Comparison of pretreatments for ethanol production from softwood

Transcription

1 DEPARTMENT OF CHEMICAL ENGINEERING LUND UNIVERSITY, LTH Comparison of pretreatments for ethanol production from softwood Comparison between the processes of steam explosion, organosolv, sulphite and ionic liquid pretreatment Josefine Hagman, Lisa Hedborn, Malin Isgren, Emelie Larsson, Patrick Mårtensson Tutors: Guido Zacchi, Chem. Eng., Stefano Macrelli, Chem. Eng., Ola Wallberg, Chem. Eng., Børre T. Børresen, Statoil

2 Acknowledgement We would like to express our sincere gratitude to the whole Department of Chemical Engineering at Lund University, LTH, which has provided great expertise during our work with the project. In particular for all support from Professor Guido Zacchi who has helped us through many complications during our work, always in a positive manner. We also want to specially thank Stefano Macrelli who has spent countless hours teaching and helped us with Aspen Plus. We would also like to thank Statoil for providing us with such an interesting project. Besides our tutors, we also want to thank Stig Stenström, Mats Galbe, Ann-Sofie Jönsson and Hans T. Karlsson.

3 Abstract The purpose with this project was to compare different pretreatment methods for ethanol production from softwood. The examined pretreatment methods are; steam explosion, organosolv, sulphite and ionic liquid pretreatment. Each of the included pretreatment methods and subsequent processes have been set up and simulated in Aspen Plus, except the ionic liquid pretreatment due to lack of data. From the Aspen simulations results for material and energy balances were obtained. The economy of the processes has been examined with respect to operational and investment costs to determine the minimum ethanol selling price (MESP). A sensitivity analysis was also performed. In a society with a growing demand for green energy the need to find renewable sources is vital. Lignocellulosic materials are a good alternative since they have rich abundance and it does not compete with food production, which is the case for ethanol production from e.g. maize and sugar canes. The main challenge with ethanol production from lignocellulosic material is that the cellulose is protected by lignin, and hence pretreatment methods are vital. The objective of the calculations was to produce m 3 ethanol annually. For this production volume, the steam explosion process needs 41.1 ton dry matter/h, the organosolv process needs 67.7 ton dry matter/h and the sulphite pretreatment process needs 51.9 ton dry matter/h. Steam explosion is the most favorable pretreatment method with respect to energy efficiency (which corresponds to smallest losses) and with respect to the energy amount of the raw material present in the ethanol. In the matter of district heating, the sulphite process was found to be most favorable, since this process had the smallest distribution of the energy input as district heating since this is a season dependent income. A significant amount of the energy in the raw material is present as lignin and lignosulfonates in the organosolv and the sulphite pretreatment processes. During pretreatment, inhibitors are formed to different extent. Lots of levulinic and formic acid was present in the organosolv process. For the steam explosion and sulphite pretreatment processes only hydroxymethyl furfural (HMF) and furfural were included in the calculations due to the fact that conversions for other inhibitors were not presented in the reference articles used for the Aspen Plus calculations. The sulphite pretreatment process resulted in the lowest amount of inhibitors. Regarding the lowest ethanol selling price it was found that steam explosion is the best pretreatment method with a MESP of 5.2 SEK/L compared to sulphite and organosolv pretreatment with a price of 9.0 SEK/L and 13.5 SEK/L. These MESPs were calculated using the lowest lignin price, corresponding to the energy content of lignin. As long as there is no market for the lignin, steam explosion is the best pretreatment method concerning the MESP. The lignin price needed to reach the same MESP as for steam explosion was calculated. Better conversions are also needed for organosolv and sulphite pretreatment to compete with steam explosion. In the sensitivity analysis it was found that basically all three investigated processes had the same sensitivity for the three investigated parameters (costs of raw material, enzymes, and capital costs). The steam explosion process resulted in the lowest ethanol price for all cases. The ionic liquid pretreatment process cannot yet be economically beneficial mainly because of the high purchasing price of ionic liquids.

4 Table of contents Acknowledgement... 1 Abstract Introduction Task description Background Composition of biomass The process Pretreatment Hydrolysis Fermentation Product recovery Economy Pretreatment methods Steam explosion Catalyst process Organolsolv Pulping based processes Ionic liquids Process descriptions General process description Feed Reactions in pretreatment SSF reactor Distillation Dehydration Biogas production CHP Process description - steam explosion process Raw material Pretreatment reactor Simultaneous Saccharification and Fermentation Distillation... 33

5 2.3 Process description - organosolv process Raw material Pretreatment reactor Regeneration of solvent Washing Simultaneous Saccharification and Fermentation Distillation Biogas CHP Process description - sulphite pretreatment process Raw material Pretreatment reactor Separation of hydrolysate, lignosulfonates and solid fraction Simultaneous Saccharification and Fermentation Distillation Process Economics General assumptions SSF reactor Distillation columns Heat exchangers Biogas plant CHP plant Results Material and energy balances Inhibitors Energy evaluation of the processes Steam explosion Organosolv Sulphite pretreatment Internal energy consumption Process economics Steam explosion process Organosolv Sulphite pretreatment... 53

6 4.5 Sensitivity analysis Sensitivity of price of lignin Ionic liquids The process Pre-pretreatment Pretreatment Separation of intermediate and contaminated ionic liquid Washing of intermediate Recycling and purification of the ionic liquid Simultaneous Saccharification and Fermentation Distillation Process equipment Estimation of ethanol price Summary Discussion Material and energy balances Energy evaluation of the processes Economy Sensitivity analysis: feed-, capital- and enzyme cost Sensitivity analysis: lignin price Conclusions References Appendix A, Material balance calculations A.1 Steam explosion A.2 Organosolv A.2.1 Feed to the CHP A.3 Sulphite pretreatment Appendix B, Aspen Plus calculations B.1 Steam explosion B.1.1 Conversions B.1.2 SSF B.2 Organosolv B.2.1 Conversions B.2.2 SSF... 84

7 B.3 Sulphite pretreatment B.3.1 Conversion in the pretreatment reactor B.3.2 SSF Appendix C, Energy evaluation calculations C.1 Steam explosion C.2 Organosolv C.3 Sulphite pretreatment Appendix D, Capital cost calculations D.1 Steam explosion D.2 Organosolv D.3 Sulphite pretreatment Appendix E, Total plant cost Appendix F, Variable and restricted cost Appendix G, Calculations for the Ionic Liquid pretreatment process G.1 Assumptions G.2 Calculations for theoretical ethanol per kg dry biomass G.3 Pre-pretreatment G.3.1 Disk milling G.3.2 Drying G.4 Pretreatment G.5 SSF G.6 Rough estimation of process streams

8 1. Introduction In this section a task, background and process descriptions are presented. 1.1 Task description The purpose of the project is to compare and evaluate four different pretreatment methods for producing ethanol from lignocellulosic raw materials. The different pretreatment methods are; steam explosion, organosolv, pulping based processes (sulphite pulping) and ionic liquids. The comparison is with respect to the different technologies and the aspect of economy, taking the whole process into account. Aspect issues like energy and mass balances, investment and operating costs and sensitivity analysis as well as main advantages and disadvantages should be considered. Furthermore, formation of inhibitors and the value/ quality of the co-products and the time for a possible introduction in industry will also be included. 1.2 Background One of today s greatest challenges is to find replacement of oil, since it is only going to last for a few more decades and furthermore for the reason that global warming today is a fact. The transport sector accounts for the largest consumption of oil. In Sweden it stands for 60%. New fuels are needed for us being able to keep our standard of living, and bio-ethanol can be one of the solutions. Bio-ethanol has many advantages; firstly it is produced from renewable feed stocks and it is carbon dioxide neutral. The vehicle exhaust is cleaner with less NOx and particles compared to petrol. Furthermore, it is easy fitted into our existing infrastructure and creates both regional balance and employments. On the other hand, the greatest disadvantage with bio-ethanol is that it is more expensive to produce and it has lower energy content per volume compared to petrol. Another disadvantage is its inflammability. Lots of different renewable feed stocks can be used including starchy materials, sugar containing agricultural products and cellulosic raw materials. There exist plants producing bio-ethanol today, for example from maize in the USA and sugar cane in Brazil. But it has been discussed if it is ethically correct to use food to produce bio-ethanol [1]. Ethanol from lignocellulosic raw materials has not yet been commercialized. The existence of physical barriers, e.g. complex structure, presence of various kinds of hexoses and pentoses, and compounds inhibiting the fermentation makes the process more complex compared with ethanol production from starchy feed stocks. Basically those factors contribute to difficulties achieving high overall ethanol yield and also high ethanol concentrations during fermentation which leads to high process costs. Furthermore, there is a high risk being the first one to invest in commercializing ethanol production from lignocellulosic material, which is one of the main reason why no industrial plants yet exist today [4]. Even though there does not exist any plant using cellulosic raw materials yet, ethanol produced from this feed stock is up and coming and the first commercialized plant is planned to be commissioned in Italy in The capacity of the plant is going to be about m 3 of ethanol per year [6]. Naturally biomass is designed to withstand attacks from the surroundings, e.g. from microbes. This means that hydrolysis is one of the greatest challenges. In spite of its complexity, lignocellulose is a good feed stock due to its low price and the fact that it doesn t compete with the use alternative being food. 1

9 The main steps in the process from lignocellulosic material to ethanol are basically pretreatment, hydrolysis, fermentation and product upgrading. Pretreatment is needed because the biomass is protected by lignin. The pretreatment opens up the structure and prepares the cellulose for enzymatic hydrolysis. In the hydrolysis the cellulose is depolymerized into its monomers. This is followed by fermentation where ethanol is produced from sugars by yeast. To obtain the concentrations needed to use ethanol as a fuel, upgrading by distillation and dehydration is needed. Lignocellulosic material contains cellulose, hemicelluloses and lignin. This means that not only C6 sugars (e.g. glucose, mannose) but also C5 sugars (e.g. xylose, arabinose) are present after the hydrolysis. This causes problems because normal baker s yeast is not capable of fermenting C5 sugars, which has led to development of genetically modified yeast which ferments C5 sugars. C5 fermentation has not yet been industrially introduced. Alternatively the C5 sugars can be separated and used as feed in an anaerobic digestion process and hence produce biogas. It is important that the microorganisms used in fermentation are robust and can withstand potential inhibitory compounds formed in the pretreatment step [1] Composition of biomass Lignocellulose is a mix of cellulose, hemicelluloses and lignin. Lignin acts as a protecting wall around the hemicelluloses and celluloses. All three compounds are complex polymers whose interactions are not fully understood [7]. The composition of softwood and hardwood is presented in table 1. Hardwood contains more hemicelluloses, hence more C5-sugars, and less lignin than softwood [1]. Another difference is that hemicellulose is acetylated to a larger extent in hardwood than in softwood [10]. Table 1: Composition of hardwood and softwood [2] Softwood (%) Hardwood (%) Cellulose Hemicellulose Lignin Extractives Cellulose is an unbranched polymer of glucose coupled together with β-1,4-linkages and a Degree of Polymerization (DP) varying between Cellulose has both crystalline and amorphous regions depending on the extent of the hydrogen bonding. The crystalline structure is very stable against enzyme- or acid attack. This glucose is the main source for ethanol production. Hemicellulose is a mixture of linear and branched polymers containing C-5 sugars; xylose, arabinose and C-6 sugars; glucose, mannose, galactose and also acetyl groups Lignin confers protection, stiffness and water resistance to the fiber. Lignin does not contain any carbohydrates but phenolic-propane units arranged in a complex three dimensional matrix [2]. There are two different classes of lignin; guaiacyl lignins and guaiacyl-syringyl lignins. The difference is the number of methoxy-groups, since guaiacyl-syringyl lignins both have methoxy groups at carbon 3 and 5, whereas guaiacyl lignins only have one methoxy group at carbon 3 [10]. 2

10 1.3 The process The overall process contains four main steps; pretreatment, hydrolysis, fermentation and product recovery. Those steps are described in further detail in this section below, and an overview of the process is presented in figure 1. Enzymes and yeast Ethanol BIOMASS Pretreatment SSF, hydrolysis and fermentation Distillation Lignin and residuals Figure 1: Process overview Pretreatment Since the feedstock is protected by lignin, pretreatment is needed to open up the structure and expose the cellulose. Today lots of different possible pretreatment methods exist. Overall generalized classifications of types of pretreatments are physical, biological and chemical. Combinations of those, e.g. physical treatment (high pressure/ temperature) followed by a chemical treatment are often more effective. The goal of pretreatment is to prepare the feedstock for enzymatic hydrolysis resulting in an increase in the sugar conversion. The composition of different kind of biomass varies. The digestibility of a given feed stock depends on properties like lignin content, the accessibility of cellulose and its crystallinity. Other important factors that determine the digestibility are the degree of polymerization of cellulose, porosity (available surface area), hemicelluloses covering cellulose and fibre strength. The ideal pretreatment results in a disrupted biomass structure, ready for hydrolysis but with no formation of sugar degradation products or compounds that inhibit the fermentation [11]. If enzymatic hydrolysis is performed without any pretreatment, only about 20% of the available sugars are hydrolysed, but with pretreatment 90% of the available sugar can be obtained and made available for fermentation [17]. Physical pretreatment One type of physical pretreatment is size reduction. The purpose of size reduction is to increase the digestibility of lignocellulosic biomass and also decrease the mass and heat transfer resistance. Methods used for mechanical size reduction are shredding, chipping (results in sizes in the range of mm), milling and grinding (results in sizes in the range of mm). They all result in higher specific area, lower DP and crystallinity. There is no need to reduce the particle size below 0.40 mm since this just result in a small effect on the rates and yields during hydrolysis. Milling of the biomass increases the ethanol yield, but the energy demand of milling is high on industrial scale. However it might be possible to perform the milling after the pretreatment. This would decrease the energy needed for milling and e.g. costs of solid liquid separation. Another 3

11 suggested method is physical pretreatment by gamma rays. The gamma rays cleave bondages, decrease the crystallinity and increase the surface area. This option is very likely too expensive on an industrial scale [11]. Biological pretreatment Pretreatment can be biological; i.e. microbes can be used to remove the lignin from the biomass. The main advantage is the low energy consumption, and the main disadvantage is that long time is needed for the processing [7]. One of the most effective and common lignin degrading microorganisms is a fungus named white rot fungi. The ligninase producing white rot fungi is placed, together with lignocellulosic material, in a container to delignify the material [8] and is subsequently put into a process for ethanol production. The use of the white rot fungi has some great advantages compared to the most effective pretreatment methods used today, but also some major drawbacks. Important advantages are as follows; generically the technique is uncomplicated, the energy input is small as are the waste streams. The reduced amount of inhibitors is also a positive feature. All these advantages together contribute to a low process cost. The major drawback is the residence time needed for storing (where the delignification occurs). The time needed depends on how much lignin one wishes to remove; the time of treatment is about days. Furthermore the degree of delignification is not effective compared to other processes used today. Those two disadvantages are major drawbacks and make this kind of pretreatment unfeasible to apply in industry. But, due to the low production cost one solution could be that civil persons with lots of space could invest in this pretreatment method, and then sell the pretreated lignocellulosic biomass to a biorefinery. Another possibility can be that companies rent storage where the fungi-pretreatment can occur combined with another pretreatment method. Also it can be combined before another pretreatment method, which could then be performed at milder conditions [9]. Chemical pretreatment There are a number of different chemical pretreatment methods for lignocellulosic material that have been suggested. They are shortly described as follows: Steam explosion (SE): treatment with high pressure saturated steam followed by a fast decrease in pressure. Steam explosion can be performed with or without catalyst and the catalyst can either be an acid or base. Hence it is both a physical and chemical method. Furthermore an auto-hydrolysis is performed by the acetic acid from the hemicelluloses. One of the drawbacks with steam explosion is the formation of inhibitors. Liquid hot water (LHW): similar to steam explosion, but saturated water is used instead of steam. The main operating cost for both steam explosion and liquid hot water is the energy needed to feed the water/ steam. Addition of acid in SE and LHW increases the dissolution rate. One disadvantage with adding an acid is the risk of corrosion. Organosolv: involves an organic solvent used to simultaneously delignify and pretreat the feed stock. A mixture of solvent and acid breaks the bonds between the hemicelluloses and lignin. Common solvents are for instance ethanol, methanol, acetone or glycols which are all mixed with water. It is 4

12 important to remove the solvent since it can act as an inhibitor in the following process steps. The recovery of the solvent is an important aspect regarding the economics; the main costs are due to solvent recovery [7]. Pulping based: can be divided into three types; sulphite, sulphate and soda pulping. The difference is the chemical used for cooking. The wood is delignified and the cellulose is digested to different extent, depending on the method used, and the conditions do not need to be as severe as when making pulp. The by-product lignin has different properties depending on the method, where lignosulfonates from the sulphite process has a market today [33]. Ionic liquids (ILs): includes salts that melt below 100 C. The reason for its moderate melting temperature (compared with inorganic salts like sodium chloride, which melts at 803 C) is due to their asymmetric organic-based nature with delocalized charges. ILs have almost negligible vapor pressure and uniquely high solvent capacities. Their ability of solving cellulose is due to its interactions with hydroxyl groups which lead to disruption of hydrogen bonds. They can dissolve large amount of cellulose during mild conditions. The major benefit with ILs is that cellulose, hemicellulose and lignin can be fractionated into separate streams and that no inhibitors are formed during this process. The major drawback is the costs of the ILs, and hence the recovery is vital [16]. Alkaline extraction: disrupts and breaks the bond between the lignin and the carbohydrates. Studies have been made using sodium hydroxide and calcium hydroxide (lime) solutions. The residence time depends on the temperature. In case high temperature (and hence short residence time) is to be used, energy consumption is an important factor in the economic analysis. The cost of lime and the lime recovery are also important factors. Ammonia fiber explosion (AFEX): the AFEX pretreatment is quite similar to steam explosion and LHW in that the media is first applied to the feed stock at high pressure followed by a pressure reduction. The main difference is that in the AFEX method concentrated liquid ammonia is used as medium. One disadvantage with this method is that it is not effective if the lignin content in the biomass is high, e.g. for wood. It is important with efficient ammonia recovery in order to have a good economic feasibility [7]. One important drawback is that ammonia is toxic. Carbon dioxide explosion: supercritical carbon dioxide, SC-CO 2, can be used to process lignocellulosic material as pretreatment. It is an environmental friendly method (with respect to chemicals) but the research done is limited. The glucose yield is generally increased with increasing pressure and temperature [17]. One advantage with the SC-CO 2 method is that the formation of inhibitors is small, and neither the lignin nor the hemicellulose is modified during the process [7] Hydrolysis The purpose of the hydrolysis is to convert cellulose and hemicellulose to monomer sugars. This can be carried out by cellulase enzymes. Hydrolysis is the least efficient part of the overall process. Depending on the feedstock, different amount of enzymes are needed, and pretreated cellulose requires 100 times more enzymes than starchy materials [2]. The accessibility of the cellulose during hydrolysis is one of the rate limiting factors [11]. 5

13 Production of enzymes, cellulases Cellulase enzymes are used to convert cellulose to glucose. They can be produced by different microbes but the best suited is the fungus Trichoderma. The fungi s purpose of producing glucose is that glucose is its nourishment, meaning that glucose enhances the growth of the organism but not the production of cellulase. For this reason, the carbon source must include an inducing sugar when producing the enzyme, e.g. lactose. If the production of enzymes does not occur on site, storage is needed and the cellulase must be stabilized against microbial contamination and protein denaturation. Production on site thereby means savings in preservatives and stabilizers [2]. The cellulases contribute to a significant extent of the overall process costs (10-20%) and it is an important challenge to reduce it, though one ought to remember that this is one of the most uncertain costs in most economic analysis. Some measures for reducing the costs is to carry out further research and improvement of cellulolytic microorganisms, improvement of hydrolytic capacity of the enzymes and optimization of the enzyme production. It is also important to have an efficient pretreatment method. If enzymes are produced on site, the capital cost is the largest contribution to the total cost which stands for 60-78% of the total enzyme production cost. This contribution ends up with an addition to the ethanol price of SEK/L ethanol (assuming steam explosion pretreatment is used, [3]). The question concerning the feasibility of on-site production of enzymes strongly depends on the price of commercial cellulase enzyme preparation, which is still very uncertain. Furthermore, the question of feasibility might also depend on the type of sugars used in the feed stock; it might be favorable to not only have C6 but also C5 sugars, since the overall ethanol yield then decreases by a smaller extent. Still, more research is needed before this can be stated as fact. According to a review made by Barta et al. [3] most authors assume that cellulases are purchased from enzyme manufactures and calculates with prices in the range of SEK/L ethanol. Hydrolysis reactions The pretreated feed stock is generally fed to a hydrolysis tank as slurry with various amounts of total solids, depending on the pretreatment method. The optimum conditions for the enzyme are 50 C and ph 5. If the hydrolysis is carried out separately from the fermentation, it proceeds for 5-7 days and the viscosity of the slurry decreases by time. At the end of hydrolysis 65-95% of the cellulose has been converted to glucose. The cellulase proteins produced by Trichoderma mainly consists of three different enzymes: cellobiohydrolases (CBH), endoglucanases (EG) and β-glucosidase (BG). Cellulases consist most of CBH followed by EG, and only less than 1% of the cellulose mixture is BG. Basically there are two reactions taking place, first the conversion of cellulose to cellobiose (1) followed by the reaction producing glucose (2). The first one is performed by both CBH and EG and the latter one is the task of BG. (1) (2) The kinetics of reaction (2) can be described by Michaelis-Menten kinetics. The small amount of available BG is a problem, due to the fact that cellobiose might inhibit both CBH and EG which means 6

