A view of the economics of converting cellulosic biomass to ethanol can provide a useful context for understanding the impact of feedstock cost and availability on competitiveness. However, it is vital to keep in mind that no cellulosic ethanol plant has yet been built, and

although various cost estimates have been published, all such analyses are just estimates (Wooley et al. 1999a; Aden et al. 2002). Thus, such information is primarily useful to help understand key cost drivers that can have a strong influence on costs, but accurate information will not be possible until several operational plants have accumulated enough of a learning curve to be reaching technology maturity. In addition, costs are highly dependent on the technology chosen for the design, and many low-cost approaches are not accessible as they are often protected as trade secrets and know-how. For this reason, this chapter will provide a basic outline of processing costs, and the reader is referred to other publications for detailed process designs and estimates while keeping in mind the approximate nature and other limita­tions of all of such analyses (U. S. DOE 1993; Wooley et al. 1999a, b; Aden et al. 2002).

A facility for processing a nominal 2000 dry tons per day of corn stover or 666,666 dry tons annually based on typical on — stream times is taken here to provide a perspective on the economics of converting agricultural residues into ethanol. In general terms, the process is based on use of dilute sulfuric acid for pretreatment and other features as outlined early in this chapter. Furthermore, the composition of corn stover is based on that reported in a coordinated study by the Biomass Refining Consortium for Applied Fundamentals and Innovation (CAFI), as summarized in Table 9.4 (Wyman et al. 2005b). This table also includes the maximum amount of ethanol that could be produced from these sugars at the theoretical limit.

Operating conditions for this analysis are as developed by Lloyd and Wyman as part of the CAFI study to achieve the highest total yields of glucose plus xylose from corn stover in a coordinated comparison to performance with other pretreatment options (Lloyd and Wyman 2005). For this approach, Table 9.5 outlines the operating conditions employed experimen­tally and the corresponding sugar yields. It is further assumed that these sugars can be fer­mented to ethanol with yields of about 92% of the theoretical maximum based on experience in industry and with the recombinant organisms employed for fermenting the five carbon sugars arabinose and xylose as well as galactose and mannose in addition to fermentation of glucose, and that 99.9% of the ethanol can be recovered in distillation and dehydration using proven technology. The lignin is burned to generate heat and power, and any extra left after heating streams and providing power in the process is exported for sale.

From this information, cash costs were estimated based on the assumed costs for feedstock, sulfuric acid, lime, nutrients, and labor outlined in Table 9.6. Enzyme costs were not included at this point because of the uncertainty in these values and will be considered later. Based on the yields reported by Lloyd and Wyman, the total estimated cost is about $0.785/gal prior to subtracting any coproduct credit for exported power. Because others have reported much lower yields based on less optimal performance data, a breakdown of operating costs are also included for lower overall ethanol yields of 82.6 and 73.6 gal/dry ton to give total costs estimates of $0.847 and $0.942/gal, respectively.

These estimates clearly show the importance of biomass costs in the economics as they dominate the overall cash costs. Thus, it is highly desirable to seek low-cost agricultural resi­dues that can dramatically cut these costs. However, several other points must be kept in mind for these rough estimates. First, they are at the plant gate and do not include transporta­tion, taxes, marketing, and a myriad of other costs that must be adsorbed before the fuel reaches the consumer. In addition, up to 50% of these totals could be added to compensate for the fact that ethanol contains about two-thirds the energy content of gasoline; however, ethanol can also be used more efficiently than gasoline in a properly optimized engine, which can make up for up to 50% of this difference (Bailey 1996). Although benefits are factored into the labor costs as 30% of wages and plant supervision and management are included, these costs do not include other overhead or manufacturing costs such as maintenance and administration or other costs that are typically factored off of capital estimates. First plants may also require more labor than assumed here.

A rough estimate of the income from sale of electricity is included in Table 9.6 based on a selling price of $0.05/kwh for power. In this analysis, about a third of the heat gener­ated by burning residues was calculated to be left for generating electricity at an assumed efficiency of 33%, that is, 1 Btu of electricity is assumed to be produced for every 3 Btu of heat available. These calculations take a penalty for the water in the lignin and other residuals by assuming somewhat more than half is water that must be vaporized during combustion. Furthermore, these values do not address the amount of electricity needed to run equipment because of the detail required for such estimates, but they also do not include any possibility of heat exchange or cascading to substantially reduce the amount of heat needed to bring the biomass feed up to pretreatment temperature or heat up the large fer­mentation beer stream to boiling for distillation. Thus, these values should be regarded as providing a rough idea of how much revenue could be gained from selling power, with the upper bound being about three times the value given and the lower bound being zero, the latter corresponding to either no power left to sell or no market into which to sell it. Of course, these values should be scaled proportionately if a higher or lower selling price is assumed for electricity.

