Energy, wealth, and human development: Why and how biomass ...

Energy, Wealth, and Human Development: Why and How Biomass Pretreatment Research Must Improve Bruce E. Dale and Rebecca G. Ong Biomass Conversion Research Laboratory, Dept. of Chemical Engineering and Materials Science, Michigan State University, East Lansing, MI 48824 DOE Great Lakes Bioenergy Research Center, East Lansing, MI DOI 10.1002/btpr.1575 Published online July 16, 2012 in Wiley Online Library (wileyonlinelibrary.com).

Abstract: A high level of human development is dependent on energy consumption (roughly 4 kW per person), and most developed countries that have reached this level have done so through the extensive use of fossil energy. However, given that fossil resources are finite, in order for developed countries to maintain their level of development and simultaneously allow developing countries to reach their potential, it is essential to develop viable renewable energy alternatives. Of particular importance are liquid fuel replacements for petroleum, the fossil resource that primarily drives commerce and economic growth. The intent of this article is to remind our fellow biofuel researchers, particularly those involved in lignocellulosic pretreatment, of these global issues and the serious nature of our work. We hope that this will inspire us to generate and report higher quality and more thorough data than has been done in the past. Only in this way can accurate comparisons and technoeconomic evaluations be made for the many different pretreatment technologies that are currently being researched. The data that primarily influence biorefinery economics can be subdivided into three main categories: yield, concentration, and rate. For these three categories we detail the specific data that should be reported for pretreatment research. In addition, there is other information that is needed to allow for a thorough comparison of pretreatment technologies. An overview of these criteria and our comparison of the current state of a number of pretreatment technologies with respect to these criteria are covered in C 2012 American Institute of Chemical Engineers Biotechnol. Prog., 28: the last section. V 893–898, 2012 Keywords: lignocellulose conversion, biofuels, pretreatment, cellulases

A Sobering Context for Biomass Pretreatment Research A nation’s per capita rate of energy consumption (power consumed) is a strong predictor of national wealth and individual human development. Approximately, 4 kW per capita power consumption is required for people to achieve a high level of development as measured by the United Nations Human Development Index (HDI), a composite metric of human development including measures of wealth, education, and health (Figure 1). If all 7 billion people on the planet were to achieve this level of power consumption (4 kW per person) the total power required would be about 28 TW (approximately 26 billion Btu/sec), roughly twice the current world power consumption (15 TW). About 80% of current world power consumption is based on fossil fuels: petroleum, coal, and natural gas.4 The developed world is rich, educated, and relatively healthy in large part because it consumes a great deal of power—in other

Correspondence concerning this article should be addressed to B. E. Dale at [email protected]. C 2012 American Institute of Chemical Engineers V

words, a great deal of fossil fuels. The developing world is trying hard to become richer, better educated, and healthier by consuming more power. For example, between 1990 and 2008, the per capita power consumption in China increased by nearly threefold and in India by nearly twofold,5 however, these rates of consumption and level of human development are still far lower than those of the western world. Any energy market ‘‘surpluses’’ that result when developed countries reduce their fossil fuel use are quickly absorbed by developing countries as they strive to improve their economies, infrastructure, and the quality of life of their citizens. At least two conclusions arise from a thoughtful consideration of these facts: (1) existing fossil energy resources will be consumed sooner rather than later as the developing world draws increasingly on these finite resources and (2) fossil energy resources are utterly inadequate to sustain national wealth and human wellbeing in the long term. Furthermore, the long term may not be all that long. Already there is strong evidence that peak oil (the maximum rate of oil production) has arrived. Total world production of oil has been constant at around 80 million barrels per day since 2004.6 Peak oil does not mean we are ‘‘running out" of oil. 893

894

Biotechnol. Prog., 2012, Vol. 28, No. 4

Figure 1. Relationship between 2008 per capita primary energy consumption and HDIs for 170 countries. The figure concept is based on a figure by Martinez and Ebenhack1; and the inset is based on a figure from the Human Development Report Office (HDRO) of the United Nations (UN).2 Data on the human development indices are from the UN HDRO2 and per capita primary energy consumption are from the U.S. Energy Information Administration.3

