MEMORANDUM

DATE:
July 1, 2014
TO:
Air and Radiation Docket EPA-HQ-OAR-2012-0401
SUBJECT:
Additional Detail on the Calculation of the Cellulosic Converted Fraction, and Attribution of Batch RINs for D-code Dependent Feedstocks

This memorandum provides additional information about the implications for cellulosic fuel producers processing biomass with an adjusted cellulosic content below EPA's cellulosic content threshold of 75%. This memorandum may also be informative to biofuel producers who generate a single type of fuel by simultaneously processing multiple feedstocks.  All cellulosic biofuel processes using biomass feedstocks must employ a process with the demonstrated capability of converting cellulosic material to biofuel. However, fuel produced using a single type of biomass with an adjusted cellulosic content greater than 75% (e.g., corn stover or switchgrass) will be allowed to generate cellulosic renewable identification numbers (RINs) for the entire fuel volume, while producers using a single type of biomass with an adjusted cellulosic content below 75% will have to apportion different D-codes to the RINs produced based on the amount of cellulosic and non-cellulosic fuel produced. In practice, such producers will treat a single type of biomass material as two separate feedstocks  -  one cellulosic feedstock and one non-cellulosic feedstock  -  and allocate RINs using existing regulations for generating fuels using multiple feedstocks. For most producers this apportionment will be straight-forward, but for producers simultaneously processing cellulosic and non-cellulosic feedstocks in a single reactor using a biochemical process, the apportionment will be more complicated and require additional record keeping and reporting requirements. The regulatory framework EPA established to verify that the number of cellulosic RINs generated by such processes corresponds to the volume of the cellulosic biofuel produced involves verifying that the calculated converted fraction (CF) applied to each portion of the feedstock material (cellulosic and non-cellulosic) is accurate. Establishing the accuracy of these determined CF values requires obtaining reasonably accurate feedstock and process data, which are used to perform a mass balance accounting of the various feedstock materials.  This allows portions of the final fuel to be attributed either to the cellulosic or non-cellulosic portions of the biomass used as feedstock. This memorandum explains:  1) why EPA has established separate requirements for various types of processes, 2) what types of data can be considered reasonably accurate to form the basis for calculating cellulosic and non-cellulosic CF values for feedstocks, and 3) how accurate feedstock CF values can be used in existing regulations to apportion RINs for a single type of fuel produced from multiple feedstocks (including single types of biomass that have less than 75% adjusted cellulosic content, and are treated as two feedstocks under EPA regulations).


I. Background
      a. The cellulosic content threshold, and the implications for feedstocks with lower cellulosic contents
EPA is finalizing a cellulosic content threshold for feedstocks to qualify for cellulosic RINs for the total volume of fuel produced.  If a feedstock has an average adjusted cellulosic content of at least 75%, all of the fuel produced from the feedstock is eligible to generate cellulosic RINs.  "Adjusted cellulosic content" is the percent of organic matter that is cellulosic material (cellulose, hemicellulose, and lignin).  In other words, the "adjusted cellulosic content" is adjusted so as not to include the inorganic (ash) components of the feedstock.  This adjustment is appropriate because the inorganic portion of feedstock is not converted biofuel.
Where a biofuel producer is seeking to generate cellulosic RINs from a single type of biomass, but the biomass contains less than 75% adjusted cellulosic content, the biomass must be treated as two separate feedstocks; one cellulosic and the other non-cellulosic.  In such cases, as well as in situations where biofuel producers convert two or more different types of biomass simultaneously, producers are eligible to generate cellulosic RINs for a percentage of the fuel produced, based on the cellulosic converted fraction and the equations in 40 CFR 80.1426(f)(3)(vi).   The purpose of this memorandum is to provide details on how the cellulosic converted fraction can be calculated, and examples of methods that may be used to generate the data needed to calculate the cellulosic converted fraction.
      b. Registration, reporting and record keeping requirements for thermochemical and biochemical processes
EPA is requiring all producers of cellulosic RINs converting cellulosic biomass to complete additional registration requirements. These additional requirements are designed to demonstrate that the process being registered is capable of performing cellulosic biomass conversion.  Additionally, as part of the engineering review, the relevant unit process equipment to perform the described conversion to biofuel will be confirmed to be both in place and operational.
Process technology plays a key role in how much of the final fuel product is derived from cellulose, hemicellulose, or lignin.  The two basic processes for converting cellulosic feedstocks into biofuel are thermochemical and biochemical.  Thermochemical processes mainly consist of pyrolysis  -  in which cellulosic biomass is decomposed with temperature to form bio-oils that can be further processed to produce a finished fuel  -  and gasification  -  in which cellulosic biomass is decomposed to form synthesis gas ("syngas") that with further catalytic processing can produce a finished fuel product.  Thermochemical processes typically convert all of the organic components of the feedstock into finished fuel, and the portion of the finished fuel derived from cellulosic material is proportional to the cellulosic content of the organic fraction of the feedstock. Therefore, EPA's additional registration requirements for thermochemical producers are intended to demonstrate how much of the feedstock is cellulosic, and that the conversion equipment to be used in the process is both in place and operational at the time of registration.  
By contrast, the biochemical process first requires the release of sugars from biomass and then the use of microorganisms to convert the sugars into fuels.  The amount of cellulosic material converted to fuel will depend on the details of the process, such as the types and combinations of enzymes used. Variations from process to process will establish different hydrolysis efficiencies for cellulosic and non-cellulosic carbohydrates.  Thus, for biochemical processes, the cellulosic portion of the fuel will not necessarily be proportional to the cellulosic composition of the feedstock. 
For in situ processes, where cellulosic and non-cellulosic material are simultaneously hydrolyzed to intermediate sugars or converted to fuel, the registration, reporting, and recordkeeping requirements, as well as the method for determining the cellulosic converted fraction, are different depending on whether the cellulosic biomass is converted by a biochemical or a thermochemical process. EPA's additional registration and record keeping requirements for in situ biochemical producers of cellulosic fuel are intended to establish and monitor the relative conversion efficiencies for both the cellulosic and non-cellulosic portions of the feedstock in order to ensure that RIN assignments have been made appropriately. 