Table 9.6 also includes a more mature performance case that obtains better yields for each step in the process, as outlined in Table 9.5 . Now, a cost of about $0.686/gal of ethanol is calculated for a feedstock cost of $60/dry ton, as outlined in this table. This estimate would drop to on the order of $0.64/gal after subtracting for sale of coproduct power, again subject to all of the caveats described above for other costs, capital recovery, and sale of power. Overall, the estimates show that high yields are beneficial, but that sale of power can some­what compensate by gaining value from the unutilized fraction. In addition, even the costs without including power and for low yields could be competitive with gasoline and ethanol from corn.

The cost estimates above do not include repaying for capital or the interest on debt to obtain that capital. However, estimating capital costs is extremely challenging, and as a result, most values can only be used as a rough guideline as to what to expect. For that reason, we will employ a guesstimate of $4.00/annual gallon for the 2000 tons/day base case to give a rough idea of the investment level required. This value is in line with estimates developed by NREL and others (Eggeman and Elander 2005; Wooley et al. 1999a). It is also consistent with the idea that a cellulosic ethanol facility combines the capital needs of a corn ethanol plant to make sugars and ferment them to ethanol with the capital demands of building a biomass power plant to burn the residual lignin and other unutilized portions of biomass to produce heat and power for the process with excess to export. In addition, we expect the ethanol facility to be somewhat more expensive than for a corn ethanol process because of the harsher condi­tions required for pretreatment and the longer times and more dilute solutions for sugar fer­mentation. Thus, about $4.00/gal appears in the right range. Yet, first projects can be more expensive than this estimate because of concerns about inexperience with the technology and resulting overdesign to ensure the project is successful. Learning curve experience will rapidly lower the cost once a facility is running through greater throughput with the extra equipment and improvements in operating conditions and biological catalysts (Goldemberg et al. 1993; Moreira and Goldemberg 1999).

Another aspect of capital costs is that they do not change linearly with scale of operation but scale as a fractional power of size. For example, employing the exponent of 0.67 often used as the norm would result in the capital cost increases by about 59% when the size of the operation is doubled instead of being 100% more expensive. Such economies of scale can be understood from the perspective that amounts of material and fabrication labor are proportional to surface area while capacity is proportional to volume of an equipment item. Furthermore, the surface to volume ratio of equipment drops with increasing size and does not increase linearly. Exponential scale factors vary with the type of equipment, as tabulated in several reference books and are almost always less than 1.0 (Peters et al. 2003). The result is that unit capital costs drop as the process throughput increases, leading engineers to favor larger-scale operations to minimize capital costs. Consequently, the assumed cost of $4.00/ annual gallon for cellulosic ethanol would drop to about $3.18/annual gallon if the capacity was doubled to 4000tons/day and the 0.67 exponent were applicable.

To counter the possibility of lower unit costs for larger facilities, it is often stated that economies of scale cannot be realized in biomass processing because of the low density of biomass crops. However, if we consider that, about 3.75 dry tons of corn stover/acre would result for a corn productivity of 150 bushels/acre given the virtual one-to-one ratio of above­ground plant to corn kernels. In addition, within the 50-mile radius typically considered to be reasonable for collecting corn and wood, this volume of corn stover production would amount to about 1.875 billion gallons of ethanol. Even if we assume that only 1.0dry tons/ acre on average can be accessed and/or removed sustainably from the field, the result would be about 500 million gallons of ethanol annually within the 50-mile radius. A few studies have clearly shown that even this low harvest rate favors conversion facilities of at least

10,0 dry tons per day, giving an annual ethanol production of about 250 million gallons (Wyman 1995). The primary limitation to building plants of this size is the high capital costs; for example, a 10,000 gallon per year operation would cost on the order of $3.0 billion or more. Although it may be possible to raise such large sums for a mature process, the risk of first applications is considered too great for most investors, and it is extremely unlikely that first projects would be this large.

