Because of the complexity of plant cell wall structure in terms of its components and organi­zation, it is difficult to decipher cell wall imaging results without knowledge of its molecular and electronic structures. Even though recent experiments using synchrotron X-ray and neutron diffraction have elucidated the crystal structures of cellulose Ia and Ip from algae and tunicate respectively (15, 16), the cellulose structures in plant cell walls remain largely unclear. However, it is known that the cross-sectional diameter of an elementary cellulose fibril is only about 3-5 nm in size (9). It has been further proposed (9) that the cellulose elementary fibril consists of 6, 12, and 18 glucan chains in the center, middle, and interface respectively, with increasing disorder. Only the six center chains can be considered truly

5 pm

(d) CfCBM3-GFP

Cellulose elementary fibril

Hemicelluloses

Figure3.10 Total internal reflection fluorescence micrograph of fresh maize parenchyma cell wall labeled by QCBM3-GFP and QCBM6-RFP. (Modified from Ding et al, 2006.) (Reproduced in color as Plate 3.) crystalline. This hypothesis is yet to be tested due to the lack of available experimental techniques. Combined ab initio and force-field molecular dynamics simulations provide a unique tool to investigate the plant cell wall structure at the molecular and electronic levels. Molecular modeling will help shed light on the true nature of biomass recalcitrance at the atomic and molecular levels, thus pointing out ways to overcome this resistance to its deconstruction. Even less is understood about the structures of hemicelluloses, and the mechanisms and energetics between hemicelluloses and cellulose interactions. Hemicellu­loses are polysaccharides consisting of mostly xylose and other minor sugars. Xylan and xyloglucan (XG) are the main hemicelluloses in plant cell walls. Xylan displays large struc­tural variation and complexity. The backbone of xylan is a linear polymer of xylose linked via the (3-(1-4) glycosidic bond. In higher plants, the linear backbone is substituted by a variety of side chains mainly a-L-arabinofuranosyl and a-D-glucopyranosyl uronic acid units (73). XG has a glucan backbone wherein up to 75% of the glucose (G) units are sub­stituted at O6 with a-D-xylose (X) (74). Some of the xylose residues are then substituted at O2 with (3-D-galactose (L), which can be further substituted at O2 with a-L-fucose (F). Previous molecular dynamics studies (75-79) of plant cell wall structure focused mainly on

crystalline cellulose structures with unrealistic sizes and dimensions. Furthermore, the ma­jority of these simulations relied only on force-field-based molecular dynamics simulations (73,74,76-78). It is known that the accuracy of the simulations of carbohydrates will depend on the quality of the force field used. Matthews and coworkers (75) used second-generation CHARMM force fields; however, their cellulose crystal size was significantly larger than the actual size in plant cell walls. It is known that surfaces and interfaces will dominate the prop­erties of nanostructured materials. It is thus possible that this diminished cellulose surface to volume ratio in Matthews’ initial work led to some unrealistic conformational and hydrogen bonding structures in cellulose. Because the properties of cellulose and hemicelluloses are dominated by hydrogen bonding interactions, the force field method is not ideal for describ­ing this type of interaction. In contrast, application of ab initio-based molecular dynamics simulations to investigate the atomic and electronic structures of crystalline cellulose Ip yielded very good agreement with X-ray and neutron diffraction results, particularly for the hydrogen bonding network (79). Ab initio molecular dynamics with CPMD (80, 81) is another promising method to investigate the structures of celluloses and hemicelluloses in regard to their interaction mechanisms and general energetics. CPMD is capable of simu­lating thousands of atoms with currently available computing power (82). Combined with classical MD, it is possible to investigate the structures of the plant cell wall with up to tens of thousands of atoms.

A full and detailed integration of science and engineering research will be needed for the continued growth of the biorefinery and biofuels. Fundamental R&D in biomass conversion must be targeted to process improvements based on technical barriers. An integrated funda­mental and applied research program in biochemical conversion must include advancements in these three critical areas.

2.42.1.1 FEEDSTOCK ENGINEERING

• Develop genomics and agronomic/silviculture strategies to maximize the yield and quality of developing energy crops.

• Design and manipulate plant cell wall composition and structure to maximize the yield of fermentable sugars.

2.4.2.1.2 CELL WALL SACCHARIFICATION

• Analyze glycosyl hydrolase structure/function as it applies to plant cell wall deconstruction.