14 that accumulation of cellobiose makes the hydrolysis slower. Also, BG is inhibited by glucose. There are different ways to solve this problem. One solution is to produce β-glucosidase in a separate fermentor, but this is quite expensive. Another solution is to run the hydrolysis and fermentation in the same device; Simultaneous Saccharification and Fermentation (SSF). Since the yeast consumes glucose this counteracts inhibition of β-glucosidase. The problem with SSF is that the optimal operation temperature for the hydrolysis-enzyme and the yeast is not the same, 50 C and 28 C respectively. A third solution to the inhibition problem is to genetically modify the fungi to produce more β-glucosidase. Another generic problem concerning the hydrolysis is the reaction rate, which decreases as hydrolysis proceeds. Roughly the adsorption equilibrium can be described by a Langmuir isotherm, which is reached after a couple of minutes. The reaction rate is initially fast but decreases by time; after 24 hours it is less than 2% of the initial. It is not fully understood why this is the case [2] Fermentation In the fermentation, sugars are converted to cellular energy by yeast at anaerobic conditions, and at the same time the by-products ethanol and carbon dioxide are produced. The glucose overall reaction is described by equation (3) below. The transformation of sugars to ethanol can also be performed by other microorganisms such as bacteria. (3) Fermentation can be done either simultaneously or separate from hydrolysis, i.e. Simultaneous Saccharification and Fermentation (SSF) or Separate Hydrolysis and Fermentation (SHF), where SSF has been shown to be less capital intensive and results in a higher ethanol yield [5]. Microbes for fermentation The bacteria Zymomonas mobilis and genetically modified Escherichia coli have shown to have higher productivity and give higher ethanol yields than yeast. The most common microorganism used for ethanol fermentation is despite this the yeast Saccharomyces cerevisiae, and it has been the subject a lot of research. The reason for this is that it is more robust against both ethanol and other inhibitors present [12]. The disadvantage with S. cerevisiae is that it is unable to ferment the pentoses (xylose and arabinose) which has been considered necessary to reach an as high yield of ethanol as possible. Strains of S. cerevisiae, which use the same pathway as natural xylose and arabinose fermenting microorganisms, and therefore are able to ferment pentoses, have been developed by genetic modification. These strains are not yet as robust as they need to be to be introduced on an industrial scale [13]. It is important to know that the sugars that are to be fermented occur to different extent in different lignocellulosic materials, e.g. the hemicellulose of softwood contains a higher proportion of mannose than hardwood. The hardwood, on the other hand has a higher proportion of xylose in the hemicellulose [10]. The pentose fermentation may thus not be as important for softwood as for hardwood raw material. Inhibition During pre-treatment and hydrolysis, a wide range of compounds that inhibit the yeast are obtained. The inhibitors are mainly divided into three groups; weak acids, furan derivatives and phenolic 7

15 compounds. It is important to identify the inhibitors and their mechanism of inhibition in order to improve the fermentation. When identification has been done specific detoxification methods can be used, alternatively one can use a modified microorganism which can withstand the inhibitors or optimization of the pretreatment with respect to fermentation can be carried out [10]. It has been observed that concentrations equal or lower than 3.75 g/l of furfural respectively 6 g/l 5- hydroxymetylfurfural (5-HMF) does not reduce the ethanol yield. It has also been shown that 6 g/l of acetic acid, formic acid and levulinic acid actually increases the ethanol yield. On the other hand, higher concentrations reduce the yield. Usually the amount of furfural decreases during fermentation, due to the fact that yeast can convert the furfural to furfuryl alcohol. This is also valid for the HMF which is converted to HMF-alcohol by the yeast. The other way around is true for acetic acid, i.e. it increases from pretreatment to SSF, which can be due to liberation of acetyl groups of the remaining hemicelluloses [7]. Pentose fermentation Pentose fermentation increases the overall yield of ethanol. In a study made by Sassner et. al. [5] the overall ethanol yield increased by 32% respectively 8% compared to the base case for Salix and spruce, when additionally pentose fermentation was carried out. In that study, steam explosion was used as pretreatment method. In the base case, only the C6 sugars where fermented. According to the study the increase of the ethanol yield could have been higher. As mentioned above strains of S. cerevisiae which are able to ferment pentoses have been developed but have not yet been introduced industrially. Hemicelluloses for biohydrogen and biogas production Another option for the use of hemicellulose, instead of pentose fermentation to ethanol, is to ferment it to bio-hydrogen. In a study by Ren et al. fermentation of hemicellulose hydrolysates to hydrogen was done by Thermoanaerobacterium thermosaccharolyticum W16. The hydrolysates were collected from steam-pretreated corn stover and consisted mainly of xylose. The fermentation produced a gas that contained over 60 % hydrogen and the molar yield was 2,11 mol H 2 /mol substrate, which is the mixed sugars present in the hydrolysates (mainly xylose). This indicates that this is a promising technique in terms of taking care of the pentoses [14]. The fermentation of the pentoses can be included in a bio-refinery concept in which the biomass can be converted to several products and useful by-products. In a study by Kaparaju et al. a bio-refinery concept using wheat straw as raw material is proposed. Ethanol is produced from the solid fibers and bio-hydrogen from the hemicellulose hydrolysates. In addition to the production of hydrogen biogas can be produced from the stillage from ethanol distillation and from the effluent from the hydrogen production [15] Product recovery Even though no full scale lignocellulosic bioethanol plant has yet been built, most of the process steps for product recovery are considered well known; e.g. distillation, evaporation and drying. Still, these unit operations need to be verified on pilot scale before full-scale plants can be built [4]. The distillation is carried out to upgrade the ethanol concentration to an azeotropic composition. Furthermore evaporation can be used to remove water and thereby concentrate the non-volatile compounds in the liquid part of the residues, which then can be used as fuel in a boiler. 8

16 Ethanol After fermentation, the ethanol needs to be separated from the fermentation broth. This can be carried out in distillation columns: an energy integrated system can contain two stripper columns and a rectification column. The reason why two stripper columns can be used is to reduce the energy usage. If the ethanol concentration in the feed to the distillation is high enough (8% w/w), only one stripper column is needed. In the stripper columns ethanol is separated from solids and non-volatile compounds, and in the rectifying column the ethanol is distilled to its azeotropic composition, 94% (w/w). The energy demand in the distillation columns is one of the highest in the entire process. Thereby heat integration is needed to reduce the energy consumption [4]. This can be carried out by using overheads vapor from the first stripper column to run the reboiler of the second stripper column before being fed to the rectifying column. Overheads vapor from the second stripper column is used in the reboiler of the rectifying column, and live steam is additionally used if needed [5]. The energy use in the distillation is strongly dependent on the ethanol feed concentration. The ethanol content in the feed is usually about 4-5% (w/w) when using lignocellulosic raw materials, compared with starch based production where concentrations about 8% (w/w) are obtained. Higher ethanol concentration leads to less energy consumption in distillation since less water needs to be removed [4]. Finally, if ethanol with higher purity than the azeotropic composition is desired, dehydration can be carried out by adsorption of the water by zeolites. Lignin Basically lignin can be separated after or during the pretreatment, depending on the pretreatment method, or after the hydrolysis or fermentation depending on if one chooses to run the process according to a SSF or SHF. Lignin can be separated by filtration and then dried until its moisture content is less than 10%. Then it can be burnt to generate steam or power. Alternatively it can be applied to high-volume low-value applications such as an ingredient in roads as filler [2]. 1.4 Economy The three main contributors to the process costs are as follows: capital investment (30-45%), raw materials (30-40%) and the cellulase enzymes (10-20%). The cost of the enzymes is one of the most uncertain parameters [3]. To reduce the ethanol production cost it is important to reach a high ethanol yield. It is also important to have a high ethanol concentration and high solid concentration in the feed to the distillation columns in order to decrease the energy demand. Process integration is also important for reducing the costs [4]. For example the process can be integrated with a combined heat and power plant, CHP. One constraint connected to district heating is the location of the plant; surplus of heat must be needed. Another option is to integrate second generation production of bioethanol with first generation, in order to use the whole agricultural harvest. The two methods could be integrated at suitable points, e.g. after the fermentation and before the distillation by means that some of the equipment could be shared [1]. The final production cost of ethanol depends on assumptions made and process design for the calculations. It varies in different techno-economic studies between 0,93-5,49 SEK/L. It has turned out that some of the most important parameters that influence the production costs are the annual capacity, the raw materials, conversion technologies and the overall ethanol yield assumed [3]. 9

17 1.5 Pretreatment methods In this section steam explosion, organosolv and pulping based pretreatment is presented Steam explosion Steam explosion is one of the most used pretreatment methods for ethanol production from lignocellulosic raw material. The method has both physical and chemical actions that make the lignocelluloses separable into its three main components cellulose, hemicelluloses and lignin. Steam explosion can be performed with or without catalyst and the catalyst can be either an acid or a base. The principle of this method is to expose the material to pressurized saturated steam followed by rapid pressure reduction in a flash vessel. This results in a breakdown of the lignocellulosic structure, an increase of surface area, hydrolysis of some the hemicelluloses and depolymerization of the lignin components. Hence the defibration can be performed more easily and with better results, such as higher conversion to ethanol [18]. As mentioned above both chemical and physical interactions are used in order to effectively break down the lignocellulosic structure. The chemical pretreatment is present in the form of hydrolysis of the glycosidic bonds contained within the hemicellulose and cellulose structure along with the removal and/or redistribution of lignin. Cleavage of acetyl groups into acetic acid as well as the acidic nature of water at high temperatures promotes further hydrolysis of the hemicellulose. The physical pretreatment happens during the rapid decompression of the system. This rapid expansion vaporizes the saturated water within the fibrils, breaks down the molecular links and leads to a lignocellulosic matrix that can be effectively hydrolyzed in the subsequent enzymatic hydrolysis [27]. It is possible to combine steam explosion with other types of pretreatment, e.g. steaming and mechanical treatment which effectively disrupts the cellulosic structure [19]. Severity factor Since the efficiency of the method varies a lot with time and temperature a unit for this called the severity factor, R 0, is often used, se equation (4). (4) In equation (4) t stands for time (minutes), T for temperature ( C) and 100 is a reference temperature. There are both advantages and disadvantages with having a high severity factor. The surface area of the material is increased, which is beneficial. Temperatures up to a certain level are beneficial since this result in a release of hemicellulosic sugars, however, an increase in temperature also results in an increase of sugar degradation and thereby sugar loss and thus an optimum temperature exists. This temperature varies with the type of raw material [18]. Shorter residence times and lower temperatures have been shown to be more favorable because the sugars, especially those from the hemicellulose, do not degrade into products that inhibit the subsequent fermentation [23]. However, a high degree of severity is required to enhance the enzymatic digestibility of the cellulose fibers, especially in softwood. The maximum yields of sugars from hemicellulose and cellulose are not reached at the same degree of severity in the pretreatment and hence an optimum severity can be found for different systems since the proportions of hemicellulose and cellulose change depending on the type of biomass. 10

18 If a high severity factor is used sugars may degrade further mainly to furfural, hydroxymethylfurfural (HMF), levulinic acid and formic acid together with other substances. The formation of degradation products reduces the overall yield, and the products may also cause inhibition in following biological process steps. On the other hand, if a low degree of severity is used, cellulose digestibility will not be enhanced, which will cause the overall sugar yield to be lowered. To overcome this problem a proposed two-step steam pretreatment was investigated in which the first step was performed at low severity and the second at high severity, and this gave better conversion to ethanol but also higher energy consumption and higher equipment costs. Catalyst process Steam explosion can be performed with or without a chemical catalyst. The chemical catalyst can either be a dilute acid or a base. To use a catalyst in the pretreatment step has been shown to require decreased temperature and residence time and to achieve optimum fractional sugar recovery of steam pretreated samples of lignocellulosic materials [23]. Several investigations have shown that acid catalysts used within the steam explosion process in dilute quantities are improving the hemicellulose hydrolysis during the pretreatment and to improve cellulose digestibility further on in the process. Acids catalysts allow a decreased retention time and temperature of current operating systems and still reach high yields. By decreasing the retention time and temperature with the addition of an acid catalyst a reduction of inhibitory compounds formed is obtained, complete removal of hemicellulose is approached and the improvement of hydrolysis later on in the production is attained [27]. When acid is used as a catalyst the biomass is usually impregnated with an acid catalyst such as SO 2 or H 2 SO 4 before entering the pretreatment reactor [23]. Softwood feedstocks, which have high lignin content and are less acetylated, have shown increased yields when dilute acids are introduced into the system. The addition of dilute acids may also potentially reduce the formation of sugar degradation products [27]. Execution of steam explosion in one or two steps As mentioned above, steam explosion can be performed in either one or two steps. When performed in one step all cellulose and hemicellulose is released at the same time. When performed in two steps the hemicellulose is released in the first step and the celluloses structure is opened up in the second step. It has been shown that the gain in performing the process in two steps is not enough compared to one step due to the extra energy requirement for separating and washing between the two steps. Also higher energy consumption is required to reheat material for the second step. The capital cost is also higher for a two-step process [23]. The major advantages with a two-step process are a higher conversion and a lowered requirement of enzymes and water in the SSF. A two step steam explosion pretreatment can be performed in different manners. One of the studied processes used acid catalyst as follows: the first step was performed at low severity with sulphuric acid as catalyst to hydrolyze the hemicellulose. In the second step, the washed solid material from the first step was impregnated with sulfur dioxide and steam pretreated, this time at high severity to enhance the enzyme accessibility. At high severity, inhibitors may form which affects the 11

19 fermentation and inhibit the yeast [24, 27]. Studies have also been made to create an optimum cellulose fraction for hydrolysis through the use of a two-step process that dissolves the structure in stages. The first step involves a lower temperature (180 C) in order to solubilize and remove the hemicellulose fraction. The second stage uses a high temperature pressurized pretreatment with temperatures up to 210 C in which the cellulose fraction is subjected to breaking down of its carbohydrate linkages. This two-step process increased the downstream ethanol yield, compared to a one step process, by increasing accessibility to cellulose structure because of the reduction of the hemicellulose fraction. Operation costs also decreased as less enzyme dosage was required due to increased accessibility of the cellulose fraction. However, increased costs of equipment needed for processing and additional energy usage in the second step is required [27]. The process A proposal for a one step process using acid catalysis can be seen in figure 2. It is possible to compress vapours from the flash tank using a vapour compressor and use the produced vapour to increase the temperature of stream coming into the pretreatment reactor. The use of a vapour compressor saves energy for the process but problems with impurities in the steam can occur [53]. Vap compressor Steam Pretreatment Steam explosion Water Heater Steam Reactor Flash Mixer Feed SO 2 Slurry Figure 2: A proposal for the steam explosion flow diagram[53] Temperature, pressure and time can vary and are investigated in many studies. Typically values between 160 and 260 C are investigated, and pressures varying between 7-48 bars for a few seconds up to several minutes are common. The chemical processes that take place in the three main components of lignocelluloses are very much dependent of temperature, pressure and steaming time [19]. Inhibitors Inhibitors in the SSF step of the process are one of the main challenges when using steam explosion. Steam explosion produces a larger amount of inhibitors going in to the fermentation step, compared to several other methods. The reason for this is that the method includes harsh conditions for the treated material. The two most common inhibitors are furfural and HMF which can react further to other components as mentioned before [7]. Drawbacks and benefits The steam-explosion pretreatment process has been a recognized technique for the pretreatment of different biomass feed stocks for long time. It is able to generate high sugar recovery while utilizing a low capital investment and low environmental impacts concerning the chemicals and conditions 12

20 being implemented [27], i.e. no hazardous chemicals could be used [18]. It also has a high potential for optimization and improvement of efficiency [27]. It is also favorable as a pretreatment method since lots of studies has been made, and several pilot and demo plants exist. Thereby it involves a smaller risk compared to other pretreatment methods. The main drawback is the energy consumption which in most cases is high since high temperature and high pressure steam is used [21]. If the raw material consists of both hardwood and softwood another disadvantage might appear since steam-explosion is most useful for agriculture and hardwood materials without acid catalyst [11]. Regarding inhibition and waste minimization, steam explosion is found to be a less desirable pretreatment approach [7]. As mentioned, dilute acids or bases in most cases will need to be added during softwood pretreatment or when increased yields are required for feed stocks that contain lower amounts of lignin. If a dilute acid is added the cost of the chemicals increases. It also affects the equipment requirements. Regarding formation of degradation products, it differs from studies if acid catalysis increase the amount or not. The catalyst is needed to be neutralized by salts and the salts in turn would then have to be separated from the system and disposed of [27]. Since lignin (as a by-product) could come to be interesting in the future another drawback could be that the lignin is not as pure as in other pretreatment methods since it in most cases has been impregnated with a catalyst and also treated under harsh conditions. Pilot plant and research A lot of research has been done in this area since it is the most common pretreatment method. At the moment, globally, there are around 100 different research and development projects about cellulosic ethanol, e.g. a demonstration plant in Örnsköldsvik which opened According to themselves it is one of the most advanced plants in the field and the process has been developed during a long time [25]. Furthermore there are other demo plants around the world, e.g. Inbicon in Denmark, Iogen in Canada and Abengoa in Spanin Organolsolv Organosolv is another of the pretreatment methods of lignocellulosic biomass that there has been a lot of research on. The method is conducted by using organic solvents at a high temperature (~180 0 C) and sometimes also a catalyst (acidic, neutral or basic) to improve the efficiency. The method works due to the fact that the lignin and the hemicellulose can be separated from the cellulose because of the solvent, catalyst and high temperature [28]. Solvents It has been shown that the choice of solvent is very important. The choice does not only affect the cost of the solvent itself but also the recovery cost. It affects the efficiency of the pretreatment and the risks of the process, since some solvents are more harmful than others which contribute to a greater risk. There are many solvents that can be used, such as low boiling alcohols (methanol and ethanol), high boiling alcohols (ethylene, glycol and glycerol), organic acids (formic acid and acetic acid), organic peracids (performic acid and peracetic acid) and acetone. The most common solvents used in studies to develop a useful organosolv process are methanol and ethanol, where ethanol is the most favourable one. The reason why those solvents are most common is because of the low 13

21 chemical prices compared to other solvents. They are also easy to recover by distillation due to the low boiling points, which is necessary due to the fact that the solvent inhibits the growth of microorganisms, subsequent enzymatic hydrolysis, and fermentation. The recovery includes separation from the dissolved lignin and hemicellulose, followed by reusing the solvent. The low boiling point also gives an economical benefit, since the amount of energy in distillation decreases [29]. Most of the tests that have been performed through the years use ethanol as organic solvent, but a significant number of tests have used methanol as well. In one study made by Zhao et al. it was found that pinewood pretreated with a 60-80% aqueous methanol solution and an addition of HCl as catalyst removed about 75% of the lignin. When the amounts and conditions were adjusted and then tested on beech wood instead 90% of the lignin was removed. In both cases almost all the hemicelluloses were dissolved. If the temperature or amount of catalyst where raised, the enzymatic digestibility increased since more lignin was removed [29]. A similar experiment was performed by Huijgen et al. on willow wood where ethanol was used instead. This led to a delignification of 59% [30]. It is difficult to determine if ethanol or methanol is the most beneficial to use in an organosolv process. As presented above they both display high delignification but they cannot be compared to each other because some corresponding data are missing and different wood types are used. However, the pros and cons for each solvent can be compared instead. Methanol performs at a lower temperature than ethanol because it has a higher solubility of lignin. On the other hand, methanol is more toxic than ethanol and forms inflammable vapours at low temperatures which means that high safety equipment is needed [29]. A reason why ethanol is more favourable than methanol is because of the safety aspect. Another reason is that ethanol is the wanted product and the separation of the solvent is hence not as critical as if methanol was used. As previously mentioned research on high boiling alcohols has also been done. The advantage with high boiling alcohols is that they perform very well at less severe conditions, e.g. lower pressure and temperature. But due to high recovery and chemical cost it s impossible for the high boiling alcohols to compete with the low boiling alcohols [29]. Only a few studies have been made on the organic acids, organic peracids and acetone since they have more drawbacks than benefits. For example, many of them are very corrosive and hence need more advanced equipment material which is to expensive [29]. Catalyst Many different catalyst types have been tested through the years, not only acidic but alkaline and neutral as well. One of the most common catalysts is sulphuric acid due to its strong reactivity and high efficiency. But due to its properties, hazardous and corrosive, other catalysts are also commonly studied [31]. In a study by Park et al. [31] three different types of catalysts were compared: sulphuric acid, sodium hydroxide and magnesium chloride. The test was carried out in a mini bomb with 50% aqueous ethanol as solvent and pitch pine as wood type. The test showed that the enzymatic digestibility was similar for magnesium chloride and sulphuric acid with a delignification of 62 and 57 %, respectively. 14

22 The experiment also showed that when 1% sodium hydroxide was used nothing happened, but when the amount was increased to 2% the digestibility of cellulose in the enzymatic hydrolyse increased to 85%. To see how the solvent and catalyst work independently one analysis was performed with only the solvent, and one with only the catalyst. This showed a very poor result for all three catalysts with a maximum result of 32% enzymatic digestibility of cellulose for sodium hydroxide. When only the solvents were used the digestibility increased with temperature and residence time but nevertheless showed poor results. This shows that in an organosolv pretreatment both solvent and catalyst are necessary since they work together [31]. As can be seen above both the choice of catalyst and solvent are very important. It also shows that the most efficient catalyst or solvent is not necessarily the best choice for an industrial process due to the fact that one must make an economical profit. Process Figure 3 shows a possible organosolv pretreatment flow diagram. First the lignocellulosic biomass enters a reactor were the biomass is exposed to solvent, catalyst at high temperature. The resulting pulp is then separated from the solvent and catalyst and subsequently washed in a warm solvent wash. The warm solvent wash is important to avoid re-precipitation of lignin in the subsequent water wash [26]. The solvent from the wash is then mixed with the solvent from the first separation and then sent further to a distillation column. After the solvent wash the solid fraction is washed with warm water to remove the solvent and catalyst that remain. After the water wash the pulp is ready to be further processed in the SSF. In the distillation column the solvent is boiled up and then reused in the reactor and solvent wash. This leaves concentrated black liquor in the bottom of the column. The black liquor contains dissolved lignin, small amounts of glucose, all pentoses, furfural, HMF, acetic acid and extractives. This multi component liquid is subsequently diluted with water to precipitate lignin. The lignin is then filtered, washed with water and dried. The filtrate from the filtration can be further processed to separate all the components from each other through distillation. The sugars from the hemicellulose solution can be sent to the SSF if the yeast is able to ferment those as well [32, 29]. 15

