Mats Galbe1 • Per Sassner1 • Anders Wingren2 • Guido Zacchi1 (И)

department of Chemical Engineering, Lund University, P. O. Box 124, 221 00 Lund, Sweden

Guido. Zacchi@chemeng. lth. se

2SEKAB E-Technology, P. O. Box 286, 891 26 Ornskoldsvik, Sweden

1 Introduction……………………………………………………………………………………………… 304

2 Flowsheeting……………………………………………………………………………………………… 309

2.1 Simulation of Ethanol Production from Lignocellulosic Materials…………………. 310

3 Process Economics…………………………………………………………………………………….. 311

3.1 Effect of Various Parameters on the Energy Demand and Production Cost 318

3.2 Lignocellulose versus Starch—a Comparison……………………………….. 322

3.3 Co-location with other Plants…………………………………………………………………….. 325

4 Conclusions………………………………………………………………………………………………. 325

References……………………………………………………………………………………………………. 326

Abstract This work presents a review of studies on the process economics of ethanol production from lignocellulosic materials published since 1996. Our objective was to identify the most costly process steps and the impact of various parameters on the fi­nal production cost, e. g. plant capacity, raw material cost, and overall product yield, as well as process configuration. The variation in estimated ethanol production cost is considerable, ranging from about 0.13 to 0.81 US$ per liter ethanol. This can be ex­plained to a large extent by actual process differences and variations in the assumptions underlying the techno-economic evaluations. The most important parameters for the economic outcome are the feedstock cost, which varied between 30 and 90 US$ per metric ton in the papers studied, and the plant capacity, which influences the capi­tal cost. To reduce the ethanol production cost it is necessary to reach high ethanol yields, as well as a high ethanol concentration during fermentation, to be able to de­crease the energy required for distillation and other downstream process steps. Improved pretreatment methods, enhanced enzymatic hydrolysis with cheaper and more effective enzymes, as well as improved fermentation systems present major research challenges if we are to make lignocellulose-based ethanol production competitive with sugar — and starch-based ethanol. Process integration, either internally or externally with other types of plants, e. g. heat and power plants, also offers a way of reducing the final ethanol production cost.

Keywords Bioethanol production • Biomass • Flowsheeting • Process economics

1

Introduction

There is no single process design offering the most cost-efficient way to pro­duce ethanol from biomass. Many factors that affect the desired product have to be taken into consideration. Regarding ethanol production, some of the most important parameters are the capital cost of the plant, the type and cost of raw material, the utilization efficiency of the materials involved in the pro­cess and the energy demand. The design of the plant, as well as its individual process steps, must be based on accurate and reliable data. These comprise both physical and chemical data and cost estimation data. It is naturally best to use data gathered from the same or a similar type of plant as the intended one. Most of the data required are available, or can be adapted and used for a new plant design. This is not the situation when lignocellulosic materials are considered as feedstock for ethanol production.

Ethanol has traditionally been produced from sugar cane and sugar beet juice [1] or from various starch-containing materials, e. g. corn or wheat [2-4]. Figure 1 shows a simplified flowsheet of an ethanol produc­tion process based on starch-containing materials. Liquefaction of the starch fraction is accomplished by adding hydrolytic enzymes (a-amylases) at tem­peratures of around 90 ° C. After the liquefaction step the starch molecules are further hydrolyzed by the addition of glucoamylases. This produces sug­ars, which are readily fermented by yeast, e. g. Saccharomyces cerevisiae, to ethanol. The main co-product is usually animal feed, consisting of the re­maining fraction of the raw material, mainly proteins and fiber, which is sometimes referred to as DDGS—distillers dried grains with solubles [5]. There is considerable experience in starch-based ethanol production, and the technology can be considered mature. The design and cost estimates of new plants are, therefore, rather accurate.

The availability of agricultural land for non-food crops and the limited market for animal feed places a limit on the amount of ethanol that can be produced from starch-based materials in a cost competitive way [6]. Ethanol production from lignocellulosic raw materials, on the other hand, reduces the potential conflict between land use for food (and feed) production and energy feedstock production. The raw material is less expensive than conventional agricultural feedstock and can be produced with lower inputs of fertilizers, pesticides, and energy. Lignocellulosic materials contain about 50-60% car­bohydrates in the form of cellulose (made up of glucose) and hemicellulose (consisting of various pentose and hexose sugars), which may be fermented to ethanol, and 20-35% lignin. The latter is the main co-product, which could be used for the production of heat and electricity or, in the longer perspective, for the production of specialty chemicals. There is thus no co-product limita­tion on the use of lignocellulosic materials for ethanol production. The only limitation is the availability of the raw material and, of course, the production

cost. During recent years, there has been a considerable increase in interest in research on and the development of the conversion to ethanol of lignocellu — losic materials, such as agricultural and forest residues, as well as dedicated energy crops.

