Batrelle
The Business of Innovation
EPA/600/R-16/028 | February 2016 | www.epa.gov/research
Suitability of Leak Detection
Technology for Use
In Ethanol-Blended Fuel Service
Prepared by
Baneiie
The Business of Innovation
Under a cooperative agreement with
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
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
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
Version: 1
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) 19
4.6 Viscosity 21
4.7 Acidity 23
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 29
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 22
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 21
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 25
Table 9. Leak Detection Technologies and Principles of Operation 27
Table 10. Suitability of Existing Leak Detection Technology for Ethanol-Blended Fuel 34
-------
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
-------
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 El5) 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).
-------
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.
-------
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 571,000' 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
-------
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 (El 5) 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
-------
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
-------
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.
-------
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, E15, E30,
-------
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.
-------
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 at25°C
-------
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
-------
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.
1.2
6 8 10
Liquid Draw Level
12
14
16
•EO
•E10
•E15
116
•E30
•E50
•E85
Figure 1. Phase Separation Plot of UV-V Measurements
4.3 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
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 17 of 36
found to be in the same range. The change in water content did not appear to have an effect on
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.
100000000
10000000
1000000 -
100000 -
10000
1000
100
10 -
1
0.00%
0.25%
0.50%
Percent Water
2.50%
5.00%
•EO
-E10
E15
-116 E30 E50 E85
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 /"-values take into account the fact that multiple comparisons are being performed by
applying the Sidak adjustment to the reported significance level.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 18 of 36
In Table 3, ap-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
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, E15, 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
£85
EO
1
NO
1
NO
1
NO
<0.0001
YES
O.OOOl
YES
<0.0001
YES
E10
1
NO
1
NO
<0.0001
YES
O.OOOl
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
O.OOOl
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.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 19 of 36
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
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.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 20 of 36
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
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.
0.82
0.80
_j 0.78 ^
E
eiO.76
•5 0.74
0 0.72 -
0.70
0.68
0.00%
-EO
0.25%
0.50%
Percent Water
-E10
-E15
116
•E30
2.50%
E50
5.00%
E85
Figure 3. Density Plot by Test Blend and Water Content
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 21 of 36
Table 5. F-Test Results of Fuel Blend Comparison for Density*
Fuel
Blend
E10
E15
116
E30
E50
ESS
EO
0.0001
YES
O.OOOl
YES
0.0001
YES
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
E10
1
NO
0.839
NO
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
E15
0.821
NO
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
116
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
E30
O.OOOl
YES
0.0001
YES
E50
O.OOOl
YES
p < 0.05 indicates a significant difference
*F-test performed after significant differences were identified using an ANOVA analysis of the
dataset.
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.
-------
1.4 -1
1.2
f*3
S^0.8
•fo.6
| 0.4 -
> 0.2
0
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 22 of 36
0.00%
•EO
0.25%
0.50%
Percent Water
•E10
E15
116
•E30
2.50%
•E50
5.00%
E85
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.
Table 6. F-Test Results of Fuel Blend Comparison for Viscosity*
Fuel
Blend
E10
E15
116
E30
E50
ESS
EO
1
NO
0.017
YES
<0.0001
YES
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
E10
0.037
YES
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
E15
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
O.OOOl
YES
116
0.001
YES
O.OOOl
YES
O.OOOl
YES
E30
O.OOOl
YES
O.OOOl
YES
ESO
O.OOOl
YES
p < 0.05 indicates a significant difference
*F-test performed after significant differences were identified using an ANOVA analysis
of the dataset.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 23 of 36
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.
Table 7. F-Test Results of Fuel Blend Comparison for Acidity*
Fuel
Blend
E10
E1S
116
E30
E50
ESS
EO
<0.0001
YES
0.029
YES
0.001
YES
<0.0001
YES
O.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
ESO
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
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 24 of 36
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.
5.15
5.1
5.05
5 -
4.95
4.9
•EO
10
•E10
15 20
Temperature, °C.
25
30
E15
116
-E30
-E50
-E85
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
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 25 of 36
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 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
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 a leak 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
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 26 of 36
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.
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);
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 27 of 36
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).
