measured with a glass thermometer at the beginning and end of each testing
day as well as monitored with a 3M Temperature Data Logger. For tests requiring strict temperature
limits, a New Brunswick Series 25 Incubator Shaker and a Lauda Proline Low Temperature Thermostat
was employed. Except when specific temperatures are required, all tests were carried out at ambient
laboratory temperature (approximately 15 to 20 °C). Class A volumetric glassware and calibrated micro-
pipettes (within the last 6 months) were used for all experiments and the accuracy of pipettes was
determined gravimetrically at the beginning of each test day when anticipated to be used that day.
Glassware was used as received, rinsed with EO and allowed to air dry overnight before next use. All
experiments were carried out in triplicate to facilitate statistical comparisons between BFW mixtures.
B2 TEST PROCEDURES
B2.1 Intrinsic Properties of BFW Mixtures
This first test set aims at determining the pertinent intrinsic properties of BFW mixtures at
different ethanol or isobutanol and water contents. After preparation (Appendix A), the BFW mixtures
were poured into a 250 mL graduated cylinder. Samples were taken from the middle of the cylinder using
a glass pipette and sent to ICFTL for measurement of acidity by ASTM D16132, density by ASTM
D40523, viscosity by ASTM D4454, and water and ethanol content by either ASTM E2035 (for water) and
ASTM D55016 or ASTM D48157 (for ethanol) depending on their anticipated water and ethanol contents.
Where appropriate, samples were analyzed for isobutanol concentration by a modified ASTM D55016.
After sampling, conductivity was measured by ASTM D26248 and density was measured by ASTM
D2879 directly in the graduated cylinder. Each intrinsic property was measured in triplicate on the same
sample.
B-2
-------
Some of the BFW mixtures had separated phases. In this case, the interest in intrinsic properties
is in the bulk fuel phase and as such, aliquots sent for analytical analysis were the bulk fuel samples.
Where possible, the dense phase (i.e., water-ethanol separated phase) was archived should the analysis of
this phase be performed. At this time, it has been determined to only analyze the fuel phase because of
the relevance to technology performance for LD, the potential non-availability of enough volume for the
analyses, and to minimize extraneous analytical costs. In some cases, such as with E30 BFWs with 2.5%
and 5.0% water, aliquots of sample from both phases were sent for analysis as the sample did not
homogenize easily.
B2.2 Coefficient of Thermal Expansion
In order to determine how temperature affects the volume of specific BFW mixtures, a series of
experiments was conducted in 10 mL-capacity glass graduated cylinders (±0.1 mL). At ambient
temperatures, 5 mL of zero water BFW mixture was added to individual 10 mL graduated cylinders and
the appropriate amount of water was added to each cylinder (Table 1) to represent BFW of different water
concentrations (0%, 0.25%, 0.5%, 2.5%, and 5.0% water) Each cylinder was capped with a ground-glass
stopper. Actual mass of BFW mixture was determined gravimetrically. The BFW mixtures were then
allowed to equilibrate for 60 minutes to 5.0°C, 10.0°C, 15.0°C, 20.0°C, 25.0°C and 30.0°C in a Lauda
Proline Low Temperature Thermostat. After each 60-minute equilibration time, the volume of the
graduated cylinder was recorded before it was returned to the thermostat.
Table 1. Volume of water added to each 10 mL graduated cylinder for Coefficient of
Thermal Expansion
BFW Sample Description
0% water
0.25% water
0.5% water
2.5% water
5.0% water
Volume of Water Added QiL)
0
12.5
25
125
250
The coefficient of thermal expansion was calculated using Equation 1:
1 tdV\
Equation 1
B-3
-------
where a is the coefficient of thermal expansion, V25 is the volume of the individual BFW mixture at 25°C
(normalization temperature) and (9V/9T) is the partial derivative (i.e., slope) of the volume vs.
temperature line as calculated by linear regression.
B2.3 Non-additive Volume Changes (Degree of Accommodation)
Because of the varying miscibility of gasoline, water and ethanol, it is expected that as an aliquot
of water is added to each of the test blends, the total volume change of the resulting BFW mixture would
be less than the volume of that aliquot, and the separated, dense phase would grow disproportionately to
the added volume of water. The relative total volume decrease is due to accommodation of polar water
molecules into the structure formed by the polar ethanol molecules (degree of accommodation).
This experiment aimed at quantifying this effect. Five (5) mL of each test blend (no water) was
added separately by pipette to 10 mL (±0.1 mL) glass-graduated cylinders; the actual mass of the test
blend was determined gravimetrically. The graduated cylinders were placed in the thermostat at 25 °C for
15 minutes for initial temperature equilibration. After equilibration, the cylinders were removed from the
thermostat and a dye solution consisting of water and McCormick Blue Food Dye (1:2,000 dilution) were
added in 250 (iL increments using a micro-pipette. The actual mass of added dye solution was
determined gravimetrically. After the addition of each 250 (iL increment of water, the graduated cylinder
was sealed with a ground glass stopper. The graduated cylinder was replaced to the thermostat for 5
minutes at 25°C, after which the total volume and the volume of the dense phase was measured. At the
time of volume measurement, a photograph of the cylinder was taken to qualitatively record the interface.
A total of 5 mL of dye solution was added in this way to each sample (total of twenty 250 (iL additions)
with measurement of volume change made after each increment.
The effect of fuel: ethanol ratio on relative volume decrease was determined by calculating the
following using Equation 2:
Equation 2
The parameter y is referred to as the degree of accommodation, AVm is the measured incremental change
in total volume with incremental dye solution addition and AVa is the incremental volume addition of dye
solution. In this way, y can be seen as the measure of the amount of ethanol accommodated within the
polar water structure which results in relative volume reduction with the addition of water to the test
blends. In practice, y is defined as the slope of the Vm vs. Va curve as calculated by linear regression.
B2.4 Interface Determination
B-4
-------
As water separates from pure gasoline, a well-defined interface is formed which can be visually
determined relatively easily and objectively; however, the interface becomes less defined when water
separates from an ethanol-gasoline mixture as the water can be absorbed by both the gasoline and ethanol
phases forming a hazy suspension. Gaining an understanding of the separated phase in different ethanol
blends is important for identifying and measuring water at the bottom of an UST. This last series of
bench-scale tests focused on establishing a method for determination of a water interface in different test
blends and mathematically defining the vertical position of the interface.
A sample of 70 mL of each test blend and 70 mL of dye solution consisting of water and
McCormick Blue Food Dye (1:2,000 dilution) were measured by glass volumetric pipette into three
individual 160 mL glass serum bottles (triplicate samples of each test blend/dye solution mixture). Serum
bottles were sealed with Teflon® septa and aluminum caps. The 160 mL serum bottles were agitated with
a New Brunswick Series 25 Incubator Shaker at 200 rotations per minute for 60 minutes to ensure
mixing. After the mixing period, the septa were pierced with a thin needle protruding to the bottom of
each of the serum bottles. The needles were equipped with a Luer-Lok fitting able to be attached to a 10
mL syringe. The serum bottles were left to rest in the incubator at 25 °C for 24 h to reach equilibrium.
After equilibration, each serum bottle septum was pierced with a second needle only to the headspace to
allow 10 mL of sample to be carefully extracted through the first needle using a 10 mL syringe. 10 mL
corresponds to approximately 1 cm liquid height which was subsequently measured to the nearest 0.1 cm.
The absorbance of the 10 mL sample was then measured at 630 nm using a Hach DR5000 UV-Vis
Spectrophotometer previously zeroed with EO. Following ASTM D745110 for mixing and measurement,
the cells were briefly and vigorously shaken to ensure homogeneity immediately before absorbance
measurements are taken. Triplicate measurements were taken and to be considered acceptable,
measurements must display a coefficient of variation of less than 10%.
This extraction and measurement procedure was repeated until the full contents of each serum
bottle have been removed (approximately 14 data points per serum bottle). In this way, the transition
from water to gasoline can be plotted using visible absorbance of the dye solution as a designation of
where the water was located in the sample. Each test blend followed the same procedure.
Table 2 summarizes the series of tests performed on the bench scale. Table 3 presents the data
collection QC assessments for the fuel properties being measured in the bench-scale testing.
B-5
-------
Table 2. Summary of the Bench-scale Test Set
Test Series
Intrinsic
Properties of
BFW Mixtures
Coefficient of
Thermal
Expansion
Non-Additive
Volume
Changes
Determination
of Interface
Description
Preparation of 35 different test
blends and BFW mixtures and
analysis of their intrinsic
properties including ethanol
concentration, water
concentration, acidity, density,
viscosity, and electrical
conductivity
Preparation of 35 different test
blends and BFW mixtures and
measurement of their volume at
different temperatures from 5.0 to
30.0 °C
Preparation of seven test blends
and measurement of volume
changes with known addition of
aqueous dye solution
Mixing 50% of the seven test
blends individually with 50%
aqueous dye solution and
measuring the height-dependent
absorbance of the resulting
mixture resulting in a height vs.
absorbance curve which can be
used as a designation of water
location
Precision
Requirements
• CV<15%for
measurements on
triplicate samples
• r2> 0.90 for volume
vs. temperature curve
• CV<15%for
measurements on
triplicate samples
• r2> 0.90 for volume
measured vs. volume
added curve
• CV< 15% for single
measurements on
triplicate samples
• CV < 10% for
triplicate
measurements of
optical absorbance on
the same sample
• CV< 25% for single,
depth-dependent
measurements on
triplicate samples of
optical absorbance
Independent
Variables
• Water
concentration
• Ethanol
concentration
• EO concentration
• Water
concentration
• Fthannl
concentration
• EO concentration
• Temperature
• Ethanol
concentration
• EO concentration
• Dye solution
• Ethanol
concentration
• EO concentration
#of
Replicates
3 each
3 each
3 each
3 each
B-6
-------
Table 3. Data Collection Quality Control Assessments of the Fuel Properties
Measured
Fuel Property
Ethanol
Concentration
Water
Concentration
Acidity
Density
Viscosity
Electrical
Conductivity
Absorbance
Temperature
(incubator)
Temperature
(water bath)
Method of
Assessment
ASTMD5501and
D4815
ASTME203
ASTMD16132
ASTM D2879
ASTM D4052
ASTM D4454
EMCEE Model 1152;
ASTMD26248
HachDRSOOOUV-
Vis
Spectrophotometer
Glass thermometer
Built-in resistance
probe
Frequency
Once per unique
BFW mixture,
once per unique
test blend and
once per
collection of EO
Once per unique
BFW mixture
during
determination of
intrinsic
properties
Zero instrument
between test
blend replicates
Once each at the
beginning and
end of each
testing day and
once during
testing
Immediately
after
temperature
equilibration and
every 30
minutes after
equilibration
Laboratory
ICTFL
ICTFL
ICFTL
Battelle
ICFTL
ICFTL
Battelle
Battelle
Battelle
Battelle
Acceptance
Criteria
RPD < 15%
between result
and target. Less
than 1% for EO
RPD < 15%
between result
and target. Less
than 0.1% for
EO
CV<15%for
triplicate
measurements
CV<10%for
triplicate
measurements
±l°Cfrom
target,
monitored with
an audible alarm
when out of
range
±0.1°Cfrom
target,
monitored and
logged with a
calibrated
electronic
thermometer
Corrective Action
Discard test blend or
BFW mixture and re-
prepare
Discard test blend or
BFW mixture and re-
prepare
First unacceptable
result: Re-test BFW
mixture. Second
unacceptable result:
Discard and re-prepare
BFW mixture and
retest. Third
unacceptable result:
trouble shoot the
instrumentation^
First unacceptable
result: Re-test samples.
Second unacceptable
result: trouble shoot the
instrumentation
Replace thermometer
First unacceptable
result: trouble shoot the
instrumentation.
Second unacceptable
result: record
temperature using
external thermometer
B3
STATISTICS FOR BENCH-SCALE TEST SETS
All BFW mixtures were prepared in triplicate and measurements made on each of the triplicate
BFW mixtures were carried out once. Statistics were calculated on each of the measurements as follows:
• Average: The average value (X) of the single measurements made on the triplicate BFW
mixtures was calculated using Equation 3 as follows:
B-7
-------
Equation 3
where X is the average value of n number of measurements, x; (i = 1,2,3)
• Standard Deviation: The standard deviation (SD) of a set of triplicate measurements made on
BFW mixtures was calculated using Equation 4 as follows:
SD =
3
iv,
Equation 4
where X and x; are defined above.
• Coefficient of Variation: The CV of a set of measurements is defined as the quotient of the
SD of that set of measurements and the average of that same set of measurements and was
calculated using Equation 5 as follows:
SD
C"=T
Equation 5
where CV is the coefficient of variation and SD and X are defined above.
• Relative Percent Difference: The RPD between a measured (or calculated) value and a target
value was calculated using Equation 6 as follows:
\X-T\
RPD=-
Equation 6
where RPD is the relative percent difference between a calculated mean, X and a target value, T.
• Coefficient of Determination: The coefficient of determination (r2) of several calculated
dependent variables with respect to their associated independent variables was calculated
according to Principles and Procedures of Statistics11 and the formulae are not repeated here.
B-8
-------
In all cases, r2 were calculated based on calculated average values of both measured
dependent and independent variables by Microsoft® Excel.
B4 PRECISION OF FUEL PROPERTY MEASUREMENTS
The precision requirements of the data collected in the Bench-scale testing are summarized in
Table 2 above and explained in more detail below.
B4.1 Intensive Properties: Acidity (pH), Viscosity, Density, Electrical Conductivity, and Optical
Absorbance
Measured triplicate values of acidity (i.e., pH), density, viscosity and electrical conductivity
measured as part of the intrinsic properties of BFW mixtures experiments were subjected to statistical
analysis. The average value, SD and CV were calculated and recorded separately for each set of
measured intrinsic properties. With respect to precision, for single measurements taken on triplicate
samples to be considered acceptable for reporting, the CV for each set of triplicate measurements of
acidity, density, viscosity, electrical conductivity and optical absorbance must be less than 15%.
The single depth-dependent optical absorbance measurements of samples collected during the
interface determination experiments were considered acceptable for reporting when triplicate
measurements on one test blend in three separate serum bottles display a CV less than 15%. No accuracy
criterion was established for depth-dependent measurements taken during the interface determination
experiment as this experiment aims at determining properties heretofore undefined.
B4.2 Extensive Properties: Volume Change
Single volume measurements taken on triplicate samples for the non-additive volume and
coefficient of thermal expansion experiments were subjected to statistical analysis. The average value,
SD and CV were calculated and recorded separately for each triplicate measurement of volume change.
With respect to precision, for single measurements taken on triplicate samples to be considered acceptable
for reporting, CV for each set of triplicate measurements of volume must be less than 15%.
B4.3 Calculated Properties: Coefficient of Thermal Expansion and Degree of Accommodation
The coefficient of thermal expansion (Equation 1) and degree of accommodation (Equation 2)
was calculated from the appropriate equations and results reported with appropriate significant figures. In
contrast, within the experimental parameters set forth, the slopes of volume vs. temperature curve (for
B-9
-------
coefficient of thermal expansion) and measured volume vs. added volume curve (for degree of
accommodation) are expected to be linear. Therefore, in order to be considered acceptable, the
coefficient of determination calculated from the average values (i.e., volume and temperature) must be
greater than 0.90.
B5
BENCH SCALE TESTING RESULTS
B5.1 Intrinsic Properties of BFW Mixtures
The density increases with higher concentrations of water as well as increasing concentrations of
ethanol (Figure 1). The density data measurements are summarized in Table 4. The conductivity results
follow a similar trend in that as ethanol and water concentration increase (Figure 2). Those data are
summarized in Table 5. Likewise, the viscosity of the fuel blends increase with increasing ethanol and
water content (Figure 3), and the data are summarized in Table 6. As seen in Figure 4, the acidity is more
variable that the other three parameters; however, in general, acidity increases as ethanol content
increases. These data are summarized in Table 7.
E30 was a difficult sample to handle as it would not completelyhomoginize, but would also not
completely separate into two phases. As such, the lower water content BFWs allowed for a composite
sample to be analyzed whereas a top and bottom phase layer sample were analyzed from the higher water
content BFWs (2.5 and % and 5.0% water). Table 8 displays all of the intrinsic properties for E30 BFWs.
(Some line colors are hard to distinguish.)
0.8200 -
0.8000 -
^ 0.7800
^0.7600
•5 0.7400
=
Q 0.7200
0.7000 -
0.6800
0.00% 0.25% 0.50% 2.50%
Percent Water
5.00%
•EO
•E10
•E15
•116
•E30
•E50
•E85
Figure 1. Contour plot of density (g/mL) for all BFW mixtures.
B-10
-------
Table 4. Summary of Density Results for the BFWs (g/mL)
1%
Water
0.0
0.25
0.5
2.5
5.0
EO
0.7222
0.7228
0.7227
0.7224
0.7230
E10
0.7617
0.7648
0.7649
0.7630
0.7624
E15
0.7643
0.7650
0.7663
0.7629
0.7618
116
0.7656
0.7658
0.7669
0.7669
0.7684
E30
0.7701
0.7708
0.7722
0.7753
0.7583
E50
0.7758
0.7766
0.7779
0.7849
0.7951
E85
0.7883
0.7927
0.7937
0.8014
0.8067
^s
Q.
u
S
•a
a
o
U
(Again, some line colors are hard to distinguish.)