II. Definition of the cellulosic Converted Fraction
EPA defines the converted fraction (CF) to be the average mass percent (typically a yearly average) representing that portion of the feedstock that is converted into renewable fuel. The CF is calculated by dividing the dry mass of the feedstock that is converted to renewable fuel by the dry mass of the feedstock that is capable of being converted to fuel. Calculating separate CF and cellulosic CF values among in situ biochemical processes can be complicated because the mass of the resulting fuel corresponding to the non-cellulosic and cellulosic portions cannot be directly measured. However, existing EPA regulations detail how RINs can be allocated where multiple feedstocks are being simultaneously processed.  This memo provides more detail on how the formulas in 80.1426(f)(3)(vi) should be used to calculate RINs. 
III. Description of possible process configurations for biochemical and "in situ" biochemical processes
Biochemical processes typically employ engineered yeast strains to ferment aqueous sugars into alcohols that can be separated and used as biofuels. Release of sugars from complex carbohydrates is facilitated by enzymatic treatment in a process called hydrolysis. The polymeric linkages in these carbohydrates can be cellulosic or non-cellulosic, and this distinction determines which enzymatic treatment will result in hydrolyzing the chemical bonds linking the sugars together. In this case, enzymes can be viewed as keys that un-lock the sugars, and only the proper key will fit a particular chemical bond's lock.  Released sugars will be converted to biofuel in the same manner regardless of whether they were derived from cellulosic or non-cellulosic carbohydrates. Therefore, hydrolysis is the critical step controlling the fraction of the finished fuel that can be considered to be derived from cellulosic material. Determining the extent to which the cellulosic portion of a feedstock was hydrolyzed can be quantified in a number of ways, but this validation is more difficult for processes that perform in situ hydrolysis of both the cellulosic and non-cellulosic feedstock portions. This is due to the fact that the fraction of sugars or fuel derived from either type of carbohydrate cannot be measured directly, but rather requires a mass balance accounting to identify the source of the sugars or fuel.  Figure 1 shows a simplified example of a biochemical process performing in situ biochemical hydrolysis, and Figure 2 shows a separated hydrolysis scenario.  
Figure 1  -  Simplified schematic of a biochemical biofuel process with in situ hydrolysis  


Figure 2  -  Simplified schematic of a biochemical biofuel process with separated hydrolysis 