Table 9. Leak Detection Technologies and Principles of Operation
Technology
[ Principle of Operation
VOLUMETRIC-BASED TECHNOLOGY CATEGORY
Automatic Tank Gauge (ATG) Systems
Magnctostrictivc
Probes
Ultrasonic or
Acoustic Methods
(speed)
Cagaejtenee Probes
Mass Buoyancy/
Measurement
Systems
Wire sensor inside a shaft 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 liquid
level
Statistical Inventory Reconciliation (SIR) Methods
Traditional SIR
Continuous SIR
A SIR vendor performs analysis of liquid level data for evidence of tank tightness. Data arc
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-RASED TECHNOLOGY CATEGORY
Fuel Sensitive
Polymers
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
Tracers
Chemical markers (i.e., tracer) are added to the product and the surrounding soil is monitored
for the tracer
Acoustic Precision
Test
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.
Vacuum /Pressure
Decay Test
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
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
-------
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 (ATG, 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.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 29 of 36
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
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-tank product detectors are non-volumetric technologies that employ
instruments designed to detect hydrocarbon product vapors in the vadose zone or backfill area around a
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 30 of 36
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.
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-tank product 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
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 31 of 36
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
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.
-------
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.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 3 3 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.
-------
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)
Capacitance
Probe
Mass Buoyancy/
Measurement
System
He
U i
e> geet dj orwratepro
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.
No longer commercially available; rarely
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.
Statistical Inventory Reconciliation (SIR) Methods
SIR -Manual
SIR -Data from
ATG
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.
Methods of Release Detection for 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.
^Water 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.
-------
Suitability of Leak Detection Technology for Use In Ethanol-Blended Fuel Service
Date: 12/31/2014
Version: 1
Page 3 5 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
Vapor Out-of-tank Methods
Tracers
Tracer must be proven compatible with the
product, not foreseen as an issue given the
available tracer compounds.
Liquid Out-of-tank Methods
Hydrocarbon
(HC) layer
Reduced petroleum content of high-E blends
may produce difficulty in forming a free phase
for detection.
Fuel Sensitive
Polymers*
When the product is not dominated by
hydrocarbons, the polymers may not react.
Acoustic Methods
Sound
Detection
(Tanks)
Sound
Detection
(Piping)
Multiple liquid phases in a UST or
piping and potential interfering
sounds will make it difficult to
identify air, water, or leaked fuel
entering the tank while under
vacuum.
No reliable database of sounds expected
during leakage. Relies on hum an
interpretation of noises during tank tightness
testing.
Interstitial Methods
Liquid Filled
Sensors -
liquid ingress*
Vacuum
Pressure
Should not be affected if liquid (product,
water, or mixture of the two) is sufficiently
dense or in sufficient quantity to trigger a
reading.
Water/Aqueous Phase Detection Methods'
Water Float*
Potential effect on operation due to miscibility
of water and ethanol-blended fuels.
Density Float*
Conductivity
Water Probe
Developed for use with E-blended fuel at the
bottom of the tank.. Wrill not float until phase
separation occurs.
Current flow increases very slowly when there
is water ingress into a tank with Low-E. This
will not work with High-E because it is
conductive.
Methods for Release Detection in Piping
Pressure
Decay*
Static method does not require exact fuel
properties
*See Appendices for testing methods and results (A, C, D, E, and F).
AWater 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.
-------
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.
-------
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
Underground Storage Tank Leak Detection System Performance (QAPP) and adhered to the
quality system defined in the ETV AMS Center Quality Management Plan (QMP)1. A
A-l
-------
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
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
A-2
-------
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 capabilities; therefore, only the appropriate tests
in the QAPP were conducted. 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 Battelie'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).