100000000 -
10000000
1000000
100000
10000
1000
100
10 -
1
0.00% 0.25% 0.50% 2.50%
Percent Water
5.00%
•EO
•E10
•E15
•116
•E30
•E50
•E85
Figure 2. Contour plot of conductivity (pS/m) for all BFW mixtures.
Table 5. Summary of Conductivity Results for the BFWs (pS/m)
%
Water
0.0
0.25
0.5
2.5
5.0
EO
192
176
177
161
164
E10
12233
31900
73578
1444
1156
E15
104722
184644
382222
8833
12556
116
5163
7531
7200
6378
5028
E30
4321111
4683333
5238889
See
Table 8
E50
9204444
8762222
9498889
11910000
13914444
E85
8304444
7883333
8064444
9894444
11172222
B-ll
-------
(Again, some line colors are hard to distinguish.)
1.4 -
1.2
1 !'
S 0.8
•5 0.6 -
o
.a 0.4
0.2
0
0.00% 0.25% 0.50% 2.50% 5.00%
Percent Water
Figure 3. Contour plot of viscosity (mm2/s) for all BFW mixtures.
•EO
•E10
•E15
•116
•E30
•E50
•E85
Table 6. Summary of Viscosity Results for the BFWs (mm2/S)
%
Water
0.0
0.25
0.5
2.5
5.0
EO
0.555
0.562
0.558
0.561
0.562
E10
0.557
0.568
0.572
0.545
0.544
E15
0.582
0.593
0.596
0.586
0.567
116
0.659
0.656
0.657
0.660
0.666
E30
0.698
0.704
0.726
0.811
0.582
E50
0.863
0.865
0.873
0.970
1.147
E85
1.085
1.114
1.130
1.223
1.332
0.0018 -
0.0016 -
0.0014
0.0012
£ 0.001
£0.0008
3 0.0006
3, 0.0004
0.0002
0
(Again, some line colors are hard to distinguish.)
0.00% 0.25% 0.50%
Percent Water
2.50%
5.00%
•EO
•E10
•E15
•116
•E30
•E50
•E85
Figure 4. Contour plot of acidity (% mass) for all BFW mixtures.
B-12
-------
Table 7. Summary of Acidity Results for the BFWs (% mass)
%
Water
0.0
0.25
0.5
2.5
5.0
EO
0.0005
0.0007
0.0008
0.0004
0.0006
E10
0.0012
0.0012
0.0011
0.0011
0.0012
E15
0.0009
0.0011
0.0010
0.0009
0.0009
116
0.0011
0.0008
0.0010
0.0009
0.0008
E30
0.0012
0.0012
0.0013
0.0012
0.0008
E50
0.0016
0.0016
0.0015
0.0016
0.0017
E85
0.0015
0.0015
0.0016
0.0015
0.0015
Table 8. Intrinsic Properties of E30
Parameter
Density
(g/mL)a
Conductivity
((iS/cm)
Acidity
(% mass)
Viscosity
(mm2/S)
Water
(%v/v)
Water
Content
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
Average Value and (CV %)
Top Layer of BFW
b
b
b
0.792 (0.564)
0.766 (0.0970)
b
b
b
d
d
b
b
b
0.00120(18.7)
0.000800 (0.00)
b
b
b
0.811 (1.70)
0.582(0.213)
b
b
b
2.46(1.74)
Bottom Layer of
BFW
b
b
b
0.802(0.155)
0.841 (0.0720)
b
b
b
d
d
b
b
b
not enough sample
0.00230(15.5)
b
b
b
1.06(0.954)
1.66(2.18)
b
b
b
4.71 (7.69)
Composite
0.772 (0.00700)
0.773 (0.0810)
0.775(0.0510)
c
c
4320000 (8.47)
4680000 (6.09)
5230000 (6.50)
6410000(17.4)
200000 (0)
0.00120(0.00)
0.00120(0.00)
0.00130(15.4)
c
c
0.698 (0.0860)
0.704 (0.692)
0.726(1.02)
c
c
0.0465 (0.355)
0.334 (2.94)
0.611 (1.38)
c
B-13
-------
Table 8. Intrinsic Properties of E30 (Continued)
Parameter
Water
Ethanol
(%v/v)
Ethanol
(% mass)
Water
Content
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
Average Value and (CV %)
Top Layer of BFW
0.556(4.61)
b
b
b
3.10(1.62)
0.725 (4.55)
b
b
b
28.7 (3.57)
9.70(0.514)
b
b
b
28.8 (3.59)
10.0 (0.470)
Bottom Layer of
BFW
15.6(11.0)
b
b
b
5.86(7.80)
18.6(11.1)
b
b
b
44.0 (2.64)
67.2 (2.60)
b
b
b
43.5 (2.49)
63.5 (2.55)
Composite
c
0.0605 (0.355)
0.432 (2.82)
0.788(1.38)
c
c
28.3 (0.0124)
29.5 (0.688)
29.7 (0.753)
c
c
29.5(0.0112)
30.3 (0.608)
30.4 (0.771)
c
c
(a) Density values reported are from Iowa Fuel Testing Laboratory
(b) No separation between hydrocarbon and water layer was evident, so a composite sample was analyzed.
(c) A homogenous composite sample was not easily obtained, instead an aliquot from the top and bottom layer of the BFW
were analyzed.
(d) Conductivity was analyzed on the composite sample, despite not being able to obtain a homogenous sample
B5.2 Coefficient of Thermal Expansion
In order to determine how temperature affects the volume of specific BFW mixtures, the test
blends were plotted as volume (mL) against the temperature (°C), for each water content. The slopes of
the lines generated are reported as the coefficient of thermal expansion in Table 9. The associated r-
squared values are listed as well as the predicted volumes at 0°C (y-intercept). All blends appear to be
impacted by temperature similarly as all have a coefficient of thermal expansion near 0.0010 mL/°C.
B-14
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 4/17/2013
Version 1.0
Page 15 of 184
Table 9. Coefficient of Thermal Expansion Data
Test Blend
EO
E10
E15
116
E30
E50
E85
Water
Content
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
5.00%
0.00%
0.25%
0.50%
2.50%
Normalized at 25 °C
R2
0.9659
0.9711
0.9357
0.9282
0.9882
0.9641
0.8906
0.9546
0.9262
0.9379
0.9726
0.9429
0.9247
0.9282
0.9623
0.9849
0.9809
0.9946
0.9642
0.9730
0.9650
0.9948
0.9676
0.9658
0.9655
0.9909
0.8864
0.8992
0.9500
0.9964
0.9041
0.9854
0.9782
0.9625
Coefficient of Thermal
Expansion (mL/ °C) (slope)
0.0010
0.0011
0.0009
0.0090
0.0010
0.0013
0.0011
0.0012
0.0010
0.0009
0.0011
0.0012
0.0012
0.0012
0.0012
0.0011
0.0011
0.0011
0.0010
0.0011
0.0012
0.0011
0.0010
0.0010
0.0009
0.0009
0.0010
0.0011
0.0011
0.0010
0.0009
0.0010
0.0010
0.0011
Predicted Volume at 0°C
(y-intercept)
0.9748
0.9716
0.9746
0.9726
0.9750
0.9964
0.9715
0.9735
0.9725
0.9759
0.9749
0.9691
0.9706
0.9713
0.9739
0.9728
0.9743
0.9716
0.9744
0.9737
0.9673
0.9730
0.9736
0.9754
0.9797
0.9756
0.9792
0.9774
0.9709
0.9752
0.9730
0.9720
0.9782
0.9745
B-15
-------
Test Blend
Water
Content
5.00%
Normalized at 25 °C
R2
0.9628
Coefficient of Thermal
Expansion (mL/ °C) (slope)
0.0011
Predicted Volume at 0°C
(y-intercept)
0.9719
B5.3 Non-additive Volume Changes
Table 10 shows as the test blends increase in ethanol content, the amount of ethanol
accommodated within the polar water structure increases which results in a relative volume reduction
upon addition of water. Similarly, as the ethanol content of the test blends increase, the growth of the
dense phase occurs at a greater rate. 116 test blend behaved similarly to E15. The degree of
accommodation was calculated by determining the slope of the lines plotted as the incremental water
volume added (|iL) by total volume measured (mL) for each test blend. The growth of the total volume
was calculated by determining the slope of the lines created by plotting the measured total volume (mL)
by the expected total volume (mL) for each test blend. The growth of the dense phase was calculated by
determining the slope of the lines created by plotting the measured dense phase volume (mL) by the
expected dense phase volume (mL) for each test blend.
Table 10. Degree of Accommodation Summary for all Test Blends
Test Blend
EO
E10
E15
116
E30
E50
E85
Growth of Total Volume
(Slope of A measured total
volume/A expected total
volume)
0.9557
0.9953
0.9915
1.0039
0.9665
0.9838
0.9510
Growth of Dense Phase
(Slope of A measured dense
volume/A expected total
volume)
1.1042
1.1867
1.1424
1.1583
1.1172
1.2736
1.9470
Furthermore, the photo in Figure 5 provides a visual representation of the un-proportional growth
of the measured dense phase to what would be expected if there was no ethanol accommodation within
the polar water structure. The photo was taken after the last water addition during the Non-Additive
Volume Experiment for E85. If there was no accommodation, the dense, water phase would measure a
volume of 5 mL, however, due to the accommodation, the volume of the dense phase is around 9 mL.
B-16
-------
Figure 5. Photo taken during Non-Additive Volume Experiment for ESS. The test was completed in
triplicate. This particular photo occurred after the last water addition. The water was dyed with
blue food coloring.
B5.4 Interface Determination
For each replicate of each test blend, one serum bottle was prepared with 70 mL of test blend and
70 mL of water dyed with blue food coloring (Figure 7). A needle was inserted to draw out ten mL of
sample from the bottom into individual sample cells for optical absorbance analysis (Figures 8 and 9).
In order to make comparisons across test blends, all data was normalized to the original height of
the fluid in the serum bottle. Five different parameters were calculated for each test blend: (1) onset of
interface, (2) location of interface, (3) supervention of interface, (4) thickness of interface, and (5)
intensity of interface. Figure 6 is provided as an example to how these parameters were measured. The
onset of the interface is intended to be the point at which the optical absorbance begins to increase and is
measured in centimeters. The location of the interface is the height (cm) at which the peak occurred. The
supervention of the interface is the height (cm) at which the optical absorbance plateaus. The thickness of
the interface is how wide (cm) the peak is between the onset and supervention of the interface. Lastly, the
intensity of the interface is the change in optical absorbance (abs) between the peak and supervention of
the interface.
B-17
-------
0.2
0.4 0.6 0.8
Midpoint of Normalized Height (cm)
1.2
Figure 6. An example of E50 test blend showing how the interface determination data was
calculated
Table 11 shows as the concentration of ethanol increases that the onset, location, and
supervention of the interface decrease in height. This observation is further supported in Figures 8 and 9.
Figure 8 represents one replicate of E10 and Figure 9 represents one replicate of E85. Figure 8 (E10)
shows that the dense, water phase is evident in the vials only until draw #8 which corresponds to a height
of about 0.630 cm whereas in Figure 9 (E85), the water phase is evident until much later, in draw #13
which corresponds to a height of about 0.220 cm. The height values were measured by affixing a ruler to
the side of the serum bottle and measuring to the nearest tenth of a centimeter the height of the fluid after
every draw. The thickness of the interface is similar for all test blends and ranges from 0.08 cm to 0.190
cm. The intensity of the interface increases from 1.00 to 19.00 abs in EO to E50, then the intensity drops
to 7 abs in E85. 116 behaves similarly to E15.
B-18
-------
Table 11. Interface Determination Summary Table(a)
Test Blend
EO
E10
E15
116
E30
E50
E85
Onset of
Interface (cm)
0.504
0.470
0.444
0.522
0.369
0.292
0.047
Location of
Interface (cm)
0.670
0.630
0.520
0.522
0.450
0.380
0.220
Supervention of
Interface (cm)
0.730
0.760
0.580
0.670
0.580
0.515
0.310
Thickness of
Interface (cm)
0.090
0.185
0.080
0.100
0.170
0.190
0.140
Intensity of
Interface (abs)
1.00
2.50
3.00
4.50
11.00
19.00
7.00
(a) All heights were normalized to the original height
Figure 7. Photo shows one serum bottle from the Interface Determination Experiment with
116.
•to **«*
Figure 8. One replicate from E10 Interface Experiment.
B-19
-------
Figure 9. One replicate from ESS Interface Experiment
B-20
-------
B6 REFERENCES
1. Quality Assurance Project Plan for Biofuel Properties and Behavior Relevant to Underground
Storage Tank Leak Detection System Performance Addendum. U.S. Environmental Technology
Verification Program, Battelle, November 2013.
2. ASTM, Dl 613-06: Standard Test Method for Acidity in Volatile Solvents and Chemical
Intermediates Ulsed in Paint, Varnish, Lacquer, and Related Products. July 2012.
3. ASTM, D4052-11 Standard Test Method for Density, Relative Density, and API Gravity of
Liquids by Digital Density Meter. 2011.
4. ASTM, D445-12: Standard Test Method for Kinematic Viscosity of Transparent and Opaque
Liquids (and Calculation of Dynamic Viscosity). May 2012.
5. ASTM, E 203-08: Standard Test Method for Water Using Volumetric Karl Fischer Titration.
November 2008.
6. ASTM, D5501-12: Standard Test Method for Determination ofEthanol andMethanol Content
in Fuels Containting Greater than 20%Ethanol by Gas Chromatography. April 2013.
7. ASTM, D4815-09: Standard Test Method for Determination ofMTBE, ETBE, TAME, DIPE,
tertiary-Amyl Alcohol and Ci to C# Alcohols in Gasoline by Gas Chromatography. November
2009.
8. ASTM, D2624-09: Standard Test Methods for Electrical Conductivity of Aviation and Distillate
Fuels. February 2010.
9. ASTM, D287-92: Standard Test Method for API Gravity of Crude Petroleum and Petroleum
Products (HydrometerMethod). June 2006.
10. ASTM, D7451-08a: Standard Test Methods for Water Seperation Propoerties of Light and
Middle Distillate, and Compression and Spark Ignition Fuels. January 2009.
11. Steel, R.G.D.a.T., J. H., Principles and Procedures of Statistics. 1960, New York: McGraw-Hill.
B-21
-------
Appendix C
UST LD Operating Principle Testing Methods and Data Results
-------
Appendix C
UST LD Operating Principle Testing Methods and Data Results
Cl LABORATORY SCALE TESTING
This section describes the materials, methods and data collection procedures for the evaluation of
operating principles central to underground storage tank leak detection (UST LD) systems in alcohol-
blended fuels. The methods were adaptations of previously established standard test procedures.1'2 These
procedures have been adapted to incorporate testing with alcohol-blended fuels. The purpose of the
laboratory-scale testing was to evaluate a select number of operating principles of UST LD technologies
in a small laboratory scale. The specific focus was to determine various performance parameters of those
operating principles in detecting the presence of fuel and detecting water ingress in four different alcohol-
blended fuels (i.e., ethanol and isobutanol). Described herein are the operating principles tested, the
laboratory scale setup in which operating principles were evaluated, the specific test procedures, and the
data to be collected. Also included is a description of how these data were reduced followed by the
results.
In reading and applying this document, it is important to distinguish the difference between the
terms technology, technology category and sensor.
• A technology is a specific product marketed by a vendor.
• A technology category is a group of technologies whose operation depends on a common
operating principle (e.g., automatic tank gauges).
• A sensor is the physical means for implementation of a specific operating principle within a
technology.
It was not the intent of the tests described herein to evaluate the ability of a specific technology or
technology category to perform in alcohol-blended fuel systems. Rather, these tests evaluated specific
operating principles for LD and water ingress detection in alcohol-blended fuels by testing sensors based
on those principles in a laboratory.
C2 SENSOR SELECTION
This evaluation focused on the appropriateness and effectiveness of the sensor operating
principles. For this reason, three technologies were selected for evaluation of five operating principles.
For this evaluation, sensors were selected:
C-l
-------
• To represent a range of operating principles (conductance and capacitance, optical principles,
and float switches with a hydrocarbon polymer sensor) and technology vendors;
• To represent a range of intended operating conditions (i.e., liquid in-tank, interstitial); and
• To use testing resources wisely with the cost appropriateness of the various sensors.
A review of candidate sensors for evaluation was conducted through an internet search and
follow-up conversations with sensor suppliers. The results of this review were incorporated into a
decision matrix provided to U.S. Environmental Protection Agency's (EPA's) Office of Underground
Storage Tanks (OUST). Further conversations were held with EPA OUST and other stakeholders
regarding the sensors selected and the sensor selection matrix approach. These conversations resulted in
the selection of the three technologies for evaluation. Table 1 lists the operating principle(s), the
dimensions and types of sensors incorporated into each technology tested. For the purposes of this
testing, these sensors served as surrogate testing technologies; i.e., operability determinations for each
sensor were extrapolated to serve as an evaluation for the operating principles on which they are based.
For this reason, this document will refer to technologies by their operating principles as shown in Table 1.
The technologies and their operating principles are described in more detail in the following sections.
Table 1. Technologies and Associated Sensors Used for Evaluation of Operating Principles
Sensor Operating Principle(s)
(Sensor Identifier)
Interstitial Optical Sensor
(Optical Sensor)
Magnetic Float Switch and
Fuel-Sensitive Polymer Sensor
(FS/FSP)
Capacitance and Conductance Sensor
(Complex impedance)
(C/C Sensor)
Dimensions
4.3 in. L x 1.5 in.
