Response to Comments Document concerning the proposed use of 

ASTM D6550-05 as an alternative test method for olefin content of
gasoline

	

On July 6, 2006, the American Petroleum Institute (API) petitioned the
Agency to allow ASTM D6550 as an alternative test method for measuring
the olefin content of gasoline.  The Agency agreed with this request,
viewed API’s petition as non-controversial, and anticipated no adverse
comments with API’s petition.  On December 8, 2008, the Agency
proposed, with a corresponding direct final rule, to allow the use of
ASTM D6550-05 as an alternative test method for olefin content in
gasoline provided: 1) ASTM D6550-05 results were correlated to the
designated test method, ASTM D1319-03ε1, and 2) users of ASTM D6550-05
make use of the following conversion equation, Volume% = 0.857 X Mass%,
to convert mass percent olefins to volume percent olefins.    In
January, 2009, the Agency received an adverse comment from the New York
State Department of Environmental Conservation (NYSDEC or DEC).  On
February, 6, 2009, the Agency published a partial withdrawal notice of
the direct final rule published on December 8, 2008, due to adverse
comments submitted by NYSDEC.

The following is a brief summary of the NYSDEC adverse comments and
API’s responses to them. 

Comments:

NYSDEC argued that the use of ASTM D6550-05 would have the potential for
producing results that are different from those produced with ASTM
D1319-03ε1, and therefore, that test results would not be consistent
with EPA’s Complex Model.  According to NYSDEC, a substantial volume
of commercial gasoline has olefin content greater than 10.5 weight
percent maximum for the proposed mass-to-volume conversion factor. 
Further, NYSDEC argued that the EPA proposed mass percent to volume
percent conversion factor was not intended for universal use, that the
use of this conversion factor as proposed on December 8, 2008, may not
be appropriate for all gasoline, and would therefore, have the potential
to underestimate olefin content when above 10.5 weight percent.       

ε1, but without the use of the mass percent to volume percent
conversion equation  as earlier proposed.

ε1 and ASTM D6550-05.  A correlation equation would then be developed
from the results of both of these test methods in terms of ASTM
D1319-03ε1 in volume percent versus ASTM D6550-05 in weight percent. 
Therefore, the applicable range of the resulting correlation from a
site-specific correlation would be consistent with the specific
refinery’s olefin content range.  

API also stated that if a test facility chooses to use an EPA-allowed
alternative test method for analysis of a batch of gasoline, the
laboratory would use the site-specific correlation equation as discussed
previously to determine the designated test method value as if the
gasoline sample had been analyzed by the EPA-designated test.  In the
case of the proposed ASTM D6550-05 alternative test method, it would be
the ASTM D1319-03ε1 determined olefin value as derived from the
site-specific correlation equation that is reported to EPA on batch
reports and used as an input to the EPA Complex Model.  According to
API, the ASTM D6550-05 result in weight percent would never be used
directly for either compliance or as an input into the Complex Model for
compliance demonstrations as described at 40 CFR Part 80.  

ε1, based on a specific site’s gasoline production and, as previously
stated, would never be used directly for either compliance or as an
input to the Complex Model.  Therefore, API recommended that EPA omit
the initial conversion of ASTM D6550-05 weight percent to volume percent
using the 0.857 factor and allow the use of ASTM D6550-05 as an
alternative test method for olefin content in gasoline provided results
are correlated to the EPA-designated test method, ASTM D1319-03ε1.   

On May 14, 2009, NYSDEC provided a response to API’s February 9, 2009
comments.  NYSDEC stated that the issue of the use of alternative test
methods was discussed several times in the 1994 Reformulated Gasoline
(RFG) Final Rule.   NYSDEC stated that in the 1994 RFG rule, “EPA saw
two issues with alternative test methods – accuracy and bias. 
Therefore, in order to evaluate an alternative test method, accuracy and
bias must be the primary considerations.  We believe that the concerns
voiced by EPA in 1994 regarding multiple regulatory test methods are
still valid today.”  NYSDEC also noted that most comments to the 1994
RFG final rule were in favor or ASTM D1319-88 as the designated test
method for the measurement of olefin content in gasoline despite the
method’s inability to accurately measure this property.  They also
added concerns about the precision of the designated test method and
added that little progress has been made in improving this method.