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 D4057- and D5854-;
volumetric blend stocks of ethanol (or isobutanol) and gasoline were prepared according to
A-3
-------
ASTM D7717-. 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 D2032 and ASTM D48151, 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
E2032
ASTM
04815s
ASTM
E2032
ASTM
04815s
ASTM
D55012
ASTM
E2031-
ASTM
D55015
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 (E15,
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
E203Z
ASTM
D48152
ASTM
E203Z
ASTM
D48152
ASTM
D55012
ASTM
E2032
ASTM
D55015
ASTM
E2031-
ASTM
D55015
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%
*Water content was measured as % mass, not % volume
A-6
-------
Table 5. Test Blend Analytical Results for Laboratory-Scale and Pressure Decay Testing
Test
Blend
EO
E15
116
E30
E50
ESS
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
1 7.61
1 7.00
17.08
28.77
46.85
83.21
Viscosity*
mm2/sec
0.555
0.5467
0.5922
0.6037
06001
0.648
0.6576
0.6947
0.8345
1 .2206
Density1
E/mL
0.7601
0.7608
0.7659
0.7681
0.7672
07681
0.7699
0.7712
0.7781
0.7827
Acidity
%
Mass
0.0008
0.0008
0.0008
0.0012
0.0012
o.ooos
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
5*313-111-21
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 QMP2 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
quality objectives (MQOs) for the methods utilized by ICFTL and Battelle Labs are described in
Table 6.
A-7
-------
Table 6. Data Collection Quality Control (QC) Procedures and Measurement Quality
Objectives (MQO) for Analytical Methods
Method Designation: Method
Title
ASTM D4815: Standard Test
Method for Determination of
MTBE, ETBE, TAME, DIPE,
tertiary -Amyl Alcohol and Ci to
€4 Alcohols in Gasoline by Gas
Chromatography-
ASTM D5501: Standard Test
Method for Determination of
Ethanol and Methanol Content in
Fuels Containing Greater than
20% Ethanol by Gas
Chromatography2
ASTM D5501: Modified to
analyze Isobutanol
ASTM E203: Standard Test
Method for Water Using
Volumetric Karl Fischer
Titration2
ASTM D1613 Standard Test
Method for Acidity in Volatile
Solvents and Chemical
Intermediates Used in Paint,
Varnish, Lacquer and Related
Products12
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
QC Procedures
Annual multi-point
calibration curve and with
newly installed column
and continuing QC check
samples every 10
samples*
Annual multi-point
calibration curve and with
newly installed column
and continuing QC check
samples every 10
samples*
Annual multi-point
calibration curve and with
newly installed column
and continuing QC check
samples every 10
samples*
QC check samples every
10 samples*
QC check samples every
10 samples*
QC check samples every
10 samples*
Daily Check
MQOs
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
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
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
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
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
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.
A-8
-------
Method Designation: Method
Title
ASTMD2624: Electrical
Conductivity11
ASTMD445: Standard Test
Method for Kinematic Viscosity
of Transparent and Opaque
Liquids (and Calculation of
Dynamic Viscosity)—
QC Procedures
Daily instrument check of
probe
QC check samples every
10 samples*
MQOs
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
D4815-, 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 D5501-,
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 E2032, 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
-------
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
Title
PEA Material
Acceptance Criteria
ASTM
D4052
ASTM
D445
ASTM
D55019
ASTM
D48158
Standard Test Method for Density,
Relative Density, and API Gravity of
Liquids by Digital Density Meter
Fluka Standard
N. 10 ISO 170257 ISO
Guide 34
Standard Test Method for Kinematic
Viscosity of Transparent and Opaque
Liquids (and Calculation of Dynamic
Viscosity)
Fluka Standard
N. 10 ISO 170257 ISO
Guide 34
Standard Test Method for Determination
of Ethanol and Methanol Content in
Fuels Containing Greater than 20%
Ethanol by Gas Chromatography
NSIT
SRM 2900
Standard Test Method for Determination
of MTBE, ETBE, TAME, DIPE,
tertiary-Amyl Alcohol and Ci to C-n
Alcohols in Gasoline by Gas
Chrom atography
NIST
SRM 2287
Modified
ASTM
D5501
ICFTL In-House Isobutanol Method
Spectrum Calibration
standard for
Isobutanol
Within 10% of the
target concentration,
repeat analysis if out of
range
ASTM
E2037
Standard Test Method for Water Using
Volumetric Karl Fischer Titration
(Procedure §10)
NIST
SRM 2287
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 QMPa 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
A-ll
-------
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
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
SharePoint 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
-------
Table 9. Underground Storage Tank Leak Detection Stakeholder Committee
Last Name
Barbery
Bareta*
Baustian
Boucher
Bradley*
Brauksieck
Brevard
Chapin
Cochefski
Cornell
Courville
D'Alessandro
Dockery
Brack
Emmington
Fenton
Fisher
Flora
Folkers
Geyer
Gordji
Henderson
Hoffman
Indest
Johnson*
Jones
Juranty*
Keegan
Kubinsky
Lauen
Marston
McKernan
McMillan
Mills
Moore*
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
Kristy
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
Renewable Fuels Association (RFA)
A-13
-------
Last Name
Moureau
Muhanna*
Neil
Nelson
Parnell
Peters*
Poxson*
Purpora
Ramshaw
Reid
Renkes
Robbins*
Rollo*
Sabo
Scheib
Smith*
Thuemling
Toms
Wilcox
Wilcox
Young
First Name
Marcel
Shaheer
Peter
Bill
Brian
Heather
Marcia
Steve
Chris
Kent
Bob
Helen
Peter
Lorraine
Jeff
Tim
George
Patrick
Craig
Ken
Greg
Company
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)
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 Standard Practice for Manual Sampling of Petroleum and Petroleum
Products. August 2011.