W x 0.5 in. H
2.5 in. D x 8.86
in. H
2 in. D x
12 in. H
Sensor Type
Qualitative
Detects liquid
(non-discriminating)
Qualitative
Detects hydrocarbons and liquid
(somewhat discriminating)
Quantitative
Detects and quantifies
hydrocarbons and water
(discriminating)
C2.1 Interstitial Optical Sensor (Optical Sensor)
The Optical Sensor uses solid-state liquid level sensing technology to detect liquid in the
interstitial space of the tank. A schematic of the Optical Sensor is presented in Figure 1 along with its
intended installation configuration and dimensions. The operating principle of this sensor is optical, in
which changes in refraction of light are detected based on the medium through which the light passes.
When liquid ingresses into an interstitial space, the refractive index of that interstitial space changes based
on the differences in refractive index between air (dry condition) and liquid (wet condition indicative of a
C-2
-------
leak). The refraction of light passing through the interstitial space is detected by the sensor and an alarm
condition is triggered. Potential issues for use in ethanol-blended fuel systems include sensitivity of the
operating principle to detect changes in the refractive index of blended fuels. This sensor has been
specifically developed for use in unleaded gasoline containing up to 85% ethanol. Unlike earlier versions
of this sensor tested in low-ethanol blended gasoline, the Optical Sensor does not discriminate between
hydrocarbon and water and therefore contact of the sensor with liquids will trigger an alarm.
C2.2 Magnetic Float Switch and Fuel Sensitive Polymer (FS/FSP)
The FS/FSP sensor is used to monitor for the presence of liquid hydrocarbons (fuel product) in
dispenser sumps. A schematic of the FS/FSP is presented as Figure 2. This sensor combines two
operating principles: magnetic float switch and hydrocarbon-sensitive polymer. The sensor has an upper
and lower liquid float for liquid detection as well as a conductive polymer strip that reacts specifically
with liquid hydrocarbons. The environmental data are transmitted to an automatic tank gauge console
where data can be collected in electronic format. Specifically, the FS/FSP transmits when liquid is
detected by means of the lower liquid float, when hydrocarbons are present by means of the polymer
strip, and when a high liquid level condition is present by means of the top liquid float. In this way
FS/FSP is able to detect hydrocarbons along the polymer strip as well as floating on top of an aqueous
layer. A potential issue for use in alcohol-blended fuel systems is the specificity of the hydrocarbon
polymer in detecting diluted hydrocarbons mixed with alcohols.
C-3
-------
Cord grip
—) 1/2" rigid conduit
-/(to Console)
— Weatherproof junction box
4.3" (L) 1-5" (W) 0.5" (H)
Sensor Switch
_-- Fiberglass tank
Figure 1. Optical Sensor
T
8.86 inch (225 mm)
Length of detection
Polymer
Strip
Liquid Float |
7.5 inch
(190.5mm)
High level
| Liquid Float
1.0 inch
(25.4 mm)
Low level
.2.5 inch (63.5 mm)
Max. dia.
Figure 2. Magnetic Float Switch and Fuel-Sensitive Polymer (FS/FSP)
C2.3 Capacitance and Conductance (Complex Impedance) (C/C)
The C/C Sensor is used primarily to determine the liquid level, to determine the vertical
fuel/water profile, and to detect ingress of water. A photograph of the C/C Sensor is shown in Figure 3.
C-4
-------
The C/C Sensor operates under the complex impedance principle which combines two operating
principles: electrical conductivity and capacitance. As the composition of the liquid between two series
of parallel plates changes, the liquid's properties of electrical conductivity and dielectric constant
(measured by capacitance) also change. These properties are combined to determine the complex
impedance of the liquid. After laboratory calibration, the water content, fuel content and alcohol content
of the liquid can be determined at various heights along the sensor. Challenges for use in alcohol-blended
fuels include specificity, accuracy, and precision of the operating principle to detect changes in complex
impedance in alcohol-blended fuels.
Figure 3. Capacitance and Conductance (Complex Impedance) (C/C)
C3
TEST SETUP
All sensors were evaluated within clear glass containers with a sufficiently large inner diameter to
accommodate the sensors without being excessively wide. The FS/FSP and C/C Sensor were tested in a
graduated cylinder and the Optical Sensor was tested in a 4-L beaker. A ruler, graduated in millimeters,
was affixed to the outside of the test containers to monitor the liquid rise height with more resolution
during the testing. An explosion-proof pump was used for the alcohol blend ingress and a peristaltic
pump delivered water into the test chamber. The fuel pump used tubing that is compatible with fuel. The
tubing was secured in place so the liquids flowed along the side of the container to the bottom without
C-5
-------
touching the technology. The fuel and water ingress rates were between 13 and 100 milliliter per minute
(mL/min) to achieve a height increase rate of approximately 5 mm/min. The rate of height increase was
calculated by taking into account the volume displacement of the technology in the test chamber. Once
the technology and ingress lines were situated in the test chamber, Parafilm® was used to cover the top of
the chamber to minimize volatilization.
Before initiation of testing, the sensor was inserted through the top of the test chamber. The
sensor configuration with respect to the test chamber (e.g., suspended, vertically resting on the bottom of
the test chamber, horizontally resting on the bottom of the test chamber) was in concert with requirements
of the vendor-supplied literature and as close to intended field-operating configuration as possible. All
sensors were operated in accordance with vendor-supplied operations manuals and guidance including
wiring, data collection and maintenance. The Optical Sensor testing was performed in a dark
environment by taking measures to minimize light as much as possible without compromising safety.
The test chamber was wrapped and the lighting in the lab was minimized.
C4 TEST PROCEDURES
The tests were designed to simulate ingress of water or alcohol-blended fuel into a dry
environment and where applicable, water ingress into an alcohol-blended fuel. For each sensor,
groundwater and four different alcohol-blended fuels (referred to as test blends from this point forward)
will be used during testing: 0% ethanol v/v (EO), 15% ethanol v/v (E15), 85% ethanol v/v (E85) and 16%
isobutanol v/v (116). The FS/FSP sensor was also tested in 30% ethanol v/v (E30) and 50% ethanol v/v
(E50). Test blends were prepared as stated in the original QAPP in 4-L or 2-L batches (Section Bl.l).3
Groundwater used for this testing was collected from the tap in Battelie's Environmental
Treatability Laboratory. The tap was opened and flushed for at least 5 minutes before the groundwater
was collected. The groundwater was collected in a 5-gallon container and a sub-sample was measured for
pH, conductivity, and oxidation/reduction potential. After collection, groundwater was poured from the
container into a 2-L graduated cylinder (±20 mL) as needed for the water ingress detection test. A
peristaltic pump and associated tubing was dedicated for the water ingress test. The water was pumped
into the test chamber at a rate of 24.5 mL/min for FS/FSP, 37.0 mL/min for Optical Sensor and 21.4
mL/min for C/C for the initial test blend detection tests. For the water ingress testing of the C/C sensor,
water was pumped at a rate of 13.9 mL/min.
The three technologies have different test procedures due to their specific abilities for detection
and discrimination. Tests conducted were dependent on the abilities of the sensor. Table 2 presents the
test matrix including the test blend, number of replicates, and tests performed. The number of replicates
C-6
-------
was determined using a power analysis which provides a 95% probability of detection in gasoline with
83% power. The Optical Sensor was tested to detect liquid without discriminating between test blend and
water (Initial Water/Test Blend Detection Test). The FS/FSP Sensor is somewhat discriminating as it has
the float switch ability to detect liquid and the polymer strip ability to detect hydrocarbons (Initial
Water/Test Blend Detection Test). There is a second float switch sensor at the top of the technology that
has the same ability as the bottom sensor, so the top float switch was actuated with fuel height for only
one of the replicates (High Detection with Water 1 Replicate Test). The C/C Sensor discriminates
between the test blend and water. Therefore, the initial liquid was introduced for detection (Initial
Water/Test Blend Detection Test), and then the technology was submerged to half of its height in test
blend and thereafter, water was allowed to ingress for a water detection test (Water Ingress Detection).
Table 2. Test Matrix for Lab-Scale Testing
Technology
Optical
FS/FSP
C/C Sensor
Test Blend
Water
EO
E15
E85
116
Water
EO
E15
E30
E50
E85
116
Water
EO
E15
E85
116
Replicates
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
Tests
Initial Water Detection
Initial Test Blend Detection
Initial Test Blend Detection
Initial Test Blend Detection
Initial Test Blend Detection
Initial Water Detection
High Detection with Water 1 rep
Initial Test Blend Detection
High Detection with Water 1 rep
Initial Test Blend Detection
High Detection with Water 1 rep
Initial Test Blend Detection
High Detection with Water 1 rep
Initial Test Blend Detection
High Detection with Water 1 rep
Initial Test Blend Detection
High Detection with Water 1 rep
Initial Test Blend Detection
High Detection with Water 1 rep
Initial Water Detection
Initial Test Blend Detection
Water Ingress Detection
Initial Test Blend Detection
Water Ingress Detection
Initial Test Blend Detection
Water Ingress Detection
Initial Test Blend Detection
Water Ingress Detection
During testing, liquids (test blends and water) was pumped to the test chamber using an
appropriate peristaltic pump from a 2 L (±20 mL) graduated cylinder reservoir. The reservoir was sealed
C-7
-------
with Parafilm® with a hole in the center for the pump tubing. The graduated cylinder was used to
periodically monitor the cumulative liquid volume pumped in the chamber during testing. Monitoring the
cumulative liquid volume pumped ensured accurate and constant flow rates to the test chamber and also
allowed for calculation of liquid height rate within the chamber.
At the completion of the tests, the technology and the liquid were removed from the test chamber.
The liquid volume without the technology was measured and then transferred into an approved waste
container. The technology was cleaned following the vendor-stated recovery procedure and monitored
for recovery time. The FS/FSP Sensor is the only sensor that required a recovery time. The test chamber
was rinsed with deionized water and then acetone before being left to dry in the ventilated room. Specific
details of the tests are described in the sections below.
C4.1 Initial Water/Test Blend Detection Test
The efficacy of each operating principle to detect groundwater and the test blends into the empty
test chamber was determined by the initial water/test blend detection test. After the sensor has been
placed inside the empty test chamber and activated for data collection as per the manufacturer
instructions, the output was monitored for a minimum of 30 minutes as a blank test to establish the
baseline signal. The specified liquid was pumped from the graduated cylinder into the test chamber
between 19.2 and 98.5 mL/min for the 2-L graduated cylinder and 4-L beaker, respectively, which
corresponds to an empty-chamber fuel height increase of approximately 5 mm/min.
It should be noted that each sensor has different dimensions and occupies a different volume
within the test chamber. In all tests, the actual liquid height was higher than that of an empty test
chamber due to the volume displaced by the sensor. Therefore, the actual liquid height was determined
through observation of the graduations on the side of the test chamber and by calculation after the testing
was complete.
Because of the difference in dimensions of each sensor and locations of sensing elements,
different amounts of fuel was pumped into the test chamber depending on the sensor tested. In all cases
the amount of fuel pumped into the system was sufficient to activate the appropriate part of the specific
technology being tested. Once the sensor activated, the initial detection test was complete. If the sensor
did not activate, the liquid height was brought to at least 20% higher than the vendor-stated actuation
height and the pump was turned off. A 60-minute wait time elapsed before the test was aborted.
-------
C4.2 High Detection
For the FS/FSP sensor, a second float switch is located at the top of the technology. It was tested
with one replicate by allowing the liquid to ingress to activation height using the same flow rate and
procedure explained above for the initial detection tests.
C4.3 Water Ingress Detection
The water ingress detection test was performed with using the C/C Sensor only. It was half
submerged in the test blend at the beginning of the test and then groundwater was allowed to ingress into
the test chamber until the sensor detected its presence. Once the sensor activated, the water ingress
detection test was complete. If the sensor would not have activated for every water ingress detection test,
the water height would have been brought to 20% higher than the vendor-stated actuation height and the
pump turned off. A 60-minute wait time would have elapsed before the test was aborted.
C4.4 Recovery Time
After the end of the test the pump was shut off and the technology removed from the chamber.
The vendor-stated recovery procedure was followed for each technology and monitored for recovery
time. The FS/FSP Sensor is the only sensor that required a recovery time. The other two sensors had
immediate recovery once removed from the liquid.
C5 SENSOR DATA AND EVALUATION METRICS
As each test proceeded, different environmental conditions prevailed within the test chamber. It
was the goal of the test to determine the operability of each sensor to produce the correct sensor output
depending on liquid present. Each sensor has different capabilities and therefore had different data
outputs. The performance parameters and evaluation metrics are the means of determining the operability
of each sensor; these are described in Table 3.
C-9
-------
Table 3. Performance Parameters
Performance Parameter
Average Detection Time
Evaluation Metric
Difference between actuation time and
test start times
Data recorded
Test start time and actuation time
calculated for each liquid
Average Recovery Time
Average of difference between
recovery and test end times
Test end time and recovery time
calculated for each liquid
Liquid Activation Height
Average activation height and standard
deviation
Liquid height level at activation,
calculated for each liquid
Specificity
% Specificity
Liquid height level at activation,
calculated for each liquid
Accuracy
(qualitative only)
Relative % Accuracy
Liquid height level at activation,
calculated for each liquid
Accuracy
(quantitative only)
% Accuracy
Liquid height level at activation,
calculated for each liquid
Precision
(quantitative only)
% Coefficient of Variation
Liquid height level at activation,
calculated for each liquid
C5.1 Liquid Detection Time and Recovery Time
Detection time was evaluated for all three sensors. During the initial fuel/water detection tests,
test blends of different alcohol concentrations and groundwater were pumped into an empty test chamber.
All of the sensors were expected to be able to detect the presence of the liquid and differentiate from the
empty condition and the liquid present condition. Because of the different configurations of the sensors,
the presence of fuel and water will be detected at different times (heights) after fuel pumping begins. The
elapsed time between the test start time and when the detector responded was the detection time for the
initial water/test blend detection test.
During the water ingress test, groundwater was pumped into the test chamber that had the test
blend filled at 50% height at the beginning of the test. Due to operating principles, only the C/C sensor
was expected to be able to differentiate the water absent and water present conditions in the test blend.
The elapsed time between the start time and when the detector responded was the detection time for the
water ingress detection test.
The recovery time was recorded from the FS/FSP Sensor console output when it ceased to be in
alarm mode. The elapsed time between the test end time and when the detector was no longer alarming
was the recovery time.
C-10
-------
C5.2 Average Detection Time and Average Recovery Time
The liquid detection time and the recovery times were reported as the average (x) and the
standard deviation (S) of the observed values for each liquid. They were calculated following Equations 3
and 4 from the original QAPP, respectively.
C5.3 Specificity
The percent (%) specificity was calculated using the following equation for each of the liquid
individually as follows:
/ x \
Specificity, % = 100x1 —
VX£/
x = mean of observed values, cm
xt = the theoretical value, cm
C5.4 Accuracy (Qualitative Sensors Only)
Accuracy for the qualitative detectors was determined by calculating percent accuracy of
replicates as follows
Accuracy, % = 100 X f-J
r = the number of positive responses
n = the number of tests for a particular liquid
C5.5 Relative Percent Accuracy (Quantitative Sensors Only)
Accuracy in measuring the liquid level was computed for each measurement made for the water
ingress detection test replicates by the following equation:
\M-D\
Accuracy, % = * 100
M
M = Measured liquid level, mm
D = Detected liquid level, mm
C-ll
-------
C5.6 Precision (Quantitative Sensors Only)
Precision was calculated as the percent coefficient of variation (%CV) for quantitative sensors only as
follows:
%CV = 100x -
S = standard deviation of n values, cm
x= mean of observed values, cm
C6 TECHNOLOGY RESULTS
Using the above performance parameters the data collected are summarized below according to
each operating principle.
C6.1 Optical Sensor Performance
The optical sensor tested is an interstitial monitoring device which is used on secondarily
contained tanks and piping. This interstitial monitor performs by utilizing a refractive index and can be
performed continuously or intermittently, and no other parameters must be monitored to adjust the
observations. Only qualitative leak determinations are possible as the sensor is not able to discriminate
between water and hydrocarbons. The sensor is expected to alarm in the presence of liquid which was
confirmed during testing. The sensor was effective at distinguishing when liquid was present regardless
of the ethanol concentration and showed an accuracy rate of 100% for all blends (Table 4). The recovery
time for the optical sensor was instantaneous upon removal from the fluid present condition for all blends
(Table 4).
Table 4. Optical Sensor Performance Summary (n=10)
Performance Parameter
Average Detection Time (hh:mm:ss)
Average Recovery Time (hh:mm:ss)
Average Activation Height (mm)
Activation Height Standard Deviation (mm)
Specificity (%)(b)
Relative Accuracy (%)
Test Blends
EO
0:01:09
0:00:03
4.9
3.1
95.1%
100%
E15
0:01:25
0:00:02
7.1
1.8
139%
100%
116
0:00:58
0:00:02
4.5
2.1
87.3%
100%
E85
0:01:21
0:00:02
7.1
1.7
139%
100%
Water
0:04:49
0:00:03
9.9
0.6
193%
100%
(a) Water was ingressed at half the flow rate of product due to limitations of the water pump
(b) Source of theoretical value (<0.2 inch) is from NWGLDE website
C-12
-------
C6.2 Float Switch Performance
The FS/FSP sensor was composed of two float switches, one is on the bottom on the sensor and is
described as the bottom float switch (Table 5), the second float switch is higher on the sensor and is
referred to as the top float switch (Table 6). Both float switches operate on the same principle where the
buoyancy of float allows the signal generated to coincide with the top of the liquid layer. The float switch
cannot discriminate between hydrocarbons and water, instead it only distinguishes between liquid present
and liquid absent conditions. Both float switches were effective at distinguishing when liquid was present
regardless of the ethanol concentration of the test blend and showed an accuracy rate of 100% for all
blends (Table 5 and Table 6). The recovery time for the float switches was instantaneous upon removal
from the fluid present condition for all blends (Table 5 and Table 6).