Further, NYSDEC reiterated its original comment that “[t]he Complex
Model was built using fuels analyzed for their olefin content using the
designated FIA method.  Therefore, the important factor is how well
values determined by an alternative test method compare to those
determined in the past by the designated FIA (ASTM D1319) method. 
Whether or not the SFC method (D6550-05) is more accurate, precise,
inexpensive or otherwise desirable over the FIA method should not be the
primary issue for this rulemaking.  To the extent that the SCF [sic]
method gives results which are different (without regard to being better
or worse) from those of the FIA, the calculated results of the Complex
Model will be different.”  NYSDEC further stated that “[o]ur
opposition to allowing ASTM D6550-05 as an alternative test method for
ASTM D1319 is NOT based on a contention that ASTM D 1319 produced more
accurate and precise estimations of the olefin content of gasoline.  Our
opposition is based on the fact that the algorithms used in EPA’s
Complex Model were developed using olefin concentrations based solely on
ASTM D1319.  Therefore gasoline regulatory programs which are dependent
on the Complex Model are also dependent on ASTM D1319.”(Emphasis in
original).

NYSDEC also argued that API’s new approach to allowing ASTM D6550-05
as an alternative test method for measuring the olefin content of
gasoline has little resemblance to that which was originally suggested
in their July 6, 2006 letter.  Further, NYSDEC stated that API’s new
approach admits the 0.857 conversion factor for mass percent to volume
percent is not appropriate.  According to NYSDEC, API now “suggest[s]
a site-specific correlation” of FIA volume percent data and SFC mass
percent data, and that while this was an improvement on their original
suggestion, it was insufficient evidence for EPA to entertain ASTM
D6550-05 as a proposed alternative test method.  

NYSDEC also described the statement by API that “a laboratory that
chooses to use any EPA-allowed alternative test method must first
develop a site-specific correlation of the EPA-allowed, alternative test
method to the EPA-designated test method”, as largely meaningless. 
According to NYSDEC, “any two sets of data can be correlated. 
Correlation is only the process of examining data to determine if there
is a linear relationship and to determine that strength of that
relationship.”   NYSDEC also stated that, “API presented no data to
support its contention that “a linear relationship was found to exist
between FIA and SFC over the zero to 15 volume percent concentration
range” and that the graphic is far too small to allow any analysis. 
NYSDEC also argued a lack of correlation because of API’s statement
that “if needed, a nonlinear fit could easily be found to cover the
zero to 20 volume percent range.” 

Finally, NYSDEC presented analysis of ASTM inter-laboratory crosscheck
data, from January 2006 to October 2008, on measurement of olefins in
gasoline that used ASTM D1319 and ASTM D6550.   According to NYSDEC, the
analysis showed that although the square of the correlation coefficient
was very good and the slope of the line was close to unity, the overall
line was a poor fit for data on olefin concentrations of below 5 volume
percent.  NYSDEC also presented an analysis of data for olefin
concentrations of 10 to 24 volume percent that indicated the linear fit
was poorer and noted that the slope of the least-squares linear-fit line
was far from unity and the intercept very high.  Finally, NYSDEC argued
that while “calibration curves” of a single lab may appear better in
terms of precision, there are no guarantees that any one lab would
produce results that are more accurate than the average of a large
number of labs subjected to statistical analysis.

On November 16, 2009, API provided a response to NYSDEC’s May 14, 2009
letter.   

ε1) because the FIA method has fundamentally poor precision, and
because a better method, SFC (ASTM D6550-05), has been developed.

In regards to NYSDEC’s comment that “whether or not the SFC method
is more accurate, precise, inexpensive, or otherwise desirable over the
FIA method should not be the primary issue for this rulemaking,” API
responded that ASTM test methods are constantly being updated and
improved to be more accurate and precise.  Round robin test programs are
routinely conducted to quantify improvements in the methods’ precision
statements.  Both industry and government labs participate in ASTM round
robin programs, as well as the ASTM inter-laboratory crosscheck data
programs with a goal of improving the accuracy and site precision of
their particular labs.  Also, a characteristic of world-class labs is
that they use the most up-to-date test methods having the best accuracy
and precision.  According to API, the purpose of EPA having allowed
alternative test methods in the past has been to capture selected
improvements in analytical technology and the concomitant improved
operability, speed, accuracy, precision.  Specific to olefin
measurements, it is the uncertainty of the olefin measurements when
using FIA that has led industry to develop the SFC method.

In regards to NYSDEC’s comment that, “to the extent that the SFC
method gives results which are different from those of the FIA, the
calculated results of the Complex Model will be different,” API
responded that this statement indicated a misunderstanding of how the
Complex Model is used in regulating fuels.  According to API, this is
because when an EPA-allowed alternative test method is used, the
alternative method result is translated into the “equivalent” result
for the EPA-designated test method using the site-specific correlation
required by EPA.   It is the “equivalent” FIA result that is entered
into the Complex Model, not the SFC result.  “The test result from the
alternative test method is never used directly for gasoline
certification or for the Complex Model calculations.  In addition, the
test results using an alternative test method that has been
well-correlated to FIA will be no more different from FIA that the
difference between two repeat tests results obtained using FIA.”
(Emphasis in original).