23 Figure 3: Proposal for organosolv pretreatment process 16

24 Figure 4: Proposal for ethanol production process using organosolv pretreatment Figure 4 shows a flow diagram of an ethanol production process with organosolv used as pretreatment method. The dashed line marks which steps that are included in the pretreatment process described in figure 3. Everything outside the dashed line is part of producing and purifying the ethanol and lignin. As can be seen above the solid fraction enters the SSF where the cellulose is converted to glucose which is further converted to ethanol. From the SSF, the residual lignin is mixed with the dry lignin from the filtration. The lignin is then dived into two streams, one enters a boiler to produce power which can be used in the process and the other lignin stream is purified to get high quality lignin. The possibility to get reactive lignin is one of the major advantages with this process. Unfortunately there is no use for this high quality lignin today, but a lot of research has been done and a possible future exists in the polymer branch. In the future this lignin might be a co-product valuable for other industries; in that case the burning step can be removed. But so far only a smaller amount is purified while most is burned to generate power [29]. 17

25 1.5.3 Pulping based processes Most of the pulp today is made by either the sulphite or the sulphate process (Kraft process). Where, as the names indicate, the chemical used is the main difference between the two. Today, the Kraft process is the most applied technology. Some pulp is produced by the soda process, which is the precursor of the Kraft process. For the sulphite process the active agents can be bisulphite or sulphite with different counter ions such as calcium, magnesium, sodium or ammonium [31]. The Kraft process on the other hand uses a mixture of sodium hydroxide and sodium sulphide together with some sodium carbonate as active chemicals [34]. In general the wood is chopped into wood chips and put in pulp-digesters together with the chemicals and then heated to break down the lignin and to different extent digest the cellulose and from this create pulp. This is also what needs to be done in a pretreatment step for ethanol production and therefore it has been investigated in this study. The sulphate process, the soda process and the sulphite process will be described in more detail below. Sulphate process For the sulphate process it is proposed that a repurposed Kraft mill should be used. In the sulphate process, white liquor is used which consists of dissolved sodium hydroxide and sodium sulphide. After pulping, the pulp and the resulting black liquor must be separated. This black liquor contains dissolved organic material, lignin and cooking chemicals. The recovery of this is a well-known cycle used in many Kraft mills today. It is first concentrated in an evaporator and then burned in a recovery boiler to produce steam and to get back the white liquor used for cooking. The pulp is washed and hydrolyzed before it is further separated from lignin and then fermented to ethanol. The last step is to concentrate the ethanol in a distillation process [35]. Some advantages with the scenario to use a repurposed Kraft mill for the process is that the supply chain of delivering the raw material is already in place, some equipment can be used, there is a trained labour force and that the recovery process of the chemicals is well known technology [36]. Today the lignin in a Kraft process is burnt, mainly due to the high sulphur content. This lignin is also hydrophobic and to make it more reactive it has to be modified [39]. Soda process The soda process is the precursor to the Kraft process. It uses % NaOH at a temperature of C depending on the raw material. It is the most employed method for the pulping of annual plants such as agricultural waste (bagasse and straw) and cultivated and/or naturally occurring bagasse and reed. Since these types of plants have lower lignin content than wood a lower amount of chemicals are needed [41]. In terms of using the soda process (with or without catalysts such as anthraquinone and p-benzo-quinone) in a biorefinery some advantages could be high pulp production rates due to short cooking times, high pulp yields and that it is possible to process any type of lignocellulosic material according to García et al [42]. The simulations made in the study showed that this process used a small amount of energy but was only focused on getting a cellulose rich solid fraction and not getting valuable by-products [42]. Another advantage could be that the lignin can be used for high value products in the future. The lignin from the soda process does not contain any sulfur and a very small amount of hemicelluloses. Since the size of the market for such products is not known, one does not know how large advantage this kind of lignin has [39]. 18

26 Sulphite process The sulphite process works due to the reaction between the lignin and the free sulphurous acid. These form lignosulfonic acid which in turn reacts with the cations to lignosulfonates. The lignosulfonates are relatively soluble in water. These can then be fragmented from the cellulose making this more accessible for hydrolysis [39]. In a mill using the sulphite process, spent sulphite liquor (SSL) is a by-product when making pulp. The pulp contains cellulose and remaining lignin and the SSL contains lignosulfonates and sugars from degraded hemicelluloses. These sugars are hexoses and pentoses in different proportion depending on which type of wood is used [37]. Incorporated ethanol production Compared to the sulphate process the sulphite process has lower yields and consequently mills using the sulphite process have to adjust the process to make it profitable. For example, the SSL can be fermented to ethanol. In addition to lignosulfonates and sugar the spent liquor also contains inhibitors to fermentation such as organic acids, furfural, 5-hydroxymethylfurfural, sulphite, phenols, wood extractives and dissolved solids, which makes it more difficult to ferment. To make the fermentation more effective different kind of detoxification methods for SSL can be used such as overliming, dilution of SSL or ph adjustment to remove some of the toxic components. Adding nutrients increases the yeast growth, but to keep the overall process as effective as possible the detoxification and adding of nutrients should be as mild and low as possible. Due to the lower content of hexoses in hardwood SSL than in softwood SSL, it is harder to ferment [37]. Ethanol production When the purpose is to make ethanol instead of paper there are different demands on the cellulose containing part. When making paper the cellulose should be preserved to get a strong pulp, while in ethanol production the aim is to degrade the cellulose as much as possible and completely separate it from hemicelluloses. Therefore different ph, temperature and chemical dosage can be used when ethanol production is the aim compared to traditional pulping. One process investigated is called sulphite pretreatment to overcome recalcitrance of lignocelluloses (SPORL) by the inventors of the process [38]. Figure 5 shows a schematic figure of how the process is supposed to work. The tests of the process have been done in laboratory scale. First the wood chips were impregnated with sodium bisulfate and sulfuric acid and then heated to about 180 C for 30 min. The spent liquor and the solids were then separated with a screen and the solids were fed to a disk refiner to prepare it for enzymatic hydrolysis. The hydrolysis was made simultaneously with the fermentation (SSF) [38]. To improve the efficiency of the process the spent liquor can be separated into lignosulfonates and dissolved sugars, where these sugars (the hexoses) also can be fermented into ethanol. This separation is done today with ultrafiltration; the problem is that the sugars and lignosulfonates have overlapping molecular weights [40]. The sugars probably need some kind of detoxification before they can be fermented. After fermentation and SSF the ethanol produced is concentrated in a distillation step. The chemicals and some of the lignin from the pulp is then involved in heat and chemical recovery which makes it possible for the chemicals to be reused in the process and producing heat for the process. 19

27 Water Wood chips Disk mill Press SSF Screen SPORL pretreatment ~30 min 180 C Spent liquor Filtration Detoxification Fermentation Separation RO/UF Distillation Lignosulfonate Chemicals Steam Ethanol Heat and chemical recovery Lignin Figure 5: Schematic flowsheet of the SPORL process [38] Co-products The lignosulfonates are co-products for which a market already exists. The most common use is as an additive in cement, where it influences the hardening as well as the structure of the hardened cement. This is one use where the value of the lignin is considered to be low, others are as dust control and phenol-formaldehyde resins. Lignosulfonates can also be used in other areas where the value is higher, such as for bricks, refractories and ceramics, battery expanders, industrial cleaners and vanillin [39]. Not all the lignin is found in the spent liquor, some of it is in the solids. This can be burnt to provide the process with steam and electricity [38]. Drawbacks and benefits Advantages with the SPORL method is that a lot of the techniques, e.g. pulp making, disk milling and chemical recovery, are used in the paper and pulp industry which makes the step to introduce it to industry scale more easy [38]. One more advantage is that the lignosulfonates already have an established use and is by this an extra income. The question is how the demand is affected if lignosulfonates is produced in larger quantities. Disadvantages with the SPORL method is that it is only one group of scientists that have been investigating the method. To make the process efficient the SSL can be fermented, but this requires detoxification which might not feasible on an industrial scale. 20

28 1.5.4 Ionic liquids Ionic liquids are salts that are liquid below 100 C [43]. The use of them is here examined as an alternative pretreatment method for bioethanol production from wood. The ILs are used to dissolve the wood and hence making the cellulose susceptible to hydrolysis. Consequently, the ability to dissolve wood is the foremost property needed in the ionic liquid. During the recent years a lot of progress in the area has been made, e.g. dissolution of wood has been accomplished, but there are still challenges to face before ILs can be used for pretreatment on a large scale [46]. Composition and properties of ionic liquids There are two kinds of ionic liquids; ILs (Ionic Liquids) and RTILs (Room Temperature Ionic Liquids) where RTILs are a subset of ILs. ILs are, as described before, salts that are liquids below 100 C, whereas RTILs are salts that are liquids at room temperature [43]. Since no heating is needed to keep RTILs in liquid form these are the most convenient and economical to use in an industrial process. ILs on the other hand require isolated tubes and heating at different locations to keep it from solidifying. ILs have high thermal stability, high electrical conductivity, wide electrochemical window, negligible vapor pressure [45] and an ability to dissolve both organic and inorganic compounds [46]. ILs can be used in many different applications such as electrochemical reactions, absorption spectroscopy, ionexchange chromatography, biocatalysis, as solvents for lignocellulosic biomass [45] and inorganic nanomaterials [46]. Ionic liquids consist of an organic cation and an organic or inorganic anion. Due to this structure a lot of combinations can be made and it has been estimated that the number of possible ILs range up to 10 9 pairs. This, combined with the fact that the chemical and physical properties are mostly determined by the type of anion, results in the possibility to produce custom made ILs. ILs can be custom made to fit different applications, and, for using ILs in a pretreatment method, the properties can be tailored to fit the substrate. For example, one can increase the dissolution of cellulose by decreasing the size of the anion [46]. The tunable properties include viscosity, melting point, polarity and hydrogen bond basicity. To be mentioned is also that the strongest interactions during the dissolution of cellulose appear between the anion in the ionic liquid and the cellulose, whereas the cation interactions are of lesser importance [45]. Ionic liquids can be divided into four groups; quaternary ammonium, N-alkyl-pyridinium, N-alkylisoquinolinium and 1-alkyl-3-methylimidazolium. 1-alkyl-3-methylimidazolium is the group of ILs that have been most examined for pretreatment of lignocellulosic materials [46]. A list of discussed ionics liquids are listed in table 2. Abbreviation [AMIM]Cl [EMIM]OAc [EMIM]Cl [MMIM]DMP Table 2: Abbreviations for discussed ionic liquids Name 1-Allyl-3-methylimidazolium chloride 1-Ethyl-3-methylimidazolium acetate 1-Ethyl-3-methylimidazolium chloride 1-methyl-3-methylimidazolium dimethylphosphite 21

29 Drawbacks and benefits The foremost benefits of using an IL for pretreatment of lignocellulosic biomass is that it enables the fractionation of lignin, hemicellulose and cellulose into different output streams and that the crystalline structure of cellulose is reduced, which results in a more efficient enzymatic hydrolysis [51]. Other pretreatment techniques require extreme conditions such as high temperatures and pressures, the use of strong acids or bases and organic solvents. Also, toxic compounds are formed in many of the pretreatment techniques. This does not apply for ionic liquids, i.e. using ILs for pretreatment does not require high temperatures or pressures and no extra chemicals are needed other than the ionic liquid. No inhibitors or by-products are formed since the components of the lignocellulosic material are simply dissolved [46]. Also, the time needed for pretreatment and hydrolysis can be significantly decreased, as described below. The ability to extract native lignin from the process can in the future become a benefit, depending on the market development for this coproduct [51]. Among the problems that can be encountered are the high viscosities of the ionic liquids. As an example, alkylimidazolium chlorides typically have viscosities larger than 2000 cp at 25 C [45] which can be compared to the viscosity of water which has a value around 1 cp at the same temperature. Some ILs are also highly hygroscopic and may have to be dried before use [49]. The subsequent enzymatic hydrolysis can be affected by residual IL through inhibition of the enzymes [46]. Ionic liquids for pretreatment of wood Cellulose is insoluble in water and most common organic liquids but can be dissolved in ionic liquids. ILs have been found able to dissolve a number of different substances containing cellulose, although the different substances are dissolved by different ILs, i.e. one specific IL can dissolve one type of cellulosic material and another IL can dissolve another cellulosic material. Some ILs can dissolve pure cellulose, i.e. in the form of cellulose-dissolving pulps from cellulose acetate, fibrous cellulose [47], microcrystalline cellulose [48] etc. Many of these ILs are also able to dissolve wood, that is, all components in the wood are dissolved [47]. Studies have also been made on extraction of lignin from wood with ionic liquids, which is pointed out as an alternative to dissolving the entire wood [45]. Also, studies have been made showing the possibility to perform hydrolysis of cellulose from wood in ionic liquid [46]. The complete dissolution of wood, the extraction of lignin and the simultaneously performed pretreatment and hydrolysis are further examined as alternatives for pretreatment of wood. The dissolution of cellulose-dissolving pulps with ionic liquids could be an alternative improvement of the pulp process described in this report, but is not examined here. Dissolution of wood The most frequently used pretreatment technique with ionic liquids is to dissolve the entire wood and then precipitate the cellulose and hemicellulose. To dissolve wood the ionic liquid and wood mixture is heated, usually to around 100 C, and mixed for a few hours. To retrieve the cellulose and hemicellulose by precipitation an anti-solvent, such as water or ethanol, is added to the dissolved wood solution. Studies of the structure after precipitation show that the cellulose no longer has any crystalline regions and is consequently only amorphous, which is, as described before, a great benefit. The precipitated cellulose and hemicellulose is separated from the ionic liquid through 22

30 filtration and then proceeds to a SHF or a SSF. A general flow diagram for the pretreatment with ionic liquids is presented in figure 6.The remaining ionic liquid, containing dissolved lignin and impurities, can be recycled through a series of steps including precipitation by anti-solvents (e.g. deionized water or alcohols), filtration, adsorption of impurities by activated carbon, organic solvent washing, purification with neutral-activated alumina and evaporation. However, there are alternatives to the described recycling which do not require as many steps. For example, supercritical fluids can be used to extract IL-soluble polymers. This alternative is probably expensive. Another option is to use anion exchange resins to isolate the ionic liquid as a salt and precipitate a part of the lignin. For lignin extraction lignin-compatible blocks in micelles could be used to transfer lignin from the ionic liquid to an aqueous phase. Also, a three phase system can be generated by the addition of K 3 PO 4 and K 2 HPO 4 to the ionic liquid after dissolution resulting in a salt-rich aqueous phase, a cellulose-rich solid phase and a phase containing the ionic liquid and dissolved lignin and thus separating almost everything but the lignin from the ionic liquid [45]. Although the amount of water that needs to be evaporated from the ionic liquid is decreased, the ionic liquid still contains a lot of lignin. If only the removal of antisolvent is of interest evaporation is sufficient since the boiling point of the anti-solvent is significantly lower than the boiling point of the ionic liquid. Wood Ionic liquid Dissolution of wood Anti-solvent Precipitation of cellulose and hemicellulose Filter IL with lignin and impurities To recycling of IL Cellulose and hemicellulose Solvent Wash To SHF or SSF Figure 6: General flow diagram for pretreatment with ionic liquids The kinetics of wood dissolution depends on the size of the wood particles. As would be expected the rate of dissolution decreases with an increase in particle size. This is consistent with the basic reason for pretreatment methods, that the wanted cellulose is protected by the lignin and hemicellulose structure. Thus, the smaller the particle the more accessible the cellulose [45]. Some ionic liquids have shown good wood dissolving qualities. It was found that [AMIM] Cl can dissolve both hardwood and softwood entirely (when examining silver fur, spruce, common beech and chestnut). But, the obtained solutions can have high viscosities resulting in severe difficulty to mix effectively [49]. Extraction of lignin The goal of studies of extraction of lignin is to find ionic liquids which selectively dissolve lignin and not cellulose or hemicellulose. Reports have shown that lignin can be extracted using ionic liquids containing acetate [45]. [EMIM]OAc is such an IL and has been shown to have interesting qualities such as high lignin solubility and low wood flour solubility, i.e. cellulose and hemicellulose solubility. 23

31 A good extraction of lignin has also been proved. When wood powder was pretreated with [EMIM]OAc at 80 C for 24 h the following hydrolysis of remaining wood flour showed a conversion of cellulose of 90 %. Another benefit with lignin extraction is that even if the cellulose is not dissolved, the crystalline structure decreases significantly [50]. In principle, the extraction of lignin is performed through mixing of wood and ionic liquid during heating. After the extraction of lignin the solid wood is filtrated and washed and is then ready to be hydrolyzed. Lee et al. [50] conclude that complete extraction of lignin is not necessary, a mere 40 % extraction is sufficient to get a conversion of 95 % of the cellulose in a following hydrolysis. They have also examined the effect of temperature and time on conversion in hydrolysis, extraction of lignin, extraction of cellulose and hemicellulose and the crystalline structure of the cellulose. It is concluded that the amount of extracted lignin increases with pretreatment temperature but that the amount of recovered wood decreases with the same. It is also stated that the crystallinity decreases with an increased pretreatment temperature. The extraction of lignin increased with increased incubation time but as said before, an extraction of lignin of 40 % is sufficient, which is completed within five hours (when T=90 C). In that current study [50], pretreatment with [EMIM]OAc increased the conversion of cellulose in hydrolysis from < 50 % up to >95 % compared with hydrolysis of nonpretreated wood [50]. It should be noted that hydrolysis of non-pretreated wood chips usually result in a conversion < 10 % and hence that the value of < 50 % for non-pretreated wood flour is high in comparison. This is due to the smaller particle size of the wood; in fact, milling of wood is considered a pretreatment technique in itself. Recycling of [EMIM]OAc was performed accordingly; after separation of delignified wood flour and lignified [EMIM]OAc by filtration the IL was washed with water to remove extracts from residual wood flour solids. The water was subsequently removed by evaporation. Recycling was tested for up to five times and showed no important decrease in efficiency. After a while the lignin should be extracted from the ionic liquid and this can be done by adding excess water, although more investigations must be done in this area to make it more efficient [50]. Figure 7 shows a flow diagram developed according to the process performed by Lee et al. [50]. The lignin is accumulated in the ionic liquid. [EMIM]OAc, lignin [EMIM]OAc Ethanol Wash Delignified wood powder Hydrolysis Fermentation Distillation 2 Water Removal of extractions from residual wood Water [EMIM]OAc, dissolved lignin, cellulose and hemicellulose Distillation 1 EtOH Figure 7: Proposed flow diagram for extraction of lignin using [EMIM]OAc 24

32 Simultaneous pretreatment and hydrolysis Another alternative emerging process option when using ionic liquids in the pretreatment is the ability to perform hydrolysis simultaneously with the pretreatment step. Both acid (liquid and solid) and enzymatic hydrolysis have been performed in ionic liquids. It has also been concluded that hydrolysis can occur in ionic liquids without the presence of a catalyst, i.e. the ionic liquid represents the catalyst [46]. Varying results have been achieved in the area. For example, a process for pretreatment and hydrolysis of lignocellulosic biomass has been developed using acid in IL with yields for xylose and glucose of 79 % and 70 % respectively. The ionic liquid used was [EMIM]Cl and the acid HCl. An advantage with this method is the speedup of the hydrolysis compared to the traditionally separate pretreatment and enzymatic hydrolysis and also that no enzymes are needed which effects the economy positively. The acid hydrolysis in [EMIM]Cl is performed satisfactorily after a few hours, compared to traditional enzymatic hydrolysis which can take days. Both HMF and furfural (by-products) are produced in this process which contradicts the commonly known benefits with using IL for pretreatment. This drawback is probably due to the combination of dissolution and hydrolysis. The components in the liquid going out from the simultaneous pretreatment and hydrolysis can be separated by ion-exclusion chromatography resulting in one stream containing ionic liquid and water, one stream containing HMF and furfural and one containing sugars [52]. Ionexclusion chromatography is probably expensive on industrial scale. For enzymatic hydrolysis using cellulases in [MMIM]DMP, experiments have shown hydrolysis yields of more than 70 % when treating corncobs. Although, problems emerge when having enzymes in ionic liquids since they to some extension are inhibited by the IL [46]. With IL-dissolved wood, when there is a continuous liquid instead of a slurry (which is the case with other pretreatment methods), the opportunity to use a solid catalyst occurs. Hydrolysis of cellulose to cello-oligomers have been performed on Amberlyst, Nafion, alumina, sulfonated zircona and zeolithes when using [BMI]Cl as ionic liquid. The reaction was selective without sugar dehydration products. However, there is a need to perform enzymatic hydrolysis after this unit operation for the oligomers to depolymerize into monomers [44]. Future perspectives Research on ionic liquids for pretreatment of lignocellulosic biomass has up till now mostly focused on laboratory studies of different variations of dissolution of cellulose and wood. The chemical properties, the dissolution kinetics and chemistry and the ability to dissolve cellulose, wood, lignin etcetera have been investigated. The use of ILs as a pretreatment method can be seen to be in its infancy. A few studies, on lab-scale, have been made on two or more process steps in series, however, not all of the steps required for an industrial process have been fully investigated. A lot of work has to be made before ionic liquids can be used for pretreatment of lignocellulosic biomass in the industry. The main issue is to make the process economically efficient. For this to happen a better understanding from a process-engineering perspective is needed to be able to single out the unit operations that need to be improved or more examined [51]. What is already clear is that the recycling of ILs and the extraction of lignin and other compounds released from the biomass from the ILs need to be examined in more detail. These unit operations have been completed successfully in lab but they need to be made more efficient to be suitable for industry [50]. How the IL waste should be dealt with has also got to be examined. [51]. 25