5. ASTM, D5854 Standard Practice for Mixing and Handling of Liquid Samples of
Petroleum and Petroleum Products. May 2010.
6. ASTM, D7717 Standard Practice for Preparing Volumetric Blends of Denatured Fuel
Ethanol and Gasoline Blendstocks for Laboratory Analysis. August 2011.
7. ASTM, E 203 Standard Test Method for Water Using Volumetric Karl Fischer Titration.
November 2008.
A-14
-------
8. ASTM, D4815 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 Standard Test Method for Determination ofEthanol andMethanol
Content in Fuels Containting Greater than 20% Ethanol by Gas Chromatography. April
2013.
10. ASTM, D1613 Standard Test Method for Acidity in Volatile Solvents and Chemical
Intermediates Ulsed in Paint, Varnish, Lacquer, and Related Products. July 2012.
11. ASTM, D4052 Standard Test Method for Density, Relative Density, and API Gravity of
Liquids by Digital Density Meter. 2011.
12. ASTM, D28 7 Standard Test Method for API Gravity of Crude Petroleum and Petroleum
Products (HydrometerMethod). June 2006.
13. ASTM, D2624 Standard Test Methods for Electrical Conductivity of Aviation and
Distillate Fuels. February 2010.
14. ASTM, D445 Standard Test Method for Kinematic Viscosity of Transparent and Opaque
Liquids (and Calculation of Dynamic Viscosity). May 2012.
A-15
-------
Appendix B
Fuel Property Testing Methods and Data Results
-------
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
the QAPP2:
a) 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.
b) 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.
c) 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.
d) 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
B-l
-------
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 D1613—, density by ASTM
D4052-, viscosity by ASTM D445-, and water and ethanol content by either ASTM E2031 (for water)
and ASTM D55012 or ASTM D4815- (for ethanol) depending on their anticipated water and ethanol
contents. Where appropriate, samples were analyzed for isobutanol concentration by a modified ASTM
D55012. After sampling, conductivity was measured by ASTM D2624n and density was measured by
ASTM D287— directly in the graduated cylinder. Each intrinsic property was measured in triplicate on
the same sample.
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
B-2
-------
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:
dV
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 (dV/dT) 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
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
B-4
-------
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 D7451— 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
Precision
Requirements
• CV<15%for
measurements on
triplicate samples
• r2> 0.90 for volume
vs. temperature curve
• CV<15%for
measurements on
triplicate samples
• T2> 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
co nc entration
• Ethanol
co nc entration
" EO concentration
• Temperature
•
concentration
• EO concentration
• Dye solution
added
• 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
ASTM D5501 and
D4815
ASTM E203
ASTMD161310
ASTMD28712
ASTMD4052
ASTM D44514
EMCEE Model 11 52;
ASTMD262413
HachDRSOOOUV-
Vis
Spe ctrophotom eter
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
±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
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
instrum entation.
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:
* = EP=i*i Equations
where X is the average value of n number of measurements, x; (i = 1,2,3)
B-7
-------
• Standard Deviation: The standard deviation (SD) of a set of triplicate measurements made on
BFW mixtures was calculated using Equation 4 as follows:
SD =
'6
i=i
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
CV=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 Statistics16 and the formulae are not repeated here.