In regards to NYSDEC’s comment that their “opposition is based on
the fact that the algorithms used in EPA’s Complex model were
developed using olefin concentrations based solely on ASTM D1319. 
Therefore gasoline regulatory programs which are dependent on the
Complex Model are also dependent on ASTM D1319,” API response was that
“[i]f such logic were followed, progress would have stopped long ago. 
We would still be required today to use the version of each ASTM test
method that was used to obtain the data for development of the Complex
Model.  For example, we would be required to use ASTM D323 for measuring
gasoline vapor pressure, which was abandoned in 1991 when ASTM D5191 was
adopted in 1992.  Also, the ASTM standard specification for gasoline
would be a very outdated D439 rather than today’s D4814.  And, we’d
be stuck with a 1990 vintage vehicle fleet for our driving needs because
that is the technology that the Complex Model is based upon.”

In response to NYSDEC describing API’s statement that “[a]
laboratory that chooses to use any EPA-allowed alternative test method
must first develop a site-specific correlation of the EPA-allowed,
alternative test method to the EPA-designated test method,” as
“largely meaningless” API responded that “NYSDEC appears to be
unaware that except for the diesel rule where EPA adopted a performance
based approach in approving diesel sulfur test methods, EPA has required
that any lab that wishes to use an EPA-allowed alternative test method
must develop and maintain such a site-specific correlation.  The
correlation requirement for each of the nine alternative test methods
currently allowed by EPA is not meaningless, but rather a federal EPA
requirement.”

API responded to NYSDEC’s comment that “API presents no data to
support the statement that a linear relationship was found to exist
between FIA and SFC over the zero to 15 volume percent concentration
range” by stating that “Figure 1 in the July 6, 2006 API letter is
reproduced below to demonstrate the linearity of the matter over the
zero to 15 volume percent range.  Note that the results in this plot
include the use of the 0.857 factor only because we chose to reproduce
the original plot shown in the July 6, 2006 letter.  The use or omission
of the conversion factor makes absolutely no difference in any
conclusions regarding the linearity of the data.”

In regard to NYSDEC’s comment that “EPA’s lab would not be able to
test fuels because correlations will be refinery-[specific] and
laboratory-specific.  Therefore, enforcement will be hindered as EPA
would not be able to recreate the correlations used,”  API stated 
that “EPA’s enforcement of its fuel standards is completely
independent of the EPA-allowed alternative test methods being used by
labs, as well as related site-specific correlations.  Whereas EPA may
use non-designated test methods for screening purposes on occasion, we
understand that all EPA enforcement is based on measurements obtained
using EPA-designated test methods.”

In regard to NYSDEC comment that they also have issues with how ASTM
analyzes round robin program data and inter-lab crosscheck data using
“robust statistics,”  API responded  that “the ASTM statistical
package used to analyze the data has been peer reviewed and accepted by
statisticians in industry and governments across the globe.”  

	Further, API noted that NYSDEC’s May 14, 2009 letter, persistently
referred to the API February 9, 2009 letter as a “new proposal.”  As
one example they cited NYSDEC’s statement that “API’s newly
suggested method has little resemblance to that which they originally
suggested in their July 6, 2006 letter.  API’s new approach admits
that the 0.857 conversion factor for mass percent to volume percent is
not appropriate.”  In response, API notes that their February 6, 2009
letter, recommended that the EPA regulation omit the initial conversion
of the D6550 weight percent to its volume percent using the 0.857
factor.  According to API, as explained in that letter, the use or
omission of the conversion factor would make absolutely no difference in
any reported olefin result in the proposed rule.  Finally, API
recommended again that EPA allow ASTM D6550-05 as an allowed alternative
to the designated test method, ASTM D1319, for measuring olefin content
of gasoline, provided the results are correlated with ASTM D1319 using a
refinery-specific correlation of FIA (volume %) versus SFC (weight%).”

EPA Response to API & NYDEC comments:

ε1 (also known as FIA, or the designated test method) and ASTM
D6550-05.   A resulting correlation equation would then be developed in
terms of ASTM D1319-03ε1 in volume percent and ASTM D6550-05 in weight
percent.  Thus, the applicable range of the resulting correlation from a
facility’s site specific correlation would be consistent with that
specific facility’s olefin content range.  Furthermore, the
requirement of correlating SFC test methods results to the FIA
designated test method produces an SFC-FIA equivalent result that is
suitable for input into the Complex Model or for meeting fuel reporting
requirements.  