It means that we have reached the maximum rate at which oil can be supplied and eventually that maximum rate of oil production must decline. As world economic growth has paralleled the growth in the oil consumption rate; a stable or declining oil consumption rate has profound, negative consequences for continuing economic growth and all the social/ political structures that depend on continuing economic growth. Seen in this light, renewable energy is not just a ‘‘nice idea.’’ Renewable energy is essential if more people, now and in the future, are to have the opportunity to develop their potential. There are three primary energy services provided by energy resources. These services include heat, electricity, and mobility. (Mobility is defined as the ability to transport our goods and ourselves.) Although we have several sources of renewable heat and electricity (e.g., solar, wind, hydro, geothermal, biomass, etc.), we are much more constrained in our potential sources of renewable mobility. At best, renewable electricity can provide about half of current mobility services, mostly for personal vehicles and light duty transport. However the current U.S. infrastructure cannot support 100% conversion of passenger vehicles to electric power. Total 2008 U.S. light duty vehicle power consumption was equal to 0.562 TW.3 In contrast, total U.S. electricity generation across all sectors was equal to only 0.472 TW.7 Also, this comparison does not take into consideration energy inefficiencies during electricity transmission and conversion to mobility. In addition, electricity is unable to service the very vehicles on which most commerce depends. This includes all aviation and ocean shipping, and unless there are significant changes to the existing electrical systems and infrastructure, most land freight (heavy truck and rail transport). From the work of Adam Smith and subsequent economists, commerce has been recognized as a key element in the trade that enables greater wealth and therefore greater human development. Currently, commerce is almost completely dependent on high-energy density liquid fuels. These

liquid fuels are, in turn, almost completely derived from petroleum. Therefore, much human wealth and development depend on petroleum consumption. Within (or even without) the context of peak oil, obtaining renewable liquid fuels should be our most pressing renewable energy priority. The only renewable source of high-energy density liquid fuels is biomass or plant matter. By far the largest, and potentially the most sustainable source of renewable liquid fuels is cellulosic (nonfood) plant biomass. Thus, the conversion of cellulosic biomass to liquid fuels is a crucial priority for research, development, and deployment.

Conversion of Cellulosic Biomass to Liquid Fuels: Why Pretreatment is Critical There are two primary routes for making liquid fuels from cellulosic biomass. One route is the thermochemical platform, which, like the oil refining industry, primarily depends on heat and chemical catalysts to produce fuels. There is also the biochemical or sugar platform. As the name implies, the sugar platform depends on producing sugars from biomass that are then acted on by microbial and/or chemical catalysts to produce fuels. Hybrids of the thermochemical and biochemical approaches are possible and are being actively developed. This editorial primarily concerns the sugar platform, but much of it may also apply to the thermochemical platform. Our opinion is that the nature of the raw material (cellulosic biomass) primarily determines the appropriate processing platform. Without going into details, it seems that softwoods are inherently more suitable for thermochemical processing, whereas grasses and crop residues are inherently better suited for conversion by the sugar platform. Hardwoods might be processed using either approach. As far as the sugar platform is concerned, we note that the biological (enzymatic/microbial) deconstruction of plant biomass is already being practiced in nature, but at a rate and


Figure 2. The relative effects of process scale, yields, concentration, and reaction rate on overall raw material, operating, and capital costs of a process plant.

with sugar yields far below those required for commercial viability. A key requirement for the sugar platform is therefore some sort of preprocessing of the cellulosic biomass that will increase the rate and yield of sugars. This preprocessing step is called ‘‘pretreatment’’ and has been an active field of research for over 100 years. In that time, many different pretreatment approaches have been explored. The interest in pretreatment research has waxed and waned (mostly waned) over this time as interest in petroleum alternatives has waxed and waned (again, mostly waned). Until a few years ago, there were about a dozen researchers worldwide whose research focused primarily on biomass pretreatment. Since then, interest in pretreatment has exploded and many new researchers have entered the field. We must do a better job at performing and reporting pretreatment research. If we do not, progress toward a better, more sustainable liquid fuel system will be greatly hindered. This editorial is written to welcome newcomers to the field and to provide what we hope are useful guidelines for performing and reporting pretreatment research. We do this because, simply put, much current (and past) biomass pretreatment research as reported in the literature is useless. It cannot be used by industry to choose between competing pretreatment processes and then to develop commercial pretreatment technologies. This is because much pretreatment research does not report key fundamental data required by industry to make these choices. The remainder of this article primarily expounds on the key fundamental data that (we believe) pretreatment research should provide.