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
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.
B-9
-------
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 completely homogenize, 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.
0.82
0.80
_j 0.78 -
E
oiO.76
£ 0.74
Q 0.72 -
0.70 -
0.68
0.00%
0.25e
2.50%
5.00%
Percent Water
-EO E10 E15 116 E30 E50 E85
Figure 1. Plot of density (g/mL) for all BFW mixtures.
B-10
-------
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
I
o.
100000000
10000000
1000000
100000
10000
1000
100
10
1
0.00%
0.25%
0.50%
Percent Water
2.50%
EO E10 E15 116 E30 E50 E85
Figure 2. Plot of conductivity (pS/m) for all BFW mixtures.
5.00%
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
£85
8304444
7883333
8064444
9894444
11172222
B-ll
-------
1.4
1.2 H
£ 0.8
T3 0.6 H
e
| 0.4
0.2 -
0
0.00%
0.25%
0.50%
Percent Water
2.50%
EO E10 E15 —116 E30 E50 E85
Figure 3. Plot of viscosity (mm2/s) for all BFW mixtures.
5.00%
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 -
| 0.0006 -
< 0.0004
0.0002 -
0
0.00%
0.25?
0.50%
Percent Water
-EO
-E10
-E15 116
-E30
E50
2.50%
E85
5.00%
Figure 4. 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
ESS
0.0015
0.0015
0.0016
0.0015
0.0015
Table 8. Intrinsic Properties of E30
Parameter
Density
(g/mL)a
Conductivity
OS/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%
5.00%
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)
0.556(4.61)
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)
15.6(11.0)
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
c
B-13
-------
Parameter
Water
Ethanol
(%v/v)
Ethanol
(% mass)
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%
Average Value and (CV °/o)
Top Layer of BFW
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
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
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
-------
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%
5.00%
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
0.9628
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
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
0.9719
B-15
-------
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.
Table 10. Degree of Accommodation Summary for the 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
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
-------
in
•*-•
'E
s
E
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.
Figure 8. One replicate from E10 Interface Experiment.
B-19
-------
Figure 9. One replicate from ESS Interface Experiment
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, D1613 StandardTest Method for Acidity in Volatile Solvents and Chemical
Intermediates Ulsed in Paint, Varnish, Lacquer, and Related Products. July 2012.
3. ASTM, D4052 Standard Test Method for Density, Relative Density, and API Gravity of Liquids
by Digital Density Meter. 2011.
4. ASTM, D445 StandardTest Method for Kinematic Viscosity of Transparent and Opaque Liquids
(and Calculation of Dynamic Viscosity). May 2012.
5. ASTM, E 203 Standard Test Method for Water Using Volumetric Karl Fischer Titration.
November 2008.
6. ASTM, D5501 Standard Test Method for Determination ofEthanol andMethanol Content in
Fuels Containting Greater than 20% Ethanol by Gas Chromatography. April 2013.
7. ASTM, D4815 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 Standard Test Methods for Electrical Conductivity of Aviation and Distillate
Fuels. February 2010.
9. ASTM, D28 7 Standard Test Method for API Gravity of Crude Petroleum and Petroleum
Products (HydrometerMethod). June 2006.
10. ASTM, D7451 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-20
-------
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.—' —
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
C-2
-------
on the differences in refractive index between air (dry condition) and liquid (wet condition indicative of a
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-2
-------
,-Cord grip
—, 1/2" rigid conduit
__x (to Console)
- Weatherproof junction box
4.3" (L) 1.5" (W) 0.5" (H)
Sensor Switch
i Fiberglass tank
Figure 1. Optical Sensor
T
8.86 inch (225 mm)
Length of detection
Polymer
Strip
Liquid Float
Liquid Float
7.5 inch
(190.5mm)
High level
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)
C-4
-------
C2.3 Capacitance and Conductance (Complex Impedance) (C/C)
The C/C Sensor is used primarily to determine the approximate 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. 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 complex impedance, measured by the C/C sensor, also
changes. 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 liquid and interface
height. Advantages include precise response to complex impedance changes in alcohol-fuel blends and
alcohol-fuel-water mixture.