• Develop improved (engineered) enzymes for advanced biochemical conversion technolo­gies and integrate them with pretreatment chemistries.

2.4.2.1.3 STRAIN DEVELOPMENT

• Apply systems biology and biochemistry to strain improvement to increase the conversion of sugars released during biomass deconstruction to ethanol and products.

• Focus on strains that will produce saccharifying enzymes and ferment the resulting sugars to ethanol.

Biomass-degrading enzyme preparations must work to convert as much of the polysac­charides in the cell wall as possible to monomers. Currently, high loadings of cellulases are needed to reach 95% conversion of cellulose in pretreated biomass to sugars in 3-5 days using simultaneous saccharification and fermentation. Cellulase preparations are expensive in the biorefinery context for two reasons historically: 1) the source of the enzymes, usually Trichoderma reesei, was costly to grow and induce and 2) the specific performance (or activity) are low compared to other polysaccharide degrading enzymes.

However, a significant breakthrough in reducing the cost to produce and use T. reesei cellulases in the biorefinery was achieved by the DOE Office of the Biomass Program funded subcontracts awarded to Genencor International andNovozymes Biotech (2000-2005). Over the period of performance of these subcontracts, this cost was reduced about 10-fold from the starting cost of about $5 per gallon of ethanol produced. We note that only a small percentage of the final cost reduction came from actually improving enzyme structure/function. The question is often asked: Howlow-cost must cellulases be to enable a new biorefinery industry? One answer may be based in considering the current cost of starch-degrading enzymes, which are about $0.01-$0.05 per gallon of ethanol produced. New amylase and glucoamylase technology introduced in 2005 will further lower these costs. For cellulase costs to approach that of starch-degrading enzymes, we must focus on considering the resistance of cellulose in plant microfibrils to deconstruction.

A deep understanding of the structure/function principles governing cell wall polysaccha — ridase action is critical. The study of cellulase action is especially challenging, considering that these enzymes function to first decrystallize cellodextrins and then hydrolyze the extracted chains to cellobiose and glucose. This process is not currently understood at the kinetic or thermodynamic level. Fundamental questions exist on the limits of enzyme activity and the action of soluble enzymes or enzyme aggregates on insoluble polymeric substrates in aqueous environments. It is possible, for example, that enzymes acting on microcrystalline cellulose are already working at the maximal rate!

The areas of poor scientific understanding presented above have clearly deterred past re­search programs aiming to reduce cellulase cost by improving performance. To summarize, the task of improving the specific activity of cellulases is complicated by our poor under­standing of 1) cellulase natural diversity, 2) cellulase active-site architecture, 3) cellulase processivity, 4) cellulose decrystallization, and 5) the cellulose structure in plants.

Most recent achievements in plant cell wall biosynthesis research resulted from the anal­ysis of Arabidopsis thaliana cell wall phenotype mutants and the availability of genome sequence data (see detailed review in Chapter 5) (3, 4). The cellulose synthase complex (CelS), known as rosettes in higher plants, was first observed using electron microscopy and the freeze-fracture sample preparation technique. Rosettes appear in hexagonal geometry with a honeycomb pattern arrayed in the plasma membrane (5,6). Rosettes are believed to be responsible for the synthesis ofelementary fibrils in most current plant cell wall biosynthesis models (7). More recently, mutant analyses [reviewed by Doblin and coworkers (4)] and immunolabeling (8) have confirmed that these rosettes are composed of cellulose synthase (CesA) proteins, and that at least three types of CesA isoforms (a1, a2, and p) are required

Cellulose elementary fibril

6 core chains

Figure 3.4 Model of plant cell wall cellulose elementary fibril and its synthesis. In this model, at least three types of cellulose synthases (CesA subunits, a1, a2, and p) are needed to spontaneously assemble the rosettes that composed of 6 x 6 CesA enzymes synthesizing 36-chain cellulose elementary fibril. The rosettes may also form arrays in the cell membrane, in this case, a number of rosettes synthesize a bundle of elementary fibril, the macrofibril. The estimated dimensions of elementary fibril are 3 x 5.5nm that agrees with direct measurement using atomic force microscopy (see also Figure 3.9). The depiction of the glucan chains is based generally on an X-ray structure of cellulose Ip. It has been proposed that the cellulose elementary fibril may contain three groups of glucan chains: in group C1 (red) there are 6 crystalline chains;in group C2 (green) there are 12 sub-crystalline chains with a small degree of disorder; and in group C3 (blue) there are 18 surface chains that are sub-crystalline with a large degree of disorder. (Modified from Ding and Himmel, 2006;Himmel etal., 2007) (Reproduced in color as Plate 1.) for the spontaneous assembly of single rosettes. The next question is — how many CesA proteins are assembled into single rosettes?