Key Data Needed from Pretreatment Research A thorough technoeconomic analysis requires certain data to determine the costs of processing raw materials to various products, the most relevant being those data that have the greatest impact on the process costs. Based on many previous analyses of chemical and refinery operations, we know that the primary cost centers of these refining facilities are: (1) the feedstock or raw material cost (typically 60–70% of the cost of production of fuels and commodity chemicals), (2) the capital equipment (upfront investment) costs, and (3)

895

operating costs, including utilities and chemicals/supplies consumed. Thus, we believe that the most important data that should be reported by pretreatment researchers relate to process yields, product concentrations, and the rate of conversion. The relative impacts of yield, concentration, and rate on the key commercial parameters of feedstock costs, capital costs, and operational costs are summarized in Figure 2. Specifically, we suggest that all pretreatment research report the following variables clearly in their papers and figures, which we will describe in greater detail in subsequent paragraphs: • Yield: kg sugars (both monomeric and oligomeric) per kg of dry, untreated biomass • Concentration: kg of sugars (both monomeric and oligomeric) per liter of the hydrolysate solution • Rate: kg of sugars (both monomeric and oligomeric) produced per liter of hydrolysate per hour • Hydrolysis catalyst loading: g enzyme protein (or acid) per kg of dry, untreated biomass (or glucan) Of all the pretreatment parameters that should be reported it is the yield of product sugars in terms of the untreated biomass input that is the most important. Unfortunately, yield is frequently unreported, unmeasured, or reported ambiguously. Feedstock costs are minimized by high yield (mass of product per mass of untreated cellulosic feedstock), but high yields also reduce capital costs per unit output (less equipment is required to process a given amount of feedstock to product) and operating costs (e.g., less stirring, heat, etc. required per unit product). In addition, sugar concentration (mass sugar per volume of pretreatment or hydrolysis liquor) is important because high sugar concentrations increase the concentrations of the resulting fuel products, thereby reducing utility costs for separation and the cost of reactors for processing the sugars to fuels. Therefore, concentration (or titer) of sugars following pretreatment and enzymatic hydrolysis is the second most important parameter for pretreatment research. We note that many pretreatments result in two or more streams leaving the pretreatment vessel. When this occurs, the concentrations of sugars in all streams should be reported. Finally, sugar production rate (mass sugar per unit reaction volume per unit time) is important because high reaction rates decrease the size (and thus the capital costs) of reaction systems. (The reaction time in the pretreatment system is generally much less than that required for enzymatic hydrolysis, but both rates should be reported). Unfortunately, based on our experience with many manuscripts we have reviewed, many new pretreatment researchers seem to believe that increased enzymatic reaction rate is all that is required of a potential pretreatment process. In fact, increased rate is only one relevant parameter, and probably not even the most important. We suggest that for enzymatic hydrolysis, the rate be reported over the initial, nearly linear portion of the enzymatic hydrolysis curve, approximately the first 24 h. Techniques are emerging to deal effectively with the much slower hydrolysis rates seen after this time,8 so we believe it is this high rate period that is most important. The catalyst loading during enzymatic hydrolysis (whether enzyme or acid) obviously affects the rate of reaction and yields. For clarity, we recommend that the catalyst loading be reported separately from the reaction rate and yield. Enzyme loading should be reported unambiguously as the mass of enzyme mixture applied per unit mass of original,

896


Figure 3. Example reporting format for important, yield, rate, and concentration data generated during pretreatment research.

untreated, dry biomass. Industrial designers of biofuel systems will typically have better access to enzyme cost information than will academic or government pretreatment researchers, so breaking out enzyme loading in this way will assist industry in evaluating pretreatment processes. A great deal of effort has been invested in improving hydrolytic enzymes and enzyme formulations. As noted above, most of this knowledge is in the hands of the enzyme companies and is generally not accessible for related pretreatment research. However, pretreatment researchers can evaluate their own new enzymes and enzyme formulations in the context of carefully defined biomass samples and pretreatment conditions and compare these with commercial enzymes under the same conditions. Aspiring enzyme discoverers and developers should recognize that a minimum of about 10 different enzyme activities is required to hydrolyze cellulosic biomass, particularly biomass that has been pretreated under conditions where the hemicellulosic sugars are not selectively removed (e.g., for alkaline pretreatments in contrast to acidic pretreatments). Because of this, new promising enzymes would need to be evaluated for their ability to replace the rate-limiting enzyme within a defined complex mixture. This is challenging to say the least.9 We suggest to our fellow researchers that a figure like Figure 3 be prepared to summarize these four key pieces of data (yield, concentration, rate, and enzyme loading) for work resulting from their own pretreatment research. As is typical of chemical engineering practice, Figure 3 uses a basis for reporting all other streams and calculations, in this case one (1.0) kg of completely dry (‘‘bone dry’’) untreated biomass feedstock. The editors of this journal have graciously agreed to allow Figure 3 to be reproduced without special permission, to make reporting these data as easy as possible.