Table 5. Bottom Float Switch Sensor Performance Summary (n=10)
Performance Parameter
Average Detection Time
(hh:mm)
Average Recovery Time (hh:mm)
Average Activation Height (mm)
Activation Height Standard
Deviation (mm)
Specificity (%)(a)
Relative Accuracy (%)
Test Blends
EO
0:07
0:00
36.1
2.1
98.4%
100%
E15
0:07
0:00
36.1
0.3
94.8%
100%
116
0:06
0:00
36.2
0.4
94.9%
100%
E30
0:07
0:00
35.9
0.2
94.2%
100%
E50
0:06
0:00
36
0.0
94.5%
100%
E85
0:05
0:00
36.1
0.4
94.8%
100%
Water
0:05
0:00
31.6
0.3
82.9%
100%
(a) Source of theoretical (1.5 inches) is from the manufacturer's specification sheet.
Table 6. Top Float Switch Sensor Performance Summary (n=l)
Test Blends
renuimaiice raiameier
Detection Time (hh:mm)
Recovery Time (hh:mm)
Activation Height (mm)
Relative Accuracy (%)
EO
0:47
0:00
205.0
100%
E15
0:39
0:01
205.0
100%
116
0:36
0:01
201.0
100%
E30
0:37
0:00
200.0
100%
E50
0:34
0:00
201.0
100%
E85
0:33
0:00
201.0
100%
Water
0:35
0:00
197.0
100%
C-13
-------
C6.3 Fuel Sensitive Polymer Performance
The FS/FSP sensor was also composed of a fuel sensitive polymer strip in addition to the two
float switches. The FSP operates on the principle where a fiber optic cable is coated with a polymer that
interacts with fuel. When fuel is present, the light passing through the cable will be affected. The FSP
can discriminate between hydrocarbons and water and the sensor alarms in the presence of fuel. The FSP
was effective at distinguishing that fuel was present with 100% accuracy in test blends of EO, E15,116,
E30, and E50. However, E85 contained too high of an ethanol content for the FSP to distinguish that fuel
was present and therefore had a 0% accuracy (Table 7). The recovery time for the FSP is not
instantaneous and requires, on average, one hour to return to its non-activated state (Table 7).
Table 7. FSP Performance Summary (n=10)
Performance Parameter
Average Detection Time (hh:mm)
Average Recovery Time (hhmrn)
Average Activation Height (mm)
Activation Height Standard Deviation
(mm)
Specificity (%)(a)
Relative Accuracy (%)
Test Blends
EO
0:05
1:11
27.5
13.1
549%
100%
E15
0:05
1:01
26.9
15.5
537%
100%
116
0:05
1:02
28.1
20.8
562%
100%
E30
0:06
0:42
32
23.4
640%
100%
E50
0:10
0:24
57.8
41.2
1156%
100%
E85
NA
NA
NA
NA
NA
0%
(a) Source of theoretical value (0.50 cm) used in calculation is from NWGLDE website
C6.4 Capacitance and Conductance Performance
The C/C Sensor operates under the complex impedance principle which combines two operating
principles: electrical conductivity and capacitance. As the composition of the liquid between two series
of parallel plates changes, the liquid's properties of electrical conductivity and dielectric constant
(measured by capacitance) also change. As the C/C sensor was the only sensor that can discriminate
between hydrocarbons and water, it was the only technology that underwent the initial detection and
water ingress performance testing. During the initial detection testing, for all blends the C/C sensor
properly activated and was able to detect the appropriate fuel/water type present (Table 8). In addition,
the C/C was able to detect water ingress when submerged in any of the test blends (Table 9).
C-14
-------
Table 8. Capacitance and Conductance Initial Detection Performance Summary (n=10)
Performance Parameter
Average detection time (mmss.O)
Average Activation Height (mm)
Average Activation Standard Deviation
(mm)
Specificity (%) 2
Relative Percent Accuracy (%) 3
Precision (%CV)
Test Blends
Groundwater
01:25.0
5.7
2.275
114.00%
11.40%
39.90%
EO
02:02.3
7.85
0.337
157.00%
19.10%
4.30%
E15
02:37.8
9.95
0.158
199.00%
36.20%
1.60%
E85
01:31.4
5.55
0.599
111.00%
14.40%
10.80%
116
02:27.2
9.75
0.425
195.00%
34.90%
4.40%
(1) Values calculated according to Table 3 in Section Bl.4.4 of QAPP Addendum 110113
(2) The theoretical detection height was estimated at 5mm for this calculation
(3) Assumed that detected liquid level is the height of the segments detecting water (0.25in * number of segments)
Table 9. Capacitance and Conductance Water Ingress Performance Summary (n=10) 1
rci lui maiice rai aineiei
Average detection time (mm:ss.O)
Average Activation Height (mm)
Average Activation Standard Deviation
(mm)
Specificity (%) 2
Relative Percent Accuracy (%) 3
Precision (%CV)
EO5
02:04.6
10.6
1.165
211.0%
39.8%
11.0%
E155
01:42.1
9.4
1.696
188.0%
32.4%
18.0%
E854
00:19.5
2.0
1.462
39.0%
98.4%
75.0%
116s
01:56.9
10.0
0.577
200.0%
36.5%
5.8%
(1) Values calculated according to Table 3 in Section Bl.4.4 of QAPP Addendum 110113
(2) The theoretical detection height was estimated at 5mm for this calculation
(3) Assumed that detected liquid level is the height of the segments detecting water (0.25in * number of
segments)
(4) Detection time is time to sensor reading 'Aqueous Ethanol'
(5) Detection time is time to sensor reading 'Water'
C7 GROUNDWATER QUALITY
At the recommendation of the UST LD Stakeholders, groundwater was used to simulate water
ingress during testing. There is a lot of variation in groundwater characteristics; therefore, the
groundwater used was generally characterized to document the water being used for testing. A sub-
sample of the groundwater was analyzed for conductivity, pH and oxidation/reduction potential (ORP)
using the appropriate meters and probes (Hach LDO meter and VWR meter with Thermo Scientific
C-15
-------
probes). Because the characteristics were reported for an understanding of the type of water only and not
to achieve certain characteristics, no DQOs were associated with these data. Table 10 presents the
average of three measurements taken on the groundwater used for testing.
Table 10. Summary of Groundwater Characteristics
Groundwater
Average (n=3)
Conductivity (u,s/cm)
1133
PH
7.62
ORP (mV)
408.1
Temperature (°C)
20.3
C-16
-------
C8 REFERENCES
1. Carnegie Mellon Research Institute Advanced Devices and Materials Group, Test Procedures for
Third Party Evaluation of Leak Detection Methods: Point Sensor Liquid Contact Leak Detection
Systems. 1991.
2. Ken Wilcox Associates, I., Standard Test Procedures for Evaluating Leak Detection Methods:
Liquid-phase Out-of-Tank Product Detectors. 1990, USEPA Solid Waste and Emergency
Response/Research and Development.
3. Quality Assurance Project Plan for Biofuel Properties and Behavior Relevant to Underground
Storage Tank Leak Detection System Performance. U.S. Environmental Technology Verification
Program, Battelle, April 2013.
C-17
-------
Appendix D
Pressure Decay Testing Methods and Results
-------
Appendix D
Pressure Decay Testing Methods and Results
1 TEST SETUP AND PROCEDURE
Limited information is available as to the impact of different ethanol/isobutanol blended
fuels on the functionality of pressure decay as a pipeline LD method. Pressure decay relies on the
concept that a pipeline containing fuel is pressurized and sections isolated to show a loss of
pressure overtime if a leak is present. This pressure decay test was focused on whether the fuel
would affect the leak rate. The pressure decay rate was associated with leak rate according to the
following equation (when temperature is kept constant):
where Q = the leak rate (cm2/min)
V = test volume (cm2)
p
= average absolute gas pressure (psi)
PI - P2 = change in pressure (psi)
T = test duration (min)
This test utilized a leak tight 1-gallon pressure vessel set up as depicted in Figure 1 . The
test was conducted individually on the same test blends utilized in the sensor testing (Deionized
[DI] water, EO, E15, E85, and 116). A pressure environment was established in the vessel (initial
pressure was 20 psig), a specific leak rate was induced (average flow rates ranged between 4-6
mL/min), and the pressure decay was monitored and timed.
D-l
-------
Pressure
Source
P IT;: i K- -f\t-f •
-^ojic
^LB
j_
Pressure
Vfessel
m>
\_Hoseeoneetjr
all valve
Figure 1. Pressure Decay Test Setup
This stepwise approach was followed to produce a plot of the decay over time for each test
blend.
1. Fill pressure chamber to the manufacturer recommended level with test blend (DI water,
EO, E15, E85,andI16).
2. Pressurize system with dry air. Initial pressure (PI) should be 20 ± 1 pounds per square
inch (psi) for each test blend.
3. Isolate system from the gas pressure.
4. Allow system to stabilize for 15 minutes. Ensure pressure remains at 20 ± 1 psi using a
mechanical pressure gauge to monitor the pressure.
5. Generate a leak using 0.1 gallon per hour rate for each test blend. Start a timer and
monitor using a metering valve.
6. Liquid product is allowed to flow out of the pipe through a valve with a flow meter and is
collected in a graduated cylinder. The amount collected is divided by the time of
collection to provide an average leak rate.
7. Monitor the change in pressure over the leak duration.
8. Stop the timer at the end of the test duration (T).
9. The test should be designed so that the total pressure change is less than 10 % of the
starting pressure.
D-2
-------
2 PRESSURE DECAY RESULTS
The pressure decay results were similar across the test blends and water. Table 1
summarizes the results and reports the average and standard deviation of the replicates. Figures 1
- 5 present plots of the change in pressure (psi) (y-axis) that was observed over the test duration
in minutes (x-axis).
Table 1. Summary of Pressure Decay Testing
Pressure Decay Rate
(psig/min) (a)
Replicate 1
Replicate 2
Replicate 3
Replicate 4
Average
Standard Deviation
Test Blends
EO
-0.0466
-0.0484
-0.045
-0.0543
-0.0486
0.00406
E15
-0.042
-0.0339
-0.0447
-
-0.0402
0.00562
116
-0.0549
-0.0535
-0.0504
-
-0.0529
0.00230
E85
-0.0242^
-0.0445
-0.054
-0.0547
-0.0511
0.00570
Water(b)
-0.0465
-0.0426
-0.0543
-
-0.0478
0.00596
(a) Pressure decay rate is the slope of decay over time
(b) DI water
(c) Replicate 1 for E85 was not included in the average or standard deviation calculations
20.5
10
20 30
Time (minutes)
40
50
* Replicate 1
y = -0.0465x+19.763
R2 = 0.9957
• Replicate 2
y = -0.0426x+19.845
R2 = 0.9963
Replicate 3
y = -0.0543x+19.851
R2 = 0.9986
Figure 2. Pressure Decay Test with three replicates DI Water
D-3
-------
20.5
10
20 30
Time (minutes)
40
* Replicate 1
y = -0.0466x+19.77
R2 = 0.9953
• Replicate 2
y = -0.0484x+ 19.878
R2 = 0.9978
Replicate 3
y = -0.045x+19.824
R2 = 0.9977
X Replicate 4
50 y = -0.0543x+19.764
R2 = 0.9971
Figure 3. Pressure Decay Test with four replicates of EO
20.5
20.0
10
20 30 40
Time (minutes)
50
60
* Replicate 1
y = -0.042x+19.687
R2 = 0.9733
• Replicate 2
y = -0.0339x+19.91
R2 = 0.9937
A Replicate 3
y = -0.0447x+ 19.904
R2 = 0.9974
Figure 4. Pressure Decay Test with three replicates of E15
D-4
-------
20.5
20.0
17.5
10
20 30
Time (minutes)
40
* Replicate 1
y = -0.0549x+19.807
R2 = 0.9967
• Replicate 2
y = -0.0535x+19.935
R2 = 0.9987
Replicate 3
y = -0.0504x+19.825
R2 = 0.9976
50
Figure 5. Pressure Decay Test with three replicates of 116.
20
40 60
Time (minutes)
80
100
* Replicate 1
y = -0.0242x+19.893
R2 = 0.9892
• Replicate 2
y = -0.0445x+ 19.741
R2 = 0.9937
Replicate 3
y = -0.054x+19.825
R2 = 0.9975
X Replicate 4
y = -0.0547x+19.789
R2 = 0.9982
Figure 6. Pressure Decay Test with three replicates of ESS
D-5
-------
Appendix E
ETV Automatic Tank Gauging Verification Test Summary
-------
Appendix E
ETV Automatic Tank Gauging Verification Test Summary
1 INTRODUCTION
In 2011, automatic tank gauging (ATG) systems were tested to evaluate their functionality in ethanol-
blended fuels1'2. A total of four (4) technologies from two (2) different vendors were tested in three (3)
fuel blends (i.e., EO, E15, and E85). The following sections provide a general description of the ATGs
tested, an overview of the testing procedure, and summarized the results and findings from the testing.
2 TECHNOLOGY DESCRIPTIONS
ATG systems are volumetric leak detection technologies that rely on various physical properties
of the storage system to generate an electronic signal that can be converted into a value representing a
volume in a tank. An ATG system consists of a probe or sensor located inside the UST and a controller
(or console) mounted in an indoor location. Descriptions of each technology are summarized below:
• Vendor A-Technology 1 (Al): Al is designed to detect and measure the level of water present at
the bottom of a fuel storage tank in conjunction with a magnetostrictive level probe and ATG
system. The probe is installed in the storage tank by suspending it from a chain such that the
bottom of the probe is near the bottom of the tank. Specific versions of the water float are
available for use in diesel fuel and (non-ethanol-blended) gasoline. This float is ballasted to have
a net density intermediate to that of water and the respective fuel present in the tank such that it
is intended to float at the water-fuel interface.
• Vendor A-Technology 2 (A2): A2 is designed to detect and measure the level of a dense phase
present at the bottom of a fuel storage tank in conjunction with a magnetostrictive level probe
and ATG system. The probe is installed in the storage tank by suspending it from a chain such
that the bottom of the probe is near the bottom of the tank. Specific versions of the water float
are available for use in ethanol blended gasoline with up to 15% ethanol. This float is ballasted
to have a net density intermediate to that of the dense phase and the respective fuel such that it is
intended to float at the dense phase-fuel interface.
• Vendor B-Technology 1 (Bl): Bl is designed to detect and measure the level of water present at
the bottom of a fuel storage tank in conjunction with a magnetostrictive level probe and ATG
system. The water float, which represents a non-volumetric test technology, is located on the
bottom of the tank where water collects as a dense phase in gasoline. As the water depth
E-l
-------
increases, the float rises and transmits an electronic signal proportional to the level of water in
the bottom of the tank. Specific versions of the float are available for use in diesel fuel and (non-
ethanol blended) gasoline. These floats are ballasted to have a net density intermediate to that of
water and their respective fuels such that they will float at the water-fuel interface. The
evaluation was performed using a standard float for use in gasoline.
• Vendor B-Technology 2 (B2): B2 is a concentric, dual-float system designed specifically for
low-ethanol blend gasoline up to E15. The float is installed at the bottom of a fuel storage tank
and is used in conjunction with a magnetostrictive level probe and ATG system. An inner float
is designed to move freely within the limits of a protective housing attached to the outer float to
respond to all phase separation compositions in these fuels. The outer float is ballasted to remain
responsive to water and water-rich compositions of phase separation. This allows the inner float
to measure the full depth of water in the case of a massive ingress (lifting both floats), while
preventing the inner phase separation float from interfering with the fuel float in the rare
situation that an unusually dense, cold gasoline is delivered into the tank. As the detected phase
separation depth increases, the float rises and transmits an electronic signal proportional to the
level of phase separation in the bottom of the tank.
3 TEST OVERVIEW
For the technology evaluation a test vessel was fabricated from a 6-ft diameter piece of a
fiberglass storage tank shell that was fitted with glass ends to allow visual observations of the conditions
within the vessel during testing. All four ATGs were installed in the vessel according to the
manufacturer's specifications.
The following three test designs were incorporated to evaluate performance parameters, which were
used to characterize the functionality of the ATG system:
1. A continuous water ingress test consisting of two parts:
• Determination of minimum detection height
• Determination of smallest detectable incremental change in height
2. A quick water dump followed by a fuel dump
The first part of test one determined the minimum detection height by introducing water into the
test vessel using two methods of ingress - with splash and without splash. The water ingress method/rate
was selected to establish conditions that impact the degree of mixing that occurs in a tank using the three
ethanol blends - EO (no ethanol), E15 (15% ethanol), and E85 (85% ethanol). Two fuel height levels
(i.e., 25% [170 gallons] and 65% [610 gallons]) were specified to establish different splash mixing
E-2
-------
regimes and diffusion columns. Once the technology reacted to the minimum water height, the smallest
increment in water height was determined by continuing to ingress water at a height increase rate of 1/16-
inch every 10 minutes. Readings from the technology along with visual measurement were recorded and
used to determine the smallest detectable increment.