A model of rosettes has been recently proposed (9) (see Figure 3.4) in which three types of interactions are needed for the spontaneous assembly of rosettes in the plasma membrane, these are p-p, at-p, and a2-p. The p-p interaction facilitates the assembly of six subunits into rosettes, as well as the array of rosettes in plasma membrane. The six subunits of rosettes are identical and each rosette is composed of one molecule of at, two molecules of a2, and three molecules of p. This simplified rosette model seems probable because of the proposed at positioning and the fewer types of interactions required for the rosette assembly in plasma membrane. There is no direct biochemical evidence yet to confirm how the rosettes are assembled, probably the zinc finger domains of CesA proteins, plasma

membrane, and microtubules are all involved. During rosette assembly, the a1 molecule from each asymmetric subunit is always indexed to the center of the rosettes to ensure the correct positioning of each of the three types of CesA proteins. The rosette assembly probably starts with the dimerization (p-p) of the N-terminal zinc finger domain of CesAs. The next step is an a-p interaction, with each a isoform interacting with two p isoforms. Here as well, there is no direct evidence yet to prove how cells control these interactions among the catalytic units. One possibility is that different protein-protein interactions (i. e., p-p and a-p) occur under different physiological conditions or in different locations in the cell. Experiments of in vitro assembly or in vivo labeling of co-expressed CesA proteins might confirm this hypothesis. Furthermore, having multiple CesA proteins co-expressed in the same cell does not necessarily mean that they are all assembled into the same rosettes (10). Each position (i. e., at, a2, and p) in different rosettes may have different CesA proteins or different CesA proteins may fit in the same type of position in one rosette. For example, there are three p positions in one subunit and 18 positions in one rosette (the p position might be occupied by different CesA enzymes).

Most researchers agree that cellulose is synthesized in the plasma membrane, whereas hemicelluloses are assembled and secreted from the Golgi vesicles (11). Figure 3.5 is a schematic diagram showing cellulose being synthesized by the rosettes that comprise 36 CesA

enzymes. This enzyme complex thus produces 36 (3 -1,4-glucan chains, which simultaneously coalesce to form the cellulose elementary fibril. Hemicelluloses are synthesized in Golgi apparatus and secreted to cell wall. The hemicellulose particle interacts with the surface of newly synthesized cellulose elementary fibrils that usually form a ribbon-like bundle (the macrofibril) (9). During cell wall expansion, the macrofibril is thought to split into single elementary fibrils, the hemicellulose particles unfold and layer upon the microfibrils where some fraction of these polysaccharides start to coat the cellulose surface though numerous hydrogen bonds.

Transportation is a significant cost, and it can be a barrier to using some feedstock resources. Transportation and handling operations can account for nearly 50% of the capital investment of a feedstock assembly system. Unlike the other unit operations in the feedstock supply system that can impart additional value to the feedstock, transportation costs simply move the feedstock to the biorefinery. Hence, reducing these costs to a minimum is vital to achieving low feedstock supply system costs.

Regardless of the transport method (e. g., truck, rail, or barge), bulk density is the key tech­nical parameter that must be addressed to decrease transportation costs. As such, methods to increase bulk density are a focus of the transportation and handling R&D. Bulk handling is also affected by feedstock rheological properties and this, too, is an area of focus.

Specific research needed to reduce transportation and handling costs includes:

• Understanding physical and rheological feedstock properties (including bulk density) as they relate to handling systems to optimize handling and transportation efficiencies

• Evaluating innovative transportation and handling methods.

The natural structure of modern plants (83) believed to contribute to the recalcitrance of biomass to chemical or enzymatic degradation include (i) the epidermal tissue of the plant body, particularly the cuticle and epicuticular waxes, (ii) the arrangement and density of the vascular bundles, (iii) the relative amount of sclerenchymatous (thick wall) tissue, (iv) the degree of lignification (14), (v) the warty layer covering of secondary cell walls, (vi) the structural heterogeneity and complexity of cell wall constituents, such as microfibrils and matrix polymers (84), (vii) the challenges for enzymes acting on an insoluble substrate (85), and (viii) the presence of inhibitors in cell walls or generated during the conversion processes to subsequent fermentations (86). These chemical and structural features affect liquid penetration and/or enzyme accessibility and activity, and thus the overall biomass conversion costs.

