From: FJSchwende

To: EPA Docket Number EPA-HQ-OAR-2008-0558

Cc: FBertram, BKolowich, JSopata

Date: April 7, 2010

Subject: D6550 – data summary and recommendation.

The ASTM Method, D6550, an SFC based method for the determination of the
olefin content of gasoline1,  has been performed in the NVFEL Testing
Services Group for approximately 3 years.  A single instrument has been
utilized for this method, a Selerity Technologies Inc, Series 3100 SFC.

Over that period of time we have analyzed a single control fluid sample,
CAAB Control 2003 (identified internally as FTAG-12221) and have
participated in the ASTM RFG Crosscheck program.  Results of these
analyses are summarized below.

Control Fluid Analysis: For every analytical run, three replicates of
the Control Fluid were run before and after the submitted samples.  All
six replicates were reported to the database unless a verifiable reason
was observed to exclude a value (for instance, depletion of the injected
sample, resulting in an erroneously low value).  For the 144 values
reported to the database, the mean mass-% olefin content was 9.9, with a
standard deviation of 0.42 (relative standard deviation of  4%).  Figure
1 shows a plot of these values versus analysis date; the red line
indicates the overall mean mass-% olefin.

Figure   SEQ Figure \* ARABIC  1 .  Control Fluid Results.  Mass-%
Olefin versus Assay Date

Submitted Samples:  During this time period the majority of submitted
samples were ASTM RFG Crosscheck samples.  These samples were analyzed
by both D6550 and D13192.  The olefin composition reported for D1319 is
expressed at volume-%, and is corrected for the oxygenate composition of
the sample.  Regression analysis of the olefin values reported by D1319
versus D6550 is shown in Figure 2.

Figure   SEQ Figure \* ARABIC  2 .  Regression of D1319 versus D6550.

In this figure, six points with uncorrected D1319 values were included
in the graph (red squares) but were not included in the regression
analysis.  Samples reported as 0.0 mass-% olefin by D6550 were excluded
from the analysis because the method requires that assay results below
the lowest standard (nominally 1 mass-%) must be reported as zero.

The regression analysis demonstrates that these two methods are highly
correlated, but does not address the nature of any method bias.  To
consider method bias, the difference between the results was calculated
(Delta = D6550 mass-% - D1319 vol-%) and graphed versus analysis date as
well as versus D6550 mass-%.  These graphs, shown in Figures 3 and 4,
demonstrate that the differences in the percent olefin values reported
by these two methods were:

random with respect to analysis date;

biased positive for olefin compositions below approximately 5 mass-%;

less than +/- 2.4 for all samples over this time period.

Figure   SEQ Figure \* ARABIC  3 .  Delta versus Date.

Figure   SEQ Figure \* ARABIC  4 .  Delta versus D6550

Recommendation:  Control Fluid results indicate that D6550 is able to
provide consistent mass-% olefin values when performed according to the
requirements of the standard test method.  Even though the basis for the
reported olefin composition is different from D1319 (mass-% versus
volume-%) the reported percentages are comparable (usually within 1
percentage point of each other and never more than 2.5 percentage points
different).  Based on these findings, I recommend the use of ASTM D6550
as an alternative test method for the determination of olefins in
gasoline.

Standard test method for determination of olefin content of gasolines by
supercritical-fluid chromatography.  ASTM D6550.

Standard test method for hydrocarbon types in liquid petroleum products
by fluorescent indicator adsorption.  ASTM D1319.