A key question is how to annualize up-front capital costs over the life of the facility, and a useful approach is based on a projected cash flow coupled with an appropriate discounting formula (Wyman 1995; Wooley et al. 1999a). However, the challenge is the choice of appro­priate parameters in this analysis which are in turn tied to many other factors such as economic lifetime of the plant and expected rate of return by the financing entity. Furthermore, the interest rate and economic lifetime will depend on the stage of technology development as high rates of return are expected for first applications while much lower rates can be negoti­ated once a successful track record is firmly established. Thus, a first project may require payout in only a few years time while mature technology may be able to pay off capital over a period of 20-30 years. Unfortunately, this classical relationship between rate of return and risk presents a major chicken-and-egg dilemma for building first-of-a-kind cellulosic ethanol facilities in that coupling high rates of return demanded by investors with a capital-heavy design to ensure successful operation will almost surely result in overall costs that are too high for the commodity fuel market. In this regard, it is vital to remember that the competi­tion is gasoline which has benefitted from over a hundred years of learning curve advances, paid off capital, substantial subsidies to ensure stable supplies from hostile regions of the world, an established infrastructure, and no consideration for societal costs associated with its use.

Return on capital could result in virtually a 100% capital charge in the first year for new technology from investors demanding fast paybacks to as low as 10%—15% if project finance could be used similar to classical utility financing. Given that fuels are a commodity business, it is unrealistic to expect very rapid payback times. If we take a 20% annual charge as likely for reasonably mature technology, the total of cash cost plus capital charges would amount to something like $1.60/gal for the case based on laboratory data in Table 9.5 and a capital cost of $4.00/annual gallon. To this, additional costs need to be added for maintenance, overhead, and the other aspects mentioned before that were not included. The resulting cost could be promising compared to gasoline when oil prices are high but not when they are at lower levels. However, given their large contribution to the estimated costs, use of a low-cost residue could improve competitiveness significantly.

A key element left out of these estimates is the cost of enzyme. In many studies on enzy­matic hydrolysis of pretreated biomass, enzyme loadings of about 15FPU/g glucan were applied to realize good sugar yields from the pretreated solids. However, at a typical specific activity of about 0.5FPU/mg protein, these loadings amount to on the order of 0.251b of protein/gallon of ethanol produced including the ethanol made from the hemicellulose sugars that are often released during pretreatment. Reports have been made of advances in enzyme technology lowering the cost to about $0.10-0.20/gal (American Institute of Chemical Engineering 20041 American Chemical Society 2005) and such a price would bring the overall cost of ethanol in our simple analysis to something under $2.00/gal. Yet, this price would require protein costs of about $1.00/lb or less, but no offers are known to sell enzyme at such a price. Thus, it appears that enzymes are still very expensive, with costs of about $1.50/gal possible, and such high costs would prevent cellulosic ethanol from being competi­tive when using conventional enzymes.

Overall, this analysis shows that low-cost ethanol is possible from cellulosic biomass if we can bring enzyme costs down, and a particularly promising route to this end is through development of organisms that can both make enzymes and ferment the sugars released, as outlined at the beginning of this chapter. In addition to reducing process steps and capital and operating costs associated with separate enzyme production, the most important advan­tage to this approach called CBP is the anaerobic production of the enzymes needed to break down cellulose and hemicellulose to sugars. This feature overcomes a major barrier to current aerobic enzyme production methods that use huge amounts of energy to compress air and agitate the enzyme production vessels intensely to promote sufficient respiration by the organ­isms to grow and produce enzymes. Furthermore, such high-energy inputs also translate into major power drains on the facility and the need to provide very expensive cooling to remove all the heat generated by such intense aeration and agitation. Thus, unless enzymes are devel­oped with much higher specific activities than historically seen, the best course to low costs appears to be to follow a CBP strategy.

In closing this section, it is vital to point out that the economics of ethanol production are very site specific and for that reason hard to generalize. Such aspects as labor rates and skill levels, biomass type and availability, competing demands for feedstocks, access to raw mate­rials, transport of raw materials and finished products to and from the site, and access to markets can have a large effect on costs. In addition, while this rough analysis has been applied to corn stover because of its relatively large production, other agricultural residues may be more attractive due to cost, susceptibility to conversion, coproduct opportunities, and yields. Thus, one size does not fit all, and careful consideration should be given to regional factors before applying rigid economic evaluations.