The second test was designed to simulate a quick water ingress rate followed by a high degree of
mixing such as might occur if a large volume of water was dumped into the tank at a 25% fill height and
then fuel was delivered to fill the tank to a 65% fill height. This test was performed using all three blends
of fuel.
4 RESULTS
A summary of the results and findings for each of the four technologies tested is presented below:
• Vendor A-Technology 1: Al responded to the water ingress when the test fuel was EO and E15,
but showed no response when E85 was used as the test fuel. The reason for the lack of response
was that no clear separated dense phase was formed in the flex fuel when water was added to the
test vessel. As a result, the performance parameters defined in the QAPP could not be determined
for this technology when E85 was employed.
• Vendor A-Technology 2: A2 responded to the water ingress when the test fuel was EO and E15,
but moved up the probe shaft to the upper fuel float when tested in E85. No clear separated dense
phase was formed in the E85 when water was added to the test vessel. As a result, the
performance parameters defined in the QAPP could not be determined for this technology when
E85 was employed.
• Vendor B-Technology 1: B1 responded to the water ingress when the test fuel was EO and E15,
but showed no response when E85 was used as the test fuel. The reason for the no response was
that no clear separated dense phase was formed in the E85 when water was added to the test
vessel. As a result, the performance parameters defined in the QAPP could not be determined for
this technology when E85 was employed.
• Vendor B-Technology 2: B2 responded to the water ingress when the test fuel was EO and El5,
but showed no response when E85 was used as the test fuel. The float appeared to be neutrally
buoyant in the E85/water mixture. The reason for the no response was that no clear separated
dense phase was formed in the E85 when water was added to the test vessel. As a result, the
performance parameters defined in the QAPP could not be determined for this technology when
E85 was employed.
E-3
-------
Currently 40 CFR, Section 280.43(a) states water detection technologies should detect "water at
the bottom of the tank," which does not address water entrained in the fuel due to increased miscibility
with the presence of ethanol. The ATG reports1'2 written after this testing state that they "did not detect
water in the test vessel containing either intermediate (E15) or high (E85) ethanol blends if the water was
suspended in the product or the water did not reach the bottom of the tank. Because of this, there is not
sufficient data to evaluate whether these technologies, when used with UST systems containing
intermediate or high ethanol blends, would indicate a potential release under every circumstance."
E-4
-------
5 REFERENCES
1. Environmental Technology Verification Report: Underground Storage Tank Automatic Tank
Gauging Leak Detection Systems, Veeder-Root Standard Water Float and Phase-Two Water
Detector. U.S. Environmental Technology Verification Program, Battelle, 2012.
2. Environmental Technology Verification Report: Underground Storage Tank Automatic Tank
Gauging Leak Detection Systems, Franklin Fueling Systems TSP-IGF4 Water Float and TSP-
IGF4P Float. U.S. Environmental Technology Verification Program, Battelle, 2012.
E-5
-------
Appendix F
ATG Simulated Leak Results
E-6
-------
Battelle
Jhe Business of Innovation
Appendix F
ATG Simulated Leak Results
This appendix is presenting data collected from an underground storage tanks (UST)
testing company and the quality of the data was not verified by the EPA or Battelle. Battelle has
no reason to suspect the result as being poor quality; it just could not be verified.
In 2013, simulated leak tests were conducted on single-walled USTs with the automatic
tank gauging systems (ATGs) as the primary method of leak detection. Tests were conducted as
part of annual monitoring system certification test by a contracted testing company at sites
servicing E10 (Premium, Mid-grade, and Regular Unleaded) and diesel fuels. Using a peristaltic
pump calibrated for the regulatory leak level, technicians remove 0.2 gallons per hour (gal/hr) of
fuel while conducting a static leak test with the ATGs. If the ATG reported a failed static test,
meaning the technology determined the tank was not tight, then the simulated leak test was
reported in the below table as a "Pass". Of the 71 tests conducted, 14 were "Inconclusive." The
majority of "Inconclusive" test results were due to the product level being below the minimum
required by local requirements for the ATG setup. Other "Inconclusive" tests were due to the
temperature change during the test being too large. These results indicate that ATGs are able to
detect leaks at the regulatory level in diesel and E10 fuels.
ATG Performance Test Results in Southern California in 2013
County in
Southern
California
Kern
Kern
Kern
Kern
Los Angeles
Los Angeles
Los Angeles
Los Angeles
# of Tests Conducted -
0.2GPH Test Results
1-Pass
2-Pass, 1 -Inconclusive
2-Pass, 2-Inconclusive
1-Pass
3 -Pass
3 -Pass
1-Pass
3 -Pass, 1 -Not Tested
Comment
All Ok.
Tank #1 and #3 Passed. Tank # 2 Mid-grade unleaded (MUL)
inconclusive due to Temp Change Too Large.
was
Tank #1 and #2 Passed. Tank #3 Premium unleaded (PUL) tested
twice, inconclusive both times due to Temp Change Too Large.
All Ok. Tank #3 PUL retested and Passed.
All Ok.
All Ok.
All Ok.
All Ok. Diesel Tank #4 was not tested due to low product level.
-------
County in
Southern
California
Los Angeles
Los Angeles
Orange
Orange
Orange
San Bernardino
San Diego
San Diego
San Diego
San Diego
San Diego
San Diego
San Diego
San Diego
San Diego
San Diego
San Diego
San Diego
San Diego
San Luis Obispo
# of Tests Conducted -
0.2GPH Test Results
1- Inconclusive, 1-Pass
3 -Pass
4-Pass
1-Pass
3 -Pass, 1 -Inconclusive
4-Pass
1-Pass, 2-Inconclusive
2-Inconclusive
3 -Pass
1-Pass
1-Pass
2-Pass, 1 -Not Tested
1-Pass, 2-Inconclusive
2-Pass, 1 -Inconclusive
2-Pass, 1 -Inconclusive
2-Pass, 1 -Inconclusive
1-Pass
3 -Pass
3 -Pass
3 -Pass
Comment
Diesel Tank #4 showed a gross increase during first test, Re-test
Passed.
All Ok.
All Ok.
All Ok.
Diesel Tank #4 percent volume was too low and caused
Inconclusive.
All Ok.
Tank # 1 - Pass. Percent Volume too low on other two Tanks and
caused Inconclusive.
Percent Volume too low on both Tanks and caused Inconclusive.
All Ok.
All Ok.
All Ok.
All Ok. Tank #1 PUL Not Tested-Product too low.
Tank #2 Regular unleaded (RUL) - Pass. Tank #1 PUL and #3
MUL percent volume too low caused Inconclusive
All Ok. Tank #1 PUL percent volume too low caused
inconclusive.
Tank #2 RUL and Tank #3 RUL - Pass. Tank #1 PUL percent
volume too low caused inconclusive.
All Ok. Tank #1 PUL percent volume too low caused
inconclusive.
All Ok.
All Ok.
All Ok.
Tapes showed that the probes detected the simulated leaks, but,
ATG did not sound an alarm. Maintenance was dispatched.
the
E-2
-------
December 31, 2014
Suitability of Leak Detection
Technology for Use
In Ethanol-Blended Fuel Service
Prepared by
Baiteiie
The Business of Innovation
Under a cooperative agreement with
wEnF\U.S. Environmental Protection Agency
-------
Suitability of Leak Detection
Technology for Use
In Ethanol-Blended Fuel Service
By
Battelle
505 King Avenue
Columbus, OH 43201
December 31, 2014
E-4
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Page 2 of 36
Notice
This report is a work prepared for the United States Government by Battelle. In no event shall either the
United States Government or Battelle have any responsibility or liability for any consequences of any use,
misuse, inability to use, or reliance on any product, information, designs, or other data contained herein,
nor does either warrant or otherwise represent in any way the utility, safety, accuracy, adequacy,
efficacy, or applicability of the contents hereof.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Page 3 of 36
1.0 TABLE OF CONTENTS
1.0 TABLE OF CONTENTS 3
2.0 LIST OF ABBREVIATIONS/ACRONYMS 5
3.0 INTRODUCTION 8
3.1 Ethanol and Gasoline Blends 9
3.2 Fuel Properties that Affect the Suitability of Leak Detection Technologies 10
4.0 FUEL PROPERTIES 12
4.1 Ethanol Content 13
4.2 Ethanol/Water Solubility in Fuel - Phase Separation 14
4.3 Conductivity 16
4.4 Dielectric Constant 18
4.5 Density (or Specific Gravity) 18
4.6 Viscosity 21
4.7 Acidity 22
4.8 Coefficient of Thermal Expansion 23
4.9 Non-additive Volume Changes (Degree of Accommodation) 24
5.0 LEAK DETECTION TECHNOLOGY OPERATING PRINCIPLES 25
5.1 Volumetric versus Non-volumetric-Based Testing Technology Categories 25
5.2 Automatic Tank Gauging System Technologies 26
5.3 Statistical Inventory Reconciliation Technologies 28
5.4 Pipeline Leak Detection Technologies 29
5.5 Non-volumetric Leak Detection Methods 29
6.0 SUITABILITY ASSESSMENT OF LEAK DETECTION TECHNOLOGIES IN ETHANOL-
BLENDED FUEL 32
7.0 REFERENCES 36
APPENDICES
Appendix A Environmental Technology Verification Fuel Property and Technology Testing
Appendix B Fuel Property Testing Methods and Data Results
Appendix C UST LD Operating Principle Testing Methods and Data Results
Appendix D Pressure Decay Testing Methods and Results
Appendix E ETV Automatic Tank Gauging Verification Test Summary
Appendix F ATG Simulated Leak Results
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 4 of 36
FIGURES
Figure 1. Phase Separation Plot of UV-V Measurements 16
Figure 2. Conductivity Plot by Test Blend and Water Content 17
Figure 3. Density Plot by Test Blend and Water Content 20
Figure 4. Viscosity Plot by Test Blend and Water Content 21
Figure 5. Thermal Expansion Plot by Test Blend 24
TABLES
Table 1. Summary of Fuel Property Data Collected* 14
Table 2. Biofuel-Water Mixture (BFW) Phase Separation 14
Table 3. F-Test Results of Fuel Blend Comparison for Conductivity 18
Table 4. Summary of Density Results for the BFWs (g/mL) 19
Table 5. F-Test Results of Fuel Blend Comparison for Density 20
Table 6. F-Test Results of Fuel Blend Comparison for Viscosity 22
Table 7. F-Test Results of Fuel Blend Comparison for Acidity 23
Table 8. Degree of Accommodation Summary for all Test Blends 24
Table 9. Leak Detection Technologies and Principles of Operation 27
Table 10. Suitability of Existing Leak Detection Technology for Ethanol-Blended Fuel 34
E-4
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 5 of 36
2.0 LIST OF ABBREVIATIONS/ACRONYMS
ANOVA analysis of variance
ASTM ASTM (American Society for Testing and Materials) International
ATG automatic tank gauge
BFW biofuel water mixture
°C degree Celsius
EO gasoline
E10 gasoline with up to 10% ethanol
E15 gasoline with up to 15% ethanol
E30 gasoline with 30% ethanol
E50 gasoline with 50% ethanol
E85 gasoline with 51 to 83% ethanol
EPA Environmental Protection Agency
ETV Environmental Technology Verification
gal/hr gallon per hour
116 gasoline with 16% isobutanol
kg/L kilogram per liter
L liter
LD leak detection
mL milliliter
mm2/s millimeter squared per second
NWGLDE National Work Group on Leak Detection Evaluations
OUST Office of Underground Storage Tanks
% percent
pS/cm picosieman per centimeter
QAPP Quality Assurance Project Plan
QA/QC quality assurance/quality control
SIR statistical inventory reconciliation
(iL microliter
UST underground storage tank
UV-Vis ultraviolet visible
E-5
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 6 of 36
EXECUTIVE SUMMARY
Suitability of Leak Detection Technology for Use
In Ethanol-Blended Fuel Service
As the use of biofuels has increased in the last decade, there has been a level of concern over the
effect that ethanol blends have on the material compatibility and operability of existing infrastructure.
The focus of this research is to determine whether leak detection (LD) technologies are functioning
properly in ethanol fuel blends. Fuels with different concentrations of ethanol have different intrinsic
properties. As new fuels with varying blends of ethanol emerge, the resulting variations in fuel properties
might affect the functionality of LD technologies. Technology to detect leaks has been required since late
1989 when UST operators were required to implement procedures to prevent and detect leaks in existing
and new USTs under Title 40 of the Code of Federal Regulations Part 280 (40 CFR 280) Technical
Standards and Corrective Action Requirements for Owners and Operators of Underground Storage Tanks
(SubpartD).
When first employed, test procedures used to determine LD technology performance were
commonly performed on USTs containing diesel fuel, in which the technologies tested generally behave
in a similar manner as they do in gasoline. LD technologies tested with one of these procedures were then
"listed" by the National Work Group on Leak Detection Evaluations (NWGLDE) as having been
evaluated by a third party in accordance with an approved leak detection protocol. Currently, the
increasing desire to use motor fuels containing ethanol, such as E15 and Flex Fuel (also referred to as
E85), has led EPA, NWGLDE, and others to question the appropriateness of use of these LD technologies
with fuels that have different properties than the fuel on which they were originally tested and for which
the test methods were designed.
Fuel property research was conducted in order to better understand how ethanol blended into
fuels in different concentrations can affect the properties of those blends. The objective of examining fuel
properties was to identify when various blends are significantly different with respect to a fuel property.
The fuel blends included EO, E10, E15, E30, E50, E85 and an isobutanol blend at 16 percent (116).
Subsequently, various LD technology categories were described with respect to operating
principle and how the change in fuel property may affect the operability of the technologies in that
category. For the purpose of this technology review, ethanol blends are categorized as low-ethanol (i.e.,
E10, and E15) and high-ethanol blends (51 to 83 percent ethanol) and categorized as:
• Technology is expected to be suitable for indicated use (GREEN).
• Technology has limitations with the indicated use (YELLOW).
E-6
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 7 of 36
• Technology is expected to not be suitable for indicated use (RED).
As all technologies are different, have different algorithms, and are influenced by human inputs
and installation, these conclusions may not be appropriate for every technology in a category. This paper
discusses the relationship between fuel properties and operating principles against the performance
standards established in the federal LD requirements. The potential negative impacts are highlighted in
the following sections for consideration. In some cases, the technology may need to be modified to
recognize these changes at the regulatory level with adjustments of threshold values and monitoring data
processing.
E-7
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 8 of 36
3.0 INTRODUCTION
Biofuels are an increasing portion of the fuel supply in the United States (US) due partially to
enactment of the Renewable Fuel Standard established by the Energy Policy Act of 2005 and amended by
the Energy Independence and Security Act of 2007. As the use of biofuels has increased in the last
decade, there has been a level of concern over the effect that ethanol blends have on the material
compatibility and operability of existing infrastructure. The focus of this research is to determine whether
leak detection (LD) technologies are functioning properly in low and high ethanol fuel blends. Fuels with
different concentrations of ethanol have different intrinsic properties. As new fuels with varying blends
of ethanol emerge, the resulting variations in fuel properties might affect the functionality of LD
technologies.
Approximately 571,OOO1 underground storage tanks (USTs) currently in service in the US have
the potential for contaminating groundwater and subsequently drinking water should they fail. UST LD
regulations were therefore created to specify monitoring requirements for detecting leaks. Technology to
detect leaks has been required since late 1989 when UST operators were required to implement
procedures to prevent and detect leaks in existing and new USTs. As a result of regulations adopted at
that time [Title 40 of the Code of Federal Regulations Part 280 (40 CFR 280) Technical Standards and
Corrective Action Requirements for Owners and Operators of Underground Storage Tanks], LD
technology was to be applied not only to the USTs themselves, but also to the piping network that
connected storage tanks and delivered fuel to dispensers. LD requirements are defined in 40 CFR 280
Subpart D.
To assist the regulated community when evaluating LD options, US Environmental Protection
Agency (EPA) developed a series of standard test procedures that cover most of the technologies
commonly used for UST LD monitoring and testing. Over the years there have been numerous additional
test procedures and adaptations of these standard EPA test procedures. The procedures are publicly
available through the National Work Group on Leak Detection Evaluations (NWGLDE)
(www.nwglde.org) and are organized according to general LD technology categories.
These test procedures have been used by technology vendors or third party evaluators to provide
information needed by tank owners and operators to determine if a LD technology meets the regulatory
requirements. Concerns regarding LD operability arise from the trend of using legacy LD technologies in
new fuel applications. When first employed, these procedures were commonly performed on USTs
containing diesel fuel, in which the technologies tested generally behave in a similar manner as they do in
gasoline. LD technologies tested with one of these procedures were then "listed" by the NWGLDE as
E-8
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 9 of 36
having been evaluated by a third party in accordance with an approved LD test procedures. Currently, the
increasing desire to use motor fuels containing ethanol, such as E15 and Flex Fuel (also referred to as
E85), has led EPA, NWGLDE, and others to question the appropriateness of use of these LD technologies
with fuels that have different properties than the fuel on which they were originally tested and for which
the technologies were designed.