33 Chemicals Process details Improvements of the chemicals involved are also of interest. Firstly, the price of the ILs needs to be lowered, i.e. the production of ILs needs to be more cost-effective and the production needs to be increased. Nowadays the production volumes are in the scale of kg which results in high prices [51]. As described before, the IL can be tailor made to fit the application, thus, more investigations should be made to improve the properties of ILs. They can be made less corrosive, the viscosity can be lowered and they can be made biodegradable. Some ILs have been investigated for biofuel production, but more combinations can be tested and their fitness can be improved. Another possible improvement is to make the removal of IL from the extracted cellulose better since the IL inhibits the following hydrolysis. Before being used in industry the biocompatibility and the ecological impact of the IL should also be examined [46]. Further, to be able to make good techno-economic calculations, the future market for native lignin has to be investigated. The potential price of native lignin has been shown to be of great importance when calculating the cost-effectiveness of the pretreatment. This is since, if it can be sold at a high price, the income can cover all the expenses for the entire pretreatment step, including costs for IL. This could result in a cost-efficient process even without improvements in recycling and reuse of the IL and without a decrease in IL-price [51]. A summary of the things that need to be improved or examined is listed below in table 3. Table 3: Summary of improvements needed before the use of ionic liquids as pretreatment method for production of ethanol from wood Improvements/developments/examinations of Process unit operations [51] Recycling and reuse of IL [46,50] Extraction of lignin from IL [44,46,50] How to deal with IL waste [51] Inhibition of cellulase activity caused by IL residues [46] Design of ILs for biofuel production [46] Lower the price for IL [44,51] Biocompatibility and ecological impact [46] Market for native lignin [51] 26

34 2. Process descriptions In this section process descriptions are presented for the steam explosion, organosolv and sulphite pretreatment processes. Those are the process setups used for the calculations. The general process description is valid for all those processes but the assumed yields and results for the different processes are presented under each section. 2.1 General process description The calculations was carried out in Aspen Plus where the whole processes was set up; i.e. using steam explosion, organosolv and pulping as pretreatment method. The reason why ionic liqiuids was left out is described in section 5. All calculations are based on an annual ethanol production of m 3. The operational time for the plant is set to be hours per year. The plant is assumed to be located in Sweden/Norway. The location of the plant makes softwood a good alternative as a raw material and it has been assumed that there is a demand for district heating at the location of the plant. All the reference articles used for conversions etc. are based on laboratory scale experiments. Aspen Plus calculations (conversions, relations for calculations of flows etc.) are presented in Appendix B Feed The raw material feed was calculated from the annual production of ethanol. The type of raw material used for calculations was softwood. The type of softwood used for calculation varies between the different pretreatment methods, since no identical composition was found in reference articles Reactions in pretreatment The main reactions that occur in the pretreatment step are the same for all pretreatment methods. Below the general reactions that occur in every process is presented. For some pretreatment methods more reactions can occur, these are presented in each description of that specific pretreatment method. The main difference is to which extent the reactions occur, i.e. the yield of the reactions. Difference in which reactions that have been simulated also depends on what data that was found in the literature. Glucan + Water Glucose Mannan + Water Mannose Xylan + Water Xylose Galactan + Water Galactose Arabinan + Water Arabinose Xylan 2Water + Furfural Glucan 2Water + HMF Glucan Levulinic acid + Formic acid Mannan Levulinic acid + Formic acid SSF reactor The SSF can be simulated either as two reactors, one for the hydrolysis and one for the fermentation, or as one reactor including both hydrolysis and fermentation. In reality the SSF is done in one reactor, and how it is simulated does not make any difference for the result. 27

35 The main hydrolysis reactions are: Glucan + Water Glucose Mannan + Water Mannose Xylan + Water Xylose Regarding the fermentation, the microorganism chosen is the yeast S. cerevisiae. The main reactions for fermentation performed by the yeast are: Mannose 2Ethanol + 2CO 2 Glucose 2Ethanol + 2CO 2 2Furfural + 2Water 2Furfuryl alcohol + O Distillation The distillation for ethanol purification is performed in two strippers and one rectifier. The strippers are run in parallel regarding the feed, which is split between the two strippers, and the overheads of the two strippers are mixed and fed to the rectifier. The feed into the strippers are preheated with the bottom product from each stripper respectively. The idea of this setup is that the condenser of the first stripper is the reboiler of the second stripper. This means that the ratio between the two streams into the strippers has to make the condenser duty and the reboiler duty in respective stripper to be equal. In Aspen Plus, the condenser and reboiler were simulated as two different blocks, but in reality it is the same unit. The strippers were designed to give an ethanol concentration in the bottom product to be less or equal to 0.5 % (wt). The total efficiencies (regarding the trays) of all the columns were set to 70 %. The heat produced in the condenser of the second stripper is used in the reboiler of the rectifier. In cases where the heat from the stripper was not enough to cover the heat demand in the rectifier the difference was supplied as external heat. When the heat from the stripper was higher than the demand in the rectifier an extra condenser was used. Additionally live steam needed was supplied from the CHP plant. In the rectifier the distillate should have an ethanol content of 92.5 % (wt) and the mass recovery of ethanol from the feed was set to 0.2 %. This means that 99.8 % of the ethanol in the feed was recovered in the distillate. To be able to set up this heat integration in the distillations columns, the pressures in the columns had to differ. The strippers had a pressure of 3 and 1.7 bar and the rectifier had a pressure of 1 bar. The numbers of trays were 24, 24 and 36, respectively. The feed into the strippers was preheated with the bottoms from each stripper respectively. The distillation was simulated as distillation between only ethanol and water. All other compounds, i.e. sugars, by-products, solid material and acids, were separated before the distillation to make the simulation possible. The separated streams were later mixed with the bottoms product from the strippers, i.e. water, to get a picture of how the bottom product would look like in reality. Furthermore in reality, there is need for a side stream around the middle of the rectifier column since fusel oils otherwise are accumulated there. The activity factor model used for all the processes was NRTL-HOC, which is a so called local-composition model. 28

36 2.1.5 Dehydration In reality dehydration is needed to obtain pure ethanol. Since the overheads ethanol concentration from the rectifier is azeotropic, it is the same for all pretreatment methods. For this reason there was no need to carry out calculation for this unit operation Biogas production Biogas was produced from the bottom products from the two strippers, containing sugars, byproducts, acid and ethanol, which could be fermented into biogas. For this purpose microorganisms had to be supplied. The microorganisms does not need to be added continuously, but only once at the start. The biogas was used as fuel in the CHP plant to generate steam and electricity for the process and an alternative would be to sell the surplus CHP In a CHP plant steam and electricity is produced. The fuel of the CHP is produced biogas (methane) as well as lignin and none converted solids and sugars. Steam at the different pressures needed for the different units in the process was produced here and bled off to the process; hence a multistage turbine was set up. Steam from the boiler is available at highest at 65 bars. The pressure of the live steam needed varies between the different pretreatments. An economizer is included in the CHP to utilize the heat of the flue gases. The electricity produced in the CHP is primarily used internally to run pumps, compressors etc. The surplus is regarded as electricity output, which can be sold. 2.2 Process description - steam explosion process All the data, composition of the biomass, conversions etc. for solving the material and energy balances for steam explosion was taken from the same reference article by Sanam Monavari et al. [23]. The chosen unit operations of the process are presented in figure 8 below. Description of the flow sheet is presented in the following section. 29

37 Steam Pretreatment Steam explosion Feed SO2 Mixer Heatexchanger Reactor 190 C, 12 min Flash 1 atm 100 C Waste water CO2 O2 SO2 SSF 36 C 72 h ph 5 WIS conc. 10 wt% NaOH Water Cellulas15 FPU/g WIS Yeast, 3g/L 1 STRIPPER Ethanol RECTTIF STRIPPER 1 Bottoms from rectifier Steam 13 bar Solid fraction, DM=40% FILTER Fresh air Heat and power production BIOGAS Water treatment Flue gas cleaning Flue gases Electricity District heat Micro organisms Salt Water Figure 8: Flow diagram for the steam explosion process Raw material The raw material used in the steam explosion is chipped spruce, (Picea abies) free from bark. The wood chips are milled and chips between 2 and 10 mm are used. The dry matter content of the wood chips is 41.7 wt %, in table 4 the content of the woodchips is presented. The feedstock is 30

38 impregnated with sulfur dioxide until the concentration in the water phase of the woodchips is 2.5 wt %. Table 4: Composition of the dry matter in the softwood (Picea abies) used for steam explosion calculations Component Amount (%) Ash 0.2 Acid-soluble lignin 3.8 Acid-insoluble lignin 28.5 Glucan 45.1 Mannan 11.8 Xylan 5.2 Galactan 2.8 Arabinan 1.1 Acetyl groups 1.6 Before the feed enters the pretreatment reactor it is preheated with flashed vapor, since less external heat is needed then Pretreatment reactor The reactor is run in batch mode at 190 C; the residence time is 12 minutes. The pressure in the pretreatment vessel is 13 bars, which corresponds to 190 C. Only the four first reactions specified in the general process description occurr in the pretreatment reactor and those were simulated. The reason why the last five reactions were left out in the pretreatment step is because those are assumed not to affect the process economics and also data was not found in the literature. The fractional conversion of the reactions in the pretreatment from the reference article is presented in table 5. The total ethanol recovery is based on the glucose and mannose content in the wood. The overall ethanol yield is 75% of theoretical based on the glucan and mannan content of the raw material. The liquid recovery occurs in the pretreatment reactor and the solid recovery occurs in the SSF reactor. Since some of the monomer sugar is obtained in the pretreatment step, and some is obtained in the SSF reactor, glucose equivalents are used to make the yields comparable in the different unit operations. For this reason, glucose and mannose recoveries (based on equivalents) for the liquid and solid phases are presented in table 5. 31

39 Table 5: Data from the pretreatment experiment Component Pretreated slurry DM (wt %) 20.1 WIS (wt%) 15.3 Lignin content (%) 45.6 Glucose recovery (%) Liquid 16.2 Solid 68.4 Total 84.6 Mannose recovery (%) Liquid 68.5 Solid 9.1 Total 77.6 The yield of sugar degradation products concerns the liquid obtained from pretreatment and is presented in table 6 together with a summary over all conversion in the pretreatment reactor. Table 6: Summary of conversions in steam explosion pretreatment Component Conversion (%) Glucan Glucose 16.2 Mannan Mannose 68.5 Glucan HMF 2.0 Xylan Furfural 4.24 After the feed has been pretreated in the reactor it is flashed at 1 bar pressure. The vapor is then heat exchanged as described above and the slurry goes further to the SSF reactor Simultaneous Saccharification and Fermentation In the steam explosion process both hydrolysis and fermentation takes place at the same time and hence was simulated as one unit operation. The batch time in the reference article was 120 h, but calculations were based on 72 h since this is more realistic on industrial scale. The temperature in the reactor is 36 C, which is a reasonable temperature for the yeast cells. The yeast concentration in the SSF is 3 g/l. A commercial cellulase mixture is used, according to the reference article, with an enzyme activity of 15 FPU/g WIS. The WIS concentration in the reactor was set to 10 % which differ from the reference article (where 5% was used). This was done since this is a more realistic concentration on industrial scale [22]. To fit the total theoretical conversion found in the literature to the fractional conversion, the conversions over the SSF reactor were calculated which is presented in appendix B.1.1. The hydrolysis conversion for glucose and mannose respectively was calculated from the given sugar monomer liquid recovery combined with given yields for recovery of the solid fraction (hence in the SSF). Furthermore the yield of the fermentation was calculated from the overall ethanol yield. See appendix B for the calculations. 32

40 The reactions that take place in the SSF reactor is almost the same as the ones described in the general description, except for that the reaction that forms xylose was not set up in the simulation. This was done since no data for this reaction was found in the literature were the other numbers was taken from and it is also assumed not to be important. One extra reaction was added since the raw material contained a small amount of acetate, which reacts in the SSF reactor. Acetate Acetic acid The reaction above and the reaction when furfural reacts with water to furfuryl alcohol were assumed to have a fractional conversion of 100 %. All the other conversions are presented in table 7. In reality also HMF is converted by the yeast to its alcohol, but since this HMF-alcohol could not be found in the Aspen Plus database it was not simulated and hence HMF was not calculated to be further oxidized. The ph before the SSF reactor was assumed to be 2.3 and sodium hydroxide was added to neutralize the feed to the SSF to reach ph 5, see Appendix B.1.2. Water was also added to the SSF reactor which was calculated from the given WIS concentrations. Before the feed entered the SSF reactor the WIS content needs to be lowered, in order to achieve this water is added. Also enzymes and yeast are added to the SSF. Calculation regarding amounts of water, yeast and enzymes can be found in Append B.1. Gases such as carbon dioxide, oxygen and sulfur dioxide are released in the SSF reactor. Component Table 7: Conversion of reactions in SSF reactor Conversion(%) Glucan Glucose 83.6 Mannan Mannose 29.8 Glucose Ethanol 90 Mannose Ethanol 90 Acetate Acetaldehyde 100 Furfural Furfalcohol Distillation In the rectifier column 16 trays instead of 36 as was described in the general process description were used during simulations. This was the optimal amount of trays which was found by manually varying the number of trays which corresponded to the lowest energy duty. The bottom streams from the stripper columns are filtered. The solid part (with dry matter content of 40 %) is further fed to the heat and power production and the liquid is fed to the biogas production. 2.3 Process description - organosolv process The data used for the calculations of the organosolv process was found in two different articles. A study by Pan et al. [28] was used for the composition of the feed stock and the conversions in the pretreatment. Another study made by Pan et al. [54], was used for the conversion in the SSF since this was performed with higher consistency which is needed on an industrial scale. 33

41 The process with the chosen unit operations is presented below, see figure 9. Description of the flow diagram as well as the conditions and assumptions made for calculations are presented in the following section. VVX Mixer EtOH H2S04 Feed Reactor 170 0, 10 atm FILTER Solids Liquid Flash atm STRIPPER RECT Lignin, 90wt% Drier Lignin FILTER VVX VVX VVX EtOH Wash EtOH 60 0 C Water Wash Water 60 0 C STRIP Ethanol Heat Electricity EtOH Water RECT FILTER Bottoms CHP Biogas Flue gas Flue gas cleaning Water treatment CO2 SSF Buffer, H2SO C ph 4.8 Yeast, 3 g/l 10% WIS 20 FPU/g cellulose Water Salt HE HE Figure 9: Flow diagram for the organosolv process Raw material The raw material feed used for the calculations is a type of softwood called Pinus Contorta, the composition is presented in table 8 [28]. The size of the wood chips fed to the pretreatment is 2.5x2.5 cm - 5x5 cm and 0.5 cm thick Pretreatment reactor The reactor is run in batch mode at 170 C, where each batch is 60 minutes. The feedstock is mixed with ethanol to a concentration of 65 wt% and sulfuric acid with a concentration of 1.1 wt%. The amount of solvent is based on the amount of feedstock with the ratio of 10 L solvent/kg dry feedstock. The pressure in the reactor is 10 bars. The reactions that occur in the pretreatment are the same as those specified in the general calculation part. The fractional conversion of the reactions is based on the known amounts of the different components in the water soluble fraction (WSF) and the ethanol organosolv lignin fraction (EOL) 34

42 based on the first reference article, see table 9. The amounts in the table are based on a feed of 100 kg into the pretreatment reactor. The rest of the components are found in the solids fraction. For the calculations of the fractional conversions see Appendix B.2.1. In the calculations all lignin was assumed to be dissolved. Table 8: Composition of the softwood (Contorta) used for the organosolv calculations Component Amount (%) Ash 5.08 Klason lignin Acid-soluble lignin 0.47 Glucan Mannan Xylan 6.55 Galactan 2.07 Arabinan 1.29 Table 9: Amount of components in the WSF and EOL streams (based on a feed of 100 kg) Component Amount (kg) WSF Glucose 4.19 Acid soluble lignin 4.76 Xylose 3.23 Mannose 5.37 Galactose 1.73 Arabinose 0.78 Furfural 2.06 HMF 2.12 Levulinic Acid Formic Acid 1.5 EOL Ethanol organosolv lignin Total Regeneration of solvent Since large amount of ethanol is used as solvent, efficient recycling of the solvent is crucial. After the pretreatment reactor there is a filter which separates the solids from the liquid. The liquid mainly contains solvent but also the soluble lignin from the pretreatment. This filter separation is needed so the liquid can enter a flash vessel. The liquids are further fed to a flash vessel where some of the ethanol and water are flashed off. The liquid (bottoms) from the flash contains precipitated lignin, ethanol and water. One more separation in a filter is carried out where lignin is separated, and further dried in a dryer until a dry matter of 90 wt% is reached. The liquids of this second filter are fed to a stripper column for recycling of ethanol. The overheads of the stripper have an ethanol concentration of about 70 wt%. The concentrated ethanol is mixed with the vapor from the flash vessel and the mixture can be reused in the pretreatment. The stripper bottoms, which mainly 35

43 contains water and some ethanol, needs in reality further purification. Hence it needs to be fed to the biogas plant, which was not considered in the calculations. This stream consists of small amounts of ethanol: hence this stream contributes to some ethanol losses Washing Before the solid fraction enters the SSF it needs to be washed with hot water, but the dissolved lignin that is still present in the solid fraction would precipitate in contact with water. The accessibility for the enzymes to glucan would in this case be lowered. To avoid precipitation, the solid fraction is washed several times with hot solvent, i.e. ethanol, which is followed by the hot water wash. The solvent used for washing is mixed with the WSF and distilled so it can be reused in the process Simultaneous Saccharification and Fermentation In the organosolv process the SSF reactor was simulated in two separate unit operations which was only for the calculations, i.e. in reality it only one unit. The hydrolysis was performed with a consistency of 10% WIS and an enzyme loading of commercial cellulase with an activity of 20 FPU/g cellulose. A ph of 4.8 in the reactor is obtained by addition of sulfuric acid. The reactions in the hydrolysis are those described in the general calculation part. The conversion from glucan to glucose is 65 % [54]. The same conversion was assumed to be valid for the reaction from mannan to mannose. The reactions in the fermentation step are the same as the ones that are presented in the general process description. The conversion from sugar to ethanol, 0.9, is the same for both glucose and mannose. This is achieved at a temperature of 36 C. The concentration of yeast is 3 g/l. The residence time of the SSF is 32h Distillation The number of trays in the rectifier was set to 12 to obtain the desired concentration of ethanol, 93.4 wt%, which is near the azeotropic composition. The number of trays was determined by studying the concentration profiles in the column Biogas In the calculations all formic and levulinic acids where separated before the digestion plant, since great quantities of those are toxic to the microorganisms CHP The CHP plant in the organosolv process differs from the other processes in the matter of district heat. Since the lignin is separated in the process and obtained as a co-product, it is not burned and used as fuel in the CHP. Hence the feed to the CHP-plant mostly consist of methane from the biogas plant, non-reacted biomass, some sugars etc. During the calculations it was found that the output heat from the CHP-plant was not sufficient enough to run the process. For this reason extra biomass was fed to the CHP plant to supply the process with all heat needed internally. This resulted in basically no district heating output from the process, but some heat exchangers in the process were still regarded as possible district heat sources; i.e. heat exchanging the rectifier overheads and bottoms. This is described further in section 4.2 Energy evaluation of the processes. 36

44 2.4 Process description - sulphite pretreatment process Most of the data, e.g. composition of the feed, conversions, concentrations, for the sulphite pretreatment process calculations was found in an article based on a study by Zhu et al. [55]. In the reference article different concentrations of sulphuric acid and sodium bisulfate were investigated. The concentrations chosen in this study are 2.2 and 8.0%, respectively hence it is the yields based on those concentrations from the article that have been used. The reason why those concentrations were chosen is because they were presented as optimal in the reference article. The process scheme with the chosen unit operations is presented in figure 10. The process scheme as well as the conditions and assumptions made for calculations are described in the following section. Feed H2SO4 NaHSO3 Reactor 180 C, 12.5 bar, 25 min Flash 1 atm 100 C Ultra filtration Dryer Hydrolysate Lignosulfonates FILTER Disk mill FILTER Water STRIP CO2 SSF 35 C ph 4.8 Wis conc. 8 wt% Solid fraction NaOH Water Enzymes 15 FPU/g WIS Yeast 2 g/l HE Heat Electricity RECT Ethanol CHP Flue gas Flue gas cleaning HE STRIP Bottoms from rectifier FILTER Solids DM= 40% Biogas Water treatment Figure 10: Process scheme for sulphite pretreatment Microorganisms Salt Water Raw material The raw material used in the simulations consists of wood chips of the softwood type lodgepole pine. The wood chips used in the experiments have a length between 6-38 mm and a thickness between 3-8 mm. Table 10 shows the composition of the raw material. The dry matter content was assumed to be 50 %. 37