However, in contrast to starch-containing materials, cellulose-containing raw materials, such as forest residues and straw, have not yet been commer­cialized in the ethanol industry. The reasons for this are several. For instance, there are physical barriers such as:

• the complex structure of lignocellulosic materials, making them recalci­trant to hydrolysis;

• the presence of various hexose and pentose sugars in hemicellulose, mak­ing fermentation more difficult; and thirdly,

• the presence of various compounds that inhibit the fermenting organism. These compounds either originate from the raw material itself, e. g. ex­tractives, or are formed during the early process steps, e. g. degradation products of sugars and lignin. This makes it difficult to reach high ethanol concentrations during fermentation, which in turn results in a high energy demand and thus high production cost.

There is a big risk involved in being the first to invest in commercialization of a lignocellulose to ethanol plant and this is the main reason why there is no full-size plant in operation today.

Interest in lignocellulose-based ethanol production has recently brought about action on high political levels. For example, in the USA, the Energy Pol­icy Act of 2005 requires blending of 7.5 billion gallons (^ 28.4 million m3) of alternative fuels by 2012 [7] and recently, in his State of the Union Address (Jan 31, 2006), the US President announced the goal of replacing more than 75% of imported oil with alternative fuels by the year 2025 [8]. The major part of this alternative fuel will probably consist of ethanol, and to be able to meet these demands this will have to be largely produced from lignocellulosic materials. In Europe the European Commission plans to progressively replace 20% of conventional fossil fuels with alternative fuels in the transport sector by 2020, with an intermediate goal of 5.75% in 2010 [9]. Bioethanol is also expected to be one of the main means of achieving this goal.

Experience in the production of ethanol from lignocellulosic materials is limited, at least using modern technology. Full-scale plants have only been run occasionally during times of war. Examples are the Bergius process (con­centrated HCl) operated in Germany during World War II, and the Scholler process (dilute H2SO4), which was used in the former Soviet Union, Japan and Brazil [10]. Thus, design and cost estimation for lignocellulosic-based processes cannot be based on reliable operational experience, but data gath­ered on lab scale, or at best on pilot scale, must be used. It is true that some of the process steps are of the same type as in a starch-based process, but there are several major differences. For example, the by-products from the various processes are not the same. Some of these are considered valuable co­products, which will contribute to the profit from the process, while others are waste materials that must be dealt with in wastewater treatment plants, or disposed of by other means.

During the past 20 years or so, a great deal of effort has been devoted to research on various areas, such as the pretreatment of raw material, enzy­matic hydrolysis of cellulose, including the production of more cost-effective enzymes, and the development of new microorganisms and fermentation techniques to ferment all the sugars available in lignocellulosic materials. An enormous amount of data has been generated (see the work by Galbe, Vikarii, Cherry, Hahn Hagerdal, and Ingram, all in this volume), which today forms the basis for techno-economic calculations. However, although the re­sults may be accurate, there is still a huge scale-up problem involved in going from batch pretreatment reactors on the liter scale, to continuous reactors of several cubic meters, and from 1- to 100-liter fermentors to vessels with a vol­ume of 1000 cubic meters or more. Issues such as material corrosion, rapid heat evolution, excessive foaming, and precipitation of solids and incrusta­tion, which may not even be considered on the lab scale may become serious problems in a full-scale process.

Pilot-scale trials have been run in several places during the past decade. The National Renewable Energy Laboratory (NREL) (Golden, Colorado, USA) has constructed a pilot fermentation facility to test bioprocessing technolo­gies for the production of ethanol and other fuels or chemicals from cellulosic biomass [11]. The Process Development Unit (PDU) of the Bioethanol Pilot Plant was set up to investigate biomass fuel and chemical production pro­cesses from start to finish on a scale of about 900 kg day-1 of dry feedstock. The plant is, however, not a fully integrated unit that can run continuously.