Current biomass conversion technology uses chemical pretreatments to remove hemicel­luloses from the microfibrils, which in turn exposes the crystalline cellulose core rendering it more amenable to the action of cellulase enzymes. In addition, pretreatment typically breaks down the macroscopic rigidity of biomass and decreases the physical barriers to mass transport. A suite of enzymes such as cellulase, hemicellulases, and accessory en­zymes are introduced to depolymerize cellulose (saccharification). During this process, the cell walls are decomposed at the molecular level, hemicelluloses and lignins are either hy­drolyzed in place or allowed to migrate, and the crystalline cellulose structures are exposed and modified. Eventually, the polysaccharides are depolymerized to monomer sugars for fermentation. The technical barriers that contribute to the high cost of current biomass conversion processes have been identified as low sugar yields and low efficiency of en­zyme performance. To overcome these problems, improvements of these processes rely on further understanding of cell wall ultrastructure and the molecular mechanisms of enzyme hydrolysis.

The lignocellulose biorefinery is envisioned to comprise four major processing steps: 1) feedstock harvest and storage, 2) thermochemical pretreatment, 3) enzymatic hydrolysis, and 4) sugar fermentation to ethanol or bio-based products. Among these processes, chem­ical and enzymatic treatments of biomass contribute to the majority of the processing cost. An acidic chemical pretreatment step is usually conducted to depolymerize and solubilize hemicelluloses (approximately 20-40% wt/wt of biomass). This step converts hemicellu — loses to monosaccharides and oligosaccharides, which can be further hydrolyzed in the later processes. Thermochemical pretreatment of biomass has long been recognized as a critical step to produce celluloses with acceptable enzymatic digestibility. Various technologies have been developed to accomplish this goal. For example, dilute sulfuric acid pretreatment at 140-200°C renders the cellulose in cell walls more accessible to enzymes (83). For dilute acid treatments (pH ~1.5), release of mono and oligomeric sugars from hemicellulose exhibits multi-modal kinetics. It is this slow monosaccharide release phase of chemical hemicellulose hydrolysis that directly relates to the high process conversion cost (87, 88). A number of researchers (88-93) have noted that the depolymerization of hemicellulose appears to be best described as a pair of parallel first-order reactions where one takes place at a fast rate and the other at a much slower rate. Pretreatment schemes based on alkaline explosive decom­pression or organic solvent extractions have also been used with considerable success (86). The alkaline process, known as ammonia fiber expansion (AFEX), leaves the hemicellulose in place, yet renders the remaining cell walls considerably more amenable to enzymatic hydrolysis (94). At moderate pretreatment severities (95), the hemicelluloses are hydrolyzed and solubilized as monomers and oligomers; however, the yields of solubilized sugars are often unpredictable, and less than ideal (96). The improvements of chemical pretreatments now focus on increasing sugar yields and reducing the severity.

The factors that govern the pretreatment process at the level of the cell wall are not clear today. However, this process undoubtedly depends on a number of factors, such as hemicellulose composition, biomass density, the presence of non-sugar components (i. e., lignin, ash, acetyl, and uronic acids), and most importantly plant cell wall structure (i. e., types of cells, ratios of primary and secondary cell walls, as well as the macromolecular structure and arrangement of cell wall polymers).