45 Table 10: Composition of the dry matter of the lodgepole pine used in sulphite pretreatment calculations Component Amount (%) K.Lignin Arabinan 1.56 Galactan 2.23 Glucan Xylan 7.63 Mannan Ash Pretreatment reactor The pretreatment reactor is run in batch mode at 180 C for 25 minutes. The wood chips are cooked in the reactor vessel together with sodium bisulfite with a concentration of 8.0 wt% and sulphuric acid with a concentration of 2.2 wt%. The ratio between the pretreatment liquor and the oven dried wood is 3. The pressure in the reactor is 12 bars. The conversions in the pretreatment reactor are were based on the composition in the hydrolysate stream, see table 11, which is given in the reference article. The amount of respective component is given as the weight corresponding to the polymer of the sugar. The calculations of the conversions are presented in Appendix B.3.1. The reactions used compared with the general process description are those including the components in the hydrolysate stream. The reaction below was added to describe the dissolution of lignin that occurs in the pretreatment. Lignin Lignin (sol) Table 11: Composition in the hydrolysate stream after pretreatment based on a feed of 1000 kg dry matter Component Amount (kg) K.Lignin (dissolved) 81.8 Furfural as pentosan 8.9 HMF as hexosan 4.4 Glucose as glucosan 21 Xylose as xylan 23.4 Mannose as mannan 67.7 The product from the pretreatment is flashed at atmospheric pressure when leaving the pretreatment reactor Separation of hydrolysate, lignosulfonates and solid fraction After the flash vessel the hydrolysate (liquid) is separated from the solids in the first filter. The solids are fed to a disk mill, due to the fact that size reduction of the wood chips is needed prior to the SSF in order to increase the digestibility of cellulose. The liquids are fed to ultrafiltration for further purification of the lignosulfonates since this is a valuable co-product. Ultrafiltration is not considered in the reference article. The concentration of the lignosulfonates after the ultrafiltration was set to be 20 wt%. After the ultrafiltration the lignosulfonates are dried until a dry matter content of 90 wt% 38

46 is reached. The lignosulfonates from the drier can be sold and the liquid from the ultrafiltration is fed to the biogas production. Furthermore, water is added in the disk mill to reach a solid content of 20% in this unit. The disk mill was not included in the simulations since numbers of the energy consumption for the given size reduction are given in the reference article. The main energy needed is supplied from the CHP, extra electricity is purchased. After the disk mill the slurry is filtered in the second filter and the water content is reduced until a solids content of 30% is reached. The solid fraction is fed to the SSF. The liquids from both filters are fed to ultrafiltration. In reality the liquid stream from the second filter also contains some of the lignosulfonates and washing of the solids is probably needed in the filter to utilize all lignosulfonates. This was however, not considered in the calculations Simultaneous Saccharification and Fermentation The SSF was simulated as one unit including both the hydrolysis and fermentation. The SSF is run in batch mode for 72 h. The temperature in the reactor is 35 C. The enzymes used are a commercial mixture of cellulase with an activity of 15 FPU/g substrate. The yeast has a concentration of 2 g/l. The consistency in the reactor is 8% WIS, which means that water is added prior to the SSF to reach this concentration. Sodium hydroxide is also added to obtain a ph of 4.8 in the SSF reactor. The ph before the SSF is 1.9 according to the reference article. For calculations of the added sodium hydroxide, water, enzymes and yeast, see appendix B.3.2. The reactions simulated are the same as the reactions presented in the general process description except for the reaction forming xylose. This reaction does not make any difference since the pentoses are not fermented, and the conversion was not found in the reference article. The conversion from glucan to glucose is 90% according to the reference article. The conversion from mannan to mannose is not given, but was assumed to be the same. The fermentation efficiency is 80.7 % of theoretical ethanol yield and this is the same for both glucan and mannan. The conversion from furfural to furfurylalcohol was assumed to be 100% Distillation The distillation was simulated as described in the general process description except from that the rectifier column has 16 trays instead of 36. This was found by changing the number of trays to find the number that gave the lowest heat duty in the reboiler. The bottoms from the strippers are filtered and the solid part is fed to the CHP and the liquid is fed to the biogas production. In the strippers, only 18 trays were used. This was decided after evaluation of the concentration profiles in the column. 39

47 3. Process Economics Economical evaluations were carried out for the steam explosion, organosolv and sulphite pretreatment processes. The study and result for ionic liquids pretreatment are presented separately in section 5. The process economics was calculated using the Ulrich-method. Firstly design specifications were identified and calculated for the unit operations. From the design specifications the module cost (C BM ) for each unit operation was obtained by diagrams of the database, see appendix D. The total plant costs could then be calculated by summing all the module costs and multiplying those with a mark-up factor of unexpected and contractor costs. The calculations followed with conversion of currency, and time-updating of the costs, and finally an investment calculation was performed. The detailed calculations are presented in appendix E. The ethanol price was calculated as the minimum ethanol selling price, MESP, i.e. the price of ethanol when the revenues are the same as the expenditures. The expenditures were the variable cost (raw material, chemicals etc.), operational cost (staff, administration etc.), stocking and periodized investment cost. The revenues were calculated from the produced ethanol, district heating, electricity and for organosolv and sulphite pretreatment also from the lignin and lignosulfonates respectively. The calculations are presented in appendix F. A sensitivity analysis was performed to examine the influence of different parameters on the ethanol. In table 12, the prices used for chemicals, feed, district heat and electricity is presented. Table 12: Price for variable costs. Variable cost DM feed (wood) SO 2 NaOH H 2 SO 4 NaHSO 3 Enzymes Yeast Electricity District heating Lignin Price 0.53 SEK/kg 1.5 SEK/kg 1.5 SEK/kg 1.5 SEK/kg 2.2 SEK/kg 4.3 SEK/kg 8 SEK/kg 350 SEK/MWh 280 SEK/MWh 1.5 SEK/kg 40

48 3.1 General assumptions All the unit operations were assumed to be made of stainless steel unless otherwise is stated. The following units/ chemicals of the processes were not included in the economy calculations: Chipping of the wood Extra pumps needed Dehydration of the ethanol from the rectifier overheads Anti-foam chemicals (for SSF) Flue gas cleaning The reason why the listed units/ chemicals were left out was because they do not differ between the processes, and the purpose of the project is to evaluate which one of the pretreatment methods that is most profitable, e.g. a comparison is of greater interest than an accurate final ethanol price The storage of raw materials and products was assumed to be for 1 week production SSF reactor It was assumed that each SSF reactor had a maximum volume of 2000 m 3. Hence the number of parallel SSF reactors needed was obtained in the calculations. The liquid part of the SSF was assumed to be 75% of the total volume. The time needed in each SSF reactor is the sum of the residence time for the batch and the emptying time. It was assumed that emptying and filling took 24 h including sterilization Distillation columns Doughnut trays were assumed to be used in the stripper columns and simple sieve trays were used in the rectifier column. The doughnut trays (packing) were assumed to be twice as expensive as the sieve trays. The distance between the trays was set differently depending on the diameter (d) of the columns as follows: d=[1-1.5] m, distance between trays=0.6 m d> 1.5 m, distance between trays=1 m Heat exchangers Generally five different heat exchanging scenarios are valid in the processes and different heat transfer coefficients and type of heat exchangers were assumed to be valid for each scenario as according to the following: Liquid-liquid heat exchange for pure compounds, (e.g. ethanol-water) k=1500 W/m 2 /K, plate heat exchanger. Liquid-liquid heat exchange with solids and sugars present in one of the liquids (e.g. glucan, glucose, mannan, mannose, lignin etc.) k=750 W/ m 2 /K, plate heat exchanger. Liquid-condensing vapor heat exchange (e.g. in the condensers of the distillation columns) k=2000 W/m 2 /K,tubular heat exchanger. Liquid-boiling liquid heat exchange for pure compounds (e.g. in the rectifier reboiler) k=2500 W/m 2 /K, tubular heat exchanger. Liquid-boiling liquid heat exchange with solids and sugars present in the liquid that is not boiling (e.g. in the stripper reboilers) k=1250 W/m 2 /K, tubular heat exchanger. 41

49 3.1.4 Biogas plant The cost of the biogas plant was calculated based on the amount of COD in the feed. The cost was 4.8 M$/ton COD/h [56]. For calculations see appendix D CHP plant The district heat from the CHP plant is assumed to be used for 6 months per year (during the winter season). The district heat is supposed to be delivered at 100 C, but was calculated at 60 C during the simulations. To compensate for this only half of the calculated district heat was assumed to be used. The cost of the CHP plant was calculated based on the total feed to the boiler. The cost was dollar/ton feed/h [56]. For calculations see appendix D. 42

50 4. Results 4.1 Material and energy balances Relevant results from the material and energy balances are presented in table 132 for the studied pretreatment methods that where computed in Aspen Plus. Calculations can be found in appendix A. For the organosolv process, extra raw material as feed supplied to the CHP-plant is included in the table. The amount of fed biomass into the pretreatment of the organosolv process was ton/h and the extra biomass fed to the CHP was ton/h, those two streams add up to the presented feed in the table. The ethanol yield (liter ethanol per ton dry matter) is based on the total amount of fed biomass. The amount of theoretical ethanol production, based on the reference articles, is only based on the amount of biomass fed to the pretreatment reactor. The amount of electricity and district heat produced from each process is the total output; i.e. heat and electricity integration of the processes was carried out. Section 4.2 Energy evaluation of the processes shows a description of how the integrations where carried out. The yields used for calculating the theoretical yield of ethanol were: Yield coefficient glucose (sugar)-ethanol 0.51 kg/kg (same for mannose-ethanol) Yield coefficient glucan-glucose (sugar) 1.11 kg/kg (same for mannan-mannose) Table 13: Result for material and energy balances steam explosion, organosolv and pulping based ethanol production. Steam explosion Organosolv Mass flow (ton/h) Energy content (MW) Heating value (MJ/kg) Raw material (dry matter) Product, ethanol Electricity District heating Enzyme comsumption 3.43 Yeast consumption 0.28 Ethanol production: 305 l/ton dry matter Theoretical ethanol production: 13.2 ton/h Yield (percentage of theoretical): 74.6 % Concentration of ethanol into the stripper column: 4.66 wt% Feed (dry matter) Product, ethanol Electricity 0 District heating 66.9 Enzyme comsumption 5.47 Yeast consumption 1.52 Lignin produced (90 wt%)

51 Ethanol production: 185 l/ton dry matter Theoretical ethanol production: 20.1ton/h Yield (percentage of theoretical): 47.2 % Concentration of ethanol into the stripper column: 2.10 wt% Feed (dry matter) Product, ethanol Sulphite pretreatment Electricity 0 District heating Lignosulfonates Enzyme 5.34 comsumption Yeast consumption 1.23 Lignosulfonates produced (90 wt%) 4.72 Ethanol production: 241 l/ton dry matter Theoretical ethanol production: 16.5 ton/h Yield (percentage of theoretical): 60.0 % Concentration of ethanol into the stripper column: 2.05 wt% The required raw material feed is largest for the organosolv process, 67.7 ton dry matter/h, followed by the sulphite pretreatment, where 51.9 ton dry matter/h is needed. The steam explosion process requires the smallest amount of feed, ton dry matter/h. The reason why a larger raw material feed is needed for the organosolv process is mainly due to the low conversion from glucan to glucose which was 65 % (same for mannan-mannose). This can be compared with the conversion used for steam explosion where the conversion from glucan to glucose was 83.6 % (mannan-mannose 29.8 %) and for the sulphite pretreatment process the conversion from glucan to glucose was 90 % (same for mannan-mannose). Furthermore extra biomass is fed to the CHP plant in the organosolv process. The electricity output is zero for both the organosolv and sulphite pretreatment processes due to large internal needs for electricity. For steam explosion the electricity output from the process is MW. The district heating from steam explosion, organosolv and sulphite pretreatment are 75.62, 66.9 and MW, respectively. Hence the steam explosion process produces the largest amount of both district heat and electricity. The energy consumption is dependent on the WIS concentration. Higher WIS leads to less water that needs to be evaporated in the distillation columns, but if it is too high it can be toxic to the yeast which sets the limit. The WIS content in the SSF in steam explosion, organosolv and sulphite pretreatment was 10 %, 10 % and 8 % respectively. Hence it would be beneficial for the sulphite pretreatment process to increase the WIS with respect to energy consumption. 44

52 The conversion obtained has to correspond to a certain experiment, which is why the yields not could be increased in the calculations. However for steam explosion the WIS concentration in the reference article is 5 % and the used value was as mentioned 10 %. This change was made due to the study [22]. The enzyme consumption is largest for the sulphite pretreatment process followed by the organosolv process, with consumptions of 5.47 and 5.43 ton/h respectively. The enzyme consumption of the steam explosion process is lowest, 3.43 ton/h. The enzyme consumption is crucial for the process economics. The reason for the large enzyme consumption is due to the feed into the SSF which is greatest for organosolv and the sulphite pretreatment processes. The enzyme concentrations used in the SSF reactor are 15 FPU/g WIS for steam explosion, 20 FPU/g cellulose for organosolv and 15 FPU/g WIS for the sulphite pretreatment. Since the amount of enzymes is based on the amount of cellulose for organosolv, this number is lower if it is set per g WIS since e.g. lignin also is included in WIS. This means that the enzyme concentration in the SSF reactors is approximately the same for the processes. Regarding the percentage of theoretical yield, based on fermentable sugars present as polymers in the biomass, the steam explosion process had the highest yield of 74.3 %. The sulphite pretreatment process had 60 % and the organosolv process had a yield of 47.2 % of theoretical. The concentration of ethanol into the distillation columns is crucial for the energy consumption of the processes. Steam explosion had the highest inlet ethanol concentration of 4.66 wt % (which correspond well to the introduction section) to the strippers, while the organosolv and sulphite pretreatment had concentrations of 2.10 and 2.05 wt% respectively. These concentrations are low. When the WIS concentration in the SSF reactor was set to 5 % for the steam explosion process, the inlet ethanol concentration to the strippers was approximately 2 wt%. Hence the WIS concentration is important as described above Inhibitors The inhibitors present in the steam explosion, organosolv and sulphite pretreatment processes are presented in this section. The inhibitors present in the steam explosion process where furfural and HMF. The reason why more inhibitors were not considered was because no yields were found in the reference article. The mass fraction into the SSF of the inhibitors where: % furfural and 0.49 % HMF in the liquid phase. Inhibitors present in the organosolv process where furfural, HMF, levulinic acid and formic acid. All furfural was set to be oxidized by the yeast in the SSF reactor and hence present as furfuryl alcohol in the outlet stream. The mass fraction into the SSF of inhibitors where: 0.16 % furfural, 0.3 % HMF, 1.31 % levulinic acid and 0.51 % formic acid in the liquid phase. In the sulphite pretreatment the only inhibitors present were furfural and HMF since only those yields were found in the reference article. The mass fraction into the SSF of the inhibitors where: % furfural and % HMF in the liquid phase. The reason why the inhibitor content is low for the sulphite pretreatment process is due to that most of the inhibitors are present in the hydrolysate stream, which is not considered to be fermented. 45

53 MW 4.2 Energy evaluation of the processes To obtain an overview of how the energy in the raw material feed is distributed in the products, energy diagrams were developed for the processes. The lower heating value for ethanol was set to 27.1 MJ/kg and 19 MJ/kg for the raw material, spruce. Regarding the electricity, it is in this section presented as comparable with heat; i.e. an efficiency of 35 % was used (i.e. the electricity output from the processes was divided by 0.35 to convert it to a corresponding heat value). The energy efficiency was defined as energy output in products (i.e. ethanol, electricity, district heating and for sulphite pretreatment also the lignosulfonates and for the organosolv pretreatment also the lignin) divided by the energy input in the raw material feed. In all of the processes no consideration was made for electricity usage for e.g. stirring in the SSF, and in reality also more pumps are needed. This means that in reality more electricity will be needed in the processes and hence the presented electricity output is a bit high. This was assumed since it is valid for all the processes and should therefore not influence the comparison. The calculations and assumptions are presented in appendix C Steam explosion The energy diagram for steam explosion is presented in figure 11. Most of the input energy is found as district heating and ethanol, and the losses of the process are rather low. 250 Steam explosion Input Output Product, ethanol (MW) 74,25 District heating (MW) 75,62 Electricity (MW) 54,37 Feed, biomass (MW) 216,7 Figure 11: Energy diagram steam explosion pretreatment To obtain the total electricity output from the process, electricity needed for the included pumps and compressors was subtracted from the electricity output from the turbines in the CHP plant. Table 14 46

54 MW shows the energy content in the major products together with the percentage of energy distribution with respect to the energy content in the inlet biomass feed. The largest contribution is district heating (34.9 %) followed by the ethanol (34.3 %). The loss of energy in the process is 5.73 % which is reasonable. The energy efficiency of steam explosion process was 94.3 %. Table 14: Result for distribution of energy input in products, steam explosion Energy (MW) % of input Ethanol Electricity District heating Losses Organosolv Figure 12 shows how much of the energy in the raw material feed that is found as electricity, ethanol, district heating and lignin. The energy in the ethanol, electricity and district heating does not add up to the energy content in the feed due to losses. The reason why there are two contributors to the raw material input is because extra biomass was fed to the CHP plant to provide the process with enough heat (steam) needed for the process. The reason for this was that the non-reacted solids, sugars, methane from the biogas plant etcetera did not supply enough energy to the process. The lower heating value used for the lignin is 26.6 MJ/kg Organosolv Input Output Lignin (MW) 119,3 Product, ethanol (MW) 74,25 District heating (MW) 66,89 Electricity (MW) 0 Feed CHP, biomass (MW) 95,77 Feed, biomass (MW) 329,5 Figure 12: Energy diagram organosolv pretreatment 47

55 The electricity needed to run pumps, compressors and the dryer (for drying the lignin) was subtracted from the electricity output from the turbines in the CHP-plant. The electricity demand was not covered by the produced electricity, extra electricity is needed to be externally purchased to the process, and hence the electricity output was set to zero (in fact 6.66 MW is needed to be purchased). The single greatest contributor to the required internally electricity is the dryer. The main sources for the heat regarded as district heat comes from the heat exchanger of the rectifier overheads (the ethanol product) and from the heat exchanger from the rectifier bottoms; where both the bottoms and the top product are condensed and cooled to 60 0 C. Table 15 shows the energy content of the main products together with the percentage of energy distribution with respect to the energy content in the inlet biomass feed. As inlet biomass feed, only the feed into the pretreatment was considered (not the extra biomass added to the CHP plant). There is a loss of MW which corresponds to 21.0 % of the feed input. This loss is greater if furthermore the extra biomass feed to the CHP plant is considered. Most of the energy is found as lignin (36.2 %) followed by the ethanol (22.5 %). The energy efficiency of organosolv process was 79.0 %, which is lower if the extra biomass feed into the CHP is considered. Table 15: Result for distribution of energy input in products, organosolv Energy(MW) % of input Ethanol Electricity 0 0 District heating Lignin Losses Sulphite pretreatment The energy diagram for sulphite pretreatment is presented in figure 13. The lower heating value used for the lignosulfonate is 26.6 MJ/kg. Most of the energy input is found in the ethanol, followed by district heating. 48

56 MW Sulphite pretreatment Input Output Lignosulfonates (MW) 31,38 Product, ethanol (MW) 74,25 District heating (MW) 45,11 Electricity (MW) 30,08 Feed, biomass (MW) 274 Figure 13: Energy diagram sulphite pretreatment When the electricity output for the sulphite pretreatment was calculated, electricity usage for pumps and compressors was subtracted from the CHP turbine output (as for steam explosion and organosolv). But unlike the other pretreatment methods also the electricity usage in the disk-mill, ultra filtration and the dryer was subtracted since those where regarded as great consumers. The dryer (superheated steam dryer) was the greatest consumer of electricity, and the presence of the dryer resulted in no electricity output for the overall process. In fact, the electricity produced in the CHP was not enough to run the process, and hence electricity is needed to be bought externally (9.83 MW). Hence, the electricity output was set to zero. Table 16 shows the energy content in the product and the co-products together with the percentage of energy distribution with respect to the energy content in the inlet biomass feed. The largest energy contribution is ethanol which contains 27.1 % of the energy input, followed by district heating which contains 16.5 % of the energy input. The losses of the process are 45.0 % and the energy efficiency of the sulphite pretreatment process was 55.0 %. Table 16: Result for distribution of energy input in products, sulphite pretreatment Energy(MW) % of input Ethanol Electricity 0 0 District heating Lignosulfonates Losses

57 MW 4.3 Internal energy consumption To get an overview of which units in the processes that contributes to the internal energy consumption figure 14 was constructed. Only the greatest contributors were considered; i.e. the energy demand of the distillation columns, dryers and pretreatment. The electricity for pumps and compressors was not considered since it was basically the same for the processes. To be able to compare the amount of electricity and heat consumed, an efficiency factor of 0.35 was used to convert the electricity in a unit comparable to heat (since electricity a more valuable sort of energy than heat). Internally energy consumption Steam explosion Organosolv Sulphite pretreatment Dryers (MW) ,17 Pretreatment (MW) 7 66,57 58,6 Destillation columns, total duty (MW) 0 31,33 49,88 46,11 Figure 14: Internal energy consumption of the processes The units included as pretreatment in steam explosion is the energy consumption for producing steam. In the organosolv process, the units included as pretreatment is the stripper (for ethanol recovery) and heating of the reactor. The stripper accounted for 7.32 MW and heating accounted for MW. Those two contributors adds up to the amount presented in figure 14. In the sulphite pretreatment process, the units included as pretreatment are the diskmill, the ultrafiltration and heating of the reactor. If those are to be separated, the diskmill accounted for MW, the ultrafiltration accounted for 3.92 MW and heating accounted for MW. Those three contributors add up to the amount presented in the figure. In the figure, the total energy duty for the distillation columns is shown. Furthermore the net duty is also of interest. The net duty, as further energy demand after the energy integration of the columns, 50