Is that All? Other Criteria with Which to Compare Biomass Pretreatment Processes These four appear to be the most important pretreatment parameters and should be reported in all thorough papers.

However, many other parameters could be reported, or at least considered, as a given pretreatment develops from a laboratory technique into a potential commercial technology. Here are other criteria that may be important to collect and report as needed, with a brief explanation of the potential importance of each criterion. 1. Potential to increase biomass density and improve logistics and transport. It is increasingly recognized that the logistics of cellulosic biomass systems will be a key constraint on the large-scale deployment of these systems. The low density and limited stability/durability of much cellulosic biomass limit logistical systems. Therefore, pretreatments that can increase the density and durability of biomass, and that can potentially be practiced in distributed systems nearer to the field, will be highly attractive. 2. Potential for other products from pretreatment systems. The cellulosic biofuel industry faces a difficult ‘‘chicken and egg’’ situation. Farmers will be slow to grow or supply biomass in the absence of a functioning industry and the industry will likewise be slow to emerge without reliable, large-scale feedstocks. Other products, with established markets, resulting from pretreatment processes would help catalyze the emergence of cellulosic biofuel systems. Animal feeds, fertilizers, and soil amendments might be particularly attractive coproducts in agricultural settings. 3. Fermentation compatibility without washing or extensive processing. Some pretreatments produce toxins that inhibit downstream fermentation. Also, the pretreatment chemicals themselves are frequently toxic and must be removed very completely. The presence of such toxins may require high usage of water and/or utilities, with additional expense. High water use may be especially problematic, given emerging world concerns about adequate freshwater supplies in the face of rising populations. It is a simple matter to include water consumption by pretreatment and washing per kg of feedstock biomass in Figure 3, and in fact we have done so.10 Many researchers are likely to be shocked by the implications of the water used in their laboratory process when translated into a potential large-scale technology.


897

Figure 4. Comparison of important criteria for 20 different pretreatment technologies.

4. Expected scalability of the pretreatment process. Some processes can simply be scaled more easily than others based on materials handling, equipment issues, and other factors. 5. Energy required. The purpose of biomass pretreatment is to produce an energy (liquid fuel) product. Thus, if the pretreatment process requires an excessive amount of energy, it is probably not viable. It seems most useful to report the energy requirement of the pretreatment process versus the lower heating value of the biomass feedstock itself. Electricity consumption should be multiplied by three because it takes roughly three units of primary energy (coal or natural gas) to produce one unit of electrical energy. 6. Capital equipment costs. As noted above, capital costs are strongly influenced by yield, concentration, and rate. However, some pretreatment processes may require unusually expensive materials or exotic alloys, further increasing the relative importance of capital costs. 7. Water use. As noted above, freshwater supplies are under increasing pressure worldwide. Our colleague Professor YJ Yuan of Tianjin University (PRC) told us that China has simply removed a number of potential pretreatments from consideration ‘‘because we don’t have the water.’’ 8. Chemical costs. Some pretreatments, such as ionic liquid-based pretreatments, require expensive chemicals and/or large amounts of chemicals that must be recovered efficiently and effectively. 9. Process control. Some pretreatments require very careful control of process operating conditions to achieve good results. For example, dilute acid treatments carried out in nominally plug flow reactors can suffer poor yields if the residence time distribution is not tightly controlled, and this can be difficult to achieve in practice. 10. Is lignin preserved? Although the primary objective of pretreatment is to obtain high sugar yields from the struc-