Although it is not fully known how many enzymes are involved in cell wall deconstruc­tion in nature, over a hundred families of glycoside hydrolases (GH) have been identified in the CAZY database (http://afmb. cnrs-mrs. fr/CAZY/fam/acc_GH. html). Three general categories ofenzymes are considered necessary to hydrolyze native cell wall materials: cellu — lases, hemicellulases, and the accessory enzymes, which include hemicellulose debranching, phenolic acid esterase, and possibly lignin degrading/modifying enzymes (97). Once the hemicellulose barrier associated with cell wall microfibrils has been compromised by chem­ical pretreatments, cellulase enzymes can be used to hydrolyze the crystalline cellulose. Crystalline cellulose is hydrolyzed by the synergistic action of endo-acting enzymes known as endoglucanases, and exo-acting enzymes, known as exoglucanases. The endoglucanases locate surface sites along the glucan chain and insert a water molecule in the (3-(1,4) bond, creating a new reducing and non-reducing chain end pair. The release of cellobiose from the cellulose is thought to occur at these new chain ends and this process considered to be the rate limiting step in cellulase action, is accomplished by exoglucanases also known as the “processive” cellulases. At this time, studies of the synergistic reaction of cellulases are primarily based on assays on purified cellulose substrate such as Sigmacell, Avicel, or bacte­rial cellulose, not cell walls (98). There is no doubt that the deconstruction of the complex structures found in cell walls require a wider range of enzymes than just the cellulases; in fact, the synergistic action of many GH family enzymes as well as whole microbial cells are likely critical, and yet poorly understood (99).

Acknowledgment

This work was supported by the US Department ofEnergy, Office ofthe Biomass Program.

The objective of bioengineering research is to acquire new understanding in broad-based aspects of applied process engineering research. Applying commercial enzyme preparation components to various pretreated biomass samples, both within and beyond the scope of established consortia, is also a key component of this research. Work will need to be extended to include comparative pretreatment analysis for multiple feedstocks (e. g., corn stover, switchgrass, and hybrid poplars) and additional pretreatment process impacts (e. g., identifying hydrolyzate conditioning requirements for different pretreatments).

The applied research program required will need to include advances in process applica­tion knowledge at two levels. The first will address process-related engineering research that converts new understanding from fundamental research to the biorefinery context. The sec­ond will use process-related engineering information to develop industry recommendations regarding process parameters, equipment, and operating conditions.

Thermal chemical pretreatments are currently necessary to enable cellulase access through the hemicellulose sheath of the plant cell wall microfibrils, thus exposing the crystalline cellulose core. This pretreatment must be just severe enough to create this access, but not so severe as to divert sugars to non-fermentable or toxic compounds (6-8). Today, the de­polymerization of arabinoxylans (hemicellulose in hard woods and grasses) in cell walls is accomplished with good results by a variety of hot acid, hot water, and alkaline treatments. Final conversion of liberated soluble oligosaccharides is often accomplished using hemi — cellulases. Soft woods contain hemicellulose composed primarily of galactoglucomannan, which liberates galactose, glucose, and mannose upon depolymerization. These sugars are all fermentable by natural yeasts.

We recognize that the capital cost of the pretreatment unit operations is a critical factor for enabling the future biorefinery. High pretreatment capital cost is primarily due to the materials of construction required by conditions of high severity. In this context, severity is based on the pretreatment acidity, temperature, and time at temperature. New combinations of biological preconditioning (before thermal chemical pretreatment) and better thermal chemical pretreatments prior to enzymatic conversion have promise for overcoming this barrier.

The reactions of plant cell wall chemical constituents and ultrastructure to pretreatments must also be understood at a more detailed level. For example, basic research is required to understand the relationships between feedstock plant structure and composition. Simply, we need to develop better chemical and enzymatic treatments. Solving the yield challenge requires the integration of the complexities of plant structure, chemical pretreatment, and enzyme action. This integrated approach is a new and critical research paradigm.

Cell walls are deposited by layers upon synthesis. Generally, primary cell walls are synthe­sized when cells grow. Secondary cell walls are deposited when cell growth has ceased. Some cells possess only primary wall, such as the parenchyma cells; however, secondary deposi­tion may occur on most of cell walls when cells age. For example, thin layers of secondary deposition are commonly observed in mature parenchyma (Figure 3.6). Also, there is only one microfibril sheet in each lamella. These thin secondary wall lamellae are measured ap­proximately 10 nm, which appear to contain only one layer of parallel-arranged microfibril. Microfibrils are rotated approximately 50° with respect to each lamella. Secondary cell walls commonly consist of three anatomical layers: the outer (S1), middle (S2), and inner (S3) layers. The thickness of each layer varies in different cell types and tissues. The S2 layer is often thickest, and sometimes contains sub-layers. Another important structure in cells with

thick secondary wall (e. g., vessel cells) is the warty layer in the inner surface, this layer is comprised of granules and amorphous structures (Figure 3.7) that restrict access by rumen microorganisms (12).