To determine the relative amounts of cellulosic and non-cellulosic feedstock that are hydrolyzed, a mass balance accounting of the converted feedstock and resulting products and co-products is required. There are a number of ways to perform a mass balance accounting for the converted portion of the cellulosic feedstock. It will be easier for producers using separated hydrolysis because the cellulosic conversion can be quantified directly from the cellulosic hydrolysis unit as indicated by the dashed boundary line in Figure 2.  In this case, complex analysis of the inputs and outputs is not required because the aqueous sugars in the hydrolysis unit's effluent are assumed to be derived from cellulosic carbohydrates.  Similarly, if cellulosic hydrolysis occurred as a separate step, and the resulting intermediates from that step were segregated and fermented separately, then all of the resulting fuel from this process would be cellulosic. A direct accounting for the converted cellulosic fraction will allow RINs to be easily attributed.
By contrast, in situ processes require a more complex mass balance accounting to confirm the portions of the biofuel resulting from cellulosic hydrolysis relative to non-cellulosic hydrolysis. Assessment methods can be designed in a number of ways, but would typically account for cellulosic conversion by either quantifying overall biofuel increases relative to non-cellulosic treatment, or by directly analyzing the structural make-up of the feedstock and residual materials, as well as the overall fuel yield, in order to determine relative feedstock conversion.  EPA is allowing producers to design analytical protocols specific to their process configuration as long as the mass balance accounting calculation method is verified by an independent third party engineer, these calculations can independently identify both a cellulosic and non-cellulosic CF, the data used to perform the calculations is regularly reported along with the calculated CF values, and the data itself is collected and reported in a scientifically reasonable manner.
               