58 are for the steam explosion process MW, for the organosolv process MW and for the sulphite pretreatment process MW. All the internally heating and electricity demands for the processes are integrated with heat and electricity from the CHP plant. 4.4 Process economics The results from the calculated investment costs, G, for the different processes are presented in table 17. The investment cost for the steam explosion process is the lowest, MSEK, and highest for the sulphite pretreatment process, MSEK. The investment cost of the organosolv process is MSEK. K is the sum of the capital cost for the equipment for the unit operations, C BM. Pretreatment method Table 17: Result of the total plant cost calculations for the different pretreatment methods (M$) (M$) (MSEK) G (MSEK) Steam explosion Organosolv Sulphite pulping The variable cost, revenues, operational and restricted costs for the processes are summarized in table 18. Table 18: Result of the variable costs, revenues, operational and restricted costs for the different pretreatment methods Steam explosion (MSEK/year) Organosolv (MSEK/year) Sulphite (MSEK/year) Variable cost Revenues Operational cost Restricted costs (stocking) The MESP of the processes are presented in table 19. The MESP of the ethanol for the steam explosion process is lowest. The MESP for organosolv and sulphite pretreatment is calculated with the assumption that the lignin and lignosulfonates from these processes was sold to a price corresponding to the heating value; i.e. a price as if it was used as fuel and burned, which corresponds to the lowest price possible. Table 19: MESP for the different pretreatment methods Pretreatment method MESP (SEK/L) Steam explosion 5.2 Organosolv 13.5 Sulphite pretreatment

59 To obtain an overview of which costs that are the major contributors to the ethanol MESP, pie charts were constructed showing the distribution of costs and revenues for the different processes. These charts are presented in the following sections Steam explosion process The distribution of costs and revenues for the steam explosion pretreatment process is presented in figure 15. The largest contribution to the costs is the capital cost and the raw material feed, which each stands for 27 % of the total costs. The thirdly largest cost is the enzymes which stand for 18 %. When the revenues were set equal to the cost, the ethanol price was calculated to 5.2 SEK/L. The greatest income is the ethanol which contributes to 84 % of the total revenues. This is reasonable since it is the main product. Electricity contributes to 9 % of the total incomes and district heat stands for 7 %. Capital cost Feed Enzymes Operational cost Others Yeast Chemicals EtOH Electricity District heat Figure 15: Distribution of the costs (to the left) and revenues (to the right) for steam explosion Organosolv The distribution of costs for the organosolv pretreatment process is presented in figure 16. The two largest contributions to the costs are the feed and the capital cost, which stands for 23 % respectively 22 % of the total cost. Feed Capital cost Enzymes Operational cost Chemicals Yeast Others Feed CHP Electricity Figure 16: Distribution of the costs for organosolv 52

60 The distribution of the revenues is presented in figure 17. The largest contribution to the revenues is ethanol. The MESP was calculated to 13.5 SEK/L which is higher than the prices calculated for the other pretreatment methods. The price was calculated when selling the lignin for the lowest price possible, 1.5 SEK/kg. To lower the ethanol price to the same level as steam explosion, which is needed if the processes is about to be comparable, the price of the lignin must be 5.3 SEK/kg. If this is the case, then lignin would be the largest revenue of the process and contribute to 68% of the total income. EtOH 13.5 SEK/L EtOH 5.2 SEK/L Lignin 1.5 SEK/kg Lignin 5.3 SEK/kg Figure 17: Distribution of the revenues for organosolv with no extra revenue from the lignin (to the left) and with the same price of the ethanol as steam explosion (to the right) Sulphite pretreatment The distribution of costs for the sulphite pretreatment method is presented in figure 18. The largest contribution to the costs is the chemical cost. The cost for the chemicals is even greater than the cost for the raw material feed. This is unreasonable and one reason for this is due to that no chemical recovery is taken into account. The MESP for this case is 13.0 SEK/L. To get a more reasonable distribution of the cost and as well lower ethanol price the chemicals were assumed to be recovered, see figure 18. This is done in reality and in this case the cost for chemical recovery was assumed to be 20 % of the original chemical cost. The chemical cost is reduced drastically and the distribution of the costs looks more similar to the other processes, with the capital cost, feed and enzymes as the largest contributors. This is the case that has been investigated further in this study. Chemicals Capital cost Feed Enzymes Operational cost Yeast Electricity Others Figure 18: Distribution of costs for sulphite pretreatment with no chemical recovery (to the left) and with chemical recovery (to the right) 53 Capital cost Feed Enzymes Operational cost Chemicals Yeast Electricity Others

61 When the cost is reduced due to chemical recovery the price for the ethanol (MESP) was calculated to 9.0 SEK/L, see figure 19, assuming that the price of the lignin was 1.5 SEK/kg, which is the lowest lignin price possible, corresponding to a solid fuel. The ethanol is hence the largest contribution to the revenues. For the sulphite pretreatment to be competitive with steam explosion the ethanol price has to be 5.2 SEK/L. To be able to have this ethanol price, the price of the lignin has to be 11 SEK/kg, see figure 19. The revenue from the lignin is in this case almost as large as the revenue from the ethanol. EtOH 9 SEK/L Lignin 1.5 SEK/kg District heat EtOH 5.2 SEK/L Lignin 11.6 SEK/kg District heat Figure 19: Distribution of the revenues for sulphite pretreatment with no extra revenue from the lignin (to the left) and with the same price of the ethanol as steam explosion (to the right) 4.5 Sensitivity analysis A sensitivity analysis was performed for the three simulated pretreatment methods (steam explosion, organosolv, sulphite pretreatment). The effect of change in raw material cost, capital cost and enzyme cost on ethanol price was examined. These parameters were chosen since they were the largest contributors to the costs for the three processes. Each parameter was changed 50 % from the reference case, i.e. the cost was increased by 50 % and decreased by 50 % based on the original calculated cost. The result of the analysis can be seen in figures In figure 20, the cost of the feed is varied, in figure 21 the capital cost is varied and in figure 22 the enzyme cost is varied. This resulted in a varying MESP in the range of SEK/L for the steam explosion process, SEK/L for the organosolv process and SEK/L for the sulphite pretreatment process. As can be seen in the figures the ethanol price for the pretreatment methods is not in the same range. A change in the sensitivity parameters does not make the functions intercept, which means that steam explosion is still the cheapest alternative in each case, and organosolv is the most expensive when lignin is just priced as a solid fuel. For all the sensitivity parameters the organosolv process has the steepest slope, i.e. it is the most sensitive for a change. The sulphite and steam explosion processes have about the same slope in all figures and hence they have about the same sensitive to a change. 54

62 MESP(SEK/L) MESP(SEK/L) Organosolv Steam explosion Sulphite Wood cost per year(msek) Figure 20: Sensitivity analysis for a change in raw material cost Organosolv Steam explosion Sulphite Capital cost per year(msek/kg) Figure 21: Sensitivity analysis for a change in capital cost 55

63 MESP(SEK/L) Organosolv Steam explosion Sulphite Enzyme cost per year(msek/kg) Figure 22: Sensitivity analysis for a change in enzyme cost Sensitivity of price of lignin When no consideration is taken of extra revenue from the lignin/lignosulfonates, steam explosion is by far the most favorable pretreatment method with the lowest MESP. This result might be a bit misleading since the produced lignin/lignosulfonates is sold for its minimum price, i.e. it is sold as pellets just to be burnt. Figure 23 illustrates the selling price of ethanol as a function of the lignin price. To be able to compete with steam explosion, the MESP of the organosolv and sulphite pretreatment processes has to decrease to 5.2 SEK/L. The price of the lignin was calculated for this ethanol price, as is marked out as the dashed line in figure 23. Organosolv reaches the same ethanol price as steam explosion when the lignin can be sold for 5.5 SEK/kg. For the sulphite process, lignosulfonates must be sold for 11.6 SEK/kg to produce ethanol for the same MESP as steam explosion. 56

64 MESP(SEK/L) MESP(SEK/L) 20 Organosolv Sulphite Lignin(SEK/kg) Figur 23: MESP dependence of the lignin price for the organosolv and sulphite pretreatment processes 57

65 5. Ionic liquids The investigations performed here on the ionic liquid pretreatment differ from the other pretreatments. The reason for this is firstly because of the lack of chemical and physical data available for ionic liquids in Aspen Plus. Some physical data can be found for specific ionic liquids from different producers, but still information about boiling point and heat capacity is missing. Secondly, the data available does not supply enough information and the investigations made have not had the scope to develop a process. Some unit operations have not yet been investigated, i.e. some have not been thought of and some only exist in theory. Additionally, no knowledge does yet exist concerning the appropriate equipment and consequently no knowledge of the optimal unit operations and conditions of the process exists either. This and the lack of essential physical and chemical properties of ionic liquids make it impossible to simulate the ionic liquid pretreatment in Aspen Plus. Consequently it has been decided that the ionic liquid pretreatment is to be examined through rough estimates and an extended literature study and discussion of the possibilities, benefits and drawbacks of the pretreatment. 5.1 The process The following text will discuss the different unit operations needed in an ionic liquid pretreatment. From the lab scale investigations made in the area, independent of variations in execution, a few unit operations have been identified. What is common in all studies is the dissolution of a lignocellulosic material in an ionic liquid. The wood subjected to dissolving is either in the form of wood chips or wood powder, dried or not dried. Since the wood going in to the process is in the form of un-dried wood chips, a pre-pretreatment must be added for the cases when the wood going in to the dissolution is either in the form of wood powder, dried wood or both. After the dissolving step the cellulose and hemicelluloses (called intermediates) need to be separated from the ionic liquid. The ionic liquid has to be regenerated and recycled and the lignin has to be taken care of. The cellulose and hemicelluloses might need to be washed before going in to the SSF. The chemical used for the wash might in turn need to be recycled or purified. All possible unit operations are displayed in figure 24 and will be investigated in the following part. Lignin Purification of IL IL Impure IL Wood chips Prepretreatment Pretreatment Separation of intermediate and contaminated IL Intermediate Ethanol Distillation SSF Washing of intermediate Figure 24: Process scheme for production of ethanol using a general ionic liquid pretreatment 58

66 5.1.1Pre-pretreatment In most studies the wood used in the experiments is in the form of dry wood powder which implies that both milling and drying is necessary prior to the dissolution in the ionic liquid, supposedly in that order. The milling and drying would then be considered part of a pre-pretreatment since the dissolution in the ionic liquid is considered as the main pretreatment. What also can be considered a part of the pre-pretreatment is the drying of the ionic liquid. It is necessary in the cases when the ionic liquid is hygroscopic and when water negatively affects the dissolution Milling Milling alone can increase the amount of cellulose susceptible to enzymatic hydrolysis which in turn can increase the conversion from < 10 % up to < 50 % [50]. It has been stated that milling itself can be un-economical [11] and if that is the case, and if milling is needed for the dissolution to work properly, any further investigation of the ionic liquid pretreatment is unnecessary. On the other hand, milling might not be needed; wood chips have been used in at least one study [49]. Several different types of mills might be used, for example hammer mills and disk mills. The hammer mill has a lower energy demand than the disk mill, but produces coarser particles [57]. To reach the particle sizes used in most studies (e.g mm [58]) a disk mill is thus needed. The high energy demand of a mill is the factor that can make it un-economical. According to Schell and Harwood [57] the energy demand for milling of wood chips with a water content of 60 % to a particle size between 0.5 and 2.5 mm is between 140 and 600 kwh/odt (Oven Dried Ton) when using different disk mills and different operation settings. The attrition mill is the one with the lowest energy demand [57]. Disk milling is also used in the sulphite pretreatment. In a study investigating the SPORL pretreatment on lab scale the energy demand for disk milling (with water injections) is measured to between 0.12 and 1.25 GJ/ODT which corresponds to 34 and 348 kwh/odt [55]. The estimated cost for disk milling per liter of ethanol for different energy demands is presented below in table 19. The calculations and assumptions can be seen in Appendix G. The cost varies between 0.06 and 1.11 SEK per liter of ethanol. As a realistic production cost for one liter ethanol is assumed to be around 4 to 5 SEK/Liter, 1.11 SEK/Liter for milling is a quite big value. The lower operational costs are more reasonable and might make the process possible for commercialization. Table 19: Operational cost for disk milling for different energy demands Energy demand (kwh/odt) Price per liter of ethanol (SEK/L ethanol) Further investigations are needed to establish whether a smaller particle size (wood powder) is of vital importance for the pretreatment to be efficient enough to make the pretreatment and following process commercially viable. Probably, the particle size has some effect on the process. An increase in particle size should result in a reduction of the efficiency in the dissolution step and consequently a reduction in sugar conversion, but this does not necessarily make it un-commercial. It could even be that it is beneficial not to mill because of the resulting reduction in investment and operational costs. 59

67 Drying Since the dissolution for some ionic liquids is affected by the presence of water, drying of both the wood and the ionic liquid might be needed [44]. However, this is not always the case. The performance of the ionic liquid [EMIM]OAc is not affected by the presence of water up to a concentration of 10 % [59]. In many studies though, both the wood and the ionic liquid are dried until they contain less than 1 % water [58], but since it is unsure whether this is needed an investigation with the scope of examining this should be performed. Calculations were made to find the water content allowable in the wood when using the IL [EMIM]OAc, i.e. which water content in the wood that resulted in a water concentration in the IL of 10 %. These calculations showed that no drying was needed since the allowable water content in the wood was higher than the water content in the wood going in to the process. The allowable water content in the wood ranged from 53 % to 74 % for different IL/biomass ratios (10 and 25, see pretreatment). It is however not favorable to have high IL/biomass yields and drying might be needed to be able to use a lower ratio, for example if the wood contains 60 % water and the wanted IL/biomass ratio is 10. Calculations on the drying were only performed for the case where a water content of 1 % is assumed to be needed. The operational costs for a simple steam dryer would be the cost for the steam used to superheat the steam in the dryer and the electrical power demand of the fan that drives the steam and evaporated water through the process. The dryer could also be connected to a district heating network which in its turn would lower the cost. The process used for calculations is shown in figure 25 [60]. Cyclone Dryer 100 kpa Dry biomass ecirculated steam C Steam (evaporated water C Fan To district heating network Wet biomass 50 % water team C Condensate to boiler Steam 1.2 MPa To waste water treatment From district heating network Figure 25: Flow diagram for a steam dryer [60] The cost for drying to a concentration of 1 % was calculated to 0.8 SEK/L ethanol, without any calculations being made neither for the gain from the delivery to the district hea ng network nor for the opera onal cost for hea ng the wood from to C (the saturation temperature for the steam in the dryer). 60

68 It can be noted that one advantage with using a steam dryer usually is that secondary steam is formed which can be used in other parts of the process, e.g. in the distillation, decreasing the operating cost. In this case however, the pressure of the steam might need to be increased for this to be possible. The consequence of an increase in pressure would be a higher investment cost for the dryer and also a risk for leakages which might lead to odor problems. As for the ionic liquid, it can be dried e.g. in a drying oven. In experiments performed by Zavrel et al. [49] the drying of the ionic liquids was performed for ten hours at 80 C in a drying oven and resulted in concentrations of 0.7 % water [49]. However, for the large process streams needed industrially an evaporator probably would be more appropriate Pretreatment The yields reached in pretreatment with ionic liquids vary greatly. They depend on type of process, temperature, concentrations, type of wood, particle size, water concentration and type of ionic liquid. They range between around 30 % and 99 % or more. No conclusion can hence be made of the yields in general for the ionic liquid pretreatment. The operational cost for the pretreatment step is represented by the cost of the ionic liquid and the energy demand for heating and stirring. The ionic liquids are currently expensive and must not be used to a larger extent than needed. However, no reports have been found stating the optimal ionic liquid/wood ratio and this hence needs to be investigated. Most experiments have operated with ratios around 20, but there are a few that differ. In the techno-economic report made by Klein- Marcuschamer et al. [51] the ratio has been varied between 1 and 10, but the scientific background to these values is unclear. A dissolution performed by Li et al. [58] uses a ratio of Higher ratios have also been used, e.g. in experiments executed by Miyafuji et al. [61] with a ratio of 33. According to BASF [45] the expected price for ionic liquids, after the beginning of large-scale production, should be around $40 per kg IL. Klein-Marcuschamer et al. [51] have based their calculations on prices ranging from $2.5 to $50 per kg ionic liquid. With the presented numbers the highest possible cost for the ionic liquid is found for an ionic liquid to wood ratio of 25, an ionic liquid cost of $50 per kg ionic liquid and when no IL is recycled. This results in a cost of approximately SEK/L ethanol. The lowest possible cost for the ionic liquid without recycling is then represented by the cost calculated from an ionic liquid to wood ratio of 10 and $2.5 per kg IL. The lowest possible cost for the IL without recycling would then be approximately 540 SEK/L ethanol. Both these scenarios result in costs which make the ionic liquid pretreatment impossible to use industrially, at least for the production of ethanol. Consequently, it can be concluded that recycling of the ionic liquid is a must. If the ionic liquid is recycled the operational cost would be the loss of ionic liquid due to that the efficiency of the recycling is not 100 %. The operational cost for the pretreatment would then only depend on the efficiency of the recycling. Operational costs for different recycling efficiencies are shown in table 20. For information on the recycling, see Recycling and purification of the ionic liquid. 61

69 Table 20: Operational costs for the pretreatment step for different recycling efficiencies (lowest possible price; IL/wood ratio of 10 and $2.5/kg IL) Recycling efficiency Operational cost (SEK/L ethanol) 94 % % % % % 2.16 As can be seen the cost can vary between 32.4 and 2.16 SEK/L ethanol for a recovery of 94 to 99.6%, respectively. The chemical cost, including all chemicals and yeast, for the steam explosion process is approximately 0.3 SEK/L ethanol. If compared to this number it can be concluded that even if the optimistic efficiency of 99.6 % for the IL recycling is applied the cost is still too high. The energy demand for heating is determined by the temperature needed in the pretreatment. The temperature needed varies between different ionic liquids and it should also vary for different particle sizes and residence times (larger particle sizes should require higher temperatures; lower temperatures should require longer residence times). The temperature also varies between the different types of pretreatment; dissolution, extraction and simultaneous pretreatment and hydrolysis. For the simultaneous pretreatment and hydrolysis the temperature needs to be adjusted to the enzymes, if enzymes are used as catalysts. There is also a difference between hardwood and softwood. More intense conditions, i.e. higher temperatures and longer residence times, are needed to pretreat hardwood because of its higher density. The higher density makes it more difficult to open up the structure [58]. To save energy the optimum temperature for each case should be found, hence requiring more investigations in the area before commercialization. The energy demand for stirring varies with the viscosity of the ionic liquid. An IL with lower viscosity is hence preferred. It might be necessary to perform the pretreatment in an inert atmosphere to prevent the hygroscopic ILs to absorb water from the air [62]. The pretreatment can be performed at atmospheric pressure Separation of intermediate and contaminated ionic liquid After the pretreatment the desired intermediate for further processing needs to be separated from the ionic liquid. The way of separation differs between the different pretreatment options. When the entire wood has been dissolved the cellulose and hemicelluloses have to be precipitated to be retrieved. As described before, this can be performed by adding an anti-solvent such as water, methanol, ethanol or acetone. The precipitated cellulose can obtain different forms; monoliths, fibers and films [44]. In any case the intermediate is in the form of a solid and thus a solid-liquid separation can be performed to separate the precipitated cellulose and hemicelluloses from the ionic liquid and anti-solvent. Suitable for this is a filter or a centrifuge [44]. In the same way the remaining wood after extraction of lignin can be separated from the IL. In the extraction of lignin some of the cellulose and hemicelluloses are dissolved in the ionic liquid. These are also precipitated through the addition of an anti-solvent and can also be separated by filtration or centrifugation. For the other pretreatment techniques it has been assumed that the separation by filtration results in two streams, one containing only liquid and the other containing 40 % dry matter and 60 % liquid. In this case a lot 62

70 of ionic liquid would then accompany the intermediate further in the process. This is not desirable firstly since the ionic liquid is expensive and secondly since the IL has an inhibitory effect on the enzymes in the subsequent SSF. The ionic liquid could be removed from the process stream containing the intermediate through the addition of water to the filtration. The amount of water that needs to be added varies between 1.5 to 2 times the amount of ionic liquid that needs to be removed from the stream. One third to half of the added water would then be found in the stream with the washed out ionic liquid. The streams going out from the filter would then comprise of one stream containing 40 % dry matter (precipitated cellulose and hemicelluloses and/or delignified wood) and 60 % water. The other stream would contain ionic liquid and water. In the case of a simultaneous pretreatment and hydrolysis the substances needed to be separated are ethanol, sugars, enzymes, acids or bases (which ever needed to adjust the ph) and ionic liquid. This can be performed, at least on a smaller scale, with an ion-exclusion chromatograph. In this case the recycling of ionic liquid and the separation of the ionic liquid from the intermediate can be performed in the same process step. Values for the recovery of the different substances after the ion-exclusion chromatograph from experiments performed by Binder and Raines [52] can be found in table 21. The use of an ion-exclusion chromatograph also results in another benefit, the removal of the inhibitory compounds HMF and furfural which are formed during the simultaneous pretreatment and hydrolysis [52]. Table 21: Experimental values for the recovery after an ion-exclusion chromatograph Recovery (%) Ionic liquid >95 Glucose 94 Xylose Washing of intermediate If washing is not performed in the separation of intermediate and contaminated ionic liquid and the stream containing the intermediate still contains IL, it has to be washed. This is due to that the IL has an inhibitory impact on the enzymes. The wash is performed by addition of an anti-solvent. Investigations made on the simultaneous pretreatment and hydrolysis show that the inhibitory effect on the enzymes vary with different ILs and different IL concentrations. For example, the presence of the ionic liquid [MMIM]DMP decreases the enzyme activity with 70 % at an ionic liquid concentration of 10 %. The enzyme activity is probably affected negatively by an increase in viscosity and ionic strength. The hydrogen bond basicity of the ionic liquid also affects the activity; a lower hydrogen bond basicity protects the enzymatic stability more effectively [46]. As has been concluded for the other pretreatments that have been examined, the cost for enzymes is quite large. A decrease of enzyme activity, which leads to that more enzymes need to be supplied, is consequently not desirable. Removal of ionic liquid is then needed. For the dissolving of wood it is the precipitated cellulose and hemicelluloses that need to be washed. As for extraction of lignin the remaining wood is the intermediate going forth towards the hydrolysis and thus is what needs to be washed. In the case of a simultaneous pretreatment and hydrolysis washing of the intermediate is neither needed nor possible. 63