This suitability assessment presents an analysis of the available information on characteristics of
ethanol-blended fuels and on LD technology operating principles to assess potential LD technology
performance functionality in ethanol-blended fuels. This assessment and related testing were performed
under the EPA Environmental Technology Verification (ETV) program Advanced Monitoring Systems
Center (www.epa.gov/etv). ETV involves a rigorous quality assurance/quality control (QA/QC) program,
engagement with stakeholders in the industry, and a peer review process. Data were collected in multiple
phases of testing following two ETV-approved Quality Assurance Project Plans (QAPPs): Biofuels
Properties and Behavior Relevant to Underground Storage Tank Leak Detection System Performance2
and Addendum3 and QAPP for Verification of Underground Storage Tanks Automatic Tank Gauging
Leak Detection Systems.4 The data are presented in Appendices A - E. Appendix F presents
supplemental data of simulated leak tests performed in the field by a reputable testing company and have
not been independently generated through ETV.
3.1 Ethanol and Gasoline Blends
Several ethanol-gasoline blends are currently in use or being considered for use as motor fuels.
E10, which represents a mixture of up to 10 percent (%) by volume ethanol with the remaining percent
gasoline, has been distributed throughout the US for several years and is the most widely used gasoline
blend in the US. E85 or Flex Fuel (between 51 and 83 % ethanol) has also emerged as a motor fuel,
although its use is much less prevalent compared to E10. A waiver under the Clean Air Act to allow
distribution of fuel containing 10 to 15 % ethanol (El5) was partially approved by EPA in 2010 and 2011
and has appeared minimally on the market. EPA has stated that E15 is suitable for 2001 and newer model
year vehicles (FR 68093 November 4, 2010 and 76 FR 4662 January 26, 2011). Other blends being
evaluated by Oak Ridge National Laboratory for material compatibility issues include various mixtures of
ethanol and gasoline up to 30 % ethanol by volume5.
For the purpose of this technology review, ethanol blends are categorized as low-ethanol (i.e.,
E10, and E15) and high-ethanol blends (51 to 83 % ethanol). Although mid-ethanol blend levels (i.e.,
E30, and E50) are included in the fuel property discussion, conclusions with respect to the technologies
are categorized as low and high blends. There are limited data on the performance of the technologies
E-9
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 10 of 36
with the mid-level blends; therefore, this review is evaluating blends that are currently in use. It should
also be noted, that if mid-level ethanol blends are offered on the market in the future, they may be
blended at the dispenser from E10 and E85 instead of having dedicated tanks for the specific blends.
Different grades (i.e., regular, mid-grade, and premium) are not considered separately in the current
review. Seasonal differences in fuel properties (mainly related to vapor pressure) and detergents or
additives are also not being considered.
In addition, an isobutanol-blended gasoline is another option that potentially will enter the
market. Isobutanol blended at 16% is an anticipated level of one of the manufacturers and the higher of
two levels attempting to be brought to market. Isobutanol can function within the current infrastructure
and ethanol production plants have the potential to be retrofitted for its production. Although not ethanol,
this alcohol may potentially enter the market and therefore is included in this discussion.
When reviewing the suitability of LD technology in ethanol-blended fuel service a challenge is
accounting for the uncertainty of knowing the actual ethanol percentage in each blend of fuel, because
fuel quality specifications allow for ethanol content variation in the blends. This uncertainty can best be
illustrated by looking at the ASTM International (ASTM) specification for E85. Pursuant to ASTM
D57986, E85 must contain between 51 and 83 % alcohol by volume. Similarly, low ethanol blends may
be subject to the same variability in ethanol content of the fuel. For example, E10 may technically
contain any ethanol percentage up to 10 % volume (although most often blended close to 10%), while
E15 contains greater than 10 volume % by volume ethanol and up to 15 % volume ethanol
(http://www.epa.gov/otaq/regs/fuels/additive/el5/). As discussed below, in addition to the physical
characteristics of ethanol-blended fuel, this allowable variation of ethanol content may produce an
unwanted impact on functionality or accuracy of the technologies.
3.2 Fuel Properties that Affect the Suitability of Leak Detection Technologies
Parties interested in LD technologies usually discuss two topics when evaluating the suitability of
a particular LD technology to be used in ethanol-blended fuel service: (1) material compatibility, and (2)
operability. The first topic, compatibility, relates to corrosiveness of ethanol and ethanol/water mixtures
on metal and plastic components of the detection system in contact with fuel or fuel vapor. Increased
microbial growth induced by ethanol is also a concern. Since this has been the subject of significant
research by Oak Ridge National Laboratory5 and others, the material compatibility aspect of technology
used in ethanol-blended fuel service will not be discussed herein. The second topic, operability, relates to
the ability of LD technology to properly function in ethanol-blended fuel service as a result of different
product characteristics than were used to originally design the equipment. Technology evaluators
E-10
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 11 of 36
generally consider two properties to be most important on the ability of existing LD technologies to
properly operate while in ethanol-blended fuel service: (1) water solubility in ethanol, and (2)
temperature. Depending on the technology operating principles, other properties that may also be
important include ethanol concentration, density, viscosity, and conductivity.
The data generated is presented in the summary of the fuel properties in Section 4 and the
operating principles of the various LD technologies categories are discussed in Section 5. Finally in
Section 6 is the suitability assessment of the various technology categories which utilized the data
presented in Appendices A - E and summarized in the main document.
E-ll
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 12 of 36
4.0 FUEL PROPERTIES
The primary fuel properties that are suspected of affecting LD system operability include:
• Ethanol content (or isobutanol content)
• Alcohol/water solubility in gasoline
• Dielectric constant
• Electrical conductivity
• Viscosity
• Coefficient of thermal expansion
Each of these properties is affected by the ethanol content in the blend, and as ethanol content
increases, other properties are affected. For example, the density of pure (neat) ethanol is greater than the
density of neat gasoline, and therefore, as the ethanol content of a blend increases, so does the density of
the blend. In a similar fashion, water solubility is greater in ethanol than in gasoline (water is essentially
insoluble in gasoline), and therefore, a blend with a greater ethanol content is able to absorb a greater
amount of water. Viscosity, conductivity, and coefficient of thermal expansion are also all greater for
neat ethanol than for neat gasoline, thereby producing higher values for each parameter as ethanol content
increases. Several other combinations of properties are also related to one another. For example, addition
of water to an ethanol-blended fuel also increases the density, viscosity, dielectric constant, and (usually)
conductivity of the blend. These interrelationships can make exact identification of property effects
complicated and difficult.
In addition to the difficulty noted above, ethanol-blended fuel may not consistently contain the
same amount of ethanol. This may be due to blending differences, volatilization, water ingress, or phase
separation. Thus, the actual value of the physical property of interest may be unknown. Furthermore,
while values for these properties are readily available for neat materials such as gasoline, ethanol, and
water, they are much less available for different mixtures of ethanol, gasoline, and water. Fuels also have
proprietary additives and detergents that have the potential to affect all of these fuel properties. All of
these uncertainties in fuel composition could contribute to potential errors during system operation.
EPA utilized the ETV program to conduct fuel property research in order to better understand
how ethanol blended into fuels in different concentrations can affect the properties of those blends. The
objective of examining the fuel properties was to identify when various blends are significantly different
with respect to a fuel property. For example, is the conductivity of E15 significantly different from E30
or is the viscosity of E10 significantly different from E85? Table 1 summarizes the fuel blends and fuel
properties data gathered from samples measured in triplicate. Fuel blends included EO, E10, El5, E30,
E-12
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 13 of 36
E50, E85 and an isobutanol blend at 16 % (116) and were prepared using the same gasoline throughout
the project. The variability of gasoline and unknown proprietary differences are common at fueling
stations; however, for the purposes of lab testing, these variables were limited by the use of one fuel for
preparing the mixtures. Detailed methods, QA/QC procedures, and results are presented in the
Appendices A and B.
4.1 Ethanol Content
As mentioned in the previous paragraph, each of the properties listed above are impacted by
ethanol content in the blended fuel. Other than compatibility, however, which is not the subject of this
suitability assessment, ethanol content does not directly impact LD technology operability. Instead, its
effect is manifested by altering listed fuel properties that impact one or more operating principles of
specific technologies. As a result of the variability of ethanol content mentioned previously, one cannot
estimate how other physical properties of the blended fuel are altered by the addition of ethanol. Without
some independent means of knowing the exact ethanol content of the blend, the true correction that may
need to be made to readings from the various technologies will not be known. Because ethanol affects
each of the physical properties noted above, this situation may impact LD technology by limiting the
ability of a technology to accurately quantify leak rates, even when a technology may still be able to
qualitatively identify that a leak is present. The regulations require technologies to identify a 0.2
gallon/hour (gal/hr) leak rate for monthly testing and a 0.1 gal/hr leak rate for tank tightness testing,
establishing a target leak detection performance level that may be influenced by these unknown changes.
For example, when ethanol content increases, so does the density of the fuel blend. The LD technology
software may not be set for the actual fuel blend density because the ethanol content of the fuel blend may
vary with each delivery. Qualitative leak determination will still be possible; however, when comparing
calculated product volumes at different periods, the volumes change with time and the true leak rate will
have the potential for more error since it is based on the assumed ethanol content (entered into the
software program) or assumed density.
E-13
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 14 of 36
Table 1. Summary of Fuel Property Data Collected*
Property
Specific Gravity
(Dimensionless)
Density (g/mL)
(15.6 °C)
Coefficient of
Thermal
Expansion
(5-30 OC1)
Viscosity 25 °C
(mm2/S)
Conductivity
(pS/cm)
Acidity
(% mass)
Gasoline (EO)
0.722
0.722
0.0010
0.555
192
0.00053
E10
0.761
0.762
0.0012
0.557
12233
0.0012
E15
0.764
0.764
0.0011
0.582
104722
0.00093
116
0.765
0.766
0.0012
0.659
5163
0.0011
E30
0.770
0.770
0.0013
0.698
4321111
0.0012
E50
0.776
0.776
0.0009
0.863
9204444
0.0016
E85
0.790
0.788
0.0010
1.085
8304444
0.0015
Triplicate samples were measured in triplicate for all properties and blends.
4.2 Ethanol/Water Solubility in Fuel - Phase Separation
The solubility of water in fuel increases dramatically as ethanol content increases. This increase
has an effect on the physical properties of the blended fuel and will have an effect on many operating
responses of LD technologies. Water is absorbed into the ethanol fraction of the blended fuel, and as
water is absorbed, density, viscosity, and conductivity increase while the coefficient of thermal expansion
remains relatively similar for the blended fuels. Tests were performed using the above test blends with
multiple levels of water content, 0%, 0.25%, 0.5%, 2.5%, and 5.0%. Test results show that some of these
mixtures became two distinct phases (S), some were semi-separated with the separation not clearly
distinguished (SS), and others were composite single-phased mixtures (C). Table 2 presents the biofuel-
water-mixtures (BFW) and the observed separation, if any. When samples were separated, analytical
results were acquired for the bulk fuel phase (top). If the dense phase (bottom) sample volume was large
enough to sample, a sample was archived for analysis, if deemed necessary. EO and 116 had clearly
separated phases (S) as they have the lowest miscibility with water.
Table 2. Biofuel-Water Mixture (BFW) Phase Separation
% Water
EO
E10
E15
116
E30
E50
E85
0.0
0.25
0.5
2.5
5.0
C
S
S
S
S
C
SS
SS
S
S
C
C
C
S
S
C
S
S
S
S
C
C
C
SS
S
C
C
C
C
C
C
C
C
C
C
C = Composite, SS = Semi-Separated, S = Separated Clearly; All at 25°C
E-14
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 15 of 36
Because water is essentially immiscible in gasoline, a very small addition of water to a UST
storing gasoline will cause a water phase to settle in the bottom of the tank. This makes it relatively
simple to determine the presence of water in USTs storing gasoline. However, E10 and E15 blends can
hold approximately 0.5% of water with mixing before phase separation occurs. As fuel temperature is
lowered, the amount of water needed before phase separation occurs is also lowered. Because water
alters the solubility of ethanol in gasoline, when phase separation occurs in E10, the separated phase
consists of an ethanol/water mixture with a density greater than ethanol but less than water. If water
entering a UST does not mix into a low ethanol-blended fuel, it will collect at the bottom of the UST,
similarly to EO. However, once the UST receives a fuel drop (that is not saturated with water),
substantially mixing the contents, the water bottom is absorbed into the fuel. With continued water
ingress, water will collect at the bottom and be detected, then disappear with each fuel delivery. This
phenomenon has been shown to render traditional water detection floats unreliable unless the float
composition density is adjusted in comparison with the density of the separated phase7'8. Another
alternative would be for the technology console to be programed to recognize this reoccurring pattern of
detected water followed by no detectable water.
As mentioned previously, water absorbed into the blended fuel will also increase the density of
the blend (as well as other physical parameters), thus making proper selection of volumetric correction
factors difficult. In addition, a certain amount of water can be absorbed in ethanol without an increase in
volume. In a large volume of stored fuel, the amount of water absorbed into the ethanol fraction of an
ethanol-blended fuel could be appreciable and could exceed the required sensitivity of the regulation [e.g.,
40 CFR280.43(a)(6) requires the measurement of any water level in the bottom of tank be made to the
nearest 1/8" at least once a month]. Therefore, an automatic tank gauging (ATG) system or other level-
based technology may be unreliable in detecting water at the bottom of a tank, because the product
volume will not accurately reflect the total volume of water that has entered a tank. Liquid level readings
may also be unreliable if a tank has multiple leak points and fuel is leaking out while water is leaking in.
As a method to characterize phase separation and define the vertical position of the interface of
various fuel blends, an experiment was conducted measuring the absorbance of fuel blend-water mixtures.
Figure 1 represents the Ultraviolet Visible (UV-Vis) measurements recorded on the 50-50 mixture of fuel
blend and water (mixed with a dye). The UV-Vis measurements were recorded on a 10 milliliter (mL)
aliquot that was drawn from the bottom of the sample vial holding the 50-50 fuel-water mixture (See
Appendix B for more detail). The plateau on the top left hand side of Figure 1 represents the dyed water
while the one on the bottom right hand side represents fuel (where dye did not reside and therefore no
absorbance was measured). For gasoline (EO) with no ethanol content it can be observed that there are
E-15
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 16 of 36
only two distinct features to the curve, plateaus on the left and right hand sides with no intermediate
peaks. This infers that the EO fuel had no transition zone or phase mixed with water. However, with the
increase in ethanol content the measurements became more complicated and the phase separation more
apparent. The following observations of the transition zone can be made from the data presented in
Figure 1.
1. A drop in absorbance value (y-axis) indicates the ethanol is absorbing into the water. With
the increase in ethanol content, more ethanol was available for absorption into the water,
which led to lower initial absorbance values.
2. For EO and 116, the fuel phase was detected at draw 8 (approximately midway up the sample
vial), as ethanol content increased in the fuel blends, the fuel phase was detected at higher
draw levels (up to 12). In other words, with the increase in ethanol content the water-ethanol
mixture was more dominant.
3. The appearance and augmentation of intermediate peaks indicates formation of a transition
phase and its broadening as ethanol content increases.
4.3
14 16
Liquid Draw Level from Bottom to Top of Vial
Figure 1. Phase Separation Plot of UV-V Measurements
Conductivity
From the conductivity plot (Figure 2) it can be observed that with the increase in ethanol content,
conductivity of the fuel increased exponentially. Also, conductivities of fuels E30, E50, and E85 were
found to be in the same range. The change in water content did not appear to have an effect on
E-16
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 17 of 36
conductivities of the fuels E30, E50 and E85, however, increase in water content beyond 0.5% lead to
drop in conductivity by two orders of magnitude for the fuels E15 and E10 and beyond 2.5% lead to a
similar trend for E30. This was due to the bulk fuel being measured since the BFW mixtures had phase
separated at these water concentrations. Similarly, EO and 116 had distinct water-fuel separation and the
bulk fuel conductivity measurements were not influenced by the water. The wide range of conductivity
readings between the test blends (with or without water) indicates that a technology operating principle
based on this property would need to operate over a large range or specify the range of operability by fuel
blend.
Figure 2. Conductivity Plot by Test Blend and Water Content
To determine if the differences between conductivities of the fuel blends were significant, an
analysis of variance (ANOVA) was performed on the dataset. The ANOVA found significant differences
existed within the dataset of fuel blend conductivity measurements. To further understand the
differences, an F-test was performed, which allowed for direct comparison between the different fuel
blends. The null hypothesis of the F-test assumes that the means of each fuel blend are equal. Rejection
of the null hypothesis of the equality of means was done at the 0.05 significance level. Rejection of the
null hypothesis is an indication at least one mean among the different types of fuel blends is not equal.
Table 3 presents the/>-values obtained from the F-test along with "YES" and "NO" to indicate yes, there
is a significant difference, or no, there is not a significant difference between the fuel blends being
compared. These p-values take into account the fact that multiple comparisons are being performed by
applying the Sidak adjustment to the reported significance level.
In Table 3, a/>-value less than 0.05 indicates a significant difference, while any value 0.05 or
greater (i.e., up to 1) indicates the difference is not significant. Almost all significant differences in
E-17
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 18 of 36
conductivity were observed between the higher alcohol-blended fuels (i.e., E30, E50, and E85) and the
lower alcohol-blended fuels (i.e., EO, E10, El5, and 116). The only exception was the comparison
between E50 and E85. Given the effect of water in these blends on conductivity, without modification,
technologies which operate on conductivity may function differently in low versus high ethanol-blends.