A 1000 kg day-1 plant, using spruce as the raw material, has been in op­eration in Ornskoldsvik in Sweden since the middle of 2004 [12]. Abengoa Bioenergy Corp. has constructed a pilot plant in York, Nebraska, USA [13] and is now constructing a demonstration scale plant in Salamanca, Spain, with an annual production capacity of 5000 m3 ethanol. This will be brought into operation at the beginning of 2007 [14]. This demonstration plant, which will be co-located with a 195 000 m3 y-1 starch-based plant, will utilize the straw from wheat and thus contribute to the overall production capacity. Furthermore, Iogen Corp. is operating a pre-commercial demonstration fa­cility, located in Ottawa, Canada, where ethanol is made from agricultural residues [15]. The plant is able to handle up to 40 metric tons of feedstock daily, consisting of wheat, oat, and barley straw, and is designed to produce up to 3 million liters of ethanol annually.

Data from these types of plants will increase the reliability of cost estimates significantly. They can also be used to identify process problems associated with continuous processing, such as the accumulation of toxic substances in various process streams, and fouling of heat exchanger surfaces. However, in most cases this will be proprietary information not available in the scientific literature.

Two process concepts have been investigated more than others regarding ethanol production from lignocellulosic materials. The main difference be­tween the two is the way in which the cellulose chain is broken apart; either dilute sulfuric acid or cellulolytic enzymes are used to hydrolyze the cellulose molecules. Figure 2 shows the main features of a dilute acid hydrolysis pro­cess. The raw material is treated with 0.1-3% (w/w) H2SO4 at temperatures normally ranging from 160 to 200 °C. It may be advantageous to perform dilute-acid hydrolysis in two steps since the hemicellulose fraction is more easily degraded than is the cellulose fraction. A disadvantage of the dilute acid process is the somewhat low ethanol yield and the necessity of using ex­pensive construction materials that are resistant to corrosion by acid at high temperatures. The acid must also be neutralized, which leads to the forma­tion of large amounts of gypsum, CaSO4, or other compounds that have to be disposed of.

An alternative to acid hydrolysis is enzymatic hydrolysis (Fig. 3). Cellu­lolytic enzymes are produced by microorganisms and have the ability to cleave off short sugar units from the cellulose chain, as described in detail

by Vikarii 2007 and Merino 2007 (this volume). The enzymatic process is op­erated at much milder conditions than the dilute acid process, which is of great importance for several reasons. The yield can be expected to be higher, the construction materials will be less costly and the formation of toxic by­products will also be reduced. However, the enzyme action suffers from being slow if the raw material is not pretreated prior to enzymatic hydrolysis. Pre­treatment can be performed in a number a ways. Depending on the type of raw material (hardwood, softwood or agricultural residue) a certain pre-

treatment method can be more or less successful. Pretreatment is described in more detail by Galbe 2007 (this volume). Fermentation can be performed either in a separate fermentor tank, a process configuration normally re­ferred to as separate hydrolysis and fermentation (SHF), or simultaneously with the hydrolysis of the cellulose chains, so-called simultaneous sacchar­ification and fermentation (SSF). If the pentose sugars are also fermented, the process is sometimes referred to as simultaneous saccharification and co­fermentation (SSCF). The downstream processing section is similar for the dilute acid hydrolysis and the enzymatic processes, or at least includes the same process steps (Figs. 2 and 3).

Simulation of processes with the aid of flowsheeting programs is an in­valuable tool in studying how changes in process design affect the overall performance of a plant. Plants operating 24/7 cannot be experimented on, since the profit loss may be considerable if an ill-planned test causes standstill for a day or two. By performing “experiments” on a plant using computers the outcome of a design change can be evaluated beforehand, which will make a change in the process less risky.

This work will focus on the process economic aspects of ethanol pro­duction from lignocellulosic materials and provide targets for where process improvements should be investigated. The enzymatic process will be consid­ered in detail, as most research over the years has been concentrated on this type of process. However, as mentioned earlier, the process suffers from the fact that process data from large production plants are very scarce. Neverthe­less, the data gathered so far on lab and bench scales can be used as input data in flowsheeting programs for comparison of various process alternatives and to help identify bottlenecks in a process. A summary of various pub­lished reports and papers will be made. Unfortunately, this is an area that has clearly been neglected by many researchers, since the number of publications is small.

2