

 From:
brian.c.mormino@cummins.com
 To:
Byron Bunker/AA/USEPA/US@EPA, Charles Moulis/AA/USEPA/US@EPA
 Cc:
bill.t.martin@cummins.com, Angela Cullen/AA/USEPA/US@EPA, jackie.m.yeager@cummins.com, Cleophas Jackson/AA/USEPA/US@EPA, jon.a.dickson@cummins.com, l.k.hwang@cummins.com, robert.a.jorgensen@cummins.com, steve.j.charlton@cummins.com
 Date:
05/24/2011 09:54 AM
 Subject:
Re: Fw: 1036.530(b) Comments on CO2 Adjustment Factor for fuel carbon-specific net energy content



Byron and Chuck, as discussed last week, please see the email below from Bill Martin on the CO2 adjustment factor. 






From:        Brian C Mormino/Corp/Cummins 
To:        Bunker.Byron@epamail.epa.gov, moulis.charles@epa.gov 
Cc:        Robert A Jorgensen/Corp/Cummins@Cummins, Steve J Charlton/Ind/Cummins@Cummins, Jackie M Yeager/Ind/Cummins@Cummins, Jon A Dickson/Auto/Cummins@Cummins, Bill T Martin/Ind/Cummins@Cummins, Cullen.Angela@epamail.epa.gov, Jackson.Cleophas@epamail.epa.gov, L K Hwang/Ind/Cummins 
Date:        04/29/2011 12:55 PM 
Subject:        Fw: 1036.530(b) Comments on CO2 Adjustment Factor for fuel carbon-specific net energy content 


Byron and Chuck, please see the notes and recommendations below from Bill Martin.  As these are detailed technical comments, we wanted to lay this out fully in an email.  Please review and let me know if you would like to setup a call to discuss, 

BCM 

---------------------------------------------------------
Brian C. Mormino
Director -- Energy Policy and Emissions Compliance
Cummins Inc. - Mail Code: 60203
500 Jackson Street 
Columbus, IN 47201
812-377-0512 (p)
812-377-0085 (f)
202-669-3597 (c)
brian.c.mormino@cummins.com


----- Forwarded by Brian C Mormino/Corp/Cummins on 04/28/2011 01:50 PM ----- 

From:        Bill T Martin/Ind/Cummins 
To:        Brian C Mormino/Corp/Cummins@Cummins 
Date:        04/28/2011 01:40 PM 
Subject:        1036.530(b) Comments on CO2 Adjustment Factor for fuel carbon-specific net energy content 


Brian, 

Please forward this to Byron Bunker, Chuck Moulis and other appropriate EPA people. 

Dear Mr. Bunker and Mr. Moulis: 

Several technical issues have been raised within Cummins with the NPRM language on the CO2 adjustment factor.   

The issues include the following: 
   1.	rounding, 
   2.	precision of reference values, 
   3.	variation of fuel analysis, 
   4.	units symbols, and 
   5.	carbon weight fraction analysis method.  

The concerns are detailed below.  Please consider these comments as you finalize the language.  If you have any question, please contact me. 

Beside these comments, Cummins also supports the comments made in the Emissions Measurement and Testing Committee meeting on March 31, 2011 recommending that a tolerance be added around the reference values where it would be permissible to not correct the measured results.  As discussed at this meeting, the intent of this allowance was to allow test facilities to avoid the additional expense of new data systems and software to track these fuel properties and apply the correction under situations where there is little variation in the carbon-specific net energy content.  Effectively these tolerances provide an option to control this source of CO2 variation by controlling the fuel properties instead of correcting for them.  The magnitude of these tolerances need to be set wider than the variation in the carbon-specific net energy content to avoid triggering an adjustment due only to variation in the fuel properties.  See the discussion below on variation of fuel analysis.     

1.  Rounding 
1036.530(b) reads: 
(b) Adjust CO2 emission rates calculated under paragraph (a) of this section for test fuel properties as specified in this paragraph (b) to obtain the official emission results. Note that the purpose of this adjustment is to make official emission results independent of small differences in test fuels within a fuel type. 
(1) ...Calculate the test fuel's carbon-specific net energy content (BTU/lbC) by dividing the net energy content by the carbon fraction and rounding to the nearest BTU/lbC. 
1036.530(b)(2) reads: 
(2) Calculate the adjustment factor for carbon-specific net energy content by dividing the carbon-specific net energy content of your test fuel by the reference level in the following table and rounding to five decimal places. 
(3) Your official emission result equals your calculated brake-specific emission rate multiplied by the adjustment factor specified in paragraph (b)(2) of this section. ... 