Table 3. F-Test Results of Fuel Blend Comparison for Conductivity*
Fuel
Blend
E10
E15
116
E30
E50
E85
EO
1
NO
1
NO
1
NO
<0.0001
YES
<0.0001
YES
<0.0001
YES
E10
1
NO
1
NO
<0.0001
YES
<0.0001
YES
<0.0001
YES
E15
1
NO
<0.0001
YES
<0.0001
YES
<0.0001
YES
116
<0.0001
YES
<0.0001
YES
<0.0001
YES
E30
<0.0001
YES
<0.0001
YES
E50
0.948
NO
p < 0.05 indicates a significant difference
*F-test performed after significant differences were identified using an ANOVA analysis of the
dataset.
4.4 Dielectric Constant
Dielectric constant is the "measure of a substance's ability to insulate charges from each other.
Taken as a measure of solvent polarity, the higher dielectric constant means higher polarity, and greater
ability to stabilize charges."9 When ethanol and water are added to gasoline the conductivity of the
mixture substantially increases and this can affect certain capacitance probes (depending on the design).
Several technology manufacturers and organizations have indicated that this change makes use of some
capacitance probes in ethanol-blended fuel service unreliable. Furthermore, the presence of a separated
phase at the bottom of a tank would produce a different dielectric constant in the separated phase than in
the fuel phase and make it difficult to determine the proper response for a capacitance probe when used
for leak detection. Legacy capacitance ATG probes are no longer offered by manufacturers; however,
this operating principle is being applied to sensors for monitoring at various parts of UST systems.
4.5 Density (or Specific Gravity)
Density of a material is often defined in terms of specific gravity. Specific gravity is the ratio of
the density of a material to the density of water (the density of water is 1 kg/L at 15°C). A material with a
specific gravity less than 1 is less dense than water, and a material with a specific gravity greater than 1 is
E-18
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 19 of 36
more dense than water. Because gasoline is a mixture of hydrocarbons, the content of one batch of
gasoline (and by extension, specific gravity) may be different than that of another batch. Density is a
parameter of inherent importance for several mass-based or pressure-based LD technologies (e.g.,
buoyancy probe, piping flow meters). Until the density difference due to mixing of different batches
comes to equilibrium, a response change in the LD technology could be interpreted as inconclusive.
Achieving equilibrium is mainly driven by the rate of temperature change after a delivery and can vary
substantially if the delivered fuel temperature is very different from the stored fuel temperature. Once
equilibrium is achieved or the rate of change is within the technology's acceptable range, the test will
complete. However, the LD technology may not be able to compensate for a density change when the
change is due to phase separation or water absorption into ethanol. In these cases, the technology may not
be able to detect a leak, or the calculated leak rate may not be accurate. Because density of a liquid varies
with temperature, the highest precision in level measurement necessitates that density be compensated for
or expressed with relation to the actual temperature of the measured liquid. Table 4 summarizes and
Figure 3 plots the density values obtained during fuel property testing of the BFWs.
Table 4. Summary of Density Results for the BFWs (g/mL)
%
Water
0.0
0.25
0.5
2.5
5.0
EO
0.7222
0.7228
0.7227
0.7224
0.7230
E10
0.7617
0.7648
0.7649
0.7630
0.7624
E15
0.7643
0.7650
0.7663
0.7629
0.7618
116
0.7656
0.7658
0.7669
0.7669
0.7684
E30
0.7701
0.7708
0.7722
0.7753
0.7583
E50
0.7758
0.7766
0.7779
0.7849
0.7951
E85
0.7883
0.7927
0.7937
0.8014
0.8067
From the data it is evident that an increase in ethanol content leads to increase in the density of
the fuel. Furthermore, the plot also reveals that the densities of the low alcohol-blended fuels (being
dominated by the hydrocarbon portion) are fairly independent of low additions of water. However, as
ethanol content dominates the blend, beyond 2.5% water content the density of E50 and E85 appear to be
marginally increasing. While that of the fuel E30 decreases after the 2.5% water content level, this is due
to the analysis of the bulk fuel after phase separation occurred.
To determine if the differences between the densities of the fuel blends were significant, an
ANOVA was performed on the dataset. The results are presented and interpreted as above in Section 4.3.
As shown in Table 5, the ANOVA found significant differences existed within the dataset of fuel blend
density measurements. All differences in density between the fuel blends were found to be significant
E-19
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 20 of 36
with the exceptions of low alcohol-blended fuels (E10 and El5, E10 and 116, and E15 and 116), again
since they are dominated by hydrocarbons. The low alcohol-blended fuels were significantly different
from the EO, so the alcohol does have an effect. With the significant differences in densities observed
between most fuel blends, technologies which utilize this principal may not be transferable between
blends.
Figure 3. Density Plot by Test Blend and Water Content
Table 5. F-Test Results of Fuel Blend Comparison for Density*
Fuel
Blend
E10
E15
116
E30
E50
E85
EO
<0.0001
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
E10
1
NO
0.839
NO
<0.0001
YES
<0.0001
YES
<0.0001
YES
E15
0.821
NO
<0.0001
YES
<0.0001
YES
<0.0001
YES
116
<0.0001
YES
<0.0001
YES
<0.0001
YES
E30
<0.0001
YES
<0.0001
YES
E50
<0.0001
YES
p < 0.05 indicates a significant difference
*F-test performed after significant differences were identified using an ANOVA analysis of the
dataset.
E-20
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 21 of 36
4.6 Viscosity
Measurement of flow through piping requires that pressure in the pipe section be monitored.
Pressure monitoring systems require knowledge of several parameters of product in the piping, including
density and viscosity. Addition of ethanol to gasoline increases the viscosity of the blend thus yielding
higher differential pressures across the flow measurement device than obtained for neat gasoline (EO).
Proper calculation of leak rate would require knowledge of the ethanol and water content of the blend or
exact determination of density and viscosity. Once again, because these liquid properties vary with
temperature and the rate of temperature change effects the ability for a technology to make a conclusive
test, the highest precision in level measurement may necessitate that they be compensated for or
expressed with relation to the actual temperature of the measured liquid.
From Figure 4 it is evident that an increase in ethanol content leads to increase in fuel viscosity
and that E85 is the most viscous among the fuels. Furthermore, the plots also reveal that fuel viscosity
measurements are fairly independent of low additions of water. However, beyond 2.50% water content,
the viscosity of E50 and E85 appear to be marginally increasing, while that of the fuel E30 decreases.
Again the E30 decrease is due to the analysis of the bulk fuel after phase separation occurred.
Figure 4. Viscosity Plot by Test Blend and Water Content
To determine if differences between viscosities of the fuel blends were significant, an ANOVA
was performed on the dataset. The results are presented and interpreted as above in Section 4.3. As
shown in Table 6, the ANOVA found significant differences existed within the dataset of fuel blend
viscosity measurements. Every fuel blend comparison was found to be significantly different, except for
the comparison between EO and E10. Without modification, technologies which incorporate viscosity as
an operating principle may not function appropriately across all the tested fuel blends.
E-21
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 22 of 36
Table 6. F-Test Results of Fuel Blend Comparison for Viscosity*
Fuel
Blend
E10
E15
116
E30
E50
E85
EO
1
NO
0.017
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
E10
0.037
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
E15
<0.0001
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
116
0.001
YES
<0.0001
YES
<0.0001
YES
E30
<0.0001
YES
<0.0001
YES
E50
<0.0001
YES
p < 0.05 indicates a significant difference
*F-test performed after significant differences were identified using an ANOVA analysis
of the dataset.
4.7 Acidity
Acidity may not have a direct relation to the operating principles of LD technologies; however, it
is included here as a measure of potential compatibility issues. Acidity of the fuel (expressed by the
ASTM method as percent mass normalized to acetic acid) remained fairly independent of its water
content, little to no change was observed with the increase in water. While EO was least acidic among the
fuels, E50 and E85 were found to be on the higher end.
To determine if the differences between acidity of fuel blends were significant, an ANOVA was
performed on the dataset. The results were presented and interpreted as above in Section 4.3. As shown
in Table 7, the ANOVA found significant differences existed within the dataset of fuel blend acidity
measurements. Of 21 comparisons made between different blends for acidity, 12 were found to be
significant and nine (9) were not, with no discernable pattern being observed between fuels blends. What
can be said is that EO is significantly different from all of the other blends tested.
E-22
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 23 of 36
Table 7. F-Test Results of Fuel Blend Comparison for Acidity*
Fuel
Blend
E10
E15
116
E30
E50
E85
EO
<0.0001
YES
0.029
YES
0.001
YES
<0.0001
YES
<0.0001
YES
<0.0001
YES
E10
0.334
NO
1
NO
1
NO
0.029
YES
0.19
NO
E15
0.932
NO
0.334
NO
<0.0001
YES
0.001
YES
116
1
NO
0.004
YES
0.029
YES
E30
0.029
YES
0.19
NO
E50
1
NO
p < 0.05 indicates a significant difference
*F-test performed after significant differences were identified using an ANOVA analysis of
the dataset.
4.8 Coefficient of Thermal Expansion
All materials expand or contract when their temperature changes. The degree of this expansion or
contraction is described by a material-specific coefficient of thermal expansion. Knowledge of this
coefficient and its use as a correction factor is imperative in making accurate liquid level determinations.
The storage temperature of fuels in USTs is constantly changing, albeit by relatively small amounts
compared to the average storage temperature. A measurement change can easily be produced by thermal
expansion/contraction under typical fuel storage conditions. Therefore, the coefficient of thermal
expansion must be known and used to make corrections to the measured fuel volume to allow accurate
storage volume determinations. Accurate volume calculations can only be obtained if the ethanol content
of a blend is known and used by a LD system. Figure 5 below presents the similar increasing trend of all
of the test blends as temperature increases. Regardless of ethanol content, the volume of fuel increased
with the increase in the temperature. The coefficient of thermal expansion for all fuels remained similar
at 0.001 (as presented in Table 1); therefore, if necessary, LD technologies have been compensating for
this magnitude of thermal expansion and most likely would not be affected by ethanol content.
E-23
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 24 of 36
Figure 5. Thermal Expansion Plot by Test Blend
4.9 Non-additive Volume Changes (Degree of Accommodation)
Because of the varying miscibility of gasoline, water and ethanol, it is expected that as an aliquot
of water is added to each of the test blends, the total volume change of the resulting BFW mixture was
less than the volume of that aliquot, and the separated, dense phase grew disproportionately to the added
volume of water. The relative total volume decrease is due to accommodation of polar water molecules
into the structure formed by the polar ethanol molecules referred to as the degree of accommodation.
Table 8 shows as the test blends increase in ethanol content, the amount of ethanol
accommodated within the polar water structure increases which results in a relative volume reduction
upon the addition of water. Results less than 1 show that the total volume is less than expected total
volume and with the exception of 116, all of the fuel bends were less volume than expected.
Table 8. Degree of Accommodation Summary for the Test Blends
Test Blend
EO
E10
E15
116
E30
E50
E85
Growth of Total
(Slope of A measured
A expected total
Volume
total volume/
volume)
0.9557
0.9953
0.9915
1.0039
0.9665
0.9838
0.9510
E-24
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 25 of 36
5.0 LEAK DETECTION TECHNOLOGY OPERATING PRINCIPLES
The standard test procedures are divided amongst five main categories of leak detection
technologies. Evaluation of operability of these technologies when applied to alternative fuel service
necessitates a basic understanding of the principles of operation of each technology category. Table 9
presents the categories and lists various technologies associated with each intended to represent the
most common methods and their operating principles within each category. In addition, Table 9
presents a brief description of the operating principle of each technology category. More detailed
descriptions of the test procedures and technologies associated with each are available on the EPA Office
of Underground Storage Tank (OUST) website1.
5.1 Volumetric versus Non-volumetric-Based Testing Technology Categories
The compendium of leak detection technologies can be delineated as being either volumetric or
non-volumetric. Each specific technology falls into one of these two categories; in some cases a
technology may apply to both categories. Table 9 shows the relationships between leak detection
technology categories and these technology types. Either type may be used to satisfy requirements of 40
CFR 280. The primary distinction between the two categorical procedures is that volumetric technologies
yield quantitative results (i.e. a reported leak rate) whereas non-volumetric technologies yield qualitative
results (i.e. only whether there is evidence of aleak or not when compared to a threshold value).
Volumetric technologies quantitatively measure leak rate from a UST based on changes in liquid
level in a tank. Various types of technology are available for measuring these changes, including floats,
load cells, and ultrasonic devices. They can be further categorized into methods that meet 40 CFR 280
requirements for precision testing; 0.1 gal/hr leak rate (e.g., tank or pipeline tightness tests) or a 0.2 gal/hr
leak rate (e.g., ATG systems or statistical inventory reconciliation [SIR] methods) respectively. Accurate
use of each volumetric technology requires knowledge of certain storage conditions and fuel properties so
that adjustments can be made to compensate for other factors that might produce a change in liquid level.
For example, the coefficient of thermal expansion must be known in order to allow volume corrections to
be made based on changes in the temperature of the stored product. Without this correction a volume
change that occurs as the storage temperature drops could be interpreted as a fuel leak or the actual
calculated leak rate may be inaccurate. Other corrections that may be necessary include fuel density
(based on temperature and ethanol content), air density (based on temperature above the stored liquid), or
the ground water level surrounding a tank.
E-25
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 26 of 36
Non-volumetric technologies make use of equipment that qualitatively identify when a leak is
occurring in a UST. While these technologies cannot be used to determine an actual leak rate in a UST
system, the signal from the technology can provide an indication that a tank might be leaking. Various
types of non-volumetric technology include acoustic measurements, water sensing equipment, external
tank monitoring systems, and interstitial sensors. These technologies can be used to detect sounds made
by fuel leaks through an orifice (i.e., tank shell), water present at the bottom of a tank, or liquids in the
interstitial space of a double-walled tank, respectively. A response from one of these technologies cannot
be used to calculate an exact volume or leak rate, but observation of a response provides the tank operator
with a clear indication that the integrity of the tank shell may have been compromised. Other non-
volumetric technologies include vapor and liquid out-of-tank monitoring in the excavated soil area or
ground water surrounding a UST. Tracers can also be used to detect the presence of a leak.
5.2 Automatic Tank Gauging System Technologies
Whereas manual tank gauging typically consists of "sticking" a UST with a long pole containing
graduated length markings, an ATG system relies on various physical properties of the storage system to
generate an electronic signal that can be converted into a value representing the volume in a tank. As
such, ATG systems are volumetric leak detection technologies.
An ATG system consists of a probe or sensor that is located inside the UST and a controller (or
console) that is mounted in an indoor location. The probe or sensor is used to generate the electronic
signal that is subsequently processed in the console to calculate volume and/or leak rate. The electronic
signal is generated in one of several ways, including:
• A float mounted to a probe (a liquid level method);
• A set of acoustic sensors to detect sound in the liquid or the air space above the liquid (a
sound transmission/reception method similar to sonar or radar);
• A load cell suspended in the liquid product (a buoyancy method); or
• A set of sensors to determine the electrical properties of a liquid (an electric
conductance/capacitance method).
E-26
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 27 of 36
Table 9. Leak Detection Technologies and Principles of Operation
VOLUMETRIC-BASED TECHNOLOGY CATEGORY
Technology
Principle of Operation
Automatic Tank Gauge (A TG) Systems
Magnetostrictive
Probes
Wire sensor inside a shaft detects presence of magnetic field, which indicates height of float
Ultrasonic or
Acoustic Methods
(speed)
Mass Buoyancy/
Measurement
Systems
Capacitance Probes
Sensor detects changes in fluid levels detecting a sound wave echo reflected from the interface
of water/fuel or fuel/air and calculates level based on speed of sound in the product
Buoyancy of probe is detected on a load cell and compared to tank geometry to calculate liquid
level
Detection is based on dielectric property of the stored liquid
Statistical Inventory Reconciliation (SIR) Methods
Traditional SIR
Continuous SIR
A SIR vendor performs analysis of liquid level data for evidence of tank tightness. Data are
collected using an ATG or by taking daily manual liquid level readings.
SIR vendor software performs temperature compensation and leak-test calculations on data
collected from designated input devices during tank quiet times.
Pipeline Methods (Piping)
Pressure Decay
Constant Pressure
Mechanical Leak
Detectors
Measures the change in pressure between the atmosphere and the pressurized product in the
line over time.
Sensors monitor change in volume at constant pressure.
Permanent installation on piping. Conducts leak tests every time the pump engages.
NON-VOLUMETRIC-BASED TECHNOLOGY CATEGORY
Fuel Sensitive
Polymers
Tracers
Acoustic Precision
Test
Vacuum /Pressure
Decay Test
Fiber optic cable is coated with a polymer that interacts with fuel. When fuel is present, the
light or current passing through the cable will be affected
Chemical markers (i.e., tracer) are added to the product and the surrounding soil is monitored
for the tracer
Detected sounds are used to identify potential leaks; an orifice is used to simulate the sound
produced as liquid or air leaks out of a system. This is accomplished using acoustic sensors
and microphones, and ultrasonic sensors and hydrophones.
Determine tank tightness by the decay rate of the vacuum or pressure established by the
method.