IV. Expectations for data collection and analysis
      a. General requirements regarding data used to calculate CF
EPA is providing clarifying information on the type of methods that may be used to generate data to calculate the cellulosic converted fraction.  Data used to calculate the cellulosic CF must be obtained using an analytical method certified by a voluntary consensus standards body (VCSB), or using a non-VCSB method that would produce reasonably accurate results. If a VCSB-approved method is not used to generate the data required to calculate the cellulosic CF for a given process, then the producer will need to show that the method used is an adequate means of generating the data by providing peer reviewed references to the third party engineer performing the engineering review at registration.  An example of an appropriate VCSB test method certified by a voluntary consensus standards body for assessing the cellulosic content of a feedstock is ASTM E1758, entitled, "Determination of Carbohydrates in Biomass by High Performance Liquid Chromatography." Since this ASTM test method has been adjudicated through the voluntary consensus standards body process we are confident in its use in practice.  An example of another type of method would be a non-voluntary consensus standards body test method such as the National Renewable Energy Laboratory (NREL) Laboratory Analytical Procedure to measure structural carbohydrates and is an example of such a methodology that is commonly presented in peer-reviewed literature.  Since this NREL method has not been adjudicated through the voluntary consensus standards body process, the user of these types of non-VCSB test methods will be required to provide peer reviewed references to the third party engineer performing the engineering review at registration to show that the non-VCSB method used is an adequate means of generating the data.  These are examples of test methods that quantify the amount of cellulose, hemicellulose, and/or lignin in biomass.  Data from these analyses can be used in a mass balance equation to determine the cellulosic CF.  In the next sections we describe these methods in more detail, as well as some other methods that can be used to determine the cellulosic fraction of a feedstock.  This information, in turn, can be used to calculate the cellulosic CF of the finished fuel.
      b. Structural Carbohydrate Analysis
One method of determining the cellulose and hemicellulose content of biomass is using a High Performance Liquid Chromatography (HPLC) structural carbohydrate analysis, also known as a dietary fiber analysis.  This type of structural carbohydrate analysis has been commonly used by researchers who are interested in the composition of a feedstock that will be used for biofuel.  For this type of test, acid hydrolysis is used to break down both cellulose and hemicellulose into monomeric sugars (i.e., cellulose is converted to glucose; hemicellulose is converted to xylose, galactose, arabinose, and mannose).  The concentrations of these sugars are measured using HPLC.  The glucose concentration can be used as a measure of the cellulose concentration in the plant, and the sum of the xylose, galactose, arabinose, and mannose concentrations can be used as a measure of the hemicellulose concentration.  The sample can also be analyzed for acetyl content, which is part of hemicellulose.  Examples of this type of procedure are the National Renewable Energy Lab (NREL) Laboratory Analytical Procedure, "Determination of Structural Carbohydrates and Lignin in Biomass" and ASTM E1758, "Determination of Carbohydrates in Biomass by High Performance Liquid Chromatography."  Although this type of method has been successfully used to measure cellulose and hemicellulose for many feedstocks, an important caveat is that the results obtained are highly dependent on the exact methods used, and small changes to the procedures can have a large effect on the results. In addition, the NREL and ASTM procedures warn that biases may be introduced if the sample has a high protein content, or if extractives have not been removed.    
Because the HPLC structural carbohydrate analysis only measures cellulose and hemicellulose, a separate method is needed to measure the lignin concentration.  A procedure for measuring the lignin concentration is detailed in the NREL method cited above, as well as in ASTM E1721 "Determination of Acid-Insoluble Residue in Biomass."  Briefly, the acid hydrolysis step that breaks down the cellulose and hemicellulose also separates lignin into an "acid soluble" portion and an "acid insoluble" portion.  The concentration of acid soluble lignin can be measured using a UV-Visible spectrophotometer.  The concentration of acid insoluble lignin can be measured gravimetrically.  
In addition to the HPLC structural carbohydrate method, near infrared spectroscopy (NIRS) can be used to determine the structural carbohydrate composition, as well as the presence of other components.  NIRS measures the chemical bonds in a biomass sample, which can be used to determine its composition.  This method is rapid and non-destructive.  In order for measurements to be made using NIRS, the spectra must be calibrated using other methods, such as an HPLC compositional analysis.  
      c. Detergent Fiber Analysis
For a determination of the total fiber content of biomass, a neutral detergent fiber (NDF) test can be used.  The NDF method dissolves non-fiber components such as proteins, lipids, and sugars.  The leftover material is the total fiber, which can be measured gravimetrically. This method is detailed in AOAC 2002.04 Amylase-Treated Neutral Detergent Fiber in Feeds.  This method is commonly used for determining the forage potential of biomass.
The NDF method can be used along with an acid detergent fiber (ADF) test and an acid detergent lignin (ADL) test to determine the amount of cellulose, hemicellulose and lignin, rather than just the total fiber content.  The ADF method breaks down hemicellulose, so that cellulose and lignin remain.  The measurement of ADF can be subtracted from the measurement of NDF to determine the concentration of hemicellulose in the sample.  The ADL test measures the lignin concentration by breaking down both cellulose and hemicellulose using sulfuric acid and ignition.  The ADL measurement can be subtracted from the ADF measurement to determine cellulose concentration.  The detergent fiber analyses are much simpler than the HPLC structural carbohydrate analysis.  However, these two types of methods do not always give the same results.  For example, one study showed that the detergent fiber methods overestimate cellulose and hemicellulose and underestimate lignin relative to the structural carbohydrate analysis.  Another study has shown that NDF and ADF cannot be used to accurately determine glucan and xylan content of the fiber portion of a feedstock.  Even though the detergent fiber analyses do not give the same measurements of cellulose, hemicellulose, or lignin as the HPLC structural carbohydrate analysis, the NDF method does give a good measure of the sum of cellulose, hemicellulose, and lignin.  For cases where only the total cellulosic content (cellulose, hemicellulose, and lignin) or the adjusted cellulosic content is required, rather than separate measurements of cellulose, hemicellulose and lignin, the NDF method is appropriate.   
      d. Quantification of Starch Content 
Measurement of the starch content can be used to indirectly determine the cellulosic converted fraction.   To measure the starch content, enzymes are used to convert the starch to maltodextrin and then to glucose.  A subsequent reaction forms a dye in proportion to the amount of glucose present. The absorbance of this dye is measured with a spectrophotometer.  Standards are used to calibrate the spectrophotometer measurements and determine the starch content of the initial sample.  Because both starch and cellulosic material can be converted to fuel, a measure of the starch content of the feedstock and residual material after conversion can give information about how much starch was converted to fuel, which would help determine the cellulosic converted fraction.  
V. Allocating RINs using the 1426 equations
      a. Description of how EPA views cellulosic and non-cellulosic components of a feedstock as 2 separate feedstocks for RIN allocation
For simplicity, EPA selected to specify that if a single plant's average adjusted cellulosic content is less than 75%, then the feedstock is considered to be two feedstocks  -  specifically a cellulosic and non-cellulosic feedstock. This designation allows the equations in Table 4 to 80.1426 to be used to allocate both cellulosic and non-cellulosic RINs for the resulting mixed batch of fuel. This will require a separate converted fraction (CF) to be calculated for both the cellulosic and non-cellulosic fractions.  
      b. Calculation of the cellulosic CF and identification of the volume of D3-RINs 
The converted fraction (CF) is a unitless average mass percent representing that portion of the feedstock that is converted into renewable fuel. This is calculated by dividing the mass of the feedstock that is converted to renewable fuel by the mass of the feedstock that is capable of being converted to fuel. For in situ biofuel producers, this calculation will require some type of material balance accounting because the fuel will be a mixture of both cellulosic and non-cellulosic fuel.
There are several possible combinations of system boundaries and analytical methods that could be used to perform a mass balance accounting of cellulosic conversion, which would offer the data needed to calculate the corresponding cellulosic biofuel production via an in situ process. Thus, there are different ways to obtain the data to be used to calculate the CF values need to perform RIN allocation. We will present an example of how to use data characterizing feedstock composition, cellulosic conversion and certain process parameters to allocate RINs. In this example we assume that an in situ process is simultaneously processing corn starch and cellulosic polysaccharides associated with corn kernel fiber to produce ethanol. 
Data characterizing the corn composition (Table 1) and process specific data to elucidate the relative starch and cellulosic converted fraction (CF) will be required to appropriately allocate RINs. For this example, the validation protocol for obtaining the cellulosic CF requires performing an analytic process that involves 1) removing starch, 2) quantifying an initial level of cellulosic polysaccharides prior to hydrolysis, 3) quantifying a final level of cellulosic polysaccharides following hydrolysis, and 4) identifying the cellulosic ethanol by calculating the difference between the initial and final levels of cellulosic polysaccharides. In this example, the CF is determined to be the identified cellulosic ethanol (Step 4) divided by the theoretical ethanol yield for the total initial amount of cellulosic polysaccharides.  
Once the amount of ethanol resulting from the cellulosic polysaccharides has been identified, data characterizing the combined feedstock composition can be used to identify the cellulosic portion of the feedstock. An analysis of a representative sample of the feedstock would be used to characterize the corn composition (Table 1). Feedstock characteristics (Table 1), accurate CF values for cellulosic and non-cellulosic conversion, and the total volume (cellulosic plus non-cellulosic volume) can be used to allocate RINs for the example batch (Table 2).
Table 1  -  Relevant feedstock characteristics and composition
                                Corn Composition
Kernel Mass (M), lb wet
61.2 lb/bu
Corn Moisture (m), % wet
15%