Rounding intermediate results is generally not a good practice.  Unless the standard setting part requires rounding of intermediate results, 1065.20(e) calls for rounding only the final result (which in this case is the adjusted BSCO2 value) just before comparing it to the applicable standard.  In this case, there appears to be no good reason to round intermediate results.  Specifications like these to rounding intermediate results require special processes in data analysis programs.  Cummins recommends that these paragraphs be amended by deleting the rounding phrases "and rounding to the nearest BTU/lbC" and "and rounding to five decimal places".   

2. Precision of Reference Values 
All the reference values appear to be rounded to three significant digits with a resolution of 1 part out of 212 to 285 (0.47 to 0.35 %).  1036.530(b) states the intent of this adjustment factor - "Note that the purpose of this adjustment is to make official emission results independent of small differences in test fuels within a fuel type.".  Based on this, the intent is to reduce variation without introducing a bias relative to the data used to set the CO2 standards.  Since this resolution is a significant fraction of the total variation in BSCO2 that it is designed to improve, the resolution of these reference values should be increased so that they do not introduce a bias that is significant with respect to this variation.   If the final selected reference value would cause a shift in the mean of the data set used to set the CO2 standard, then the standards would need shifted accordingly.  An analysis from the supplier of the diesel certification fuel used at Cummins Technical Center gives a carbon-specific net energy content of 21,266 Btu/lb.  This is 0.31% above the reference value of 21,200.  The resulting adjustment factor is 1.00311.  This suggests that the BSCO2 data that Cummins submitted for diesel engines would need an upward adjustment of 0.311% to be comparable to the results we would get using this adjustment factor.  Cummins recommends that the reference values be recalculated based on the average of the fuels used for the CO2 test results used to set the standards and express the final value with at least one more significant figure. 

3. Variation of Fuel Analysis 
The variation of the fuel analysis needs to be considered relative to the total observed variation in BSCO2.  This topic was discussed at  the Emissions Measurement and Testing Committee meeting on March 31, 2011.  At that meeting Cummins raised concern over whether inclusion of the CO2 adjustment factor might increase the total observed variation in BSCO2 due to variation in the fuel analysis.  Chuck Moulis, EPA, indicated he reviewed three analysis of their diesel cert fuel and it showed very little variation.  He estimates 2-3% variation is possible.  He estimated larger variation for gaseous fuels.  He agreed to look at adding a range over which no correction is required.  In setting an appropriate no-adjust tolerance, the variation in the carbon-specific net energy content needs analyzed.  The following is  an analysis for diesel fuel. 

1036.530(b)(1) references ASTM D240-09 for measuring the net energy content for liquid fuels.  This ASTM standard gives the reproducibility in section 12.1.2  to be 0.40 MJ/kg (172 Btu/lb).  This ASTM standard defines this value as follows: 
"The difference between two single and independent results, obtained by different operators working in different laboratories on identical test material, would in the long run, in normal and correct operation of the test method, exceed the values shown in the following table only in one case in twenty." 
Effectively this value of 0.40 MJ/kg represents two standard deviations of the distribution of the differences of independent observations.  The variance in the differences is two times the variance in the individual observations.  Expressing the reproducibility in terms of  95% of variation in the net energy content requires dividing by the square root of two, resulting in a reproducibility estimate based on individual observations of the net energy content of 0.28 MJ/kg (122 Btu/lb).   This value results in a relative reproducibility of 0.66 % of the nominal net energy content of 18,400 Btu/lb for diesel fuel.   

ASTM D5291-10 is used for determining carbon weight fraction.  This ASTM standard gives the reproducibility of carbon fraction in section 13.1.1 as a formula based on the mean value.  This ASTM standard does not define this value.  It references an interlaboratory study documented in ASTM Research Report, RR:D02-1289, which was not available on-line.  Assuming this interlaboratory study followed ASTM E691-09, then the meaning of reproducibility is the same as ASTM D240-09.  At the nominal carbon weight fraction of 87.0 % for diesel fuel, this formula results in an estimated reproducibility of 2.4 % or a relative reproducibility of 2.8% of the carbon weight fraction value.  Dividing by the square root of two, yields the reproducibility based on individual observations of carbon weight fraction of 1.7 % or a relative reproducibility of 2.0% of the carbon weight fraction.   