71 5.1.5 Recycling and purification of the ionic liquid As the price for ionic liquids is high the reuse of it is essential to make the process economical, as concluded in Pretreatment. Different processes contain different intermediates and might thus require different separation techniques. As described before an ion-exclusion chromatograph could be used in the case of a simultaneous pretreatment and hydrolysis for the purification of the ionic liquid [52] (see Separation of intermediate and contaminated ionic liquid). As for the other methods; dissolving of wood and extraction of lignin, the most used technique is anti-solvent washing followed by filtration and the removal of the anti-solvent. The anti-solvent wash is another name for the precipitation of cellulose and hemicelluloses. Thus, the first and second steps of this recovery cycle have already been described. The removal of the anti-solvent can be achieved through different techniques of evaporation. Since the IL is a low-boiling substance an evaporator can be used, e.g. a rotary evaporator. Different types of dryers might also be appropriate, perhaps in combination with the evaporator [58]. The mixture might also be fractionated through distillation [44]. In experiments using a rotary evaporator followed by drying under vacuum at 40 C it was found that up to 100 % of the IL could be recovered, although this depended on several process parameters [58]. The type of anti-solvent used is of importance to the efficiency of the recycling. The amount of recovered ionic liquid reaches 100 % only after the first cycle when water is used as anti-solvent. After that, it decreases rapidly. If methanol is used the amount of recovered IL decreases from 99 % after the first cycle to 96 % after the fourth cycle, i.e. the decrease in efficiency is lower for methanol. It can be noted that through the cycles lignin is accumulating in the IL, lowering its dissolving efficiency. It might also be the reason for the decrease in recovery efficiency of the IL. Regarding the choice of anti-solvent, the efficiency of the pretreatment and the subsequent hydrolysis reach higher values when water is used than when methanol is used as anti-solvent. Consequently, methanol might be preferred in the case of a large scale process to lower the cost for loss of IL. On the other hand methanol is more expensive than water [58]. It should be noted that the choice of anti-solvent also affects the energy demand for the recycling of the IL. Since the anti-solvent is to be evaporated a lower boiling point and heat of evaporation is wanted since it results in a lower cost for the evaporation. This would again promote the use of methanol as anti-solvent. An obvious drawback of methanol is nevertheless that it is toxic whilst water is not. Other possible anti-solvents are ethanol and acetone. Ethanol is not toxic and has a lower boiling point and heat of vaporization than water, though the effect of ethanol on the recycling is not known. Ethanol is also produced in the process and therefore no additional chemical has to be bought if this is used as anti-solvent. A problem that occurs in the recycling through anti-solvent washing of ionic liquids is that lignin mostly, but also cellulose, hemicelluloses and other compounds accumulate in the IL [58, 50]. The problem with this is that the efficiency of the hydrolysis decreases [58]. However, the process could become economical anyway. If approximately all of the IL is recycled and if the reduction of the pretreatment efficiency is not substantial, the gain made through the reduced loss of ionic liquid combined with the possibility to reuse the IL for a great number of cycles could cover the cost for the loss in pretreatment efficiency. However, the investigations made on recycling of IL have not tested the effect on the IL for more than five cycles and the question is what happens after that. It is not 64

72 known whether the decrease in the efficiency for pretreatment and hydrolysis is substantial or not in the long run. As have been shown before (see Pretreatment) the cost for the ionic liquid is already high when it is assumed that the entire volume never has to be replaced, only what is lost in the recycling. If assumed that this is impossible and that the ionic liquid indeed has to be replaced after a certain number of cycles the cost would be even higher. However, the lignin could be removed from the ionic liquid in several ways, although complicated and in need of simplification to be industrially applicable. It can be noted that they also are quite expensive, especially for large process streams. The removal of lignin could be performed through series of steps. These steps include precipitation by anti-solvents (anti-solvent wash) followed by filtration. Subsequently the IL-mixture is subjected to adsorption by activated carbon which removes impurities. Following this is organic solvent washing, evaporation and purification with neutralactivated alumina. There are alternatives to the series of steps presented above, for example supercritical fluids can be used to extract IL-soluble polymers such as lignin. An anion exchange resin can also be used to isolate the IL as a salt. The lignin would then be partially separated. If the IL is hydrophobic a temperature controlled micellization-transfer-demicellization shuttle between the IL and an aqueous phase can be used. The lignin would then be carried from the IL to the aqueous phase [45]. Yet another technique is presented by Tan et al. [62]. After an extraction of lignin the ionic liquid containing the lignin is washed off from the un-dissolved pulp (remaining wood) using sodium hydroxide. The ionic liquid and dissolved lignin is then subjected to a drop-wise addition of hydrochloric acid resulting in a precipitation of lignin. The lignin could then be removed by filtration, washed and dried. The yield of recovered lignin is at least 93 % [62]. Theoretically, the separation could also be executed in a chromatograph. It should also be noted that if it is made possible to industrially separate lignin from the ionic liquid, the lignin can be sold as a byproduct, e.g. as a pellet. It is discussed by Marcuschamer et al. [51] that if the semi-pure mixture of lignin constituents that is obtained by the method developed by Tan et al. [62] can be valuable, the selling of this as a by-product could reduce the production cost for ethanol even down to 0 SEK/L ethanol. This of course depends on the selling price of the lignin, as discussed for the other examined pretreatment processes. In a future continued examination of the economics of the ionic liquid pretreatment process a similar investigation to that made for the sulphite and the organosolv pretreatment processes on what selling price that is needed for the lignin to make the process profitable can be made, see Sensitivity of price of lignin. An alternative to the anti-solvent washing method is the use of a three phase system. The three phases are formed by adding solutions of K 3 PO 4 and K 2 HPO 4 after dissolution of a lignocellulosic material in the IL. The three phases contain a salt-rich aqueous phase, a solid phase rich in cellulose and hemicelluloses and an IL-rich phase which contains most of the lignin. The positive effect of this separation method is that there is less water in the IL that needs to be evaporated, thus saving energy [45]. However, additional chemicals need to be supplied to the process and calculations are therefore needed to determine whether it is economical or not. It should be noted that the added chemicals also need to be taken care of, should they be recycled or disposed of? Moreover, the IL still contains lignin and the problem with accumulating lignin is not removed by the use of this recycling method. The efficiency of the recovery is the variable of greatest interest. As was seen in the

73 Pretreatment the efficiency of the recovery determines whether the process has a chance to be economical or not. A few efficiencies from different studies are displayed in table 22. Table 22: Summary of recovery-values for the recycling of ionic liquids Recovery of ionic liquid (%) Purification method >95 % Ion-exclusion chromatography [52] 95.4 % Evaporation and drying, anti-solvent = water (mean value) [58] 97.7 % Evaporation and drying, anti-solvent = methanol (mean value) [58] 94 % % Unknown [51] Simultaneous Saccharification and Fermentation The SSF step should differ compared to the other pretreatment methods in that there are less inhibitors present. This would lead to a decrease in enzymes needed and thus a lowered cost. Another result of the decrease in inhibitors is that a higher WIS concentration could be used, and then higher ethanol concentrations would be reached and the investment cost for the SSF would be lowered. However, this is the conclusion in theory. In a study performed by Lee et al. [50] an enzyme concentration of 34 U/ml was used to reach the high yields of circa 90 % for the hydrolysis. This is a quite high enzyme concentration and hence contradicts the theory, if the high concentration was needed to reach the high yields. This is however not certain and as has been concluded before, more investigations are needed to be sure that this is the case. It has been concluded for the other pretreatment methods that the enzyme cost is of great importance for the resulting ethanol price and hence it would be important to optimize the enzyme consumption for the ionic liquid pretreatment process as well. The cost for enzymes for the other examined pretreatment processes varies between 1.18 and 1.88 SEK/L ethanol. In the case of a simultaneous pretreatment and hydrolysis the process step prior to distillation would be fermentation instead of SSF. The drawback of this pretreatment compared to the other ionic liquid pretreatments is that inhibitors are formed. If these are not removed more yeast will be needed than for the other ionic liquid pretreatments Distillation If the concentration of ethanol going out from the SSF is higher than for the other pretreatments, due to a higher WIS concentration, the investment and operational costs for distillation would be lower. If no inhibitors are present they would not have to be removed. Otherwise, the distillation step should look approximately the same as for the other pretreatment methods. 5.2 Process equipment The process equipment needed for the ionic liquid pretreatment should not differ much from the other pretreatment methods. The only process step not found in any of the other pretreatment methods is a dryer (if needed). Possibly, some of the equipment used in the recycling of the ionic liquid might differ too, depending on which recycling method that is used. Regarding the sizes of the process vessels, they should not be greater than the ones used in the other pretreatment methods. The size of the vessel used for pretreatment might however differ because of the larger process flow due to the IL. If assumed that the extraction of lignin and subsequent process 66

74 works properly and can be used for the production of ethanol, the raw material needed annually would be lower than that for the steam explosion pretreatment. If the yields produced by Lee et al. [50] for pretreatment and hydrolysis complemented by yields for fermentation and distillation taken from the organosolv process are used the annual stream of raw material can be estimated to approximately tonnes of wood chips. This is lower than for any of the other pretreatment processes and would in turn result in that the process vessels could be smaller and the capital cost lower. For this annual flow of wood, the flow of ionic liquid varies between 270 and 670 tonnes/h. This can be compared to the flows of ethanol used for the organosolv pretreatment of 500 tonnes/h. The WIS concentration can be higher for the ionic liquid pretreatment process than for the others if less or no inhibitors are formed. This would mean that less water needs to be supplied and moreover that the process streams would be smaller. The estimated flow of raw material is however quite rough but it can be concluded that the capital cost for the ionic liquid process should not exceed the capital cost calculated for the other examined pretreatment processes. Hence, the cost per liter of ethanol from capital cost should vary between 1.8 and 3.0 SEK/L ethanol. 5.3 Estimation of ethanol price If the lowest possible values obtained by estimations for milling, drying, ionic liquid, enzymes and process equipment are added together a minimum ethanol selling price of 6 SEK/L is obtained. This value however, does not include costs for stocking, personnel, administration etc. It also assumes 99.6 % recirculation of the ionic liquid and an industrially possible recycling method, which at the moment is not possible. 5.4 Summary From the current lab-scale studies a few possible industrial process steps can be considered to be required; pre-pretreatment, pretreatment, separation of intermediate and contaminated ionic liquid, washing of intermediate and recycling and purification of the ionic liquid. The pre-pretreatment could include milling of the wood chips to smaller particle sizes and drying of both the ionic liquid and wood. The specifics of these steps might be questioned; there is not sufficient knowledge in the area yet to state anything. A few steps might even be unnecessary or look totally different in the future, depending on conclusions of future studies. Following the steps connected to the ionic liquid pretreatment are the SSF and distillation to finally get ethanol. No greater differences in the SSF and distillation step compared to the other examined pretreatment processes were found. Regarding the process equipment, a process using the ionic liquid pretreatment should not differ much from the other examined pretreatment processes. The same kind of unit operations are needed, except for in one case; a dryer is not used in any of the other pretreatment processes. Also, the equipment needed for recycling of the ionic liquid might differ. Rough estimates show that the flow of raw material into the process is lower than for the other examined pretreatment methods. The result is however quite rough but it can be concluded that the capital cost should not exceed the capital cost calculated for the other examined pretreatment processes. What might alter this is however the capital cost for recycling. Investigations of the operational costs (chemical costs here included in operational costs) for each process step showed that the ionic liquid pretreatment process for production of ethanol not yet can be economically beneficial. The largest problem lies with the chemical cost for the ionic liquid which 67

75 can vary between and 2.16 SEK/L ethanol, depending on if the ionic liquid is recycled or not, how efficient the recycling is, what the price of the IL is and what IL/wood ratio that can be applied. The ionic liquid pretreatment could be economically beneficial if: 1. The IL/biomass ratio can be 10 or lower 2. The price for the IL is $2.5/kg IL or lower 3. The IL can be recycled for a substantial number of cycles, i.e. if a method for the removal of lignin is found 4. An effective recycling method for the ionic liquid is found which is not too expensive and can handle large process flows 5. Each process step is optimized 6. The hydrolysis can be performed with normal enzyme concentrations 68

76 6. Discussion In this section, discussions regarding the results for material and energy balances, energy evaluation, economy and the sensitivity analysis are presented. 6.1 Material and energy balances A significantly greater amount of feed is required for the desired ethanol production in the organosolv and sulphite pretreatment processes compared with the steam explosion process. There are two main reasons why the biomass needed for organosolv is large. Firstly it is due to the low conversion from glucan to glucose for organosolv, which was 65 %. Hence it would be beneficial for the organosolv process if this conversion could be increased. It would be interesting to carry out the calculations with higher yield and see how this affects the economics. The second reason for the high biomass demand is the extra biomass fed to the CHP plant to cover the heat required for the process. The low conversion for the organosolv process, which results in a high biomass feed, reflects a lot of the result; e.g. the enzyme consumption is large, the percentage of the theoretical ethanol yield is low etcetera. Different raw material was used in different processes and differed to a small extent in composition. The raw material used in sulphite pretreatment had a lower glucan and mannan content and hence more raw material is needed. There is no electricity output for either the organosolv or sulphite pretreatment processes due to large internal consumptions, mainly due to the drying of the lignin and the lignosulfonates. Another reason is that the lignin and lignosulfonates are a co-product and not available for burning in the CHP. If the dryer was not considered, and the lignin and lignosulfonates would have been sold as solutions,this would have resulted in an electricity output for both processes. The question is if it is more beneficial (from an economical point of view) to sell the lignin and lignosulfonates as solutions and hence skip the dryers, or if it is actually more valuable to invest in the dryer and sell the lignin and lignosulfonates dry. This issue can only be stated by carrying out calculations for both cases and compare. This will however require knowledge about the price difference for wet respectively dry lignin. For the reason that lignin is obtained as product in the organosolv and sulphite pretreatment and not supplied as fuel, the CHP plant might not have been the ultimate process integration for those processes. Perhaps there is better ways to integrate those processes. The inhibitors present in the processes might not reflect the reality as well as it could have if more yields would have been available in the reference articles. However, with the yields used, large quantities of levulinic and formic acid were produced in the organosolv process, again due to the great amount of biomass fed. This indicates that the experiment in the reference article was carried out at harsh conditions. Moreover this indicates that one could expect the yields of glucan and mannan to glucose and mannose to be high, which was not the case. Hence this might be one great drawback for the organosolv process; even though running the process during harsh conditions the yields are still low. 69

77 Regarding steam explosion and the sulphite pretreatment processes, only conversions of furfural and HMF were reported in the reference articles, which might be beneficial for these. It was found that steam explosion pretreatment resulted in greater amount of those inhibitors than the sulphite pretreatment. The amount of inhibitors present in the sulphite pretreatment was smallest for all processes ( % furfural and % HMF in the liquid phase) which is beneficial for the process. The reason for the low amount of inhibitors is because of that most of the inhibitors are present in the hydrolysate. If the hydrolysate is to be fermented the inhibitors will affect the fermentation efficiency. All the reference articles used for simulations were based on laboratory experiments. This is a disadvantage since scale up does not assure the same conversions as on laboratory scale. For the organosolv process, the calculations for the biogas plant were performed without consideration of levulinic acid and formic acid, i.e. those were separated before the biogas plant. The reason for this is because those are toxic to the microorganisms. In reality these components must be separated before the biogas plant and a technical solution must be found for this purpose. The separation unit has not been considered in the calculations of the economics. In the sulphite pretreatment process some of the mannose is present in the hydrolysate after pretreatment. The hydrolysate is not fermented but is fed to the biogas plant. This leads to that the biogas plant will be larger than if all mannose was fermented. This also means less conversion to ethanol since not all mannose is going to fermentation. If all the mannose is fermented this would result in a smaller biogas plant and a reduced raw material feed. 6.2 Energy evaluation of the processes The energy efficiencies of the total processes was highest for the steam explosion process (94.3 %) followed by the organosolv process (79.0 %). The sulphite pretreatment process was found to have the smallest energy efficiency (55.0 %). This means that the losses were greatest for the sulphite pretreatment followed by the organosolv process. However, one have to remember that only the energy of the fed biomass was regarded as input for the organosolv process. If the extra biomass fed to the CHP plant for this process also would have been considered the energy efficiency of this process would have been lower. The losses were defined as the energy output that was not included in any of the main products; ethanol, district heating, electricity, lignosulfonates and lignin. This means that if the process internally needs much electricity, this was subtracted from the electricity output and more losses occur. The same is true for the heat; if much steam is needed for the processes this was integrated and less district heating was obtained. The electricity required in the dryer in both the organosolv and the sulphite pretreatment were large, and this resulted in a significantly negative effect of the energy efficiency of the processes. If the dryer was not considered both the organosolv and the sulphite pretreatment processes would have an output of electricity. Another possible solution of integrating the dryer would have been to integrate this unit with steam from the CHP instead of electricity. This would probably be more beneficial since generally electricity is more valuable and expensive than heat. This alternative would result in less district heating produced. 70

78 MSEK Generally it is favorable to have as much of the energy input as possible from the raw material feed present in the ethanol since this is the main product. In the steam explosion process, 34.3 % of the energy input from the feed was found in the ethanol, for organosolv it was 22.5 % and in the sulphite pretreatment it was 27.1 %. Hence the steam explosion process is the most favorable process in this aspect followed by the sulphite pretreatment process. In steam explosion, 34.9 % of the energy input of the feed was distributed as district heating. For organosolv it was 20.3 % and in the sulphite pretreatment it was 16.5 %. District heating is an uncertain income since the request depends on the seasons; the demand is greater in the winter compared to the summer. Therefore it is unfavorable to have a process that depends to a greater extent of district heating. Hence the sulphite pretreatment process is most favorable in this matter. The reason why the input energy distribution of district heating is small for the sulphite pretreatment and organosolv processes is due to the fact that a lot of the lignin was not burnt in the CHP, but exported as lignofulfonates and as pure lignin products (both with dry matter of 90 wt%). Production of green (environmental friendly) electricity is favorable. The electricity produced from the CHP in all pretreatment methods can be viewed as green-electricity and hence it is favorable to have a high electricity production. In the organosolv pretreatment, the energy present as lignin accounted for 36.2 % of the energy input and in the sulphite process the energy present as lignosulfonates accounted for 11.5 %. Hence both processes depend on the market of lignin and lignosulfonates (organosolv to a greater extent). A significant amount of the energy input is present in those co-products. 6.3 Economy The annual cash flow for steam the explosion, organosolv and sulphite pretreatment processes is illustrated in figures The costs and the revenues per year for each process are shown in million SEK. Generally the costs are lowest for the steam explosion pretreatment process and hence the ethanol price for this process is the lowest Figure 26: Steam explosion, EtOH 5.2 SEK/L 71

79 MSEK MSEK Figure 27: Organosolv, EtOH 13.5 SEK/L, Lignin 1.5 SEK/kg Figure 28: Sulphite pretreatment, EtOH 9.0 SEK/L, Lignin 1.5 SEK/kg The main reason why the price for ethanol from the steam explosion process is lowest is because it has the highest conversions to ethanol. This means that for the same amount of ethanol produced less raw material feed is needed and hence the whole plant is smaller, which means lower capital cost. The plant size is also affecting other things such as operational cost and stocking since this is estimated from the capital cost. The raw material feed flow is affecting e.g. chemical and yeast consumption. Another reason why the plant has to be larger for both the organosolv and sulphite processes compared to the steam explosion process is that more complicated processes are needed. By just looking at the process schemes one realizes that more unit operations are needed which also gives a need for more staff etc. The calculated MESP for the three processes may appear unfair since there is more knowledge about steam explosion both generally and at the department. This has affected the calculations and estimations in several ways for example that the WIS concentration and residence time in the SSF was changed for the steam explosion process thanks to experience at the department. A drawback with this comparison of the ethanol price for the three methods is that a minimum selling price for lignin is used for the organosolv and sulphite pretreatment processes. Since it is expected for the lignin to have a market in the future and that it already exists to some extent for the 72

80 lignosulfonates the selling price will hopefully be higher. Those methods are then going to be a better alternative to steam explosion. Another uncertainty factor with both the organosolv and sulphite pretreatment processes is that more chemicals are used and the recycling process is not fully evaluated and hence this cost could also be lower than the ones presented in the figures above. The chemical recycling for sulphite pretreatment was assumed to be 80%. In reality this might be higher but since no investigation was done regarding the cost of recycling this was considered to be a good approximation. To get a more accurate result a thorough examination of the chemical recovery must be done. The MESP will in reality be higher for all pretreatment methods. The aim of this study was to make a comparison between the pretreatment methods and costs that will be the same has been left out from the study, e.g. the dehydration. Adding this will lead to a higher cost for the production of ethanol. 6.4 Sensitivity analysis: feed-, capital- and enzyme cost The sensitivity analysis is performed with the heating value as the price of lignin/lignosulfonates for the organosolv and the sulphite pretreatment processes. This is one reason to the higher MESP for these pretreatment methods compared with the steam explosion process. It is hard to make any conclusions from the sensitivity analysis of the feed-, capital- and enzyme cost because of this. If a selling price was known for lignin/lignosulfonates the sensitivity analysis might have rendered more information. The information from the sensitivity analysis is that the pretreatment methods are about as sensitive to a change in the sensitivity parameters. 6.5 Sensitivity analysis: lignin price The organosolv process produces pure lignin which in the future might be a valuable co-product to ethanol. Right now there is no commercial use for the lignin but at the moment a lot of research is performed on the lignin and the use of it. The sulphite pretreatment produces lignosulfonates and a market already exists for it today. The large difference in lignin price (5.3 SEK/kg for the organosolv process, 11.6 SEK/kg for the sulphite pretreatment process) is explained by the higher amount of dissolved lignin when organosolv pretreatment method is applied to the raw material compared to the sulphite pretreatment. In the organosolv process all lignin is assumed to be dissolved and in the sulphite pretreatment only 30 % is dissolved. In reality not all lignin is dissolved, the conversion is about 80%, which is still significantly higher than for sulphite pretreatment. This is the main reason why the organosolv lignin requires a lower price than the sulphite pretreatment lignosulfonates to reach the same MESP as for the steam explosion process. 73