2.2.2.1 Introduction

Basically, biochemical conversion is the liberation and fermentation of sugars from biomass feedstocks. The challenge is to efficiently convert the carbohydrate portion of the biomass to sugars, or “saccharify” it, and ferment the impure sugars to ethanol with a robust mi­croorganism. In this process, the lignin component of the biomass provides the heat and power needs of the process. This process shows great promise for producing ethanol cost effectively with high yields and minimal environmental impact.

There are two primary routes for saccharification: 1) acid hydrolysis, with either concen­trated or multiple stages of dilute; and 2) pretreatment followed by enzymatic hydrolysis. In the 1980s, DOE evaluated the long-term potential of each process (31) and although at the time acid hydrolysis technology was further developed and appeared less expensive, com­paring progress and future potential suggested that enzymes offered greater opportunity for ethanol cost reduction in the long run (32). Acid hydrolysis technologies are certainly feasible and in proper niche situations they are being pursued to commercialization.

Enzyme hydrolysis requires a pretreatment to generate an intermediate material that can be effectively digested by enzymes. Dilute acid pretreatment of corn stover followed by enzymatic hydrolysis can achieve more than 90% conversion of cellulose to glucose (33). Various pretreatment methods have been suggested; most use heat coupled with a chemical catalyst such as an acid, base, or other solvent. Recent advances (34) suggest that “accessory”

Products

Byproducts

enzyme systems such as hemicellulases could lead to low-severity and low-cost pretreatment processes in the future. Although currently it appears that dilute-acid-based approaches give the best overall performance over the range of feedstocks envisioned for biochemical conversion, other approaches such as alkaline approaches also show considerable promise; and more development is needed in the pretreatment area to meet cost performance goals. Wyman and coworkers (35) give a good recent review on the comparative performances of the leading pretreatment technologies under development.

A representative block flow diagram of a biochemical conversion route to convert lignocel — lulosic biomass to ethanol using dilute acid pretreatment followed by the enzymatic hydroly­sis approach is shown in Figure 2.4. The process also includes ancillary supporting operations such as feedstock interface handling and storage, product recovery, wastewater treatment, residue processing (lignin combustion), and product storage not shown in Figure 2.4.

The feedstock is delivered to the feed-handling area for size reduction and storage. From there, the biomass is conveyed to pretreatment and conditioning. In this area, the biomass is treated with a dilute sulfuric acid catalyst (the current leading pretreatment technology) at a high temperature for a short time. This hydrolyzes the hemicellulose to a mixture of sugars (i. e., xylose, arabinose, galactose, mannose, and a small amount of glucose) and other compounds. In addition, the pretreatment step makes the remaining biomass more accessible for later enzyme saccharification. A conditioning process then removes byproducts from the pretreatment process that are toxic to the fermenting organism.

In hybrid saccharification and co-fermentation (HSF), the pretreated solids (now pri­marily cellulose) are saccharified with cellulase enzymes to form monomeric glucose. This requires a couple of days, after which the mixture of sugars and any unreacted cellulose is transferred to a fermenter. An inoculum of fermenting microorganism is added, and the sugars are fermented to ethanol. Meanwhile, the enzymes are used for further glucose pro­duction from any remaining biomass, which is now at conditions optimal to fermentation. After a few days of fermentation and continued saccharification, nearly all the sugars are converted to ethanol. The resulting beer (or low-concentration ethanol) is sent to product recovery.

Products

Byproducts

Product recovery involves distilling the beer to separate the ethanol from water and resid­ual solids. An azeotrope of water and ethanol is converted to pure ethanol using vapor-phase molecular sieves. Solids from the distillation bottoms are separated and sent to the boiler (called residue processing). Distillation bottoms liquid is then concentrated by evaporation using waste heat. The evaporated condensate is returned to the process, and the concentrated syrup is sent to the burner.

Part of the evaporator condensate, along with other wastewater, is treated by anaerobic and aerobic digestion. The biogas (which is high in methane) from anaerobic digestion is sent to the burner for energy recovery. The treated water is suitable for recycling and is returned to the process.

The solid distillates — the concentrated syrup from the evaporator and biogas from anaer­obic digestion — are burned in a fluidized bed combustor to produce steam for process heat. The majority of the steam demand is in the pretreatment reactor and distillation areas. Generally, the process co-generates enough electricity to use in the plant and to sell to the grid. A detailed description of the conversion process described above is provided by Aden and coworkers (17).