Kernel Mass, lb dry
52.0 lb/bu
Starch, % of kernel dry
75.0%
Cellulosic polysaccharides, % of kernel dry
4.0%
       
RIN allocation for the cellulosic and non-cellulosic portions of the biofuel can be performed by applying the RIN volume (VRIN ) equation.
VRIN=EV∙VS∙FED-codeFEstarch+FEcellulose
Where EV is the equivalence value for the renewable fuel (in this case it is 1 since the fuel is ethanol), VS is the standardized volume of the renewable fuel produced, and FE is the feedstock energy value.  Note that this equation will be applied separately to calculate the RIN volume for each D-code. The feedstock energy (FE) is characterized by the FE equation. 
FE=M∙(1-m)∙CF∙E
Where M is the mass of the feedstock, m is the feedstock moisture content, CF is the converted fraction and E is the energy content of the components of the feedstock converted to fuel (Table 2). 

Table 2  -  Example: assumed inputs, intermediate values, and calculations
                  Example in situ RIN Allocation Calculation
Assumed feedstock for example (M), lb wet

Total standardized volume of fuel (VS), gal
                                                                      1,000,000
                                                                               
                                                                         64,158


Note: 77,000 Btu/gal ethanol

                                    Starch
                                   Cellulose
                                   Comments
Feedstock, thousand lb dry / batch
                                                                          637.5
                                                                             34
Account for moisture content and composition (Table 1).
 
Energy Content (E), Btu/lb dry
                                                                          7,600
                                                                          7,600
Theoretical feedstock energy assuming complete conversion.

Converted Fraction (CF)
                                                                           0.95
                                                                           0.35
Calculated using method approved at registration.

Feedstock Energy (FE), thousand Btu
                                                                      4,602,750
                                                                         90,440
Obtained by multiplying the dry feedstock, E, and CF.

RIN Volume (VRIN), gallons
                                                                         62,922
                                                                          1,236
Number of D6 and D3 RINs that would be assigned respectively for this example batch.

For the example presented, 1,000,000 pounds of wet corn feedstock was processed to produce 64,158 gallons of ethanol. Based on the feedstock characteristics (determined using an appropriate analytical method as discussed previously) for the given reporting period, the wet feedstock was composed of 637,500 pounds of starch and 34,000 pounds of cellulose on a dry mass basis. By establishing and using the dry mass of these components for subsequent calculation, the moisture content term in the FE equation simplifies to 1. Multiplying the dry mass of each feedstock by its corresponding converted fraction (CF) and energy content (E) yields the feedstock energy (FE). Using the resulting FE values for starch and cellulose respectively, and applying the VRIN equation for allocating RIN volumes to batches with these two feedstocks yields the RIN volume for each type of D-code. Note that since the fuel produced is ethanol, the energy equivalence value (EV) is 1.
This example shows how information about the wet feedstock mass, total fuel output, feedstock characteristics, and calculated process parameters can be used to allocate separate RINs for an in situ process converting both starch and cellulose to ethanol.   