Dry Interstitial Integrity Monitoring Technologies
Vacuum /Pressure
Decay Monitoring
Technology uses an integral vacuum pump or pressurized system to continuously maintain a
partial vacuum or pressure within the interstitial space of double-walled tanks and double-
walled piping. Method is capable of detecting breaches in both the inner and outer walls of
double-walled tanks or double-walled piping
Wet Interstitial Integrity Monitoring Technologies
Liquid Filled
A liquid solution is used to fill the tank or piping interstice. The dual-point level sensor system
monitors the liquid level in the interstitial reservoir and sounds an alarm if the liquid level is
either too high (ingress of liquid) or too low (egress of liquid)
Sensor - liquid
ingress
Varies depending on the type of sensor and comes in multiple forms. Most examples include
use of refractive index or float switch
E-27
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 28 of 36
Table 9. Leak Detection Technologies and Principles of Operation (Continued)
Technology
Principle of Operation
Water Detection Technologies (A TG, Non-volumetric, Sensors)
Water Float
Density Float
Conductivity Water
Probe
Buoyancy of float allows the signal generated (magnetic field or capacitance) to coincide with
the top of the liquid layer based on the liquid density in comparison to the float density. These
floats are specifically designed for water detection and the density difference between water
and the fuel product.
Buoyancy of a float signals changes in product that compares density data changes over time to
assess the change in product quality due to water ingress. This float is sensitive to the aqueous
phase detection found in ethanol-blended fuels.
The probe detects water by measuring current flow when water contacts the probe. Used with
certain acoustic methods
Regardless of the method employed, the signal generated by any of these technologies is
combined with a specific set of other data (entered by the owner or operator) and processed to calculate a
volume of liquid in in the UST. The console contains a processor that compares calculated volumes at
different times (during which the UST is not dispensing or receiving fuel) to determine if any observed
difference is due to a leak or some other factor.
Depending on the ATG system in use, the associated processor must "correct" the calculated
volume for other tank conditions. For example, the volume derived from liquid height obtained using a
float system, electrical property, or acoustic sensor must be adjusted for liquid expansion or contraction
produced by changes in temperature of the stored liquid. Similarly, the result obtained from a pressure,
buoyancy, or sound velocity reading must incorporate a liquid or air density factor (which also varies with
temperature) to accurately calculate volume. Given the proper inputs, ATG systems will yield
information on volume of stored fuel and on calculated leak rates during a leak tests.
Most probes used for ATG systems are also equipped with a water float. The water float is
located on the bottom of the tank where water may collect as a denser phase than the fuel. As the water
or water phase (water-ethanol mixture) height increases, the float rises and transmits an electronic signal
proportional to the level of the denser phase in the bottom of the tank. The inventory measurement would
also register an increase in volume given water ingress, although the quantified amount may not be
accurate depending on the water solubility of the fuel and proportion of ethanol in the fuel.
5.3 Statistical Inventory Reconciliation Technologies
SIR technologies, which can be either volumetric (quantitative) or non-volumetric (qualitative),
rely on the comparison of manually or automatically-collected liquid level data and fuel delivery and
dispensing (sales) records. Statistical evaluation of the data and records is performed, usually by a vendor
E-28
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 29 of 36
or with a vendor software program, to determine if the stored volume reconciles with deliveries into and
out of a tank. A discrepancy in the volumes may then be reported as a leak or some other event. SIR is
subject to potential sources of human and measurement error when collecting or recording the records. In
addition to errors in metering the fuel delivery and dispensing volumes, storage tank volumes may change
between readings due to temperature differences, fuel transfer between manifold tanks, fuel volatilization,
or introduction of water into the UST. Traditional SIR does not "correct" for these variables; however
continuous SIR has multiple input devices and can compensate for these variables.
5.4 Pipeline Leak Detection Technologies
Pipeline leak detection can be conducted using volumetric or non-volumetric methods.
Volumetric methods use fluid flow instrumentation to monitor flow rate of a moving fluid through the
underground piping of a UST system at one or more locations, or the static pressure in a sealed pipe
system. Flow measurement devices are usually based on pressure; however, these devices could also use
a displacement piston or graduated cylinder instead of a pressure-based measurement device. The liquid
within the piping is non-compressible, and therefore, a single flow measurement or a comparison of the
flows at different locations will indicate if a leak has occurred along the piping. By necessity, several
properties of the conveyed fluid must be known to correctly convert the measurement into a flow rate.
Critical parameters needed by most non-compressible flow monitoring systems include fluid density and
viscosity. Even without these parameters comparison of the pressures at different monitoring points can
indicate the presence of a leak. The rate cannot be accurately determined without product-specific data.
Friction losses may also need to be calculated in high-volume or long piping sections before a leak can be
confirmed. Static pressure devices installed on a non-leaking pipe section should show the pressure is
maintained over the duration of the test. Temperature correction may be needed if the product
temperature is susceptible to change during the test, as this will produce product expansion or contraction,
which in turn will change the static pressure.
5.5 Non-volumetric Leak Detection Technologies
Vapor-phase out-of-tankproduct detectors are non-volumetric technologies that employ
instruments designed to detect hydrocarbon product vapors in the vadose zone or backfill area around a
UST. The technology relies on the high volatility of some chemical components of gasoline and the
ability to measure them at low concentrations. Thus, sampling the "soil gas" surrounding a UST or
within the tank top sump, for example, for gasoline components such as benzene or toluene can be used to
detect UST system leaks. The fuel leak rate, however, cannot be quantified using this method.
E-29
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 30 of 36
A variation of this technology is an external tracer. In this system a volatile tracer compound is
added to the product stored in a UST, and the tank backfill around the UST is monitored for this tracer.
The tracer must be able to become completely mixed into the product, yet be volatile enough to separate
from the fuel after a release from the tank and migrate through the tank backfill to a monitoring location
where it is collected and later analyzed in a laboratory by gas chromatography - mass spectrometry.
Liquid-phase out-of-tankproduct detectors are non-volumetric technologies that employ
instruments designed to detect a free-product layer on the water table in an observation well near a UST
or on water collected in a dispenser sump, for example. Free-product detectors are used commonly in site
remediation monitoring wells and rely on the immiscibility of petroleum products and water. Gasoline
that leaks from a UST and intercepts the water table will rise to the top of the water column in an
observation well and be detectable as a layer of product on top of the water. Although leaks can be
detected using these detectors, the leak rate cannot be determined.
Acoustical methods (not to be confused with the ultrasonic ATG technology) make use of an
acoustic sensor to detect the sound of fuel leaking out of a UST or water or air leaking into a tank. If
desired, a tank can be placed under a slight negative pressure test condition to induce air flow into the
tank. Interfering sounds must be eliminated to use this technology, and only qualitative leak
determinations are possible. In addition, if the ground water level is above the bottom of a UST, water
may enter the tank without an audible sound. Therefore, these technologies include a water detection
component. One kind is based on conductivity and referred to as a conductivity water probe. Current
flow is measured by a gauge when water ingress contacts a probe while under vacuum. In ethanol-
blended gasoline, it is difficult to determine water ingress due to minimal conductivity of the transition
zone between low ethanol-blend gasoline and phase separation (as discussed in Section 4.2), and will not
work in high ethanol-blends due to the high conductivity of the high ethanol blend.
Interstitial integrity monitoring is a technology used on secondarily contained tanks and piping.
Dry interstitial monitoring is performed in one of two ways: (1) a vacuum or pressure is induced in the
interstitial space, and the pressure differential is monitored in the space, or (2) a sump (or reservoir) is
connected to the interstitial space to allow liquid leaking into the space to collect and be detected by
liquid detection systems. Wet interstitial monitoring is performed with the interstice full of liquid
(usually brine) with a change in liquid level indicating a release into or out of the interstice. These options
can be performed continuously or intermittently, and no other parameters must be monitored to make
adjustments based on the observations.
Traditional water detection technologies make use of the insolubility of water in non-ethanol
blend gasoline (immiscibility) and are specifically calibrated to detect the density of water. The
E-30
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 31 of 36
unexplained presence of water in a tank is an indication of a potential leak and must be investigated.
When water sinks to the bottom of a UST and forms a separate layer, a float where density is greater than
gasoline but less than water can be used to generate and send a signal to an ATG console. Because these
technologies are now needed to function in a wide range of fuel densities, a traditional water float will be
too dense to float on the interface layer between the aqueous phase and ethanol-blended fuel.
Aqueous phase density floats, water detection technologies that are calibrated for aqueous phase
detection, are density-based technologies that address concerns with ethanol-blended fuel and its ability to
absorb water. When enough water is absorbed, the ethanol and water separate from the hydrocarbon
phase and settle to the tank bottom. The density of this water-ethanol bottom; however, is less than that
of water alone, and as a result, traditional water floats do not consistently detect this aqueous phase.
These newly developed technologies employ either a float with a density sensitive to ethanol-water
mixtures, or a sensor to directly measure the density of the ethanol-water mixture at the bottom of a tank.
E-31
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 32 of 36
6.0 SUITABILITY ASSESSMENT OF LEAK DETECTION TECHNOLOGIES IN
ETHANOL-BLENDED FUEL
Most LD technologies have not been evaluated when in ethanol-blended fuel service; however,
many are used in E10 fuel service and if not relying on conventional water floats to detect water ingress,
are performing appropriately in the field. As a result, observations on the suitability of LD technology
with respect to its operability in ethanol-blended fuel service are based on stakeholder input, laboratory
tests, and hypotheses involving critical fuel properties. Table 10 presents an assessment of the suitability
of several LD technologies with respect to operability. Some technologies are expected to operate
properly in ethanol blended fuels due to their somewhat simple operating principles. For example, a
piping pressure decay system is expected to work properly with any non-compressible fluid provided that
adequate temperature monitoring is also conducted. This is because the technology represents a static
system that can only be affected by loss of fluid or expansion/contraction of the fluid. On the other hand,
the interaction of some technologies with critical fuel parameters, or the interaction of the fuel parameters
themselves, makes the operability of some technology uncertain. For example, while most parties believe
that a fuel float-based technology should be able to detect changes in liquid levels, some questions exist
as to whether the simultaneous loss of fuel and ingress of water will be adequately detected. Water
absorption into ethanol may or may not produce a change in liquid volume, and if water does not drop to
the bottom of the tank, ingress is not expected to be detected. As the ethanol content increases in the fuel
blend, water-fuel interactions and water-ethanol detection becomes more problematic.
As discussed previously, Table 10 provides observations for low ethanol content (low-E, up to
15%) and high ethanol content (high-E, E51 - E85) fuel blends. The question being posed by technology
category with respect to operating principle is:
• Is the Technology Capable of Detecting a Leak at the Regulatory Level? This criterion assesses
whether the response generated by the technology is expected to allow the user to derive the
correct conclusion regarding a leak or no-leak condition while operating in a UST at the
regulatory level.
The three possible suitability assessments were developed to the above question based on input
from stakeholders (NWGLDE, regulators, testing company representatives, and technology vendors).
These assessments are identified in Table 10 according to color coding, include the following:
• Technology is expected to be suitable for indicated use (GREEN). The operating
principle of the technology is such that no major limitations or interferences are expected to
exist when employed in the listed service as compared to gasoline service.
E-32
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 33 of 36
• Technology has limitations with the indicated use (YELLOW). One or more of the
principles upon which the technology operates is not expected to be suitable when employed
in the listed service. Without modification, the technology may or may not operate properly.
A series of tests could be conducted to demonstrate that the technology performs as expected
in the listed service.
• Technology is expected to not be suitable for indicated use (RED). One or more
principles upon which the technology operates is unsuitable when employed in the listed
service.
As all technologies are different, have different algorithms, and are influenced by human inputs
and installation, these conclusions may not be appropriate for every technology in a category. This paper
discusses the relationship between fuel properties and operating principles against the performance
standards established in the federal LD requirements. The potential negative impacts are highlighted in
the previous sections for consideration; however, in most cases, a change in liquid level will be detected
whether it decreases due to a leak or increases due to water intrusion. In some cases, the technology may
need to be slightly modified to recognize these changes at the regulatory level with adjustments of
threshold values and monitoring data processing.
E-33
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 34 of 36
Table 10. Suitability of Existing Leak Detection Technology for Ethanol-Blended Fuel
LD Category
and Technology
Is the Technology Capable of
Detecting a Leak/Water Ingress at the
Regulatory Level?
Low-E
(up to 15%)
High-E
(51 to 83%)
Comments
VOLUMETRIC METHODS
Automatic Tank Gauge (ATG) SystemsA
Magnetostrictive
Probe*
Ultrasonic or
Acoustic
Methods (speed)
Mass Buoyancy/
Measurement
System
Capacitance
Probe
Gasoline-ethanol-water has unknown
properties and therefore may not be able
to accurately diagnose the extent of a
leak. In addition, multiple liquid phases
in a storage tank will make it difficult to
derive an accurate dielectric constant for
each observed phase. Although
capacitance is being used in other LD
technology categories, the traditional
capacitance ATG probes are not
expected to operate properly.
Fuel properties are needed; liquid level
changes will most likely be detected. Water
ingress detection may have limitations when
traditional water floats are used.
Fuel properties are needed; liquid level
changes will most likely be detected. Water
ingress detection may have limitations when
traditional water floats or conductivity water
probes are used.
Fuel properties are needed; liquid level
changes will most likely be detected. Water
ingress detection may have limitations when
traditional water floats are used.
No longer commercially available; rarely
used.
Statistical Inventory Reconciliation (SIR) Methods
Traditional SIR
Continuous SIR
Comparing a change in condition using
regularly collected data; assumes no changes
in data collection process. Fuel properties
are needed; liquid level changes will most
likely be detected.
Pipeline Methods (Piping)
Pressure Decay
Constant
Pressure
Mechanical
Leak Detector
Dynamic methods require fuel properties
(coefficient of thermal expansion, viscosity)
to calculate or compare against a threshold;
properties should remain constant in a given
piping system, so if known, the methods
should operate properly.
AWater detection is a requirement of ATG systems that was evaluated separately in this paper.
*See Appendices for testing methods and results (A, C, D, E, and F).
Technology is expected to be suitable for indicated use.
Technology has limitations with the indicated use.
Technology is expected to be not suitable for indicated use.
E-34
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 35 of 36
Table 10. Suitability of Existing Leak Detection Technology for Ethanol-Blended Fuel (Continued)
LD Category
and
Technology
Is the Technology Capable of
Detecting a Leak/Water Ingress at
the Regulatory Level?
Low-E
(up to 15%)
High-E
(51 to 83%)
Comments
NON-VOLUMETRIC METHODS
Fuel Sensitive
Polymers*
Hydrocarbon
(HC) layer
Tracers
Acoustic
Precision Test
Vacuum/
Pressure Decay
Test
When the product is not dominated by
hydrocarbons, the polymers may not react.
Reduced petroleum content of high-E blends
may produce difficulty in forming a free phase
for detection.
Tracer must be proven compatible with the
product, not foreseen as an issue given the
available tracer compounds.
Not effected by fuel properties; however, no
reliable database of sounds expected during
leakage. Relies on human interpretation of
noises during tank tightness testing.
Measuring a change of vacuum or pressure
overtime. Static method does not require exact
fuel properties.
Dry and Wet Interstitial Monitoring Technologies
Vacuum/
Pressure Decay
Liquid Filled
Sensors -
liquid ingress*
Should not be affected if liquid (product,
water, or mixture of the two) is sufficiently
dense or in sufficient quantity to trigger a
change in the static reading.
Water/Aqueous Phase Detection Technologies'
Water Float*
Density Float*
Conductivity
Water Probe
Potential effect on operation due to miscibility
of water and ethanol-blended fuels.
Developed for use with E-blended fuel at the
bottom of the tank. Will not float until phase
separation occurs.
This will not work with High-E because it is
highly conductive.
'See Appendices for testing methods and results (A, C, D, E, and F).
^Water detection is a requirement of ATG systems that was evaluated separately in this paper.
Technology is expected to be suitable for indicated use.
Technology has limitations with the indicated use.
Technology is expected to be not suitable for indicated use.
E-35
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 36 of 36
7.0 REFERENCES
1. EPA Office of Undground Storage Tanks. December 17, 2014; Available from:
http://www.epa.gov/oust/.
2. Quality Assurance Project Plan for Biofuel Properties and Behavior Relevant to Underground
Storage Tank Leak Detection System Performance. U.S. Environmental Technology Verification
Program, Battelle, April 2013.
3. Quality Assurance Project Plan for Biofuel Properties and Behavior Relevant to Underground
Storage Tank Leak Detection System Performance Addendum. U.S. Environmental Technology
Verification Program, Battelle, November 2013.
4. Quality Assurance Project Plan for Verification of Underground Storage Tank Automatic Tank
Gauging Leak Detection Systems. U.S. Environmental Technology Verification Program,
Battelle, 2011.
5. Kass, M.D., Theiss, T.J., et al., Intermediate Ethanol Blends Infrastucture Materials
Compatibility Study: Elastomers, Metals, and Sealants. March 2011, Oak Ridge National
Laboratory.
6. ASTM, D5798 Standard Specification for Ethanol Fuel Blends for Flexible-Fuel for Automotive
Spark-Ignition Engines. 2014.
7. Environmental Technology Verification Report: Underground Storage Tank Automatic Tank
Gauging Leak Detection Systems, Veeder-Root Standard Water Float and Phase-Two Water
Detector. U.S. Environmental Technology Verification Program, Battelle, 2012.
8. Environmental Technology Verification Report: Underground Storage Tank Automatic Tank
Gauging Leak Detection Systems, Franklin Fueling Systems TSP-IGF4 Water Float and TSP-
IGF4P Float. U.S. Environmental Technology Verification Program, Battelle, 2012.
9. Illustrated Glossary of Organic Chemistry, [cited 2014 June 10]; Available from:
http://www.chem.ucla.edU/harding/IGOC/D/dielectric constant.html.
E-36
------- |