tural carbohydrates (cellulose and hemicellulose), it will be important to preserve as much of the lignin as possible for practical use. At a minimum, the fuel value of the lignin should be preserved. 11. Extremes of pressure and temperature. Pretreatments that use extremes of temperature and/or pressure are more expensive than ones that operate at low or moderate temperatures and pressures. 12. Toxic and/or hazardous chemicals. As with extremes of temperature and/or pressure, use of toxic and/or hazardous chemicals in the pretreatment will increase costs. Hydrofluoric acid is one example of such a hazardous pretreatment chemical. 13. Wastes produced. If the pretreatment generates waste streams, these streams must be handled and disposed of, incurring further costs. 14. Effectiveness of pretreatment to all feedstocks. A few pretreatments work well on all feedstocks, but others are more limited in their range of feedstocks. For example, many pretreatments do not deal well with softwoods, whereas grasses and crop residues are more susceptible to a variety of pretreatments. To illustrate this broader range of pretreatment evaluation criteria, we have prepared Figure 4 to visually summarize the current state of 20 different pretreatments. There are many more pretreatments that could eventually be included in this figure. We do not give detailed references for each pretreatment and each criterion but instead have drawn on our experience and general knowledge of the field to prepare this table. Each criterion is assigned a color code based on its relation to a defined numerical metric or whether it is able or not to satisfy the given criterion (yes/no/maybe or unknown). In general, the color green means that the criterion is satisfied, and the color red means that the criterion is not met. A yellow marker in the criterion space means that

898


either the criterion is marginally satisfied or that insufficient information exists to make an assignment at this time. It is remarkable how much yellow appears in this table, which is mostly due to insufficient information available to make well-founded assessments. We suggest that pretreatment researchers, in addition to determining yield, concentration, and rate as needed, focus their attention on filling in this missing information. In doing so, we will help to advance the field of pretreatment research and make largescale, low-cost cellulosic biofuels a commercial reality.

Summary Human wellbeing depends on access to large amounts of energy. Fossil energy carriers, especially petroleum, the dominant source of liquid fuels for mobility, will become increasingly scarce, expensive, and volatile in price. Low cost and sustainable liquid biofuels are most likely to come from cellulosic biomass. In turn, cellulosic biomass requires effective, economical pretreatment to release sugars. The editorial suggests ways in which those involved in pretreatment research can help our world transition to a more sustainable energy future by performing more useful, more broad-based pretreatment studies.

Acknowledgments This work was funded by the DOE Great Lakes Bioenergy Research Center (DOE Office of Science BER DE-FC0207ER64494).

Literature Cited 1. Martı´nez DM, Ebenhack BW. Understanding the role of energy consumption in human development through the use of saturation phenomena. Energy Policy 2008;36:1430–1435. 2. United Nations, Human Development Report Office. Human Development Index (HDI). Available at: http://hdr.undp.org/en/ statistics/hdi/ accessed on November 17, 2011. 3. International Energy Outlook 2011; U.S. Energy Information Administration: 19 September 2011. 4. Key World Energy Statistics 2011; International Energy Agency: 2011. 5. U.S. Energy Information Administration. International Energy Statistics. Available at: http://www.eia.gov/cfapps/ipdbproject/ IEDIndex3.cfm accessed on November 17, 2011. 6. BP Statistical Review of World Energy June 2011; BP: 2011. 7. Annual Energy Review 2011; U.S. Energy Information Administration: 19 October 2011. 8. Jin M, Gunawan C, Uppugundla N, Balan V, Dale BE. A novel integrated biological process for cellulosic ethanol production featuring high ethanol productivity, enzyme recycling and yeast cells reuse. Energy Environ. Sci. 2012;5:7168–7175. 9. Banerjee G, Car S, Scott-Craig JS, Borrusch MS, Bongers M, Walton JD. Synthetic multi-component enzyme mixtures for deconstruction of lignocellulosic biomass. Bioresour Technol. 2010;101:9097–9105. 10. Garlock RJ, Balan V, Dale BE, Ramesh Pallapolu V, Lee YY, Kim Y, Mosier NS, Ladisch MR, Holtzapple MT, Falls M, Sierra-Ramirez R, Shi J, Ebrik MA, Redmond T, Yang B, Wyman CE, Donohoe BS, Vinzant TB, Elander RT, Hames B, Thomas S, Warner RE. Comparative material balances around pretreatment technologies for the conversion of switchgrass to soluble sugars. Bioresour Technol. 2011;102: 11063–11071. Manuscript received May. 16, 2012, and revision received Jun. 12, 2012.