81 7. Conclusions For production of m 3 ethanol annually, the steam explosion process needs 41.1 ton dry matter/h, the organosolv process needs 67.7 ton dry matter/h and the sulphite pretreatment process needs 51.9 ton dry matter/h. Steam explosion is the most favorable pretreatment method with respect to energy efficiency (which corresponds to smallest losses) and with respect to the energy amount from the feed present in the ethanol. In the matter of district heating, the sulphite process was found to be most favorable, since this process had the smallest distribution of the energy input as district heating. The amount of produced district heating was greatest for the steam explosion process, but the amount of produced district heating in the other processes was in the same range. The produced district heat might have been set for too low temperatures during the calculations, hence only half was set to be sellable for this reason. A significant amount of the energy input is present as lignin and lignosulfonates in the organosolv and the sulphite pretreatment processes; hence those processes depend on the market of those co-products. The steam explosion process was the only process which had electricity output from the total process. Regarding present inhibitors, lots of levulinic and formic acid was present in the organosolv process. For the steam explosion and sulphite pretreatment processes only HMF and furfural were present in the calculations due to the fact that conversions for other inhibitors not were presented in the reference articles. The sulphite pretreatment process resulted in the lowest amount of inhibitors. The steam explosion was found to have the greatest amount of energy input present as ethanol, followed by the sulphite pretreatment. The electricity output for the organosolv and sulphite pretreatment processes were zero mainly due to high internal consumption since dryers was needed for purifying the lignin. Regarding the lowest ethanol selling price it was found that steam explosion is the best pretreatment method with a MESP of 5.2 SEK/L compared to the sulphite and organosolv pretreatment processes with a MESP of 9.0 SEK/L and 13.5 SEK/L. These prices were calculated with the lowest lignin price (a price corresponding to the heating value, as if it was used as fuel and burned) and a sensitivity analysis was performed to investigate what selling price lignin must have for the organosolv and sulphite pretreatment processes to compete with steam explosion. The prices were calculated to 5.5 SEK/kg and 11.6 SEK/kg respectively. The conversion of dry matter to ethanol is highest for steam explosion pretreatment, which resulted in lowest amount of raw material needed. This results in less capital cost, less chemicals etc. As long as there is no market for the lignin steam explosion is the best pretreatment method concerning the MESP. Better conversions are needed for organosolv and sulphite pretreatment to compete with steam explosion. In the sensitivity analysis it was found that basically all three investigated processes had the same sensitivity for the three investigated parameters (costs of raw material, enzymes, and capital costs). The steam explosion process resulted in the lowest ethanol price for all cases. 74

82 The ionic liquid pretreatment process cannot yet be economically beneficial mainly because of the high purchasing price of ionic liquids. If 100 % of the ionic liquid could be recycled for a substantial amount of cycles the process could be beneficial but since no industrial process to recycle the ionic liquid with removal of lignin exists this is not possible. If made possible however, and if native lignin becomes valuable (as discussed for the other examined pretreatment processes), the lignin might become the main product, reducing the production cost of ethanol to 0 SEK/L ethanol and making the process beneficial. There exists pilot plants for steam explosion which is the most studied pretreatment method of the ones considered in this project. Since a commercial plant is being commissioned in Italy in august this year, steam explosion is the pretreatment method that is closest to an industrial breakthrough. There also exists pilot plants for organosolv, and hence this process might also soon be industrially introduced. The unit operations of sulphite pretreatment are well known in the pulp-industry, which is beneficial if the process is to be introduced in industry. Regarding ionic liquids lots of research and time is needed before a possible industrial introduction. 75

83 References [1] Guido Zacchi 2009.Ethanol as a Transportation Fuel. ourse material in Energy and Environment, Department of hemical Engineering, Lund University. [2] J.S. Tolan. Alcohol production from cellulosic biomass: The Iogen process, a model system in operation, Iogen Corporation, Ottawa, Canada, p [3] Zsolt Barta, Krisztina Kovacs, Kati Reczey, Guido Zacchi Process design and Economics of On- Site Cellulase Production on Various Carbon Sources in a Softwood-Based Ethanol plant. Enzyme Research, volume [4] Mats Galbe, Per Sassner, Anders Wingren, Guido Zacchi Process Engineering Economics of Bioethanol Production. Advances in Biochemical Engineering/Biotechnology, vol. 108, p [5] Per Sassner, Mats Galbe, Guido Zacchi Techno-economic evaluation of bioethanol production from three different lignocellulosic materials. Biomass and Bioenergy, vol. 32, issue 5, p [6] chemicals-technology.com, [7] Carolina Conde-Mejía, Arturo Jiménez-Gutiérrez, Mahmoud El-Halwagi A comparison of pretreatment methods for bioethanol production from lignocellulosic materials. Process Safety and Environmental Protection [8] [9] Caixia Wan, Yebo Li Microbial pretreatment of lignocullulosic biomass with ceriporiopsis subvermispora for enzymatic hydrolysis and ethanol production. Thesis, ISBN [10] Eva Palmqvist, Bärbel Hahn-Hägerdal Fermentation of lignocellulosic hydrolysates II: inhibitors and mechanism of inhibition. Bioresource Technology, vol. 74, issue 1, p [11] Valery B. Agbor, Nazim Cicek, Richard Sparling, Alex Berlin, David B. Levin Biomass pretreatment: Fundamentals towards application. Biotechnology Advances, vol. 29, issue 6, p [12] João RM Almeida, Tobias Modig, Anneli Petersson, Bärbel Hähn-Hägerdal, Gunnar Lidén, Marie F Gorwa-Grauslund Increased tolerance and conversion of inhibitors in lignocellulosic hydrolysates by Saccharomyces cerevisiae. Journal of Chemical Technology and Biotechnology, Vol. 82, Nr. 4, p [13] Antoine Margeot, Bärbel Hahn-Hagerdal, Maria Edlund, Raphael Slade, Frédéric Monot New improvements for lignocellulosic ethanol. Current Opinion in Biotechnology, Vol. 20, Issue 3, p [14] Nanqi Ren, Guangli Cao, Aijie Wang, Yuhong Zhu, Bingfeng Liu Hydrogen production from hemicellulose hydrolysates by Thermoanaerobacterium thermosaccharolyticum W16. Journal of Biotechnology, Vol. 136, p. S422-S423 [15] Prasad Kaparaju, María Serrano, Anne Belinda Thomsen, Prawit Kongjan, Irini Angelidaki Bioethanol, biohydrogen and biogas production from wheat straw in a biorefinery concept. Bioresource Technology, Vol. 100, Issue 9, p [16] Haregewine Tadesse, Rafael Luque Advances on biomass pretreatment using ionic liquids: An overview. Energy and Environmental Science, issue 10, p [17] Naveen Narayanaswamy, Ahmed Faik, Douglas J. Goetz, Tingyue Gu Supercritical carbon dioxide pretreatment of corn stover and switchgrass for lignocellulosic ethanol production. Bioresource Technology, vol. 102, issue 13, p

84 [18] Encarnación Ruiz, Cristóbal Cara, Paloma Manzanares,Mercedes Ballesteros, Eulogio Castro Evaluation of steam explosion pre-treatment for enzymatic hydrolysis of sunflower stalks. Enzyme and Microbial Technology 42, p [19] Mohammad J. Taherzadeh and Keikhosro Karimi Pretreatment of Lignocellulosic Wastes to Improve Ethanol and Biogas Production: A Review. International Journal of Molecular Science, p [20] Mohammed Moniruzzaman Effect of Steam Explosion on the Physicochemical Properties and Enzymatic Saccharification of Rice Straw. Applied Biochemistry and Biotechnology, vol. 59, no. 3 [21] Mats Galbe, Guido Zacchi Pretreatment of Lignocellulosic Materials for Efficient Bioethanol Production. Biochemical Engineering / Biotechnology, vol. 108, p [22] Kerstin Hoyer, Mats Galbe, and Guido Zacchi 2010.Effects of enzyme feeding strategy on ethanol yield in fed-batch simultaneous saccharification and fermentation of spruce at high dry matter. Biotechnol Biofuels 3: 14. [23] Sanam Monavari, Andrea Bennato, Mats Galbe, Guido Zacchi Improved one-step steam pretreatment of SO 2 -Impregnated softwood with time-dependent temperature profile for ethanol production. Biotechnology Progress, Vol. 26, Issue 4, p [24] Johanna Söderström, Linda Pilcher, Mats Galbe, Guido Zacchi Combined Use of H2SO4 and SO2 Impregnation for Steam Pretreatmentof Spruce in Ethanol Production. Applied Biochemistry and Biotechnology, vol [25] [26] Xuebing Zhao, Keke Cheng, Dehua Liu Organosolv pretreatment of lignocellulosic biomass for enzymatic hydrolysis. Applied Microbiology and Biotechnology, vol. 82, no. 5, p [27] Gary Brodeur, Elizabeth Yau, Kimberly Badal, John Collier, K. B. Ramachandran, Subramanian Ramakrishnan Chemical and Physicochemical Pretreatment of Lignocellulosic Biomass: A Review. Enzyme Res [28] Xuejun Pan, Dan Xie, Richard W. Yu, Dexter Lam, Jack N. Saddler Pretreatment of lodgepole pine killed by mountain pine beetle using the ethanol organosolv process: Fractionation and process optimization. Ind. Eng. Chem. Res., 46 (8), p [29] Xuebing Zhao, Keke Cheng, Dehua Liu Organosolv pretreatment of lignocellulosic biomass for enzymatic hydrolysis. Appl Microbiol Biotechnol, 82: , DOI /s [30] Wouter J. J. Huijgen, Arjan T. Smit, Johannes H. Reith, Herman den Uil Catalytic organosolv fractionation of willow wood and wheat straw as preteratment for enzymatic cellulose hydrolysis. Journal of Chemical Technology and Biotechnology, vol. 86, issue 11, p [31]Nahyun Park, Hye-Yun Kim, Bon-Wook Koo, Hwanmeyeong Yeo, In-Gyo Choi Organosolv pretreatment with various catalysts for enhancing enzymatic hydrolysis of pitch pine (Pinus rigida). Bioresource Technology, vol 101, iss. 18, p [32]Xuejun Pan, Neil Gilkes, John Kadla, Kendall Pye, Shiro Saka, David Gregg, Katsunobu Ehara, Dan Xie, Dexter Lam, Jack Saddler Bioconversion of hybrid poplar to ethanol and co-products using an organosolv fractionation process: Optimization of process yields. Biotechnology and Bioengineering, vol. 94, issue 5 [33] Ullmann s Encyclopedia of Industrial hemistry, Pulp:.3. ulfite process, WILEY ISBN: [34] Ullmann s Encyclopedia of Industrial hemistry, Pulp:.3.. Kraft ( ulfate pulping, WILEY ISBN:

85 [35 ] Ronalds Gonzalez, Trevor Treasure, Richard Phillips, Hasan Jameel, Daniel Saloni Economics of cellulosic ethanol production: Green liquor pretreatment for softwood and hardwood, greenfield and repurpose scenarios. BioResources, Vol. 6, Issue 3, p , 17p [36] Yongcan Jin, Hasan Jameel, Hou-min Chang, Richard Phillips Green Liquor Pretreatment of Mixed Hardwood for Ethanol Production in a Repurposed Kraft Pulp Mill. Journal of Wood Chemistry and Technology, Volume 30, Issue 1 [37] Steve S. Helle, Tony Lin, Sheldon J.B. Duff 8. Optimization of spent sulfite liquor fermentation. Enzyme and Microbial Technology 42, p , [38] G. S. Wang, X. J. Pan, J. Y. Zhu, R. Gleisner, D. Rockwood Sulfite pretreatment to overcome recalcitrance of lignocellulose (SPORL) for robust enzymatic saccharification of hardwoods. Biotechnology Progress Biotechnology Progress, Volume 25, Issue 4, pages [39] William O.S. Doherty, Payam Mousavioun, Christopher M. Fellows Value-adding to cellulosic ethanol: Lignin polymers. Industrial Crops and products,volume 33, Issue 2, Pages [40] José Augusto Restolho, António Prates, Maria Norberta de Pinho, Maria Diná Afonso Sugars and lignosulphonates recovery from eucalyptus spent sulphite liquor by membrane processes. Biomass and Bioenergy, Volume 33, Issue 11, Pages [41] Ullmann s Encyclopedia of Industrial hemistry, Pulp:.3.. oda pulping of annual plants, WILEY ISBN: [42] Araceli García, María González Alriols, Rodrigo Llano-Ponte, Jalel Labidi Energy and economic assessment of soda and organosolv biorefinery processes. Biomass and Bioenergy, Volume 35, Issue 1, Pages [43] Farnoush Faridbod, Mohammad Reza Ganjali, Parviz Norouzi, Siavash Riahi and Hamid Rashedi Application of Room Temperature Ionic Liquids in Electrochemical Sensors and Biosensors. Ionic Liquids: Applications and Perspectives. ISBN [44] H. Olivier-Bourbigou, L. Magna, D. Morvan Ionic liquids and catalysis: Recent progress from knowledge to applications. Applied catalysis A: General, vol. 373, iss. 1-2, p [45] Mauricio Mora-Pale, Luciana Meli, Thomas V. Doherty, Robert J. Linhardt, Jonathan S. Dordick Room Temperature Ionic Liquids as Emerging Solvents for the Pretreatment of Lignocellulosic Biomass. Biotechnology and bioengineering, vol. 108, no. 6, p [46] Chun-Zhao Liu, Feng Wang, Amanda R. Stiles, Chen Guo Ionic liquids for biofuel production: Opportunities and challenges. Applied Energy, vol. 92, p [47] Richard P. Swatloski, Scott K. Spear, John D. Holbrey, and Robin D. Rogers Dissolution of cellulose with ionic liquids. J.AM.CHEM.SOC., vol. 124, p [48] Hao Zhang, Jin Wu, Jun Zhang and Jiasong He Allyl-3-methylimidazolium Chloride Room Temperature Ionic Liquid: A New and Powerful Nonderivatizing Solvent for Cellulose. Macromolecules, vol. 38, no. 20, p [49] Michael Zavrel, Daniela Bross, Matthias Funke, Jochen Büchs, Antje C. Spiess Highthroughput screening for ionic liquids dissolving (ligno-)cellulose. Bioresource Technology, vol. 100, iss. 9, p [50] Sang Hyun Lee, Thomas V. Doherty, Robert J. Linhardt, Jonathan S. Dordick Ionic Liquid- Mediated Selective Extraction of Lignin From Wood Leading to Enhanced Enzymatic Cellulose Hydrolysis. Biotechnology and Bioengineering, vol. 102, no. 5, p [51] Daniel Klein-Marcuschamer, Blake A. Simmons, Harvey W Blanch Techno-economic analysis of a lignocellulosic ethanol biorefinary with ionic liquid pre-treatment. Biofuels, Bioproducts & Biorefining, Modelling and Analysis, 5:

86 [52] Joseph B. Binder and Ronald T. Raines Fermentable sugars by chemical hydrolysis of biomass. PNAS, march 9, 2010, vol. 107, no. 10, p [53] Deepak Kumar, Ganti S. Murthy Impact of pretreatment and downstream processing technologies on economics and energy in cellulosic ethanol production. Biotechnology for Biofuels, vol. 4:27 [54] Xuejun Pan, Claudio Arato, Neil Gilkes, David Gregg, Warren Mabee, Kendall Pye, Zhizhuang Xiao, Xiao Zhang, John Saddler, Biorefining of softwoods using ethanol organosolv pulping: Preliminary evaluation of process streams for manufacture of fuel-grade ethanol and co-products. [55] J. Y. Zhu, Wenyuan Zhu, Patricia OBryan, Bruce S. Dien, Shen Tian, Rolland Gleisner, X. J. Pan. Ethanol production from SPORL-pretreated lodgepole pine: preliminary evaluation of mass balance and process energy efficiency. Applied Microbiology and Biotechnology, Volume 86, p [56] Techno-economic evaluation of 2nd generation bioethanol production from sugar cane bagasse and leaves integrated with the sugar-based ethanol process, Stefano Macrelli, Johan Mogensen and Guido Zacchi Biotechnology for Biofuels 2012, 5:22 [57] Daniel J Schell, Chuck Harwood Milling of lignocellulosic biomass, Results of pilot-scale testing. Applied Biochemistry and Biotechnology 159, Vol. 45/46 [58] Bin Li, Janne Asikkala, Ilari Filpponen, Dimitris S. Argyropoulos Factors affecting wood dissolution and regeneration of ionic liquids [59] BASF The Chemical Company, [60] Depertment of chemical engineering, LTH [61] Miyafuji H, Miyata K, Saka S, Ueda F and Mori M Reaction behavior of wood in an ionic liquid, 1-ethyl-3-methylimidazolium chloride. J Wood Sci 55(3): (2009). [62] O Suzie S. Y. Tan, Douglas R. MacFarlane, Jonathan Upfal, Leslie A. Edye, William O. S. Doherty, Antonio F. Patti, Jennifer M. Pringle, Janet Scott Extraction of lignin from lignocellulose at atmospheric pressure using alkylbenzenesulfonate ionic liquid. Green Chem., vol. 11, p [63] Department of Chemical Engineering, LTH, Lund University [64] Andritz, vendors of pretreatment reactor [65] Ola Wallberg, Anders Holmqvist, Ann-Sofie Jönnson Ultrafiltration of kraft cooking liquors from a continuous cooking process. Desalination, Volym 180, Nummer 1, pp [66] Per Sassner, Guido Zacchi Integration options for high energy efficiency and improved economics in a wood-to-ethanol process. Biotechnology for biofuels, vol 1, p.4 [67] &fsb=y&catid=8, [68]

87 Appendix A, Material balance calculations m 3 ethanol was set to be produced annually. The density of ethanol is kg/l =789 kg/ m 3.The operational time for the plants are assumed to be h/year. For calculations of the theoretical ethanol production, the following yields were used: Yield coefficient glucose (sugar) ethanol, Y SE =0.51 kg/kg Yield coefficient glucan-glucose (sugar), Y GS =1.11 kg/kg Amount of dry matter in feed kg/s Yield coefficient polymers that is ought to be C6 sugars (i.e. mainly glucan and mannan) based on the total amount of biomass Y C6polymer (kg/kg) A.1 Steam explosion Y C6polymer = kg/kg (from reference article) Conversion of the product was calculated as the amount of ethanol that is produces per ton of dry matter in the feed. A.2 Organosolv Y C6polymer = ( ) kg/kg (from reference article) (only the biomass fed to the pretreatment reactor) Conversion of the product was calculated as the amount of ethanol that is produces per ton of dry matter in the total feed to the process. 80

88 A.2.1 Feed to the CHP Feed of extra methane = kg/h Effective heating value for methane = 50 MJ/kg Effective heating value for wood (feed) = 19 MJ/kg This corresponds to: Hence the total amount of feed to the overall process is: A.3 Sulphite pretreatment Y C6polymer = kg/kg (from reference article) Conversion of the product was calculated as the amount of ethanol that is produces per ton of dry matter in the feed. 81

89 Appendix B, Aspen Plus calculations All conversions etc. were calculated from the reference articles. B.1 Steam explosion B.1.1 Conversions The conversion to ethanol is based on the theoretical amount that could be produced from glucan and mannan % of the dray matter content in the feed consist of the two component, due to the following calculation. Glucan 45.1% 45.1/56.9= 0.79 Mannan 11.8% 11.8/56.9= % is the total theoretical conversion and are due to the used article 75 %. From article, liquid for PT and solid for SSF Those numbers are picked from the article and are valid over the whole process and not just over the SSF step. We want the numbers based over the wanted stream and nor the one based on original. The overall convention is then: PT SSF 82

90 B.1.2 SSF The flows in to the SSF are going to be calculated, except for the flow named SolidPT, which is already defined. The flows are named in the calculation due to the picture below. Water Given in the reference article: 5 %WIS in the SSF reactor. Since this value is too low for industrial production 10 wt% WIS in the SSF is used instead. NaOH The ph before SSF is assumed to be 2.3 and in the SSF reactor the ph is 5, for neutralization NaOH is added. ph=-log[h + ] ph before SSF : 2.3=-log[H + ] [H + ]=0.005 mol/l ph in SSF: 5.0=-log[H + ] [H + ]=10-5 mol/l, 50 % Yeast Given: Concentration of yeast, 3 g yeast/l Enzymes Given: 15 FPU/g WIS, 150 FPU/g enzyme B.2 Organosolv B.2.1 Conversions 83

91 B.2.2 SSF The flows in to the SSF are going to be calculated, except for the Ptsolids flow, which is already defined. The picture below shows which flows that are calculated. Water Given: 10 wt% WIS H 2 SO 4 The ph before SSF is assumed to be 2.3 and in the SSF reactor the ph is 4.8, H 2 SO 4 = 98 wt% = = [H 2 SO 4 ] = 18.4M ph=-log[ ] ph before SSF : 2.3=-log[ ] [H + ]= mol/l ph in SSF: 4.8=-log[ ] [H + ]=7.9*10-6 mol/l 84