C3
Figure 3. Capacitance and Conductance (Complex Impedance) (C/C)
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
C-5
-------
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
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 B1.1).1
Groundwater used for this testing was collected from the tap in Battelle'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
C-6
-------
matrix including the test blend, number of replicates, and tests performed. The number of replicates 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
C-7
-------
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
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 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 at the lowest segment changed enough to exceed a gross cutoff for signaling
fluid change. 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
Average Recovery Time
Liquid Activation Height
Specificity
Accuracy
(qualitative only)
Accuracy
(quantitative only)
Precision
(quantitative only)
Evaluation Metric
Difference between actuation time and
test start times
Average of difference between
recovery and test end times
Average activation height and standard
deviation
% Specificity
Relative % Accuracy
% Accuracy
% Coefficient of Variation
Data recorded
Test start time and actuation time
calculated for each liquid
Test end time and recovery time
calculated for each liquid
Liquid height level at activation,
calculated for each liquid
Liquid height level at activation,
calculated for each liquid
Liquid height level at activation,
calculated for each liquid
Liquid height level at activation,
calculated for each liquid
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 —
\Xt/
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
/r\
Accuracy, % = 100 X I-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:
IM-DI
Accuracy, % = ——— * 100
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 = 100
(S\
x -
\xJ
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 Piiramclcr
Average Detection Time (hh:mm:ss)
Average Recover)' Time (hh:mm:ss)
Average Activation Height (nun)
Activation Height Standard Deviation (mm)
Specificity (%)M
Relative Accuracy (%)
Test Blends
KO
0:01:09
0:00:03
4.9
3.1
95.1%
100%
K15
0:01:25
0:00:02
71
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%
\Vatcr<">
0:04:49
0:00:03
99
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 (%)<•>
Relative Accuracy (%)
Text Illenels
F.O
0:07
0:00
36. 1
2,1
98.4%
100%
F.I 5
0:07
0:00
36.1
0.3
94.8%
100%
116
0:06
0:00
36.2
0.4
94.9%
100%
FJO
0:07
0:00
35.9
0.2
94.2%
100%
F.50
0:06
0:00
36
0.0
94.5%
100%
ESS
0:05
0:00
36. 1
0.4
94 S° t,
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 Blenils
renonnance rarameier
Detection Time (hlrinni)
Recovery Time (hh:mm)
Activation Height (mm)
Relative Accuracy (%)
KO
0:47
0:00
205.0
100%
K15
0:39
0:01
205.0
100%
116
0:36
0:01
201.0
100%
K30
0:37
0:00
200.0
100%
K50
0:34
0:00
201.0
100%
K85
0:33
0:00
201.0
100%
Witter
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 (hh:mm)
Average Activation Height (mm)
Activation Height Standard Deviation
(mm)
Specificity (%)«
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%
£85
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 complex impedance, measured by the C/C sensor, also changes.
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 types 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 (mm:ss.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%
£85
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 B1.4.4 of QAPP Addendum 110113
(2) The theoretical detection height was based on the sensor window estimated at 5mm from the bottom 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)l
Test Blends
j'enormance parameter
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%
I165
01:56.9
10.0
0.577
200.0%
36.5%
5.8%
(1) Values calculated according to Table 3 in Section B1.4.4 of QAPP Addendum 110113
(2) The theoretical detection height was based on the sensor window estimated at 5mm from the bottom
(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)
C-15
-------
using the appropriate meters and probes (Hach LDO meter and VWR meter with ThermoScientific
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 (us/cm)
1133
pH
7.62
ORP (mV)
408.1
Temperature (°C)
20.3
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-16
-------
Appendix D
Pressure Decay Testing Methods and Results
-------
Appendix D
Pressure Decay Testing Methods and Results
Dl 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)
5
= average absolute gas pressure (psi)
P 1 - 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
-------
V \_Hose C
toUIValv*
Hose Collector
Pressure
Source
Pressure
Vfessel
Figure 1. Pressure Decay Test Setup
This stepwise approach was followed to produce a plot of the decay overtime for each test
blend.
1. Fill pressure chamber to the manufacturer recommended level with test blend (DI water,
EO, E15, E85, and 116).