	The Agency agrees with API’s recommendation that there is no need to
convert olefin content measured by ASTM D6550-05 in weight percent to
volume percent because the olefin result as determined by SFC (ASTM
D6550-05) is correlated to FIA (ASTM D1319-03ε1) based on a specific
site’s production and never used directly for compliance or as an
input into the Complex Model.  Thus, in today’s proposal, EPA is
omitting the initial conversion of olefin content measured by ASTM
D6500-05 in weight percent to volume percent using the 0.857 factor, as
earlier proposed on December 8, 2008.

ε1 and ASTM D6550-05.  In order to compensate for any bias that may
exist between these two analytical test methods, EPA is proposing to
require ASTM D6550-05 to be correlated to ASTM D1319-03ε1, its
respective designated test method.   Refiners or importers that choose
to use ASTM D6550-05 for the measurement of olefin content in gasoline
must report the correlated result, on a site-specific basis, with the
designated method, ASTM D1319-03ε1, when using the proposed EPA-allowed
alternative test method, ASTM D6550-05.   

®

²

S

T

´

µ

¹

ä

å

ù

º

葠ː摧㬉A

hø

h™*

h™*

hø

  hŠW

h™*

 hÌ

 hÌ

葠ː摧坅

摧祪t

葠ː摧䓛¥

摧䔾Â

摧ᄊ

h

? linear correlation between these two methods, but to show that a
linear relationship could be established between these two methods using
a robust fuels data set on olefin content from the ASTM crosscheck
program.  The Agency believes that this illustration of a linear
relationship between SFC and FIA from the ASTM crosscheck program is
sufficient evidence to illustrate that a linear relationship may be
established between these two test methods.  Moreover, if a linear
relationship can be established between two analytical test methods with
fuel samples representing producers on a “global” scale, then one
would expect that a linear relationship may be established between these
two test methods on a site specific basis, where olefin content variance
in gasoline fuel samples would not be expected to be as varied, compared
to fuel samples from an ASTM crosscheck program.  

	The Agency does not believe that the adoption of alternative test
methods for compliance measurements hinders its enforcement ability. 
Rather, the Agency believes that the adoption of alternative test
methods used for compliance measurement provides flexibility to the
regulated industry by allowing methods that may be less expensive,
increase speed of analysis, or may be more accurate or precise.  We also
believe that our approval of this alternative test method would not be
inconsistent with the Agency’s posture in the RFG rule.  This is
because, as discussed previously, the requirement of correlating
EPA-allowed alternative test method results to its respective designated
test method compensates for any bias that may exist between the
alternative and designated test method.   Additionally, as also
previously discussed, the liner relationship would likely assure
accuracy.  Moreover, while the Agency may use non-designated test
methods for enforcement screening purposes, it uses the designated test
method specified in the regulations for each particular fuel parameter
for purposes of enforcement.

	

Further, we believe today’s proposal is consistent with past Agency
practice because the Agency currently allows eight alternative test
methods for the measurement of gasoline or butane properties provided
their results are correlated to the fuel parameter’s respective EPA
designated test method.  These alternative test methods are: ASTM D5453,
ASTM D6920, ASTM D3120 and ASTM D7039 for sulfur in gasoline, ASTM D4468
and ASTM D3246 for sulfur in butane, ASTM D1319 for aromatics in
gasoline, and ASTM D4815 for oxygenate content of gasoline.  In
addition, ASTM D4294, ASTM D6920 and ASTM D5453 are approved alternative
test methods for measuring the sulfur content of 500 ppm diesel fuel
provided its test results are correlated to the designated test method,
ASTM D2622. 

	Finally, although not the subject of this proposed rulemaking, EPA
intends to establish a performance-based test method approach (PBTM)
rule which would provide criteria for the qualification of alternative
test methods.  

 See Air Docket # EPA-HQ-OAR-2008-0558-0002.

 This conversion equation was validated in ASTM D6550 for up to 10.5
weight percent olefins maximum in California gasoline meeting California
Air Resource Board Phase 2 gasoline specifications.

 See Air Docket# EPA-HQ-OAR-2008-0558-0005.

 74 FR 6233, February 6, 2009.

 See Air Docket# EPA-HQ-OAR-2008-0558-0007.

  See Air Docket# EPA-HQ-OAR-2008-0558-0008.

 (59 FR 7716, February 16, 1994)

 See Air Docket# EPA-HQ-OAR-2008-0558-0009. 

 See 40 CFR § 80.46(a)(3)(i) through § 80.46(a)(3)(iv).

 See 40 CFR § 80.46(a)(4)(i) through § 80.46(a)(4)(ii).

 See 40 CFR § 80.46(f)(3)(i).

 See 40 CFR § 80.46(g)(2)(i).

 See 40 CFR § 80.580(c)(2).

 PAGE   

 PAGE   15 