These two properties are determined by independent methods, so the combined reproducibility of the carbon-specific net energy content can be estimated as the root-sum-square (RSS) of the relative reproducibility of the components.  The combined relative reproducibility is 2.1 % of the carbon-specific net energy content.  This analysis indicates that the dominate source of variation is the measurement of carbon weight fraction.  This variation is comparable to the total observed variation in BSCO2 seen on diesel engines.  This variation analysis indicates that the total variation in adjusted BSCO2 across all labs is at risk of being higher than the unadjusted BSCO2 due to variation in the measurement of the carbon-specific net energy content. Based on this simple analysis, Cummins recommends that the CO2 no-adjust tolerance for liquid fuels be +/- 2.1 % of the reference value.  If the reference values are not recalculated as described above, then the no-adjust tolerance for liquid fuels should be increased to +/- 2.5 %.  Allowing an optional no-adjust tolerance based on the expected variation in the carbon-specific net energy content will give test facilities the flexibility to choose whether to include this adjustment depending on whether the adjustment factor improves their variation in BSCO2.  To avoid biasing the results due to selectively choosing this option based on the results of the fuel analysis, labs should be required to set their no-adjust tolerances before the fuel analysis is performed.  They should be allowed to select any two sided (+/-) no-adjust tolerance from zero up to the maximum allowed.   Allowing the test facility to select a lower no-adjust tolerance gives them the flexibility to optimize the no-adjust tolerance for minimum BSCO2 variation given the variation in their fuel properties and BSCO2. 

4. Units Symbols 
Per NIST SP 811, which is reference by 1065, the proper units symbol for the British Thermal Unit is Btu.  Cummins recommends that all occurrences of "BTU" be change to "Btu". 

NIST SP 811 lists six different energy units conversions associated with this unit symbol varying from 1.05435 to 1.05967 E+03 Joules, a range of 0.5%.  To eliminate this ambiguity, either the specific Btu unit needs to be specified in the CFR language or the reference values need to be converted to a more well defined energy unit like Joule.  Since fuel analysis reports typically give net energy content in units of Btu, the specific Btu to be used must be based on the unit used to by the labs analyzing fuel samples to avoid any bias due to selecting the wrong Btu unit.   1036.530(b)(1) cites ASTM D240-09 for the analysis of liquid fuels. Section 11 of ASTM D240-09 recommends reporting results in MJ/kg and notes conversion factors to Btu/lb.  Based on section 3.1.3, which gives a 1 Btu = 1055.06 J, it is clear that this standard is using the International Table Btu.  Per NIST SP 811 (and ASTM D3588-98), the exact conversion factor for the International Table Btu is 1.055 055 852 62 kJ. Cummins recommends that the Btu unit symbol be retained and that the following sentence be added to clarify the meaning of this unit.  "For purposes of this section, the unit symbol Btu refers to the International Table Btu and its exact conversion factor is 1.055 055 852 62 kJ." 

NIST SP 811, section 7.5 (attached below) strongly discourages the mixing of information with units and units symbols.  The unit symbol used in this NPRM "BTU/lbC" does just that.  The quantity description (carbon-specific net energy content) makes it clear that the mass unit in the denominator (lb) is associated with the mass of carbon.  This is a style issue.  Cummins recommends that the unit symbol "lb" be used instead of "lbC" to conform more closely with NIST SP 811. 

5. Reference for determining carbon weight fraction 
1036.530(b)(1) references ASTM D240-09 for determining both the net energy content and the carbon weight fraction for liquid fuels.  This  ASTM standard only addresses the energy content.  A different standard needs to be sited for carbon weight fraction for liquid fuels or allow it to be determined using good engineering judgment like gaseous fuels.  Consider ASTM D5291-10 for diesel fuels.  Note that ASTM D5291-10 does not cover more volatile fuels like gasoline, see section 1.4. 

Bill Martin
Director, Emissions Development
Mail Code 50180
Cummins Technical Center
Cummins, Inc.
Box 3005
Columbus, IN  47202-3005
812-377-7411
812-377-7050 (fax)
bill.t.martin@cummins.com 

From NIST Special Publication 811, 2008 Edition, Guide for the Use of the International System of Units (SI) 