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
-------
D2
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
ESS
-0.0242(c)
-0.0445
-0.054
-0.0547
-0.0511
0.00570
Water*)
-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
50
4 Replicate 1
y = -0.0465x+19.763
R2 = 0.9957
• Replicate 2
y = -0.0426x+19.845
R2 = 0.9963
A Replicate 3
y = -0.0543x+19.851
R2 = 0.9986
10
20 30
Time (minutes)
40
Figure 2. Pressure Decay Test with three replicates DI Water
D-3
-------
20.5
20.0
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
* Replicate 1
y = -0.042x+19.687
R2 = 0.9733
• Replicate 2
y = -0.0339x+19.91
R2 = 0.9937
Replicate 3
y = -0.0447x+ 19.904
R2 = 0.9974
60
Figure 4. Pressure Decay Test with three replicates of E15
D-4
-------
20.5
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
El INTRODUCTION
In 2011, automatic tank gauging (ATG) systems were tested to evaluate their functionality in
ethanol-blended fuels1^—. 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.
E2 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): B1 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
increases, the float rises and transmits an electronic signal proportional to the level of water in
E-l
-------
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.
E3 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
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-
E-2
-------
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.
E4 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 El5,
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 E15,
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.
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 reports12' — written after this testing state that they "did not detect
water in the test vessel containing either intermediate (El 5) or high (E85) ethanol blends if the water was
E-3
-------
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."
E5 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 ETVAdvanced Monitoring Systems Center, Versions. U.S.
Environmental Technology Verification Program, Battelle, April 2011.
4. ASTM, D4057 Standard Practice for Manual Sampling of Petroleum and Petroleum Products.
August 2011.
5. ASTM, D5854 Standard Practice for Mixing and Handling of Liquid Samples of Petroleum and
Petroleum Products. May 2010.
6. ASTM, D7717 Standard Practice for Preparing Volumetric Blends of Denatured Fuel Ethanol
and Gasoline Blendstocks for Laboratory Analysis. August 2011.
7. ASTM, E203 Standard Test Method for Water Using Volumetric Karl Fischer Titration.
November 2008.
8. ASTM, D4815 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 Standard Test Method for Determination of Ethanol andMethanol Content in
Fuels Containting Greater than 20% Ethanol by Gas Chromatography. April 2013.
10. ASTM, D1613 Standard Test Method for Acidity in Volatile Solvents and Chemical
Intermediates Used in Paint, Varnish, Lacquer, and Related Products. July 2012.
11. ASTM, D4052 Standard Test Method for Density, Relative Density, and API Gravity of Liquids
by Digital Density Meter. 2011.
12. ASTM, D287 Standard Test Method for API Gravity of Crude Petroleum and Petroleum
Products (HydrometerMethod). June 2006.
13. ASTM, D2624 Standard Test Methods for Electrical Conductivity of Aviation and Distillate
Fuels. February 2010.
14. ASTM, D445 Standard Test Method for Kinematic Viscosity of Transparent and Opaque Liquids
(and Calculation of Dynamic Viscosity). May 2012.
15. ASTM, D7451 Standard Test Methods for Water SeperationPropoerties of Light and Middle
Distillate, and Compression and Spark Ignition Fuels. January 2009.
16. Steel, R.G.D.a.T., J. H., Principles and Procedures of Statistics. 1960, New York: McGraw-Hill.
17. 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.
18. Ken Wilcox Associates, I., StandardTest Procedures for Evaluating Leak Detection Methods:
Liquid-phase Out-of-Tank Product Detectors. 1990, USEPA Solid Waste and Emergency
Response/Research and Development.
E-4
-------
19. 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.
20. 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
-------
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 (ATG) systems 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
Los Angeles
Los Angeles
# of Tests Conducted -
0.2gal/hr 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
1- Inconclusive, 1-Pass
3 -Pass
Comment
All Ok.
Tank #1 and #3 Passed. Tank # 2 Mid-grade unleaded (MUL) was
inconclusive due to Temp Change Too Large.
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.
Diesel Tank #4 showed a gross increase during first test, Re-test
Passed.
All Ok.
F-l
-------
County in
Southern
California
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 -
0.2gal/hr Test Results
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
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, the
ATG did not sound an alarm. Maintenance was dispatched.
F-2
------- |